YouTube Transcript:
How to simulate PCIE / IEEE path on PCB + Everything you need to know | Explained by Bert Simonovich

Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.

Video Transcript

to create this video I partnered with design

design

Con in this video we are going to talk

about how to simulate Channel and when I

say Channel it means for example if you

are designing PCI Express and you would

like to be sure uh these uh uh tracks

are routed properly and PCI Express

signal quality is going to be okay

between your processor and your board or

connector or anything else then you can

use this video to actually learn how to

do it and also what is important for

this kind of design am I right B that's

what we are going to talk about yes yeah

uh okay so this is the simulation what

we will be doing later but we have to

start with

something more uh more like uh what does

it mean

Channel a channel would be something

like a chip to a module [Music]

[Music]

and for instance uh this chip here you

know it would be the chip going through a

via through a trace in the PC board to a

connector to another little PC board in

this module mhm so basically channel is

not uh only like PCB traces or something

channel is basically the connection

whole connection which may include

like in our case it will be everything

will be basically on P but whole channel

is like from Chip through balls or paths

and then V tracks through connector

that's why we call it Channel correct

yes okay yeah

typically but when we do simulations can

we simulate everything or what do we

usually S no we well to build a channel

you need to model each individual piece

and then you put every you will put

everything together and that'll be the

channel okay basically and then you

simulate the whole Channel end to end

that's what we will do later correct

yeah we'll be doing doing some of that

exactly and why is this important why we

would like to simulate

Channel well if uh for instance on no on

this slide here there's diff for each uh

there's different standards so like PCI

will have a standard and they'll have

some metrics to test the channel against

to perform this one for instance is i e

802.3 ethernet standard and I just pull

this particular because that's what I'm

going to focus on and that's why loss is

typically important so in this ethernet

Channel there's what they call it chip a

chip to module um

specification and in the document you'll

find find like this 400 gig uh it's like

56 gig uh Pam 4 type signaling and the

channel has got like 16 links going and

it's going from a host chip to a PCB to

what they call a module and the

module could be something that plugs in

that goes to some other box or something

so they have this chip to module spec

and as you can see here uh they have

some uh loss metrics on it so on the

host board you've got like s and a half

DB um budget you've budgeted 1.2 DB for the

the

connector and your insertion loss of the

module has got a budget of one and a

half DB basically so so basically this

means uh when we go to highr

frequencies uh it's not only about some

kind of like damaging the signal but

also uh also when we use higher

frequencies signal is going

to be smaller and smaller and smaller so

basically that's what uh this is helping us

to um to

meet uh because if we don't meet this uh

loss if we for example make this uh PCB

traces too long it means the module may

not work because signal which will

arrive to the module will be too small for

example

um yeah at the end of the day you need

to have some typical signal

so for instance let's just look at just

a simple transmission line you know this

represents the loss of

of the transmission line so there's two

different parts of it there's like the

dialectric loss of the material and the

conductor loss of the conductor then

there's surface roughness of the

conductor that contributes to more

loss um and I've just broken down this

loss here so this is what we call the

insertion loss versus

frequency and this is the loss of uh

like all sinewave frequencies up to 50

gigahertz this is the loss of say the

transmission line or a channel will look

you know similar to this but this is

just a simple transmission mine and this

just broken down to this red one is

Conductor loss the blue is dialectric

and smooth smooth uh smooth uh copper is

this pink and then when you you add

roughnesses here mhm so basically the

green one is kind of real right loss in

channel in real Channel yes right our

channel was right yeah if we just say

our channel is a transmission line right

now without anything else yes that's

what we would be looking at so this is

basically the reason maybe why someone

would like to simulate their channels

because they would like to be sure they

have enough margin to meet the required

maximum loss okay right so now if you

look at this picture for why is loss

important if we you know we started with

here if this is the transmission line

that would

represent this uh line in the PCB

without the vs

okay so we need with the vs and the

transmission line from here to here um

from the chip to this connector this 7

and A2 DB has to include two vs and a

transmission line for 1328

GHz sorry for for the 1328

GHz right at

13.2 gigahertz yeah we so in other words

we'd go back well you know roughly like

this this one's talking about roughness

if we don't consider roughness you can

see here if you simulated just with red

for instance you measured 14 or

1.28 it's close enough but when you

added roughness you have a Delta of 3

and A4 DB and if you haven't accounted

for or understand the roughness or pick

a foil that is smooth you know smoother

roughness when you build the board you

could end up with uh we've got uh

what is it 16% increase in Jitter 48% of

the eye height you can see with it so

that's why loss is is important per se

and channel modeling is important to

ensure that when you build it you're

going to end up with uh enough eye

opening in it then I'm I'm really

curious to see how the uh material will

influence it when we will do the

simulation yeah okay so what else we

need to know before we do

simulation okay we need to know so why

is loss important we talked about the

channel what it means and we want a good

signal back eventually at this little

chip that's on this board that plugs in

you know we want a good signal

there so in this ethernet spec spec

um this is like in a data center or te

telecommunication office

they have racks of

equipment and

realistically this is an example uh

taken from like s Tech they have a nice

picture of the Interior but what you can

see here practically you'll have a

Central Chip and it's got to connect to

all these modules across the front so

really that budget you've got 7.3 D DB

from the chip to this outside end uh

module and similarly on this side and

typically typically uh in these boxes

that Trace length could be anywhere from

8 to 9 Ines by the time it reaches there

so you you got to meet the 7 and a half

DB uh from from there to there for sure

the modules that's anybody's module can

now plug into into this connector from

the front so you know that has a lot

budget of 1 and A2 DB so you have to

make your own model of 1 and A2 DB loss

in a connector but basically if you're

designing this board here you've got to

fit here or if you're a A supplier of

just the modules you have to meet one

and a half DB on your board to plug in

you see what I'm saying and these

modules they're like uh Optical plugs so

there's this this module looks here and

then uh they have these plugs that plug

in the front and connects to another

rack of equipment over

here so that's how you connect racks to

racks is through uh through the plugs

and through like a cable and they could

be optical cable that could be uh an

electrical copper kind of cable to

connect it the modules are kind of

changing the signal to to the physical

interface which will then continue

further right modu is just something

what yeah what is interfacing to the

processor on one side on the other side

something else yeah so you know the real

channel is really from here to the end

of basically end of the connector so

we're worried about 7.3

DB you know the other the other person

that's applying the module they have to

meet one and a half yeah yeah yeah

so we simulate we'll build the whole

Channel 7.3 or whatever it is uh up to

you know this whole thing I'm curious so

there is some kind of standard interface

between the processor and these modeles

that's the I standard so these are

single ended signals or what they are no

they're they're typically at these high

speeds differential yeah they're differential

differential

serial singleand it would not work fun

yeah no no no and here's like here's for

instance like this is the channel so

part of that I E spec they have this uh

what they call host and in order to evaluate

evaluate

this they have what they call this host

compliant board and this host compliant

board will plug in to one of these

socket that's nice here right and then

you would hook up your uh VNA or test

equipment what they call this test Point

test Point

1A and then they will have uh an S

parameter mask for instance that you

have to meet so your your loss of the

whole Channel at tp1a has to be above

this line right and then there's the

return loss which is the reflections has

to be below this line essentially so

essentially we we have to build build a

channel to meet a spec and for the

example we're going to talk about today

is I E standard um because that's really

high speed and it has a lot of good demo

type of thing to to show that thing so

really we want to meet the channel has

to be above this line and return loss

really to be below um but that in itself

doesn't always mean that channel is

going to work or not work sometimes you

could have a you know things to be above

this line but it may not work because of

other reasons so I'm going to talk about

Channel operating margin which is a a a

new kind of a metric that you can apply

on the channel and it'll go a little

deeper uh it'll take other metrics and

really sort of stress the channel and

the channel operating margin gives you

an output number if you meet the number

you pass or fail thing so I'll be

showing some of that as well

okay so so yeah that's the thing there

um and then I'm going to show basically

well how we go about first uh you know

go about starting to model it well first

of all you'll need to define a stack up

you have to choose

material um and then you know you start

going from there uh but before you go

there you know when you make a stack up you

you

know the Symmetry typically when we

fabricate the board has to be

symmetrical so if we look at a

cross-section we typically have a core

or a prepr in the center uh and then

they build symmetrically above and below

um the center line and typically we like

to have equal weights of copper on the

cores uh equal thicknesses on it as we

build up so that's just general

guidelines for uh you know a good

stackup and we have good ground and

power planes they're adjacent to one

another so you know if we look at these

two planes here one would be power one

would be ground for instance um but this

is just a very high level view of

building a stock up um the next thing is

you know we want to do is compare the

dialectric material properties the DK DF

this is here I pck yeah so here I pick

DK is the dialectric constant and DF is

the dissipation factor and DF represents

uh how it how the loss is affected by

the material the dialectric material and

your camera Switched Off oh switched off

again yeah I'm sure I don't know it

switched off by itself I didn't touch

anything I don't know why

anyway um so compar dialectric material

properties the DF is the

um it really affects the loss of the

channel from a dialectric perspective so

we want something with a very low DF so

I've chosen two materials to

compare uh and I've chosen isolar

materials here because the information

is really read readily available online

they have very good website you go you

can pick materials and everything's

there some other laminate suppliers

they'll only show you um what I call

this data sheet and it only has the DK

and DF at a certain frequency it's not

what we use to build a stackup um and if

you need that other info you have to

contact the supplier or register and

whatever and then they'll send it

separately and it's usually got some

confidentiality asso assciated with it

so I didn't want to show any of that so

I chose Isola because they're they they

have everything wide open only for an

example I have a question so because it

looks like the numbers are very similar

like like 10 gahz is

3.92 DK 100 me MHz is 4.24 so which one

would you choose do you need to choose

the one for the signal well the the DK

is not as

the DK for the for

loss it's obviously important DK is more

important for getting your

characteristic in yeah right so I chose

something that the

dks well they're not really that you

know that close I mean this is near Four

we're near three here right so

uh yeah and and here you see the DF it's

0 Z

015 and the DF here is

02 right order of magnitude difference

right so this is your typical fr4 type

material for lack of better term you

know they probably got stuff worse but

this an4 material have a dec of around

four at at one gigahertz we'll be

looking at stuff at 10

gigahertz our comparison because that's

13.28% D DF may be even more important

DK is Just for information which you use

to calculate for example impedance but

DF that's the parameter which is

important for very high speed signals

you would like to have it as low as

possible Right for loss for the loss

aspect now DK is important as well because

because

uh a smaller

Decay will allow you to have a board for

the same

impedance when you work it through so

you know it does have an effect if

you're worried about thickness of your

board now for instance you have to look

at that too and number of layers that

you have you have a lot of layers and uh

and you have a thickness requirement um

then you start looking at DK and you

know a lower DK allow you to have a

thinner overall board based basically

for the same impedance so DK is still

important you know but for loss it's

mainly the the dissipation Factor D okay

so you know so you start off basically

looking at you know the two I mean

sometimes depending on your design you

might be able to get away with an fr4

depending on the length of your Channel

or anything else so just throwing it you

know throwing out

A4 may not be but if you're building

something here for long length you'll

see you know you're not going to

entertain anything like an fr4 you

looking so this is this is also again

important so basically uh sometimes

people may think I'm designing for very

high speed or I have to use expensive

materials but if the length of the

tracks is very very small then it still

may work perfectly fine even with

standard materials maybe right maybe

okay and all dep like everything signal

Integrity it depends right

okay okay we can move to the next slide

what do you have on the next one yeah so

you know this is where you typically

start we start like with a stackup of

the board and you know choose some

material like I picked this one because

it's it's a relatively new material and

it's it'll be good for the latest

standards type thing so this would be

typically something you would use at you

know 56 or even 112 gig now would use

something like this m right but other

other laminate suppliers have similar

materials and the dks and DFS are on the

same range right it's not just unique to

their but for this is very good for the

example here okay so the next thing we

need to look at because we I I mentioned

that roughness is

important um part of the data sheet here

here uh when you go on the website

they'll have the full data sheet and

part of the full data sheet will show um

the copper foil type that's used on the laminate

laminate

itself so for HR 370 HR I've highlighted

here the HTE grade three or RTF reverse

treated foil or options so I I picked uh

you know I went online and find a data

sheet of

HTE and uh and reverse treated foil um

you could also contact the supplier and

ask well who's the foil supplier and

they'll tell you then you can go to the

foil supplier and get the data sheet of

the foil and here you'll have the

roughness parameters that you need later

on for our

simulation so you need to gather this

info as part of your comparison so for

Isola we can only get a foil in HTE or

RTF and you can see HTE the roughness of

the foil for 18 Micron which is a half

ounce or 1 o foil it's either five

microns or seven so it's very rough foil for

for

HTE and you can see the roughness of the

profile is shown in these pictures here

there's two sides of the foil this

laminate side gets bonded to the core

material and the other side they often

call resist or shiny

side if I look at this picture this core

they have the the rough side and then

the pree is this other side of the foil

and that's what uh is showing basically

here so most of the time they only give

you the laminate side of the foil that

goes to the core because when they build

the PCB the board shop now is going to

uh add additional roughness to the prea

side of the foil yeah I I've seen it I

will show it in in the video what I'm

preparing like before actually it goes

to oven they have to do some they have

to make it rougher so it sticks better

together yes yes

exactly yeah the board shop that's the

board shop needs to make it rougher um

so that when they put this all together

you know this pree think of it as the

glue that glues these cores

together and you need to have some sort

of roughening on the on the foil itself

to get better bonding uh when they press everything

everything together

together

okay so that's really this resist side

okay you can see that side is typically

smoother but

now when you look at this mlsg that's

typically a an RTF reverse treated foil

that means they they put additional

treatment on the smooth side like here

and then that will get bonded to the

core and then this resist side is just

you know normally this roughness so

that's typically how they show this

picture here you know at the bottom

that's typically the Linate side and

over here the smoother side or or the

treated side is this bumpier side on the

top here so that's typically you know

how they they refer to it regardless um

you know you can see that the reverse

treated foil rsz is is uh much

bigger it's okay three it's only three

okay three versus five or seven see so

typically you want to use foil with a a

lower roughness so you know certainly

wouldn't choose HTE foil when I built

the stackup I would specify if I was

using uh 370 HR I would use the

smoothest that I can which is the

RTF okay any regardless if you have two

options you pick the lower roughness

option really that's what I'm trying to

show here when we do simulation we will

try we try different reference to see

how I will use different material well

for instance for is Sola I will use

basically the numbers that I here okay

right and if I use Isola I'll use the

roughness from a data sheet okay and

you'll see the difference obviously the

whole the whole it's it's not just like

one little thing I mean playing with

rough this all you're going to do is if

I go back to here all you're going to

see is this line go closer or whatever

here it's it's just like the rougher it

is the more loss you're going to get

excuse me from the foil so you know

that's step two is really now picking

you know looking at the roughness of the

foil for it and and Gathering the data

because you're going to need this data

when we do the channel modeling

so we're at the Gathering stage now help

decisions the next step if we build a

stock up once we've chosen a material

like say we chose 370 HR or terreen once

we've chosen which one and here I'm just

comparing the two

two

um they have what they call these

dialectric constant

dissipation uh DF tables they're

Construction table is what we call them

and this is what the board shop uses to

to make their stock up they will get

numbers from here not from

here okay okay so what is the difference

why we would like to well when we build

it um if I blow up one of if I blow up

this one a little bit

more you can see a typical construction

table if I start on the left this column

here is what we call the glass style

so this describes the weave pattern of

the of the fiberglass

because for PCB um lamin it's it's not

just uh resin it's made up of resin and

fiberglass cloth do you have picture

picture

uh do I have a picture but not a picture

just hold a

second I got better than a picture I

have a model

model okay

okay

so it's getting bummed up because of the

yeah but I see it I see it so this is

like this is the FI this would represent

the fiberglass weave that's in uh in

your laminate your

PCB when you buy a core material with

copper on both

sides the dialectric is got fiberglass

weave like

this and then let's just

say let's just call this the you know

the the uh the stuff so they uh the the

resin for instance you know and

everything is put together like this so

if you can envision your resin is all

mixed in with

this it's what it call you know that's

what it is

so each one has a different glass style

so a 1035 weave will Define how wide

these pieces are the different weave

patterns okay so that can be like also

bigger space between them or yes M

exactly right so there is inconsistency

bigger space there is there is a

basically when signal travels yeah it

will see like resin and the fiber resine

fiber so what happens when they spread

it the actual pieces that are coming

here they actually spread the Little

Fibers to fill in the space so now you

know it's still got a a shape to it but

it's not totally open like this the

glass is more spread out so they call

that spread

weave and usually this 1035 numbers or

1078 it'll Define what this weave looks

like right so smaller is better then or

what well spread weave you you want to

you don't want to you want to fill the

spaces yeah okay you don't big spaces I

don't have really slides to show this I

have other presentations where I have

talked about it certainly for sure but

not I think we talk about this in our

previous presentation I will link it I

think so so let's not focus too much on

the weave itself it's going to be

different weav but more

importantly this tells how much uh resin

content is in in the dialectric versus

glass so here it says you know of this

of the thickness each one of these have

another thickness too so before I get

into the

resin each one of these glass dials will

have a a a thickness of the laminate and

that's important when you go to build

your stackup because you want to build

up um the different thicknesses so you

can calculate impedance in the whole bit

so you know once you start doing your

impedances you say I need this much

thickness so you'll look on this table

and say oh I need a I'll need one layer

of this and another layer of this and

it'll build up so board shops use this

it has the thicknesses and it has the DK

and DF at the different frequencies of

Interest right and we need this DK and

DF when we do this impedance

calculations so that's why these tables

are important they're used when we

actually build the stackup and uh so I'm

really focusing on you really need this

data to build a stackup and then when I

go to doing a stackup for

example this would be a typical stackup

and you'll see

here excuse me

drink okay so I just made a um a

hypothetical stack up just for this demo

it's um it's uh 10 layers top layer to

bottom is 10 layers and it's very it's a

very simple one just for the

demonstration it's showing a core in pre

or sorry in the middle is uh

um two planes in the

middle uh should be plain plane between

five and

six there's two planes there and then I

have like signal

three is a highspeed layer and Signal

eight will be another right now if you

add more layers you you could add more

signal layers but they'll be pretty

similar to this so for the demo I'm

restricting it to just uh uh signals on

a short layer and signals on a longer

layer through the board for instance on

layer eight uh we're going to look at

the difference of the Via performance

for instance um but before we get to

that you basically would do this and

here is like board shops will use a 2d

field solver many board shops use polar

um fuel

solver uh I use ER for doing the things

to calculate my uh

impedances and if I'm building a stock

up to validate things so I just did a

hypothetical one like board shops would

do each layer that has impedance they'll

do a a polar simulation for it so this

is an example of of the top layer micro

strip um the layer width is 8 mil wide

we want 50 ohm in p

uh for single-ended I didn't show the

single-ended one and for

differential uh 100 plus or minus 10%

that's what our spec is so you play with

the different thicknesses that you need

and here you can see 2 *

1067 the dialectric

thickness is 4.6 when you put it

together okay and then I got 3.65 from

the table and the DF from the table at

10 GHz

right then the Press thickness after you

press it so the pree for instance we

want pree here the pree you know two

times 1067 ends up being 4.6 Ms thick

but after they press the board all down

together uh the Press thickness you know

is roughly the uh thickness of the uh of

the copper that fills in the space so

you lose resin for pressing basically

and the board shop their process they

will they will know what the Press

thickness is in Mills so I just

approximated this right now for the

demonstration a

4.5 and how do they know they have some

kind of the software what they use it

can calculate how much space will be

filled up with the yeah so let me let

let's go back to here yeah I know what

what will happen basically the resin it

it will flow into empty spaces between

the exactly right tracks right so you

know I you know you can estimate it's

based on amount of copper balance that's

on on the on your layer so after your

pattern that you etch all the copper

away it's going to have amount of

percent copper and the less copper you

have you know on there with all your

tracks the more resin you're going to

lose so that's why they have you know

this thieving stuff where they add extra

copper to fill to balance the the thing

out you you make it roughly like say 50%

so you know the board shops they can

calculate how much volume they're going

to lose of resin that they have and

that's where this resin content comes

into play so high resin cont content

means things will flow you know flow

more and everything else so board shops

will know what it is I don't you know

when I do the thing I really start here

as a rough point from SI perspective

I'll start here then I'll work when I

have this proposal for instance I'll

give it to now the board shop that I'm

going to be fabricating with and I'm

saying okay please sanitize my numbers

and give me back what you think these

things should be and what the impedances

will be right and we go back and forth

this way it's not like I just hand it

over you need to when you're doing highspeed

highspeed

stuff you know you need to go back and

forth with your fabrication I know this

right you can't work in isolation and

then give it to the board shop or you

can't say to the board shop just give me

something you know you need to do some

stuff ahead of time I would like to I'm

sorry for interrupting I would like to

uh talk a little bit about when people

have to start to worry about all these

small numbers because sometimes we make

this kind of video and then people are

designing I don't know a microcontroller

board and they start playing with copper

roughness or something and uh uh so when

they really should be so detailed

example well for me the way I work

is everything over a gigahertz I you

know I want to

be um I want to be more involved if

you're worried about loss and impedance

Reflections you know if you're if you're

worried about

impedance you need to know these numbers right

right

you know

the certainly certainly anything o over

like 20 gigs or let's say pcie let's say

gen uh one two gen

three gen three above you start need to

start paying attention to Gen 3 is what

frequency I think that's 16 16 gen one

is no wait a

minute gen Gen 3 is eight gen or I think

a 16 right I have to I have so basically

about maybe five gahz yeah well I I

typically do it right I have another

question before we continue so where did

you get these numbers for copper foil

DF oh oh this ah okay that's a good

question for coer copper for not for preak

preak

yeah so copper right so this

is okay let me best best to show this up in

here I can blow this up a little bit

bit

so that DK DF that

represents uh the resin that fills the

space uh between the copper

traces between the copper traces you're

not going to have any uh

fiberglass right

fiberglass it's mixed in when you press

it oh I understand so it's not the D and

dec of the copper itself but it's on

it's the resin on the layer where the

yes that's yeah very good point that you

brought that up uh because when you go

to do the modeling you can see here

that'll represent the the resin in here

that's the

reer right and that's the DK DF in here

okay so yes so when you have like this

is a strip line when you have strip line

means you have a reference plane on top

and a reference plane on the bottom

between them signals are there so when

you fabricate a board pretty much you're

going to have three decs you're going to

have the dec of the core like H1 you're

going to have the dec of uh they call it

H2 here and that includes this but

you're going to have the DK you know

here and then you're going to have the

DK in between they're going to be different

different

numbers and the core you get from the

construction table uh like here

construction tables will have core data

and construction table I didn't show the full

full

thing there's a table for the prepregs

as well like 11035 pree will have

numbers I didn't have a slide for all of

the but they're Bas you can see here the

pree 21067 the prepreg is uh

3.65 right and then uh here the 21080

it's a different one want to get the

right thickness for the impedance you

can see the DK is uh

3.99 right and then the

copper um DK is hard to estimate so it's

not so what I do is I go to um the

construction table and and I look at the

prea not the core data but the prepr

data I'll pick the pre cig that has the

highest resin content in the whole table

right if it's 76 or 80% I'll pick that

as a number to put in

okay you know it's not a huge deal if

you have if this is half ounce copper

it's not going to make a huge deal so

depends how precise you want to be most

people just use the same number as the

prepreg layer right and you'll be close

enough you're not it's not going to be a

drastic change in impedance right um for

me I try to put every little bit into it

so okay so you use the highest resin

because basically that will be the

closest to the 100%

resin pretty much that's what I say

right and you

know uh the it's like uh you know Eric

baton our friend Eric likes to say an

answer now is sometimes better than a

good answer late so you know you know

it's not going to be as high as the core

or or even what the the real pre-ra so

you're going to you know it's going to be

somewhere less than or less than that

right because the resin typically has

the dks are lower than than the copper

itself or the

the dialectric okay I think is it

everything about this table or we can

maybe move to next slide yeah yeah so so

I did a stack up for 370 HR and then I

did something similar for

tg400 okay so I tried to have something

roughly the same thickness as it's going

to be and I chose materials to be that

way to get impedance but what you'll

notice the difference is um is the line

width in space okay so here you see

3.59 is needed in 10 mil wide

separation right for the thing um to get

like a 13 mil pitch between the two

versus uh here I can get a a a wider

wider mine width for the same 10 mil space

space

okay what's Peach oh the pitch is the

between the two traces Center to Center

ah okay I understand okay yeah so uh I

put the pitch column in because

sometimes people just do pitch and don't

put space right some sometimes they have

line width and Pitch then you have to

subtract the two to get the space

because when you do this modeling this

tool wants a space right it doesn't want

pitch other tools may want pitch right

so I put both in the

so you can calculate either obviously

this is adding the two together you'll

get the pitch M right anyway but the

nice the important thing here is you

see part of the loss of your channel the

narrower line WDS uh will have more loss

too okay

so you know it's not just the roughness

but the the width of the trace will also

contribute to the loss so so here you

can see well we can achieve 3.59 we're

going to have a narrower Trace anyway

compared to here

so the so we see here that that lower

DC the lower DC numbers we're using

allows us to get a wider trace for the

same you know and wider it mean there

will be lower loss loss as well so we're

benefiting is that as well so DK when we

go back earlier as is another thing why

it's still important right for impedance

but it does factor into the loss as well

and you can see I tried to make the

stack up so they're comparable for 62

mil thick or what is it 1.5 millimeter

board right rough okay so this is what

we will need for the simulations take up

yeah I tried to make the board

thicknesses the same you know as if

you're trying to make a versus B in all

respects you can't do it exactly but

close enough right yeah so so I've done

I've done the two things

so now comes uh common roughness models

that are used in the simulation tools

that are out there so if we go back to

our equation that we started with for

the transmission line we have the

dialectric loss that's the DF basically

Works into it then the insertion loss of

the conductor

itself and then they have this

correction factor which we just call k k surface

surface

roughness this K applies to the

conductor so you apply this roughness

model to the conductor loss

loss

so basically there's three different

models out there popular models there's

some other ones too but the hammer stud

model is been basically the first model

that really has been dealing with

roughness uh for for decades and back

down to originally the work from uh

Morgan uh did a lot of work in 1949 I

believe and then hammerstad Jensen they

took it further they came up with this

uh uh correction factor equation simple ksr

ksr

to here and this Delta symbol is really

the roughness of the peak to Peak and

they basically modeled the roughness as

a triangular profile of that and and

when you took that the peak-to peak

height of that triangular profile and

they said that was the RMS value of the

Triangular profile okay and that's what

so we can use that RZ number that you

had in this table okay MH or when we put

that in so you'd use RZ divided by 2 <

tk3 and you'd put that number in and

that's been fine that model has been

good for number of years uh but it

started to lose steam uh between 3 and

15 gigahertz for accuracy

okay and then around

2009 time frame

that's when uh this hurry snowball model

came onto the

scene uh it proved to be very accurate

past the 15 all the way you know 50

gigahertz you can get it to match really

well the difficulty with the hurry

snowball model was um you needed to get

to relied on getting uh scanning

electron microscope pictures of the

roughness profile and that's what we see

here is the roughness profile and then

you know you'd assume from this you can

measure a hexagonal Arrangement and then

measure these little little balls uh and

then number of spheres very complicated most

most

people you wouldn't be able to do it

most people use the hurry snowball model

um what if they had measured data you

know they measured something and then

they would fit these parameters till

they got a good fit to that model okay

and then they would use those numbers um

for their future

modeling so that's that's fine but if

you're doing a design from the start it

doesn't really help you um the hammer St

would work because you have the RZ

numbers so around 10 years ago now I

started investigating is there a way to

um come up with something similar and

I've come up with what we now call The

Cannonball hoay model it's based on this

Cannonball stacking of equal spheres 14

equal spheres they're stacked in an

arrangement like they used to stack

cannonballs and basically we still use

the same IDE as hammerstad we use the

rsz peak to Peak uh

number in it and eventually the equation

simplifies to just this and the nice

thing about it it has the Simplicity of

hammerstad and the accuracy of the huray

model um in there and I've done several

papers since over the last 10 years on

all this

modeling all on my website certainly or

the detail about this is there but

suffice to say right now now uh if you

have RZ the radius of the sphere that

you need for this huray is simply 06 time

time

RZ and the area of the flat a flat is

simply 36 time the radius squared okay

we don't have to worry too much about it

because many tools now have uh the huray

model built in okay

and this table I put together is uh is

showing popular tools that use a version

so like you know you can use the

Cannonball huray model some tools have

it directly in the tool so polar SI 9000

e they were actually the first company

to adopt my model in it which is very

nice and similarly Mentor hyperlinks now

seans basically they have a cannonball huray

huray option

option

z0 uh Z solver you know these these

people here they include it directly so

all you really need to do is put RZ

number in okay okay in that tool okay

now we go to Anis like hfss type thing

or Cadence tool requires uh what they

call a surface ratio and nodule radius

parameters for their Hur model and

really that's just this part of the

equation so the nice thing about my

Cannonball huray model

is uh this Sr for these two tools will

always be

4.9 always it cancels out to that it's a

constant and then you're left with just

an RZ you just put the RZ and you

calculate R by

606 okay I've got this so for the other

software it's

similar yeah you can use these numbers

which are in the

table yeah we can move to the next slide

because I I'm a little bit worried we

will run out of okay okay so where do we

get four numbers where I talked about

you go from data sheet that's okay now

now the point is uh now that's uh that's

fine for the laminate the PCB shop I

said before they add roughness to the

foil to the shiny side so typically you

know you find out where do we get that's

called the oxide or oxide alternative

that they put on and you know you can

get that from the board shop or if you

know what they have or from other data

but you need that number to put on the

other side of the foil when you do the

simulation that's all I'm really

saying and I guess we can go now to a

demo okay transmission line model right

okay so we have the St up we have all

the numbers and everything what we need

now we have

to fill it up our simul go through this

exactly so that's a thing so when I'm

going to do there is I'm going to start

instance and uh so when you first start

up polar you're presented with like a

screen like this right this is before we

start so people can use also different

software correct but oh oh absolutely

absolutely we're not focus promoting po

or something yeah yeah yeah yeah you can

use any 2D solver you want um

I'm I'm showing a demo for Polar because

mainly that's what I use and a lot of

board shops use it so it's good but the

process is very similar and you'll see

polar not only does stackup but it also

does the transmission line model that

we'll need okay okay so so we'll start

with that so we end up with like a 2d

solver uh like this is what other

software will have separately for

instance so you'd put in the parameters

from the

stackup uh if I get if I pull the bo it

up I I reduce this down a little bit just

just

to just to show I think this is the TG

oops wrong one

one

yeah this is a yeah this is the TG so

you already loaded

yeah yeah so you would put in you put in

the numbers from H1 for instance H1 is

uh 5.1 thickness here you put in the

thicknesses and the Press thicknesses

you'd be dealing with so this column

you'd use you know to put in these

different numbers here right for H2 and

the spacings and

things and line wids

what about this upper and lower line

width oh yeah so this represents the

edge Factor so when they etch out the

pattern you know the etching will

control that angle so you know different

software tools will ask oh what's the

they ask for an angle to put in so you'd

have to calculate the angle polar says

oh W1 W2 so

typically you you know for just a normal

process you might lose this is a typical

thing so it's point three I put uh one

four about a mill yeah one m one oh the

nice thing about polar too it

automatically converts to microns if you

like using microns I'm used to the mill

so okay it's that so we'll talk about

that I would like to just add uh for

anyone who doesn't know exactly why this

is happening I believe it's because the acid

acid

will will longer be on the surface then

it's going down and down now that's why

it will be like this yeah so but yes

you're right you have your circuit

pattern on it when they start etching that

that

copper you know if they the etching

controls that so you know they can very

good board shops can make pretty Square

traces but you know so you can play with

that number to see the effect you know on impedance for instance if I no no you

on impedance for instance if I no no you don't need to change it don't change

don't need to change it don't change anything okay but you can see you know

anything okay but you can see you know you can that's why the solver you can do

you can that's why the solver you can do the what if analysis right and you can

the what if analysis right and you can use that to Define in your fabrication

use that to Define in your fabrication notes now what chch angle you want if

notes now what chch angle you want if it's important I'm pretty sure they

it's important I'm pretty sure they would be very happy yeah but you know

would be very happy yeah but you know have that flexibility so that's really

have that flexibility so that's really what's showing here W1 W2 and the other

what's showing here W1 W2 and the other thing that's important is um when you do

thing that's important is um when you do a field

a field solver the wide side of the trace will

solver the wide side of the trace will always be on the core

always be on the core yes okay so when you're looking at the

yes okay so when you're looking at the your

your stackup it's important to identify what

stackup it's important to identify what the core layer is you know and do things

the core layer is you know and do things uh correctly properly here because this

uh correctly properly here because this is one picture sometimes uh you know it

is one picture sometimes uh you know it could be flipped when you look on the

could be flipped when you look on the actual layer the core is the other way

actual layer the core is the other way around right and you know you're not

around right and you know you're not paying attention so H1 always makes sure

paying attention so H1 always makes sure it's core and with polar you have to add

it's core and with polar you have to add uh you know this prepreg layer plus the

uh you know this prepreg layer plus the thickness of the trace because it shows

thickness of the trace because it shows H2 is to the bottom other software well

H2 is to the bottom other software well will go to the top of the

will go to the top of the trace okay as an H2 mhm so that's just

trace okay as an H2 mhm so that's just something you know unique to Polar but

something you know unique to Polar but something to be aware of regardless you

something to be aware of regardless you just need to know your so once you do

just need to know your so once you do that you know you work this to get your

that you know you work this to get your impedance that you want basically okay

impedance that you want basically okay another question so the separation

another question so the separation region the electric that's the one what

region the electric that's the one what we were talking about okay the last one

we were talking about okay the last one the last one r e r yeah that's exactly

the last one r e r yeah that's exactly that's that's that resin okay right

that's that's that resin okay right you'd fill that in and polar it has all

you'd fill that in and polar it has all different types of uh you know whatever

different types of uh you know whatever you have if you have coded micr strip

you have if you have coded micr strip you have all this stuff so that's why

you have all this stuff so that's why it's a useful tool for stackup design I

it's a useful tool for stackup design I I wanted to talk about these numbers

I wanted to talk about these numbers because I remember seeing these tables

because I remember seeing these tables many times and I was like oh usually I

many times and I was like oh usually I only know I don't know six of these

only know I don't know six of these paramet I have no idea what are the

paramet I have no idea what are the other parameters what to fill up that's

other parameters what to fill up that's why I wanted to

why I wanted to talk about this a little bit more sure

talk about this a little bit more sure okay is there anything no there is okay

okay is there anything no there is okay I ask everything we can move okay okay

I ask everything we can move okay okay fine now once you have your stack up you

fine now once you have your stack up you got your impedance and everything so now

got your impedance and everything so now we want to um you know model the length

we want to um you know model the length of Trace that we want to use how did you

of Trace that we want to use how did you get the impedance you just press the

get the impedance you just press the button or what was there there's there's

button or what was there there's there's these tabs here just calculate okay yeah

these tabs here just calculate okay yeah lossless

lossless calculation really gives you just the

calculation really gives you just the impedance of the cross-section but you

impedance of the cross-section but you clicked on the calculate button where

clicked on the calculate button where the imped is yeah well no at the bottom

the imped is yeah well no at the bottom at the bottom there's these

at the bottom there's these tabs here yeah but I mean how did you

tabs here yeah but I mean how did you calculate differential impedance you

calculate differential impedance you press the calculate button oh okay once

press the calculate button oh okay once you put all these numbers in you click

you put all these numbers in you click the button okay and it and it calculates

the button okay and it and it calculates okay it's straightforward and if you

okay it's straightforward and if you want to know more like when you click

want to know more like when you click more it gives you more uh more stuff it

more it gives you more uh more stuff it gives you the differential it gives you

gives you the differential it gives you the odd mode impedence even mode common

the odd mode impedence even mode common mode

mode impedance you know effective dialectric

impedance you know effective dialectric con it gives you the the near and cross

con it gives you the the near and cross talk coefficient number um coupling

talk coefficient number um coupling percentage it's got a bit of extra stuff

percentage it's got a bit of extra stuff cool but really after you put in the

cool but really after you put in the numbers you calculate right in it okay

numbers you calculate right in it okay so the next thing is uh so now we've got

so the next thing is uh so now we've got our

our impedance now we want to have different

impedance now we want to have different lengths of transmission lines okay but

lengths of transmission lines okay but before you do that you need to uh go

before you do that you need to uh go into the uh because your dialectric

into the uh because your dialectric won't be constant Decay over the whole

won't be constant Decay over the whole frequency right it's it's just not so

frequency right it's it's just not so you have to go and put in what they call

you have to go and put in what they call a

a CAO dialectric model and regardless it

CAO dialectric model and regardless it it has that feature so you just edit it

it has that feature so you just edit it and then You' put the DK for the three

and then You' put the DK for the three regions in here H1 you put the DK that

regions in here H1 you put the DK that you had on the first

you had on the first sheet DK and then here's where you put

sheet DK and then here's where you put DF right in it so you put DK DF and the

DF right in it so you put DK DF and the frequency 10 GHz right so for these

frequency 10 GHz right so for these three regions you put in the DK and

three regions you put in the DK and everything and now you can hit calculate

everything and now you can hit calculate and in in a moment here you'll see U you

and in in a moment here you'll see U you can see here uh the dialectric is not

can see here uh the dialectric is not constant over

constant over frequency right at 10 gigahertz it's uh

frequency right at 10 gigahertz it's uh 3.1 but at 50 gigahertz for instance

3.1 but at 50 gigahertz for instance it's uh less than that so this shows you

it's uh less than that so this shows you how the dialectric will behave over

how the dialectric will behave over frequency based on the model you

frequency based on the model you selected yes a caal CAO dialectric mod

selected yes a caal CAO dialectric mod that's important if you have a constant

that's important if you have a constant you know it just it's just

you know it just it's just flat for modeling you don't want that

flat for modeling you don't want that you want a causal causal model okay and

you want a causal causal model okay and uh so it has that so you have to put

uh so it has that so you have to put these numbers in in order to get your

these numbers in in order to get your transmission line model that we're going

transmission line model that we're going to use so we close that and here's where

to use so we close that and here's where we add the roughness now has the

we add the roughness now has the roughness here and we en if we don't

roughness here and we en if we don't enable then then you have to guess what

enable then then you have to guess what these numbers are right when you put

these numbers are right when you put enable it now you put the roughness on

enable it now you put the roughness on each side like I show here this is from

each side like I show here this is from the data sheet as I told you you know

the data sheet as I told you you know the mat side 1.1 from the data sheet and

the mat side 1.1 from the data sheet and then the drum side I picked 1.5 from

then the drum side I picked 1.5 from oxide from that sheet that I showed okay

oxide from that sheet that I showed okay so you put those two numbers in hit

so you put those two numbers in hit calculate and then it puts the right

calculate and then it puts the right numbers in here that you want

numbers in here that you want and hit apply so now we're uh we're

and hit apply so now we're uh we're we're pretty much there and then here uh

we're pretty much there and then here uh so what line lengths do we want so

so what line lengths do we want so here's we're in Mills or if you want to

here's we're in Mills or if you want to put Micron if you want to put millimeter

put Micron if you want to put millimeter it just changes everything automatically

it just changes everything automatically for you so we're we're one inch right

for you so we're we're one inch right now or a th000 Ms and if I hit calculate

now or a th000 Ms and if I hit calculate it's uh it's going to run through just

it's uh it's going to run through just quickly and we'll see we'll see how

quickly and we'll see we'll see how things will show up here M and so these

things will show up here M and so these are basically the graphs what we could

are basically the graphs what we could see in your presentation yes so you you

see in your presentation yes so you you Ed you set up all these numbers you set

Ed you set up all these numbers you set the length what you needed and now it

the length what you needed and now it will tell us for this specific length

will tell us for this specific length what we

what we said how much loss we will have for each

said how much loss we will have for each frequency yeah so when this finishes

frequency yeah so when this finishes it'll display things on the graph here

it'll display things on the graph here and uh

it's doing a second pass because there's two traces it's fine okay you said

two traces it's fine okay you said minimum frequency 10 mahz maximum 50 g

minimum frequency 10 mahz maximum 50 g up to 50 so I'm simulating between yes

up to 50 so I'm simulating between yes between 10 and 50 gigahertz and I have

between 10 and 50 gigahertz and I have like 2,000 steps in between so it's

like 2,000 steps in between so it's going to calculate everything it's going

going to calculate everything it's going to take the dialectric and DF here and

to take the dialectric and DF here and it's going to take the roughness and

it's going to take the roughness and it's going to plot it all over oh this

it's going to plot it all over oh this is very nice actually wow you go so here

is very nice actually wow you go so here you see the green will show the loss for

you see the green will show the loss for dialectric

dialectric oh the red will show smooth conductor

oh the red will show smooth conductor loss with no roughness just the

loss with no roughness just the conductor by itself the blue will show

conductor by itself the blue will show smooth attenuation when you combine the

smooth attenuation when you combine the dialectric and and the copper and no

dialectric and and the copper and no roughness you'll get blue and then

roughness you'll get blue and then conductor loss with roughness is uh just

conductor loss with roughness is uh just just the conductor with roughness is

just the conductor with roughness is yellow but the most important is this

yellow but the most important is this final one where it puts everything

final one where it puts everything together and here you can see you know

together and here you can see you know you can go from the thing here if I

you can go from the thing here if I click roughly where I want it'll show up

click roughly where I want it'll show up here on the box here uh we're at

here on the box here uh we're at 1.03 you know at that frequency me so

1.03 you know at that frequency me so this is the good material I don't

this is the good material I don't remember or this is this is the good one

remember or this is this is the good one mhm this is the TG because these numbers

mhm this is the TG because these numbers are very

are very low yeah the DB is very low and it's one

low yeah the DB is very low and it's one inch too this is per inch okay so you

inch too this is per inch okay so you play this you know this is very nice for

play this you know this is very nice for the display but the very nice thing

the display but the very nice thing about s polar SI 9000 because you're

about s polar SI 9000 because you're already in the tool is now I can save

already in the tool is now I can save this in a touchstone file or S

this in a touchstone file or S parameters M which I'm going to use now

parameters M which I'm going to use now to the channel model M right so

to the channel model M right so basically s parameter is something is

basically s parameter is something is the model of our tray in our specific uh

the model of our tray in our specific uh stup and then when we do simulations we

stup and then when we do simulations we will basically work with real PCB track

will basically work with real PCB track that's the model what we will generate

that's the model what we will generate now right exactly they're called

now right exactly they're called touchtone format we'll take we'll take

touchtone format we'll take we'll take we'll take Touch Tone files for the PCB

we'll take Touch Tone files for the PCB traces uh with do one for the Vias for

traces uh with do one for the Vias for connectors the connector company will

connectors the connector company will supply a touchone file for the connector

supply a touchone file for the connector so you put all these together to build

so you put all these together to build your channel you see but you need to

your channel you see but you need to make models of your each part first so

make models of your each part first so so here you can save uh you can export

so here you can save uh you can export touchtone format you know you'd give it

touchtone format you know you'd give it the file name and in a in and when you

the file name and in a in and when you have a differential pair it'll always be

have a differential pair it'll always be S4 p as an extension M okay and you just

S4 p as an extension M okay and you just give it a you just give it a name what

give it a you just give it a name what you want and a number of steps and where

you want and a number of steps and where you want to save it and it'll export it

you want to save it and it'll export it but I'm not going to do it now because

but I'm not going to do it now because it we already have it oh yeah okay it's

it we already have it oh yeah okay it's fine but I want to make another point

fine but I want to make another point with polar recently in the later

with polar recently in the later versions they now have a way is a format

versions they now have a way is a format for multiple line links okay and you can

for multiple line links okay and you can save you pick the direct where you want

save you pick the direct where you want to save but now you can save uh one like

to save but now you can save uh one like one

one inch

inch 2,000

2,000 3,000

3,000 4,000 5,000 and so you have all these in

4,000 5,000 and so you have all these in one model and you can connect which one

one model and you can connect which one you want well no what this will do when

you want well no what this will do when you export now it'll save Touchstone

you export now it'll save Touchstone files for one inch 2 inch three okay you

files for one inch 2 inch three okay you can have up whatever you want so if you

can have up whatever you want so if you have 10 you can have 10 different

have 10 you can have 10 different touchdown fun and I did do that prior so

touchdown fun and I did do that prior so I have saved this this is very very

I have saved this this is very very useful if you your camera is off again

useful if you your camera is off again yeah if you want to do what if uh what

yeah if you want to do what if uh what if analysis right you want to build a

if analysis right you want to build a Channel with different mics you can do

Channel with different mics you can do it all at once now with it instead of

it all at once now with it instead of having to go here and do oh one inch run

having to go here and do oh one inch run a simulation you know two inch run a

a simulation you know two inch run a simul now you can basically do a series

simul now you can basically do a series of them and save all those at one during

of them and save all those at one during one time very useful let we see I would

one time very useful let we see I would like to point out uh when you use

like to point out uh when you use different uh simulators I know they can

different uh simulators I know they can generate this kind of uh as models for

generate this kind of uh as models for specific track in your PCV but I have

specific track in your PCV but I have never seen uh options for like setting

never seen uh options for like setting all the kind of different parameters

all the kind of different parameters what we could set here in this polar

what we could set here in this polar usually in standard

usually in standard simulators you can just use the standard

simulators you can just use the standard parameters for PCB

parameters for PCB stup yeah different tools have different

stup yeah different tools have different capabilities yeah but this is

capabilities yeah but this is interesting because for example I was

interesting because for example I was always thinking about how you include

always thinking about how you include this uh this

this uh this uh graph for different frequencies uh in

uh graph for different frequencies uh in these kind of models because when you

these kind of models because when you when you when you set up Stack Up in

when you when you set up Stack Up in standard simulator you only set it for

standard simulator you only set it for one specific frequency you can't set it

one specific frequency you can't set it for all the

for all the frequencies well no that's right that's

frequencies well no that's right that's why in here for instance all simulators

why in here for instance all simulators will ask what's the DK and DF at one

will ask what's the DK and DF at one frequency and it will make a model yeah

frequency and it will make a model yeah but then when you generate the uh s

but then when you generate the uh s model or S parameter you have it uh you

model or S parameter you have it uh you have the correct loss for specific

have the correct loss for specific frequency based on this graph no oh yeah

frequency based on this graph no oh yeah the total loss most things will most

the total loss most things will most tools will just give you this blue line

tools will just give you this blue line right this thing is you know kind of

right this thing is you know kind of nice to know but you don't really use

nice to know but you don't really use these numbers yeah but I think most

these numbers yeah but I think most tools will just use one straight line

tools will just use one straight line and that's all you do when you do the

and that's all you do when you do the Touchstone file it'll only give you okay

Touchstone file it'll only give you okay so this is not incl touch okay I okay I

so this is not incl touch okay I okay I understand okay I mean it's I mean you

understand okay I mean it's I mean you could well you won't be able to separate

could well you won't be able to separate it from the touchone right so what

it from the touchone right so what frequency touchone generates the S

frequency touchone generates the S parameter for for the minimum or maximum

parameter for for the minimum or maximum or middle from 10 to 50 gahz our

or middle from 10 to 50 gahz our Touchstone file will be right and then

Touchstone file will be right and then what for the electric loss value it will

what for the electric loss value it will use

use then pardon

then pardon so when it will generate the S parameter

so when it will generate the S parameter model yes for this all these frequencies

model yes for this all these frequencies what the electric loss it will use will

what the electric loss it will use will it use different dialectric loss for

it use different dialectric loss for different frequency yeah okay yeah

different frequency yeah okay yeah that's why this table you know that's

that's why this table you know that's why when we when we calculate this at

why when we when we calculate this at each frequency this is the DK

each frequency this is the DK DF yeah that it's using but what I

DF yeah that it's using but what I wanted to point out normally standard

wanted to point out normally standard simulators will not use different you

simulators will not use different you only fill up

only fill up one uh well if it's a CAO model like

one uh well if it's a CAO model like Dakar model or or Debby model

Dakar model or or Debby model or okay if it's asking you for a

or okay if it's asking you for a frequency of the DK and DF it's the same

frequency of the DK and DF it's the same as this okay but I never seen like

as this okay but I never seen like select which model you would like to use

select which model you would like to use there is just put here the

there is just put here the value uh

value uh well for instance if you use

well for instance if you use symor you know they have similar okay

symor you know they have similar okay one ads will have similar okay SS all

one ads will have similar okay SS all the tools have oh maybe maybe I just

the tools have oh maybe maybe I just didn't notice because I had no idea what

didn't notice because I had no idea what does it mean yeah yeah you need to make

does it mean yeah yeah you need to make sure that you choose a causal dialectric

sure that you choose a causal dialectric model and when you choose that option in

model and when you choose that option in the tool it will then ask you what is

the tool it will then ask you what is the DK DF and what frequency okay and

the DK DF and what frequency okay and then it'll use it'll know okay after

then it'll use it'll know okay after okay perfect okay yeah and the roughness

okay perfect okay yeah and the roughness you know they will have their own panel

you know they will have their own panel to come up and ask that was what I

to come up and ask that was what I showed you before they want oh I I guess

showed you before they want oh I I guess I never I never really went into so much

I never I never really went into so much details because as I said I didn't know

details because as I said I didn't know what does it mean and what to say so

what does it mean and what to say so maybe the other tools what I use they

maybe the other tools what I use they actually have it I just notice that

actually have it I just notice that because I never knew what does it mean

because I never knew what does it mean that's why we are creating this video so

that's why we are creating this video so everyone now

everyone now knows means and how to set it up yeah be

knows means and how to set it up yeah be aware that's all you

aware that's all you know it's just to be aware of of what

know it's just to be aware of of what you need to to do for highend high

you need to to do for highend high multi- gigabit stuff you need to

multi- gigabit stuff you need to understand it to get accurate simulation

understand it to get accurate simulation I mean every tool will give you a number

I mean every tool will give you a number at the end of the day how does that

at the end of the day how does that match to what your thing is built right

match to what your thing is built right that's why we do our signal integrity

that's why we do our signal integrity and modeling is to predict so we need

and modeling is to predict so we need good models and uh anyway this is the

good models and uh anyway this is the process this is my process I follow uh

process this is my process I follow uh Works fairly well and I don't just I

Works fairly well and I don't just I mean I have other tools as well I own

mean I have other tools as well I own Symbio as well um you know and I use

Symbio as well um you know and I use Symbio for different things you know but

Symbio for different things you know but if I'm building something from scratch

if I'm building something from scratch here this is very Qui

here this is very Qui it gets and more important if you have

it gets and more important if you have two different tools polar is is very

two different tools polar is is very user friendly um very quick and you can

user friendly um very quick and you can use it now to sanitize for instance a

use it now to sanitize for instance a symor tool somewhat right or another

symor tool somewhat right or another tool you know if you're getting

tool you know if you're getting something or if you have another tool

something or if you have another tool that extracts your layout your file you

that extracts your layout your file you know you want to sort of get a feel for

know you want to sort of get a feel for is that have I got something realistic

is that have I got something realistic there's a lot of other tools you can

there's a lot of other tools you can compare them yeah yeah you know it's a

compare them yeah yeah you know it's a it's a good way to have two different

it's a good way to have two different tool sets in your

tool sets in your toolbox not to say oh this one's better

toolbox not to say oh this one's better or worse it's not the case it's to

or worse it's not the case it's to validate that you haven't made

validate that you haven't made a a push button wrong mistake there's a

a a push button wrong mistake there's a lot of tools you have to go through

lot of tools you have to go through panel and panel of think it's very easy

panel and panel of think it's very easy to over over have an oversight of

to over over have an oversight of something and you'll always get an

something and you'll always get an answer but if you're only trusting the

answer but if you're only trusting the one everybody's human they make an error

one everybody's human they make an error yeah so it's nice to validate something

yeah so it's nice to validate something I actually I know the polar software is

I actually I know the polar software is really good because many PCB

really good because many PCB manufacturers they will send you

manufacturers they will send you screenshots from Polar and I wanted to I

screenshots from Polar and I wanted to I think uh I met them on some exhibition

think uh I met them on some exhibition PCB was or something like this I think

PCB was or something like this I think they always go there and I I asked for

they always go there and I I asked for price and it was quite expensive because

price and it was quite expensive because they always I don't know if they still

they always I don't know if they still sell it

sell it like kind of whole package which

like kind of whole package which includes lot of information for PC

includes lot of information for PC manufacturers because I asked them if

manufacturers because I asked them if they could make verion for uh for like

they could make verion for uh for like PCB layout engineers and at that time

PCB layout engineers and at that time they told me no no like cheaper version

they told me no no like cheaper version for you know

for you know oh well you know this is Si 9000 now

oh well you know this is Si 9000 now polar has you know their stackup planner

polar has you know their stackup planner and everything else it doesn't have all

and everything else it doesn't have all this capability either oh so maybe they

this capability either oh so maybe they have different versions now so if

have different versions now so if someone would like to have a look they

someone would like to have a look they they can check their website I think

they can check their website I think they have prices also on website I don't

they have prices also on website I don't remember yeah so you know if they're

remember yeah so you know if they're doing just stackup there's a version I

doing just stackup there's a version I think you have to go to find out

think you have to go to find out certainly you know the 2D solver is all

certainly you know the 2D solver is all you need to do to make a stack up if you

you need to do to make a stack up if you want to do signal Integrity

want to do signal Integrity modeling now SI 9000 is a better choice

modeling now SI 9000 is a better choice because it offer a lot more

because it offer a lot more capabilities you can build your

capabilities you can build your transmission L okay so we have we have

transmission L okay so we have we have the S parameter model what next okay so

the S parameter model what next okay so pretty much I think we're you know I've

pretty much I think we're you know I've saved all the files and I did the same

saved all the files and I did the same exercise for the fr4 the 370 HR so I

exercise for the fr4 the 370 HR so I have a number of different lengths for

have a number of different lengths for both okay so we basically have now our

both okay so we basically have now our transmission line models that we want

transmission line models that we want for the two okay okay could good yes so

for the two okay okay could good yes so I'm going to stop there and then I'm

I'm going to stop there and then I'm going to come up here again so where are

going to come up here again so where are we now ah okay so now let's talk about I

we now ah okay so now let's talk about I want to model

want to model VS all right and for that

VS all right and for that U I'm going to use

U I'm going to use ads um you can use ansis hfss you can

ads um you can use ansis hfss you can use symor they have nice tool there's a

use symor they have nice tool there's a number of tools out there you can model

number of tools out there you can model 3 D okay I'm going to use ads because

3 D okay I'm going to use ads because that's you know the tool and I'm you

that's you know the tool and I'm you know kind of operating at it and I'm

know kind of operating at it and I'm familiar with

familiar with so

so so I I have this already set up uh for

so I I have this already set up uh for it so I you know it takes a bit of not a

it so I you know it takes a bit of not a huge amount of time to set up but it's

huge amount of time to set up but it's it's fairly quick but all 3D tools

it's fairly quick but all 3D tools eventually you'll end up with a 3D model

eventually you'll end up with a 3D model of it so

of it so for here first of all with uh ads for

for here first of all with uh ads for instance I'm going to

instance I'm going to pull they have what they call like the

pull they have what they call like the substrate you have to create a substrate

substrate you have to create a substrate to

to uh basically Define what the material

uh basically Define what the material properties are similar to Polar so you

properties are similar to Polar so you build your stack up and everything here

build your stack up and everything here so here you have a prea and right now I

so here you have a prea and right now I got 370 HR

got 370 HR for you go in into there I have to pull

for you go in into there I have to pull this window from there and you put in

this window from there and you put in and here you see here's where uh you can

and here you see here's where uh you can use ads to do your stack up as well

use ads to do your stack up as well right it's it'll do the same job and be

right it's it'll do the same job and be just as accurate but now when you in ads

just as accurate but now when you in ads for instance you you click the

for instance you you click the dialectric and you go into this panel

dialectric and you go into this panel and here's where you would enter your uh

and here's where you would enter your uh DK and D M mhm right so youd put in DK

DK and D M mhm right so youd put in DK is 3.29 for core uh 378 Char you can

is 3.29 for core uh 378 Char you can have different you can add the materials

have different you can add the materials what do you use in your sck up exactly

what do you use in your sck up exactly you can have as many as you want and you

you can have as many as you want and you just choose which ones you want to put

just choose which ones you want to put in so here's where you put your DK DF

in so here's where you put your DK DF here real and 10 Delta 10d right and

here real and 10 Delta 10d right and make sure the type is Fenson Georg re

make sure the type is Fenson Georg re that means it'll be CA Oh yeah this is

that means it'll be CA Oh yeah this is what I just mentioned yeah that's where

what I just mentioned yeah that's where I didn't know yes now I see now I see is

I didn't know yes now I see now I see is it okay yes now you put in the frequency

it okay yes now you put in the frequency at 10 gigahertz right and don't worry

at 10 gigahertz right and don't worry about low and high just keep that there

about low and high just keep that there it's just 10 gig so that's where most

it's just 10 gig so that's where most tools will have something like this a

tools will have something like this a panel that comes up you know to put in

panel that comes up you know to put in the properties basically right and uh

the properties basically right and uh you apply it so I'm just going to stop

you apply it so I'm just going to stop cancel this just mhm so basically you've

cancel this just mhm so basically you've done it here's your stock up here you're

done it here's your stock up here you're just saying it's got copper and all your

just saying it's got copper and all your thicknesses so you really make an

thicknesses so you really make an equivalent stack up in the tool okay

equivalent stack up in the tool okay it's called the substrate file on it so

it's called the substrate file on it so I'm just going to stop that and when you

I'm just going to stop that and when you when you do it now when it does the

when you do it now when it does the simulation it gets that data and makes

simulation it gets that data and makes the model but now to build the model so

the model but now to build the model so that's the substrate so now we get into

that's the substrate so now we get into the Via itself all right

the Via itself all right and so I picked the Via stock here and

and so I picked the Via stock here and then now we start at the barrel so the

then now we start at the barrel so the barrel tells how long the whole via is

barrel tells how long the whole via is it goes through the whole board so layer

it goes through the whole board so layer one to layer 19 we're drilling all the

one to layer 19 we're drilling all the way through the board okay that's where

way through the board okay that's where it starts with uh and that's all we have

it starts with uh and that's all we have to do and then we want a back drill

to do and then we want a back drill eventually from the bottom side okay and

eventually from the bottom side okay and on the right hand side you can see uh

on the right hand side you can see uh you know a little thing will show how

you know a little thing will show how much backr and here it's not very much

much backr and here it's not very much back drill because you

back drill because you specify uh then me I see stop right now

specify uh then me I see stop right now I specified a stub length of 10 ms Max

I specified a stub length of 10 ms Max right so if you put it like zero it

right so if you put it like zero it would probably back drill yeah yeah

would probably back drill yeah yeah right so if you want to you know

right so if you want to you know normally or just try to try to change

normally or just try to try to change the end layer I don't know to number two

the end layer I don't know to number two or three we should see the back drilling

or three we should see the back drilling longer well well here here this this

longer well well here here this this panel here Barrel oh okay that's okay I

panel here Barrel oh okay that's okay I understand it goes all the way through

understand it goes all the way through right now pad stack the pad stack now

right now pad stack the pad stack now defines the uh drill

defines the uh drill diameter instance you put the drill

diameter instance you put the drill diameter of the drill not the finished

diameter of the drill not the finished hole diameter the actual drill diameter

hole diameter the actual drill diameter you have to put in a difference and then

you have to put in a difference and then you have to put the back drill diameter

you have to put the back drill diameter that you want to clear away so I had put

that you want to clear away so I had put 10 mil drill in this model um you can

10 mil drill in this model um you can put 8 mil whatever drill that you're

put 8 mil whatever drill that you're using for your model and and if you do

using for your model and and if you do what if analysis you can do if I use the

what if analysis you can do if I use the eight mil drill what is the impedance

eight mil drill what is the impedance eventually but I use 10 mil 15 mil to

eventually but I use 10 mil 15 mil to clear away after the back drill so it's

clear away after the back drill so it's a 15 mil drill it'll clear all that cop

a 15 mil drill it'll clear all that cop rad at the center to Center distance of

rad at the center to Center distance of the Via that's where your pitch now of

the Via that's where your pitch now of the Via comes so I picked a 1 millimeter

the Via comes so I picked a 1 millimeter uh pitch via

uh pitch via okay you pick that and it picks you know

okay you pick that and it picks you know picks it all up and then

picks it all up and then uh here you put uh layers with the feed

uh here you put uh layers with the feed I don't yeah see this is layers with

I don't yeah see this is layers with feed is uh pad diameters so with a 10

feed is uh pad diameters so with a 10 mil drill I use 20 mil

mil drill I use 20 mil pad and

pad and uh and if you don't have a feed coming

uh and if you don't have a feed coming in uh what is the the diameter so I I

in uh what is the the diameter so I I don't want any I don't want any pads on

don't want any I don't want any pads on the other layers yeah I see layers so I

the other layers yeah I see layers so I put the same as the drill

put the same as the drill size there's no pads unless there's a

size there's no pads unless there's a trace coming to it so you basically

trace coming to it so you basically that's your pad why it's not pled what

that's your pad why it's not pled what does it I would expect it should be

does it I would expect it should be plated oh yeah you can plate you put

plated oh yeah you can plate you put plating in thing and you put the

plating in thing and you put the thickness of the plate

thickness of the plate basically for for the analysis for the

basically for for the analysis for the Via here it won't make uh it's mainly

Via here it won't make uh it's mainly the outer D outer diameter with

the outer D outer diameter with relationship to I understand yeah so you

relationship to I understand yeah so you can put that if you want but yeah ites

can put that if you want but yeah ites yeah I understand

yeah I understand because it goes inside of the hle

because it goes inside of the hle basically it doesn't really influence

basically it doesn't really influence what's happening around okay that's

what's happening around okay that's right that's

right that's right uh and micras I'm not using it but

right uh and micras I'm not using it but you can specify if you had microa build

you can specify if you had microa build a model whatever it is now here's the

a model whatever it is now here's the feed layers this is where it's okay here

feed layers this is where it's okay here so here you can change it for example

so here you can change it for example only go to two to layer three or

only go to two to layer three or something and we should exactly so I

something and we should exactly so I started with right now the layer here is

started with right now the layer here is uh the feed one is the top so single

uh the feed one is the top so single layer one and feed two is layer eight

layer one and feed two is layer eight right here where where I picked it you

right here where where I picked it you know based on the

know based on the stackup try to change it I just like to

stackup try to change it I just like to see if it moves yeah it does there goes

see if it moves yeah it does there goes and big drilling go okay we can see it

and big drilling go okay we can see it now perfect that's the nice thing when

now perfect that's the nice thing when you have the back drill option turned on

you have the back drill option turned on you can see how it uh helps visualize

you can see how it uh helps visualize seeing where things are okay and making

seeing where things are okay and making sure that you're going to have a back

sure that you're going to have a back drill you know 10 m sort of thing plus

drill you know 10 m sort of thing plus you can uh the view here you can like

you can uh the view here you can like anything else is

anything else is front you can do some

front you can do some measuring of of of

measuring of of of things know you click here type of thing

things know you click here type of thing you can you can double Che everything

you can you can double Che everything whatever if you want you have that

whatever if you want you have that capability

capability you know you can use that to you know

you know you can use that to you know measure the type of stuff

measure the type of stuff there okay okay okay I understand there

there okay okay okay I understand there so it's straight pretty straightforward

so it's straight pretty straightforward and stitching vs that's where

um yeah I've seen on the top yeah I noticed there were there are two ground

noticed there were there are two ground vs yeah yeah you don't have I mean I put

vs yeah yeah you don't have I mean I put them in you normally you have it

them in you normally you have it normally this would be like a 1 mm BGA

normally this would be like a 1 mm BGA for high speed you you'd probably have a

for high speed you you'd probably have a a pair and then you'd have ground uh

a pair and then you'd have ground uh with it typically right

with it typically right so this is typically what you'd see in a

so this is typically what you'd see in a 1 millimeter pitch

1 millimeter pitch BGA that's what I tried to ambulate

BGA that's what I tried to ambulate here with a model okay and now we have

here with a model okay and now we have to create the model the s s parameters

to create the model the s s parameters or something yeah so once once we have

or something yeah so once once we have our model built we have our stack up we

our model built we have our stack up we have all the materials

have all the materials then you just click the

then you just click the simulate now it takes a while to

simulate now it takes a while to simulate through you click the normal

simulate through you click the normal simulate button it'll go through

simulate button it'll go through simulation and eventually it'll uh let

simulation and eventually it'll uh let me just push that up a minute eventually

me just push that up a minute eventually you'll end up with the results here okay

you'll end up with the results here okay like when you hit simulate it'll go

like when you hit simulate it'll go through and each one of these things

through and each one of these things it'll simulate each frequency and it'll

it'll simulate each frequency and it'll take you know a few minutes to to go

take you know a few minutes to to go through right all of it eventually then

through right all of it eventually then it spits out the results so each

it spits out the results so each simulation you'll have here so let's

simulation you'll have here so let's just pick it's finished now the

just pick it's finished now the simulation now I can show the S

simulation now I can show the S parameter results here there's the

parameter results here there's the results of the S parameters of of that

results of the S parameters of of that and not only that we can see the TDR

and not only that we can see the TDR results so s parameters will show us the

results so s parameters will show us the loss and TDR will l will show us

loss and TDR will l will show us impedance yeah TDR is the impedance here

impedance yeah TDR is the impedance here M here is the here is the impedance of

M here is the here is the impedance of the

the Via after

Via after simulation right here you know it comes

simulation right here you know it comes down to about you know

down to about you know 95 type of thing which is it's okay

95 type of thing which is it's okay right so you know you play around with

right so you know you play around with this until you get what you feel you

this until you get what you feel you want as as the uh and then how do you

want as as the uh and then how do you save it as uh

save it as uh paramet yeah so after you have let's

paramet yeah so after you have let's just show the TDR plot

just show the TDR plot again so here is a single ended of each

again so here is a single ended of each p&n for instance and then you have mix

p&n for instance and then you have mix mode and this is

mode and this is differential okay now that you have this

differential okay now that you have this um now

um now you

you file go file you can export

file go file you can export X and then you know you open a dat dat

X and then you know you open a dat dat display and you fill in a panel just to

display and you fill in a panel just to export and it'll export the S parameter

export and it'll export the S parameter of of

of of this okay for so I've done that already

this okay for so I've done that already I have S parameters of uh vas for uh the

I have S parameters of uh vas for uh the two different dialectric properties or

two different dialectric properties or two different

two different materials two different ISO the4 and the

materials two different ISO the4 and the high speed I've got via models for for

high speed I've got via models for for both okay now we are going to connect

both okay now we are going to connect everything together so so that's that's

everything together so so that's that's the V model uh pretty much and here you

the V model uh pretty much and here you can see comparing for instance TDR

can see comparing for instance TDR simulation results uh and here you you

simulation results uh and here you you bring now the model into ads by itself

bring now the model into ads by itself just to look at the V I'll show you when

just to look at the V I'll show you when we build the channel but nor now you can

we build the channel but nor now you can check both models and here you can

check both models and here you can compare the red one is uh using a short

compare the red one is uh using a short via for instance mm I see I see in the

via for instance mm I see I see in the pictures only from layer one to I don't

pictures only from layer one to I don't know Layer Two or something right and

know Layer Two or something right and blue represents you can see long versus

blue represents you can see long versus short okay my model so the file that the

short okay my model so the file that the file that we Sav is a short file and

file that we Sav is a short file and this this is one of the one of the

this this is one of the one of the materials you know for it um anyway so

materials you know for it um anyway so that's that's good for the Via

that's that's good for the Via simulation now here's an important part

simulation now here's an important part dialectric and

dialectric and isotropy and this is Recent research

isotropy and this is Recent research that I've been

that I've been doing and it's important especially when

doing and it's important especially when you're modeling

you're modeling Vias uh because the DK that you find in

Vias uh because the DK that you find in the DK tables are not the dks that you

the DK tables are not the dks that you want to use for Via modeling

want to use for Via modeling necessarily so just a very brief brief

necessarily so just a very brief brief thing on what an i tropy means is All

thing on what an i tropy means is All Glass reinforced laminates are

Glass reinforced laminates are anisotropic meaning uh the E Fields the

anisotropic meaning uh the E Fields the properties are different depending on

properties are different depending on the access axis uh generator because

the access axis uh generator because Tres are routed like this but vas go

Tres are routed like this but vas go right right so if I look at these three

right right so if I look at these three uh pictures here on the left starting on

uh pictures here on the left starting on the left if I put copper plates on the

the left if I put copper plates on the dialectric and this is how your

dialectric and this is how your dialectric is normally you got the weave

dialectric is normally you got the weave like this and if you have your copper

like this and if you have your copper laminate on the top and bottom and

laminate on the top and bottom and you're doing a transmission mine Trace

you're doing a transmission mine Trace right it's this way and when when you

right it's this way and when when you energize that thing the electric Fields

energize that thing the electric Fields between the copper e fields are

between the copper e fields are perpendicular so it's in the Z

perpendicular so it's in the Z Direction okay if I move the plates the

Direction okay if I move the plates the other end and energize it now the fields

other end and energize it now the fields are in X Direction or if I put it this

are in X Direction or if I put it this way they're in the y direction so we

way they're in the y direction so we assume that X and Y are going to be

assume that X and Y are going to be pretty much the same so we just call it

pretty much the same so we just call it I call it dkx y so the fields are in

I call it dkx y so the fields are in plain we call it in plane with with the

plain we call it in plane with with the the fiberglass weed and this way the

the fiberglass weed and this way the fields are out of plane soic in Z they

fields are out of plane soic in Z they go basically through the uh fibers but

go basically through the uh fibers but in the X and Y they can travel through

in the X and Y they can travel through fibers or through pre only right and the

fibers or through pre only right and the way the things are you know it's how the

way the things are you know it's how the the things are the capacitance is

the things are the capacitance is different so when it's this way each

different so when it's this way each layer because you have all these

layer because you have all these different layers of glass and everything

different layers of glass and everything you know like these

you know like these layers the capacitances are basically in

layers the capacitances are basically in series with one another so the DK is is

series with one another so the DK is is basically the capacitance it's the ratio

basically the capacitance it's the ratio of capacitance of the dialectric

of capacitance of the dialectric compared to when you replace the

compared to when you replace the dialectric with air that's how you get

dialectric with air that's how you get DK and when you when you calculate

DK and when you when you calculate capacitances in series it's like uh it's

capacitances in series it's like uh it's like they're they end up in parallel so

like they're they end up in parallel so anyway there's equations to calculate

anyway there's equations to calculate the thing and E Fields the other way uh

the thing and E Fields the other way uh they're they're kind of like in parallel

they're they're kind of like in parallel now so

now so so they calculate this in series anyway

so they calculate this in series anyway that's the topic of uh I'm doing design

that's the topic of uh I'm doing design con paper uh understanding that and uh

con paper uh understanding that and uh it'll describe it but here's a nice

it'll describe it but here's a nice picture showing the effect so when um a

picture showing the effect so when um a transmission line these red this red

transmission line these red this red energy that's the electric Fields as

energy that's the electric Fields as it's propagating at 20

it's propagating at 20 gigahertz so you can see we want dkz mhm

gigahertz so you can see we want dkz mhm we can see the difference like when it

we can see the difference like when it travels with the line it's like nice but

travels with the line it's like nice but when it travels through there are these

when it travels through there are these like I don't know it's going out little

like I don't know it's going out little bit between the right layers or

bit between the right layers or laminates right and what I'm getting at

laminates right and what I'm getting at is if you had dkz like say the data in

is if you had dkz like say the data in the data sheet or

the data sheet or whatever it depends on how they measured

whatever it depends on how they measured the dialectric MH it could be measured

the dialectric MH it could be measured the E Fields end up being in plane or

the E Fields end up being in plane or out of plane it depends on the test

out of plane it depends on the test fixture of the test method so for

fixture of the test method so for instance you know the test method if it

instance you know the test method if it gives a dkz for instance the trace

gives a dkz for instance the trace impedances will be right when you

impedances will be right when you calculate the 2D solver but your V

calculate the 2D solver but your V impedances will be

impedances will be wrong similarly if the if the material

wrong similarly if the if the material properties reported in the data sheet is

properties reported in the data sheet is XY then the V impedance will be right

XY then the V impedance will be right but dkz will be wrong I understand so

but dkz will be wrong I understand so I've developed a way if you know how the

I've developed a way if you know how the dialectric was

dialectric was measured initially and reported in the

measured initially and reported in the data sheet or the DK tables if you know

data sheet or the DK tables if you know how that was

how that was measured I've developed a way to

measured I've developed a way to calculate uh the the the the

calculate uh the the the the opposite so now you'll have

opposite so now you'll have both and you would use the right one

both and you would use the right one depending on what model you want here's

depending on what model you want here's these popular test methods used in the

these popular test methods used in the industry really you know uh that's what

industry really you know uh that's what laminate suppliers use to test and

laminate suppliers use to test and report so they have like the popular one

report so they have like the popular one is this clamp strip line so these three

is this clamp strip line so these three here they will be out of plane if if the

here they will be out of plane if if the data sheet says it's reference to this

data sheet says it's reference to this IPC number split cylinder or split post

IPC number split cylinder or split post you know that the numbers are

you know that the numbers are XY okay and you could use things

XY okay and you could use things appropriately and that's really what uh

appropriately and that's really what uh my sessions going to be about

my sessions going to be about uh it's at 12:15 at design Con on

uh it's at 12:15 at design Con on Wednesday okay if someone doesn't know

Wednesday okay if someone doesn't know what design con is is really cool

what design con is is really cool conference actually I've been there this

conference actually I've been there this year and a lot of uh lot of topics are

year and a lot of uh lot of topics are about very high speeed designs and there

about very high speeed designs and there is PD and especially there are many very

is PD and especially there are many very famous people and everyone can meet

famous people and everyone can meet there like all the I think I've done

there like all the I think I've done actually couple of videos with some of

actually couple of videos with some of the people who will be there who are

the people who will be there who are always there like Steve Eric battin yeah

always there like Steve Eric battin yeah uh you will be there then everybody yeah

uh you will be there then everybody yeah I think

I think uh uh was I forgot what's the name of

uh uh was I forgot what's the name of the S Sor uh Creator Yuri Yuri we'll be

the S Sor uh Creator Yuri Yuri we'll be there yeah yle from symor yeah syberian

there yeah yle from symor yeah syberian he'll be there everybody uh um Steve

he'll be there everybody uh um Steve Sandler yeah I mentioned St yeah he'll

Sandler yeah I mentioned St yeah he'll be there all all the ones uh you know

be there all all the ones uh you know that you've you've had on your show or

that you've you've had on your show or it's it's the premier conference for uh

it's it's the premier conference for uh pretty much sigal integrity and design

pretty much sigal integrity and design that kind of work and power Integrity

that kind of work and power Integrity too uh they have and also design con

too uh they have and also design con will have Automotive things they brought

will have Automotive things they brought in uh last couple of years related so

in uh last couple of years related so it's it's a good conference um many

it's it's a good conference um many people go year to year um so I will be

people go year to year um so I will be there too I will be there this will be

there too I will be there this will be first time since uh 2020 so I'll be back

first time since uh 2020 so I'll be back for you because I was there this year

for you because I was there this year yeah you went this year I didn't go uh

yeah you went this year I didn't go uh this year but I I I'm doing this paper

this year but I I I'm doing this paper and I'll be going everything's booked

and I'll be going everything's booked ready it's actually now it's two months

ready it's actually now it's two months away yeah two months today it'll be

away yeah two months today it'll be finished yeah so yeah I already booked

finished yeah so yeah I already booked uh same hotel where I was before very

uh same hotel where I was before very far away from the from the conference

far away from the from the conference because all the accommodation where the

because all the accommodation where the conference is it's like $200 per night

conference is it's like $200 per night that's not for

that's not for me but you'll have a booth there too

me but you'll have a booth there too right yeah yeah yeah yeah this year

right yeah yeah yeah yeah this year which is good yeah okay we can move okay

which is good yeah okay we can move okay yeah this is just the slide that went

yeah this is just the slide that went through and this was comparing the M

through and this was comparing the M this is comparing the implications of

this is comparing the implications of anisotropy so if there was 6% difference

anisotropy so if there was 6% difference in DC depending how it was measured the

in DC depending how it was measured the implication of your

implication of your via so if you measured with uh you know

via so if you measured with uh you know the blue

the blue one is

one is here and say oh I everything's balanced

here and say oh I everything's balanced nice and that was really using dkz but

nice and that was really using dkz but really it's dkx y we need and when you

really it's dkx y we need and when you built the board oh your impedance is not

built the board oh your impedance is not going to be 90 95 or 96 there it's about

going to be 90 95 or 96 there it's about uh what two or three ohms difference

uh what two or three ohms difference here right is it going to make big

here right is it going to make big difference it's what is it going to make

difference it's what is it going to make big difference well we'll see

big difference well we'll see okay you know I'll show some simulation

okay you know I'll show some simulation but uh let's get to that all right to

but uh let's get to that all right to build the Channel model so here here

build the Channel model so here here sort of is where we

sort of is where we are okay you can make it full screen

are okay you can make it full screen because it's very very

because it's very very small yeah so

small yeah so let's make the thing

let's make the thing so what I've done is I've already

so what I've done is I've already created a schematic but it's easy to

created a schematic but it's easy to create something from scratch so just

create something from scratch so just say a new just quickly just say a new

say a new just quickly just say a new cell create

cell create schematic end up something like here so

schematic end up something like here so you just call up like an S parameter

you just call up like an S parameter thing you know and it doesn't look like

thing you know and it doesn't look like that so now you know first of all then

that so now you know first of all then you double click it and I guess you find

you double click it and I guess you find the S parameter models what we created

the S parameter models what we created and the S parameter might be you know uh

and the S parameter might be you know uh let's just call it

let's just call it uh uh differential you see this is TG

uh uh differential you see this is TG 300 3,000 mil okay now it changes

300 3,000 mil okay now it changes to uh what we want plus display you can

to uh what we want plus display you can put on with the file name which is nice

put on with the file name which is nice because you can now see this is the file

because you can now see this is the file name of the S parameter so this is

name of the S parameter so this is basically the one which we generated

basically the one which we generated from Polar Polar exactly so I have as I

from Polar Polar exactly so I have as I said I have all different lengths of

said I have all different lengths of traces in one inch increment and I have

traces in one inch increment and I have a half inch one as well so you know this

a half inch one as well so you know this is how you start you you get one for

is how you start you you get one for your

your this is the Via for the transmission

this is the Via for the transmission line and uh you can copy and paste for

line and uh you can copy and paste for instance here and now you double click

instance here and now you double click it and pick the file

for differential via shorta okay right Escape so now you have these and the

Escape so now you have these and the nice thing with ads you can take

nice thing with ads you can take shortcuts you can kiss it and then pull

shortcuts you can kiss it and then pull it right and now you copy this and paste

it right and now you copy this and paste it this way but now but you you didn't

it this way but now but you you didn't change it to Via it's still 3,000 yeah

change it to Via it's still 3,000 yeah oh sorry I didn't change it to Via yet

oh sorry I didn't change it to Via yet it yeah okay skip I I clicked it but I

it yeah okay skip I I clicked it but I didn't hit

didn't hit okay see when you go fast that's what

okay see when you go fast that's what what happen it'll show up eventually

what happen it'll show up eventually this is differential via now you have to

this is differential via now you have to hit okay yeah okay now it says different

hit okay yeah okay now it says different via short right uh now you know you copy

via short right uh now you know you copy this paste it but now you can't just

this paste it but now you can't just connect here because one in so Escape so

connect here because one in so Escape so for instance one and three is the top of

for instance one and three is the top of the V two and four are the inner layer

the V two and four are the inner layer right now if we want to connect here we

right now if we want to connect here we have to swap it I guess we have to swap

have to swap it I guess we have to swap it and uh ads they have this mirror

it and uh ads they have this mirror about y AIS it swaps so I'll just kiss

about y AIS it swaps so I'll just kiss it here

and pull it out this way and here you can see two and four

can see two and four connects different it's two and four is

connects different it's two and four is the inner layers basically you know so

the inner layers basically you know so that's how you really build up your

that's how you really build up your channel now you have your via your

channel now you have your via your transmission line

transmission line via have another one for your connector

via have another one for your connector that you'd put in and then you'd have

that you'd put in and then you'd have another one of the of the actual PCB

another one of the of the actual PCB module so I'm just going to close this

module so I'm just going to close this right now and that's what you end up

right now and that's what you end up with when we do that right so where

with when we do that right so where where did you get the other

where did you get the other modules which other models oh other

modules which other models oh other models so for the module you will get

models so for the module you will get one from module

one from module manufacturer okay yeah okay I understand

manufacturer okay yeah okay I understand so what we've done already we've made a

so what we've done already we've made a model of the vs that we want we made

model of the vs that we want we made models of the transmission line here

models of the transmission line here what I did was because I have a a 1

what I did was because I have a a 1 inch version I made this model 11 in so

inch version I made this model 11 in so I can I have one inch increments plus

I can I have one inch increments plus you know so I can go in steps of one

you know so I can go in steps of one inch so if I wanted to have 11 inches

inch so if I wanted to have 11 inches instead of making 11 inches I just use

instead of making 11 inches I just use plus one right similarity here I have a

plus one right similarity here I have a half inch one because I want to go two

half inch one because I want to go two and a half the have okay now the next

and a half the have okay now the next next thing is the connector M you have

next thing is the connector M you have to go to the connector vendor like uh

to go to the connector vendor like uh Molex for instance or

Molex for instance or samtech uh or um Tao they will have

samtech uh or um Tao they will have connector models parameter models and

connector models parameter models and you have needed to get that and uh

you have needed to get that and uh that'll be the model for your

that'll be the model for your connector that you're going to use in

connector that you're going to use in your channel and then uh the

your channel and then uh the module this represents back I'll just

module this represents back I'll just throw this up quickly this just to go

throw this up quickly this just to go back to near the beginning where we had

back to near the beginning where we had our

our picture that represents this one and a

picture that represents this one and a half DB so I had created a a model that

half DB so I had created a a model that gave me one and a half DB L okay I

gave me one and a half DB L okay I understand just a transmission line

understand just a transmission line model that I've done right just to

model that I've done right just to represent the

represent the module okay and what what what is the

module okay and what what what is the next

next [Music]

[Music] yeah these oh this is a a nice feature

yeah these oh this is a a nice feature of

of ads it'll convert the differential to uh

ads it'll convert the differential to uh sorry convert single-ended to

sorry convert single-ended to differential easily so we put a Balon

differential easily so we put a Balon Transformer so that's basically like uh

Transformer so that's basically like uh termination or something it's 50 ohms to

termination or something it's 50 ohms to 100 ohm so it allows us to look at this

100 ohm so it allows us to look at this channel differentially m quickly right

channel differentially m quickly right otherwise you need to use equations to

otherwise you need to use equations to to to convert from okay okay because

to to convert from okay okay because because your source is uh single-ended

because your source is uh single-ended and your termination is single-ended so

and your termination is single-ended so you need this model to make it

you need this model to make it differential yeah so balance transform

differential yeah so balance transform is nice to convert from single-ended

is nice to convert from single-ended differential okay so okay that's the

differential okay so okay that's the nice feature about ads It's Quickly able

nice feature about ads It's Quickly able to put that in and now when I look at s

to put that in and now when I look at s parameters it'll be differential already

parameters it'll be differential already I don't have to do funny things okay

I don't have to do funny things okay okay and you need to have this

okay and you need to have this controller and here's where you specify

controller and here's where you specify the frequency from 10 controller is one

the frequency from 10 controller is one of the blocks there yeah so you pull if

of the blocks there yeah so you pull if you want to do s parameter simulation

you want to do s parameter simulation you go under you know your pallet here

you go under you know your pallet here and you pick simulation s

and you pick simulation s parameters and you pick a

parameters and you pick a controller okay

controller okay M okay basically that's what it is and

M okay basically that's what it is and then in the controller the start

then in the controller the start frequency you put 10 stop we want 50 and

frequency you put 10 stop we want 50 and step size of 10 megahertz okay I'm

step size of 10 megahertz okay I'm curious let's simulate something yeah so

curious let's simulate something yeah so right now I've got a transmission line

right now I've got a transmission line Channel or sorry our channel is with TG

Channel or sorry our channel is with TG 400g and 370

400g and 370 HR and I've used uh this is worked out

HR and I've used uh this is worked out to be 11 in and you'll see why uh after

to be 11 in and you'll see why uh after we Sim and this is 2 and a half inches

we Sim and this is 2 and a half inches so what this represents we can get uh 11

so what this represents we can get uh 11 in Long here but only 2 and2 in long

in Long here but only 2 and2 in long you'll see you'll see why after I hit

you'll see you'll see why after I hit Sim uh I want to shorten

right so I'll hit simulate and it'll it'll go through and it'll plot the

it'll go through and it'll plot the thing so you didn't see anything change

thing so you didn't see anything change because nothing changed so I'm going to

because nothing changed so I'm going to just turn that

just turn that off hit

simulate I've got a message thing it didn't like me doing

message thing it didn't like me doing that

that what I'll do is I'll turn one

off just so that you see that there's some

some difference okay okay see the thing was

difference okay okay see the thing was there so normally you'd have your two

there so normally you'd have your two channels there I'm going to turn it back

channels there I'm going to turn it back on

on again and uh now I'll run it now just

again and uh now I'll run it now just watch here you see you'll see you run

watch here you see you'll see you run the

the simulation now you see the blue and the

simulation now you see the blue and the red okay okay which one is blue which

red okay okay which one is blue which one is red well okay so

one is red well okay so just a bit more explanation is uh

just a bit more explanation is uh remember back here I showed uh these are

remember back here I showed uh these are the limit

the limit lines right this is what we have to meet

lines right this is what we have to meet the spec insertion loss and return loss

the spec insertion loss and return loss okay so this is where we are here you

okay so this is where we are here you use an equation ads you have an equation

use an equation ads you have an equation you can plot this these limit lines okay

you can plot this these limit lines okay and now uh your results will will show

and now uh your results will will show up on the same graph so uh the red one

up on the same graph so uh the red one is uh what I have for the T line that's

is uh what I have for the T line that's 11 in of the good

11 in of the good material so what it's saying is at 11

material so what it's saying is at 11 inches I'm you know I'm right on the

inches I'm you know I'm right on the meeting right but you know it says oh

meeting right but you know it says oh that's good I I should be okay with it

that's good I I should be okay with it right um on this hr3 you know 3 70

right um on this hr3 you know 3 70 HR with Just 2 and2 in uh I'm close to

HR with Just 2 and2 in uh I'm close to the Limit if if I change this

the Limit if if I change this to put this to one inch for instance no

to put this to one inch for instance no no no put there 10 I'm curious similar

no no put there 10 I'm curious similar length as the good material

oh all right so 370 HR I had 10 in I had 11 in right oh yeah yeah yeah so I'll

11 in right oh yeah yeah yeah so I'll put 10 in of 370

put 10 in of 370 HR H you don't have five there is only

HR H you don't have five there is only 5,000 is maximum 5,000 wait a minute I

5,000 is maximum 5,000 wait a minute I should have I thought I had a

should have I thought I had a 10 you can set the another one for

10 you can set the another one for another five yeah I can put that let's

another five yeah I can put that let's just put that to 5,000 right and the

just put that to 5,000 right and the other one you can change also the other

other one you can change also the other one to 5,000 exactly exactly Okay so

one to 5,000 exactly exactly Okay so cancel so what I'll do

is I had picked the Via I didn't Pi the right yeah

right yeah so so here this is 370 HR so a shortcut

so so here this is 370 HR so a shortcut because I have these names pretty much

because I have these names pretty much the same I can just make that five right

the same I can just make that five right and just add a zero here MH now that's

and just add a zero here MH now that's that's pick those models now and and

that's pick those models now and and change also the other one 5 five so we

change also the other one 5 five so we have

have same

okay I I'm very curious how big difference it is between the good and

difference it is between the good and bad

okay oh

wow right good point good that you ask you

right good point good that you ask you see there you go it's

see there you go it's significantly less so for for

significantly less so for for short 2 in it was okay it was fine yes

short 2 in it was okay it was fine yes but for 10 in it's completely failing

but for 10 in it's completely failing the standard4 material for 10 in you

the standard4 material for 10 in you have to use special really good material

have to use special really good material that says here you could not build this

that says here you could not build this box with 370

box with 370 HR at least to go to to the end units

HR at least to go to to the end units maybe maybe if things were short here

maybe maybe if things were short here there or if you can mix layers maybe

there or if you can mix layers maybe some inner layers those layers might be

some inner layers those layers might be all right with the other depends your

all right with the other depends your stackup it's a tradeoff that you do and

stackup it's a tradeoff that you do and that's what signant

that's what signant modeling at the front end helps you do

modeling at the front end helps you do makes you do those wh ifs to see what

makes you do those wh ifs to see what will work what won't work right this is

will work what won't work right this is why but you have to know the little

why but you have to know the little details to do it it's not just put a

details to do it it's not just put a bunch of numbers in that's why I tried

bunch of numbers in that's why I tried to emphasize the different parts that

to emphasize the different parts that affect the losses in in things just go

affect the losses in in things just go back just to clarify how do we know it's

back just to clarify how do we know it's failing because the blue line is very

failing because the blue line is very far below the black line which are the

far below the black line which are the limits have to be above the block

limits have to be above the block line to to be considered a pass MH so

line to to be considered a pass MH so here our losses are so much we got so

here our losses are so much we got so much loss that the signal not good

much loss that the signal not good signal will actually arrive to the

signal will actually arrive to the model right it's not going to work

model right it's not going to work so uh what I want to do is go back to

so uh what I want to do is go back to revert to the saved one just because mhm

revert to the saved one just because mhm and the return loss is basically same

and the return loss is basically same return loss is B on the kind of it

return loss is B on the kind of it doesn't matter really about material

doesn't matter really about material it's about

it's about the structure or it's just that our

the structure or it's just that our transmission lines they're they've been

transmission lines they're they've been designed for 100 ohms so and they're

designed for 100 ohms so and they're perfect 100 Doms okay but the

perfect 100 Doms okay but the vas What's Happening Here is the Via and

vas What's Happening Here is the Via and the connector they're 3D structures

the connector they're 3D structures they're not going to be 100 ohm exactly

they're not going to be 100 ohm exactly you know how the V was 95 93 ohm so a

you know how the V was 95 93 ohm so a lot of this stuff is caused by the Via

lot of this stuff is caused by the Via Reflections and that's why the impedance

Reflections and that's why the impedance of the Via is important because you know

of the Via is important because you know you could affect the return loss in a

you could affect the return loss in a sense so this is one aspect and you know

sense so this is one aspect and you know you can say oh well no we in this case

you can say oh well no we in this case oh we've met this line yeah we're okay

oh we've met this line yeah we're okay you know but unfortunately you know with

you know but unfortunately you know with 56 gig and now 112 gig just because

56 gig and now 112 gig just because you've met the mask not necessarily mean

you've met the mask not necessarily mean you met certain aspects of the

you met certain aspects of the spec and that's why they've developed

spec and that's why they've developed this channel operating margin uh way I E

this channel operating margin uh way I E and the oif optical interface

and the oif optical interface for they're starting to use uh what they

for they're starting to use uh what they call Channel operating margin and

call Channel operating margin and basically it's a it's like a mat lab

basically it's a it's like a mat lab type script that'll run and you have

type script that'll run and you have like an Excel file that it just reads

like an Excel file that it just reads parameters in uh and those parameters

parameters in uh and those parameters are based on the spec that you're

are based on the spec that you're working on and it goes in more detail

working on and it goes in more detail it'll work out the Jitter and everything

it'll work out the Jitter and everything else it's all part of it I don't

else it's all part of it I don't understand internals of the software I

understand internals of the software I just just use it as a tool and it'll

just just use it as a tool and it'll output past fail what you need so what

output past fail what you need so what is the in for the

is the in for the software pardon what is the input for

software pardon what is the input for the software well the input of software

the software well the input of software is uh the channel okay so you generate

is uh the channel okay so you generate Channel it means like as parameter of

Channel it means like as parameter of the whole Channel exactly what this

the whole Channel exactly what this whole channel right I'm just going to

whole channel right I'm just going to show this right now

show this right now basically uh I'm going to hide this I'm

basically uh I'm going to hide this I'm going to talk about now channel model

going to talk about now channel model using commn mhm okay so this is a little

using commn mhm okay so this is a little bit different I see beginning and end is

bit different I see beginning and end is little bit

little bit different the channel model it it's

different the channel model it it's different so let me get into that for a

different so let me get into that for a minute so this is what we've done that s

minute so this is what we've done that s parameters to check against the S

parameters to check against the S parameters it's frequency domain stuff

parameters it's frequency domain stuff okay we do that when we run that com or

okay we do that when we run that com or if we want to look at eye

if we want to look at eye diagrams you would do the same thing use

diagrams you would do the same thing use the use this channel simulator like we

the use this channel simulator like we picked here we're going to simulate in

picked here we're going to simulate in channel channel model Channel Sim and

channel channel model Channel Sim and now we have a a uh no connector whatever

now we have a a uh no connector whatever whatever one we want in this pallet and

whatever one we want in this pallet and you have a a driver and a receiver

you have a a driver and a receiver whatever model you want you put here so

whatever model you want you put here so I picked the differential TX and RX and

I picked the differential TX and RX and if you have the model of the

if you have the model of the seres you know IIs Ami model from the

seres you know IIs Ami model from the chap manufacturer from the chip you

chap manufacturer from the chip you could you do the same thing I don't have

could you do the same thing I don't have anything for that but if you did you

anything for that but if you did you would run it the same way except you

would run it the same way except you would you know use the Ibis Ami

would you know use the Ibis Ami option here's where you do the settings

option here's where you do the settings right here because right now it's for

right here because right now it's for Comm when you do bit by bit or

Comm when you do bit by bit or statistical it wants to have an Ibis Ami

statistical it wants to have an Ibis Ami model from the manufacturer but I'm just

model from the manufacturer but I'm just going to do Cal right now so it's going

going to do Cal right now so it's going to do Cal simulation so cancel that and

to do Cal simulation so cancel that and basically it's the same channel model

basically it's the same channel model here but you get rid of those Balon

here but you get rid of those Balon Transformers because you want to connect

Transformers because you want to connect in as you would a PN real real stuff

in as you would a PN real real stuff yeah so now this is your channel it's

yeah so now this is your channel it's the same channel As Above So this top

the same channel As Above So this top one is the uh 400g and this bottom one

one is the uh 400g and this bottom one is the 370 HR right but the but one

is the 370 HR right but the but one doesn't have beginning I know we not

doesn't have beginning I know we not doing both

doing both together okay I just turn everything off

together okay I just turn everything off and just restrict it to one thing only

and just restrict it to one thing only okay so the first one is

okay so the first one is what first one is the here okay okay

what first one is the here okay okay first one is the better

first one is the better material MH there this way it's it's

material MH there this way it's it's better let's let's just we don't need

better let's let's just we don't need this

this anymore for now I'll just blow this up

anymore for now I'll just blow this up right

right now this

now this simulation when I click

simulation when I click it I don't know today it just seems to

it I don't know today it just seems to take a long time and maybe when we're on

take a long time and maybe when we're on this video it may take even longer

this video it may take even longer because of the bandwidth of the

because of the bandwidth of the channel but generally you'll click it

channel but generally you'll click it and what I'll do is um I'll let it run

and what I'll do is um I'll let it run if it finishes then we're still talking

if it finishes then we're still talking no no no don't don't run it because

no no no don't don't run it because maybe yeah your computer will slow down

maybe yeah your computer will slow down you you already generated something yeah

you you already generated something yeah so how do you so you run it and then

so how do you so you run it and then what it will generate the file or

what it will generate the file or something yes so what it will do is

something yes so what it will do is um what I had done quickly because I was

um what I had done quickly because I was hopefully anticipating so what I'm going

hopefully anticipating so what I'm going to do is just just for a second I had

to do is just just for a second I had made a bit of a screen capture just

made a bit of a screen capture just before the call because I was testing

before the call because I was testing this out to make sure things would would

this out to make sure things would would work out yeah

work out yeah so I'm going to show when you

so I'm going to show when you click when you click the the Run button

click when you click the the Run button it takes a while and and it goes through

it takes a while and and it goes through the whole thing and as it's running

the whole thing and as it's running you're going to get different popup

you're going to get different popup windows that show up MH and these popup

windows that show up MH and these popup windows they're useful for different

windows they're useful for different things they're not as important

things they're not as important necessarily they just tell they tell you

necessarily they just tell they tell you about different things voltage bath the

about different things voltage bath the most important thing we're looking at is

most important thing we're looking at is these commm results in this green and

these commm results in this green and this red box when everything finishes it

this red box when everything finishes it spits out basically these two boxes and

spits out basically these two boxes and that's mainly what it's a pass

that's mainly what it's a pass fail now the reason for the two boxes is

fail now the reason for the two boxes is when it does this Channel model

when it does this Channel model simulation if I just go back to to this

simulation if I just go back to to this picture uh at the

picture uh at the beginning it only uh your channel model

beginning it only uh your channel model is only from via to this module at the

is only from via to this module at the end okay now the Comm also now uh has

end okay now the Comm also now uh has Provisions you can add a package model

Provisions you can add a package model if you have it of the package to do the

if you have it of the package to do the simulation if you don't have the package

simulation if you don't have the package they have a long what they call Long

they have a long what they call Long package and short package means got more

package and short package means got more loss in the longer versus the shorter

loss in the longer versus the shorter the Comm will run two simulations with a

the Comm will run two simulations with a short package called case one and with a

short package called case one and with a long package case

long package case two uh when this final red one has both

two uh when this final red one has both in it but you know it's the same numbers

in it but you know it's the same numbers it's just see bottom but the nice thing

it's just see bottom but the nice thing when it plots both is you you can see oh

when it plots both is you you can see oh the first case passed the second case

the first case passed the second case didn't okay it says pass

didn't okay it says pass pass well sorry ah okay fail eh fail

pass well sorry ah okay fail eh fail okay I see yeah and this is eye height

okay I see yeah and this is eye height IE height

IE height was and and becc is a vertical eye

was and and becc is a vertical eye closure and it had you know there's a

closure and it had you know there's a specification that it has to meet it

specification that it has to meet it pass but this effective return loss now

pass but this effective return loss now this is a relatively new thing in the

this is a relatively new thing in the com uh for chip to module you know it's

com uh for chip to module you know it's around 11 DB as a spec and here uh it

around 11 DB as a spec and here uh it passed effective return loss case one uh

passed effective return loss case one uh in case two uh I height

in case two uh I height failed the high height for uh the BS

failed the high height for uh the BS spec is 32 molts so here it failed and

spec is 32 molts so here it failed and uh and uh the other one passed mhm so

uh and uh the other one passed mhm so this comma feature is directly

this comma feature is directly integrated in the

integrated in the ads okay so I just made a screen capture

ads okay so I just made a screen capture because when it runs it spits out all of

because when it runs it spits out all of these windows you would see all of that

these windows you would see all of that um I would I would go now to my slides

um I would I would go now to my slides because I had summarized things a little

because I had summarized things a little better okay in the slides

better okay in the slides here so what's important here is to see

here so what's important here is to see uh to see this this red one is the tg400

uh to see this this red one is the tg400 it's the good one we got 11 Ines and we

it's the good one we got 11 Ines and we got two and a half inches here and you

got two and a half inches here and you can see that we've met the guidelines

can see that we've met the guidelines here pretty much okay here the return

here pretty much okay here the return loss was pretty good but it still can

loss was pretty good but it still can fail for the right longer package but

fail for the right longer package but you can see

you can see here when we run the Comm and we have

here when we run the Comm and we have the short package uh in in in it it

the short package uh in in in it it passes okay but when we have the long

passes okay but when we have the long package it

package it failed right even

failed right even though it seemed to match here because

though it seemed to match here because this line in the spec doesn't include

this line in the spec doesn't include packages see and if I come back here and

packages see and if I come back here and we look when it when the com runs it

we look when it when the com runs it plots the S parameters too but this

plots the S parameters too but this dotted line that's showing here now now

dotted line that's showing here now now this is the long package that's in there

this is the long package that's in there right it's the equivalent of the package

right it's the equivalent of the package it puts in when it runs C okay so all

it puts in when it runs C okay so all I'm saying that's why they have now have

I'm saying that's why they have now have this Comm metric because some channels

this Comm metric because some channels would pass the S parameter mask but

would pass the S parameter mask but still fail and other channels might fail

still fail and other channels might fail this mask but still be able to

this mask but still be able to work right so that's why we have Channel

work right so that's why we have Channel operating margin in these standards now

operating margin in these standards now for certainly 56 gig and 112 gig and 224

for certainly 56 gig and 112 gig and 224 coming up you know in the future

coming up you know in the future there'll be more of the C so ads has a

there'll be more of the C so ads has a built-in it it runs on mat lab but you

built-in it it runs on mat lab but you don't need to have mat lab it has a

don't need to have mat lab it has a runtime in it so it'll go and it has the

runtime in it so it'll go and it has the uh it has the spreadsheets as well part

uh it has the spreadsheets as well part of ads but if you don't have ads

of ads but if you don't have ads it'll if you have mat lab you can run

it'll if you have mat lab you can run mat lab uh Native mat lab as long as you

mat lab uh Native mat lab as long as you have the full

have the full Channel but before you do that you have

Channel but before you do that you have to make this into Just One S parameter

to make this into Just One S parameter of of the whole Channel M I understand

of of the whole Channel M I understand you can do thats what is interesting

you can do thats what is interesting even the standard fr4 material pass when

even the standard fr4 material pass when it's

it's short when it's short exactly but you

short when it's short exactly but you only have like two and a half Ines yeah

only have like two and a half Ines yeah but still I think that's really good to

but still I think that's really good to know because oh yeah because some people

know because oh yeah because some people they they they really would like to buy

they they they really would like to buy or use the very expensive materials and

or use the very expensive materials and it it may not be necessary right or mix

it it may not be necessary right or mix and match if the material is compatible

and match if the material is compatible some layers may be able to be fr4 type

some layers may be able to be fr4 type right yeah we need to do simulation

right yeah we need to do simulation first but or lower grade material maybe

first but or lower grade material maybe not the highest uh TG maybe it's uh you

not the highest uh TG maybe it's uh you know FR 408 which is like a little

know FR 408 which is like a little better right you might be able to get or

better right you might be able to get or Panasonic Meg four for instance you

Panasonic Meg four for instance you might be get away instead of Meg six Meg

might be get away instead of Meg six Meg seven you see what I'm

seven you see what I'm saying you that's where it's important

saying you that's where it's important where you compare your materials before

where you compare your materials before and try and choose do this what if

and try and choose do this what if analysis first you know there's a lot of

analysis first you know there's a lot of things you can do on your own and you

things you can do on your own and you need to do this when you're architecting

need to do this when you're architecting your system especially even pcie cards a

your system especially even pcie cards a lot of them are pcie but they still have

lot of them are pcie but they still have a front module right but they're very

a front module right but they're very small so you might have a fpga talking

small so you might have a fpga talking to a module but because a pcie card is

to a module but because a pcie card is so small it might only be two inches

so small it might only be two inches long right yeah then you may not need uh

long right yeah then you may not need uh this big stuff yeah so you need to do

this big stuff yeah so you need to do this work to make that decision and

this work to make that decision and that's what signal

that's what signal Integrity Channel modeling is about is

Integrity Channel modeling is about is to make the decision costeffective

to make the decision costeffective decision this this was so cool but I I

decision this this was so cool but I I really like this a lot I for a very long

really like this a lot I for a very long time I wanted to make video like this so

time I wanted to make video like this so thank you so much for preparing all

thank you so much for preparing all these materials and

these materials and explaining yeah you're welcome it's uh

explaining yeah you're welcome it's uh it's very high level View and we went

it's very high level View and we went through quickly and everything but uh

through quickly and everything but uh yeah

yeah it's um each one of these you know you

it's um each one of these you know you could spend a lot of time and more

could spend a lot of time and more detail into it um but you know when we

detail into it um but you know when we do when we do these pizza box designs

do when we do these pizza box designs this is basically the

this is basically the process that I go through basically when

process that I go through basically when someone is designing large pcbs larger

someone is designing large pcbs larger systems like Dimension large dimensions

systems like Dimension large dimensions they definitely will need to simulate

they definitely will need to simulate when they run about 5

when they run about 5 gahz yeah yeah yeah the the higher

gahz yeah yeah yeah the the higher frequency you go the less luck you

frequency you go the less luck you have the more you need to simulate you

have the more you need to simulate you know okay so thank you so much I really

know okay so thank you so much I really hope uh everyone finds this uh super

hope uh everyone finds this uh super useful and when they are designing I

useful and when they are designing I don't know PCI Express generation

don't know PCI Express generation 20 then they will know how to simulate

20 then they will know how to simulate their boards even as you say PCI

their boards even as you say PCI Generation 3 PCI Express Generation 3

Generation 3 PCI Express Generation 3 still may need this kind of simulation

still may need this kind of simulation if tracks are very long that's right but

if tracks are very long that's right but but PCI you know as I say you may have

but PCI you know as I say you may have pcie on there gen 3 or Gen 4 but even

pcie on there gen 3 or Gen 4 but even gen five and six they're it's up in the

gen five and six they're it's up in the same range now right now you're going to

same range now right now you're going to have to think about this that now the

have to think about this that now the spec will be different it's not

spec will be different it's not necessarily this but you know the the

necessarily this but you know the the thing is same you build the channel the

thing is same you build the channel the same how you simulate after to meet the

same how you simulate after to meet the spec is unique to the to the standard

spec is unique to the to the standard yeah yeah and I learned a lot about

yeah yeah and I learned a lot about these PCB parameters and in polar what

these PCB parameters and in polar what everything you could say and I I had no

everything you could say and I I had no idea you can do uh you can generate the

idea you can do uh you can generate the S parameters from from Polar for example

S parameters from from Polar for example and so yeah okay so thank you so much B

and so yeah okay so thank you so much B again and uh

again and uh let's let's meet you in person in few

let's let's meet you in person in few months yeah in two months we'll uh we'll

months yeah in two months we'll uh we'll meet in person for sure thank you for

meet in person for sure thank you for having me here it's uh I really enjoyed

having me here it's uh I really enjoyed it I hope uh things work out and uh be

it I hope uh things work out and uh be good good to see you I I like this a lot

good good to see you I I like this a lot so thank you okay you're

so thank you okay you're welcome and uh everything thank you very

welcome and uh everything thank you very much for watching this video by the way

much for watching this video by the way we are preparing some very testing

we are preparing some very testing tutorials so if you don't want to miss

tutorials so if you don't want to miss them hit the Subscribe button if you

them hit the Subscribe button if you want you can also check out our fedel

want you can also check out our fedel online courses where you will find

online courses where you will find everything important from basic board

everything important from basic board design up to Advanced Hardware design

design up to Advanced Hardware design and PCB layout the link is in the

and PCB layout the link is in the description that's all for this video

description that's all for this video thank you again don't forget to leave

thank you again don't forget to leave your comments and see you next time bye

Click on any text or timestamp to jump to that moment in the video

Most transcripts ready in under 5 seconds

One-Click Copy125+ LanguagesSearch ContentJump to Timestamps

Paste YouTube URL

Enter any YouTube video link to get the full transcript

Most transcripts ready in under 5 seconds

Get Our Chrome Extension

Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.

Add to Chrome — Free

Works with YouTube, Coursera, Udemy and more educational platforms

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube

←

→

↻

https://www.youtube.com/watch?v=UF8uR6Z6KLc

YoutubeToText

←

→

↻

https://youtubetotext.net/watch?v=UF8uR6Z6KLc

YouTube Transcript:How to simulate PCIE / IEEE path on PCB + Everything you need to know | Explained by Bert Simonovich

Video Transcript

Paste YouTube URL

Transcript Extraction Form

Get Our Chrome Extension

Get Instant Transcripts: Just Edit the Domain in Your Address Bar!

YouTube Transcript:
How to simulate PCIE / IEEE path on PCB + Everything you need to know | Explained by Bert Simonovich