Model Copper Roughness Properly, a Discussion with Bert Simonovich
Bert Simonovich is founder of Lamsim Enterprises and an expert in innovative signal integrity and backplane solutions. He holds two patents and author of several publications. Anyone involved with the design and fabrication of high-speed circuit cards and backplanes or people who needs to estimate PCB transmission lines losses will find value in the paper Bert shared at DesignCon 2018, “A Causal Conductor Roughness Model and Its Effect on Transmission Line Characteristics”.
Listen in to Join Judy Warner and Bert discuss copper roughness and the paper he presented at DesignCon 2018.
Listen to the Podcast:
- From a manufacturing perspective, the rougher the copper the better
- But for high speed, the rougher the copper the higher the loss
- If you don’t model it correctly you cannot simulate it accurately
- The Huray model dilemma - where do you get the parameters to use for the model?
- Last year’s paper was about correcting the dielectric constant due to roughness
- New model is dubbed the Cannonball-Huray model
- The effort is collaborative, building on each others’ research i.e. Rogers, Polar
- As speed goes up, material properties are more important i.e. Fiber weave effect
- Accurate modeling is very important especially with new standards and tighter margins
Links and Resources:
Trade In Your Outdated PCB Design Tool & Unlock 45% OFF Altium today!
Hi everyone this is Judy with the OnTrack podcast welcome back. If you are new to our podcast make sure to subscribe on your favorite RSS feed you can follow me on LinkedIn please.
Please. I love having friends on LinkedIn or at Twitter I'm @AltiumJudy and as far as Altium goes you can follow us on LinkedIn Twitter or Facebook.
So today I am at Design Con 2018 at the Santa Clara Convention Center and I'm delighted to have my dear friend Bert Simonovich who hails from Canada and so we're gonna chat with Bert today about the paper he submitted and all the things he's involved in. I also want to share with you about our fun history in Hobart and how we got to know each other. So Bert, welcome to our podcast and we only get to see each other maybe once or twice a year but I'm always so happy to see you.
Likewise, likewise Judy.
I'm glad you showed up this year and I'm glad I could get you. Last year I couldn't get you because you were so busy with your papers and everything; you were desperate, you were booked to the gunnels last year so I'm glad we got to run into each other.
So, to give you a little bit of background about Bert... well let me start from when Bert and I met. So what was it 2006 six or 2010?
Yeah 2010, something like that.
Yeah, so Bert was sort of in a transition period in his career and we met on LinkedIn and then we actually met at the show at Design Con and we connected through social media and we're old enough you know, and respectable enough in our careers that we didn't feel like that was creepy or anything. So we connected and we met upstairs; the top of the escalators here at Design Con, and we got acquainted and just networked and boy oh boy we didn't know the number of ways that our past would continue to cross over the years so, we hit it off immediately and Burt knew an awful lot about backplane and tell me again, you worked for a huge company up in Canada?
Okay so yeah we live in Ottawa area and I started my career at Bell Northern Research which was the research and development arm of Northern Telecom or eventually North Dell in Bell Canada.
There you go,
so it's essentially the same as like the AT&T; down here - researching, development and then that later got to fold it into the Nortel umbrella and I was there my entire career. It was 32 years until 2009 when they filed for bankruptcy and then my career ended…
And that was kind of a shock. Then I was fortunate that I was able to you know - we weren't in bad shape financially - so it didn't hit us like it hit others. So I've considered myself fortunate, but what I missed was the day-to-day and and not knowing what the future was because at that time there was the downturn economy so it's hard for everybody and certainly harder for older ones and we were able to get out and fortunately got our pensions early and that kind of kept us going.
Yeah so that's right about the time I met you, not long after that right, and so you're like: mmm what am I gonna do now? I have all this knowledge and expertise how am I gonna apply it and so on your own dime you flew down here. We met and I was hoping that we were just networking to see how we could help one another professionally and whatnot, since that time - boy oh boy.
IMS came up from the UK to Montreal and we met again there.
That's right we met again there. We met at IMS so then, all of a sudden, our lives started weaving because I was in the RF and microwave space from the bare board side, and then I was doing the guest blogging for Microwave Journal and then you got to know all the folks - or you knew those folks in that space and then you got to know the Microwave Journal folks who have since launched the Signal Integrity Journal and now you're on that editorial review or review board for that magazine?
And now, I think you just told me, this is your first year now that you're in the Technical Review Committee?
Technical Program Committee Yeah.
Right for Design Con so you actually reviewed the papers then that were being subpoenaed by some papers?
Did they divvy them up?
Yes they basically divvy it up so there's quite a few reviewers for it to get the diversity and not just one person, several people review the same paper as well.
Right good. So, not only is he reviewing papers he's also submitting papers. So I want to say congratulations that you are a finalist for best paper.
Oh thanks Judy.
Well I don't know that’s third time in a row now?
Five! You've been nominated that many times or you've submitted, but you've been nominated that many times. I see, so you are like a rockstar! So, tell us about the paper that you submitted this year.
Okay. This paper is actually a follow-up from last year's paper that I presented and last year after I presented it, there was a great discussion afterwards. It was the end of the day and people - there were no more papers after - and people that were interested stayed and we had a good little wrap of things and were discussing some of my results. And although I was close there was some doubt as to actually what it was, why there was still differences and my co-author Vladimir he had an idea that it's possibly due to causality of the metal.
And Vladimir works for a mentor and so he had this idea and I had data and my technique and we decided to collaborate and see how well it would work. So he incorporated his algorithms into the tool and we checked it out and that brought the data exactly right on.
Wow! So this paper is kind of a wrap or closing the loop at the end of last year's paper basically?
That's correct, it's kind of a follow-up in a way and it kind of answers the question that we were suspecting. - the people who were discussing as a group were suspecting - so that was very satisfying when I had that. So we said, well we've got to publish this now so we'll do that.
So it was a copper roughness issue?
Well it's part of modeling called copper roughness, to model copper roughness properly and get the right model afterwards so you can do the simulations.
Isn’t that hard? I mean it seems like copper roughness, which is the roughness - for anyone that's not clear on this of our listeners - copper roughness is between the laminate and the copper. Not on the topside of them it's on the bottom side where it's being bonded to the laminate but it's not always the same?
No the thing is with roughness from a PCB manufacturing perspective, the rougher the better.
Right because there's adhesion.
Right peel strength. Nowadays, for high speed past 10 gigabit, now everything starts to matter and now people are talking 25 gigabit - 56 gigabit all the stuff lost, now becomes a factor. So it turns out the rougher the copper the higher the loss, so how do you model that? If you don't model that correctly you can't simulate it in advance.
So the bottling techniques in accuracy - so there's been different models out there one that's been very popular is the Hury model that was presented back in 2010 because I remember going to that presentation. Basically it was dubbed the ‘snowball model’ because when you look at the roughness profile under an electron scanning microscope, it looks like a stack of snowballs type of thing.
Uh-huh and that's where it got coined the snowball model but it's the Hury model and it's very accurately shown that you get very good results measured to simulation. The problem was where do you get the parameters to feed that model? Back then to get it, if you tuned the parameters you could get an excellent fit to the loss but that doesn't help for tools so how do I get that? Unless you're building a board or measuring the board to get the parameters so you can use those again if you're going to build that same geometry in a new design. So that's very time-consuming - pretty accurate - but not practical for smaller companies. If you need a good answer now, rather than a good answer later as Eric Bogatin always says, that never helps you. So, one of my early contract positions, because I started going in consultancy after Nortel. One of my clients they wanted to do some analysis on dielectrics but they also wanted to include their roughness and I knew nothing about roughness at that time. So I started to do my own research into this so that I could try and do the work, and found that it was all over the place in a way.
Anyway, I completed the analysis not even based on what I knew at the time and it was okay, but that seed sort of stuck. So I started in it and once you get an interest in something then try to go further and further. So it's been a pet project of mine so I decided to think about the Hury model and say: well, is there a way to get these parameters based on datasheet value only? If I know the roughness from a datasheet the, height of the roughness, can I get the radius though? So I thought about it and you know, I thought about: okay, this is the stack of spheres. If I stack them up can I determine the radius of the sphere from that stack? So you know it's basic geometry of things and eventually you can get that number if you know this height. So I thought - just out of curiosity - I'm gonna try it and when I did, I got reasonably good results, almost right off the bat. I'm not perfect, but I can tweak things a little bit more and I'm saying okay… That's how it kind of began so it's been over a few years, off and on. I did a paper earlier in 2015, when I first presented some of it, and that was quite successful and every year it's been kind of incremental and testing it against other data and then eventually other software came out like simBR and I started having success with the model there and that was great and so it kind of went. So lately, after that, then I started thinking about I never get the the actual delay property. So in order to model things, you need proper loss and then also the the delay property or the phase. So then I thought well, perhaps, how does the roughness come into it? Because I noticed the rough with the copper I get different phases for the same material. Anyway so I worked out some some things and that's what last year's paper was about, to correct the the dielectric constant due to roughness...
It was really last year's paper.
Uh-huh and so that that worked out very well and when I correct that, I get very good results with it so that's it in a nutshell how I got from there to here.
Wow. So I know you've worked also with Polar instruments. Tell us about how that came about?
Yes so last year Polar had introduced the Hurray model into their tool and I hadn't really met Martin but in his newsletters I always got it and he announced that they were going to come out with the remodel and a Polar tool so I emailed Martin and said; look let me check it out to see how well it works. So they gave me a trial thing just to check it out and I tested it out against my data and there was just a couple of hiccups, but we sorted them out and now the tool - it's a great tool - we can get the parameters now basically from data sheet to fill in. So we've dubbed the model really like a ‘cannonball’ Hury model because it's a combination of both.
Well Martin spoke, I think I shared this with you earlier that in October Martin Godean from Polar presented a talk at Altium Life in Munich and I was chairing the room he was in and he pulls up this stack of cannonballs - you and I hadn't talked about this - then he starts citing all this work by Bert Simonovich. I nearly fell out of chair, like my friend Bert? So I really love that Bert, you've like been such a collaborative force in this industry about these models and you know that you reached out to talk to Polar and really made it better. Not so much in a self-serving way but it's good for the industry really, it's good for designers right?
I think so.
And so I really appreciate that generosity of spirit that you have and it's what I liked about you when I first met you and here we are all these years later who knew.
So, what's next Bert? Tell us first of all, where can our listeners find your papers because I'm sure they'll want to get their hands on those?
Yes so I published papers that I have on my website at my company is http://lamsimenterprises.com.
Well I'm sensing we'll try to put that in text somewhere I relate to this podcast so folks can find that and then they can find your consulting services there as well but then find all your papers there. Did Design Con aggregate the papers that come out of the show?
Oh yeah they have the proceedings.
But only the attendees get their proceedings right? So they can come to your website is the bottom line, right?
Right, so anyways so for my papers, I have my papers that I published on my website and it's not just Design Con I’ve also done EDI Con papers and that's part of the signal integrity journal. There's an offshoot of that - is EDI Con - and I attended that the last two years, they were the first two. So there's Boston that was great time. Actually you went to the first one?
Yeah I went to the first one I can't believe how our lives have just crossed so many times. We just met each other here and went hi.
And you met and you met my wife.
Yeah I met your wife on that trip.
Bert and I got to go to Fenway Park and have a private party at night and overlooking the fields and they brought out... what are those? I'm such a sports nut.
The trophies from the World Series.
- thank you the World Series trophies. They brought those out so we could you know take photo ops with and we had great food and we got to take over Fenway for a few hours and I met Bert's wife who's just a doll and we had a great time. So who knew when we met how our paths would cross?
But I think Design Con in general is about is networking, meeting new people, learning new things spreading the seed of papers. Like I was mentioning even last year's paper - Rogers took that and then confirmed the data or the theory part of it and then they presented results this year.
It’s such a great collaborative effort and I don't know I find it very refreshing because you know, like well I have a secret and if you give me a lot of money it might tell it to you. This is much more open handed and collaborative. I liked what you said earlier today - which was somebody will present something and then an offshoot will take off and then it'll take off and then someone will build on that and bring it back as something else. It's really the industry together - collaborating.
That's also for mine like 2010 with the snowball model, who knew? That room was packed to the rafters and out - that presentation from Paul Hury and Eric Bogaton’s and he presented that year and I was just dumbfounded and I'm going, well I don't understand it but that's cool, and that who knew that that would be the seed for my work, my research work.
And and then went into Polar and Rogers and it's just like an idea that sprung legs.
And that's just my example. One example of many you just look around - like Eric likes to say - come drink from the fire hose, that's the Design Con there’s just so much stuff and it's there and it's just a great place for that kind of thing.
It is a great show. So, a couple closing items. Number one what would you say for circuit board designers, are some of the kind of greatest challenges they're facing today? Looking forward into emerging technology is there any specific - one or two things maybe - that are sort of plaguing them at this point? Or things maybe they need to onboard and that they are not taking right now?
Just in general as the speeds are going up the actual material properties are really becoming a factor. A big thing is fiber - we've talked a lot about that - fiber weave effect and the industry is looking how to mitigate that effect. A lot of nice papers over the last few years, talking about that and again there's a lot of things on the roughness, getting good modeling because it's important now. And now with everything like the new Pam-4 standards, 56 gigabit. The margins are tighter, so fabricating boards is more demanding.
Yeah it ain't what it used to be I can tell you. I started in the board fabrication industry in 1984. Oh my gosh that seems like such easy stuff compared to today. I was just speaking with John Belushi from ASC and I said it used to be you just got your board, your fab drawing you had your sort of mechanical, here's the size of the footprint, this is the size of the trace plus or minus this, and off to the races you go. Now it's like a board fabricator has to think about everything they're doing and how that is going to impact. You can go to a board shop, make a board and be well within IPC standards and its’ still not performing to the designers intent.
I think it's not just the board shop, I think sort of the industry as well, you have to specify what you want properly in your fabrication notes.
What fabrication notes. What I'm saying is documentation, I think is an issue.
Yes and and the thing I'm seeing is if you want consistent results you have to nail down your design. Basically nail it down and not let the fabricator substitute material for different weaves. For instance if you have a stack up that you first designed it's got to be included verbatim on the fabrication notes. So you must use this material, this weave.You can’t substitute, and you can't play around with all the line widths and space to do things different. By specifying things on the notes that restricts what they can do you should end up getting what you want otherwise you may not.
And you had told me earlier in your consultancy business that you will talk about modeling certainly, but then you also give them the added benefits of giving them some guidelines of how to spec in these.
Yeah so typically a lot of my business is design reviewing boards before fabrication. After they do the design like Altium Board would lay it out, so now there's a design review at the end so some of the clients they'd like to get some review over that visual review and I do that sort of thing for them. But as part of that, I review the stack up even from the board shop because the stack up people are human and could make an error. Typo or a lot of times they're just templates and they don't cut and paste properly or something, and it may not be correct. I haven't found anything that wasn't correct but I do that anyways, just check it, double check that the impedance is correct in things. But on top of that I also say that I'm going to review their fabrication notes to ensure there's consistency there and usually I find that a lot of these are smaller companies or startups. They're not aware of what's needed or they have one set of fab notes and they have the same design all over. They just use it over and over but that could add confusion in the shop.
At the end of the day part of the design review it's also part of a bit of educating as well to some clients so they realize some of the importance and they appreciate that as well.
Yeah and it is so sorely needed and people don't, like you said especially smaller companies, they don't always have the resources to go out and spend endless hours in a board shop getting educated or whatever the issue is. Or educated about signal integrity or whatever it is, so I think what you're offering in your business and in that niche is really valuable so that's great.
Well my friend I'm gonna let you get back out on the show floor. I'm sure you have a lot to do today I'm sure you're a busy guy so yeah I so appreciate you spending a little time with me and I'm so glad we got to connect I know you're getting so busy - such a big shot these days - it's been a delight thank you.
Actually, the reason I knew you were here, I saw something on LinkedIn and saw you were responding to somebody and I go: wait a minute is Judy here? So then I LinkedIn messaged you last night and I said, are you here? and yeah so let’s connect today.
See there's a lesson to everybody, LinkedIn really is useful. Gosh you and I've met a lot of friends through LinkedIn. I'm so glad to see my friends. I don't get to see you enough so I'm glad to see you whenever I get to see you, I'm not gonna complain. So best to you, congratulations on your nomination again.
Thank you Judy and the same to you.
And have a great show and I look forward to catching up with you next time.
This has been Judy Warner with the OnTrack podcast please remember to subscribe on your favorite RSS feed and we look forward to talking with you next time and always remember to stay OnTrack.