Podcast: Play in new window | Download
Subscribe: Apple Podcasts | RSS
Welcome, Eric Bogatin!
- Eric is the Dean of Signal Integrity (SI) Academy, now part of Teledyne Lecroy
- He has also been teaching engineering for many years, including at Colorado University.
- Learning SI “on the streets” can’t replace central (essential) principles like understanding and applying Maxwell’s equations.
- The website that started it all is called, “Be The Signal“, thought up by Eric’s wife.
- Three tools in the toolbox
- Rules of thumb
- Approximations (formulas)
- Numerical simulation tools
- Eric posted over 25 “Rules of thumb” articles on EDN.
- One example (and the first posted) was how bandwidth is (.35/rise time).
- Dave talked about similar principles in his video about oscilloscope bandwidth:
[tube]https://www.youtube.com/watch?v=F-ZDiGmLvTs[/tube] - “Sometimes an OK answer now is better than a good answer late”
- One reason Teledyne bought LeCroy is because of the matching of the need (and solution to) Indium Phosphide technology. This was in the 100 GHz real time scope that Shahriar reviewed/tore down:
[tube]https://www.youtube.com/watch?v=U3w_EWgGQuk[/tube] - A good general 2D field solver is Polar Instruments, though Eric was careful to point out that there is a wide variety of capabilities and price points for these pieces of software.
- The continued shrinking of circuit boards means that gridded (crosshatch) planes are in the market, especially for flex/consumer products. This helps with Signal Integrity.
- Instantaneous impedance is what the signal “sees” at each step. Characteristic impedance is expanding the instantaneous impedance to the entire line.
- Grace Hopper was one of the first programmers and was an admiral in the US Navy. She was a big fan of the ENIAC and tried to integrate them into programs. She carried a piece of string to help illustrate the distance a signal travels in air in a short amount of time.
- Signals travel roughly 15 cm per 1 nanosecond in FR4.
- Another rule of thumb involves ratios: get a 50 ohm line with a 10 mil width to 5 mil to dielectric ratio. It’s the aspect ratio that’s important.
- Layer stackup is often left to fab, which can be dangerous. Doubly so if you’re depending on stack up because your power and ground planes are in the inner layers of your board.
- Yet another rule of thumb is to keep spacing twice the line width to reduce crosstalk.
- Making board bigger to guarantee it works is “buying insurance”
- 10MHz and above range is where you need the intuition. A handful of design rules will get you to 100 MHz. Beyond that, you need to pay attention and begin considering simulation.
- Eric will soon be writing Nuts and Volts articles about Arduino.
- SI academy was took the in-person classes and put them (mostly) online. They are now offered to individuals and corporations on subscription basis.
- The CU classes are also available on the site, as are some other example classes.
- It is interesting seeing where MOOCs are going, especially since very few are charging for courses.
- Chris doesn’t like how testing / exams are handled right now. The SI academy doesn’t have testing.
- Classes on the resume get you through the door, even if they are MOOCs. The interest level is what interviewers are looking for.
- Eric has written a large variety of books
- Signal Integrity Characterization Techniques (free online)
- Signal and Power Integrity Simplified (Prentice Hall)
- Out of print books:
Dave, you probably meant that the first programmer was Ada Lovelace, instead of LadyAda. I mean, she was a lady too, but these days, the term is ambiguous, due to Limor Fried using the nick 🙂 Anyway, really nice episode!
This was an incredibly enlightening episode. I cannot thank Mr. Bogatin enough for sharing his wisdom. And thank you to both Dave and Chris for another excellent podcast and interview.
Mr. Bogatin, with the utmost respect, your voice reminds me of Dogbert (AKA Chris Elliot) from the Dilbert cartoon. It’s more likely the evil genius intonation than the timbre. It may just be my bad hearing, but it is certainly meant as a compliment 🙂 Thank you so incredibly much for your time. Be the electron.
Great episode. On the topic of MOOCs, I wonder if a model could emerge where you pay to sit an exam from any course at a local testing centre/university. The course provider prepares an exam, forwards it to the testing centre, they then certify the identity of the person that took the exam, and send it back for marking. There are problems with that, but you could add checks and balances. E.g. If a testing centre has significantly higher results that other centres, something fishy might be going on. Interesting problem to think about.
Probably one of the better ways to solve it. I have also heard about a webcam program that has someone watching you (well, multiple people at once). Like proctoring an exam in a classroom, just digitally. Not sure how they do the verification of each person though.
My point/criticism was a bit broader than that and was kind of a critique of all education systems: Why are we testing so much? If you’re teaching someone code, have them write code. If you’re teaching someone signal theory, put them in a lab. Some problem sets and math are important, I get that. But the testing of knowledge synthesis, not just retention needs to be the focus (especially with Google in everyone’s pocket). I think projects are the best way to do this.
Totally agree with that. Students spend all semester creating and learning with text books and the internet, then you’re put into an artificial environment without access to information that’s nothing like the real world just for an exam. I can see use of tutorials throughout the semester to make sure they can understand the material and aren’t falling behind, but if you’re training someone to design electronics, the metric used to assess them should be, can they design electronics.
An interesting aside is that exams are pretty much the only reason calculators are still manufactured. I have a free app that exactly replicates my calculator from uni, but because they can’t use phones in exams, students still have to buy a calculator. Exams are broken.
it be interesting to know how inductive loop battery charging techniques have on circuit board designs.
as the whole project board maybe in a powerful magnetic field when charging the battery. ground planes.esc
Is there any possibility to getan either Paul Horowitz or Winfield Hill on the show and talk about the art of electronics? It would be a very interesting episode.
Great interview! Of all the guests, Eric Bogatin and Shahriar (Signal Path) have been might favorite!
Unfortunately most of the talk is over my head, but I’m hoping to someday watch Eric’s video lectures (at the moment $600 is a bit pricey for me ). Henry Ott and the late Bill Kimmel / Daryl D. Gerke courses are also not cheap (~$1500), so maybe it’s just an expensive field to learn. I’ve seen some other resources but not sure:
– Udemy has an inexpensive VLSI Signal Integrity course, but it seems to be more for practical chip design, and not so much practical board design.
– Douglas Brooks Safari Online PCB Signal Integrity LiveLessons looks good but no pricing info
– Howard Johnson will offering his lectures on his website, but no info when
BTW, did I miss where else you would put the power and ground planes if not in the middle? He seemed to suggest it’s a bad idea, but what’s the better approach?
Awesome guest. I’m glad to see there’s room for physicists in electronics.
How is it that we have two engineers and a “physicist” talking about signal integrity, but the Audio quality is shit? Pops, clicks, distortion and Eric sounds like he is talking on drive-through menu speaker. WTF?
First time listening perhaps? This really wasn’t bad. We try, man. We try.
Hi Chris, I have been following Dave Jones on the EEVblog for about a year now, but I just recently started listening to The Amp Hour after Dave mentioned it in one of his videos a few weeks ago. I have listened to five episodes now and one thing that bothers me is the inconsistency in both audio quality and playback quality…Let me explain, I typically download the entire episode and then listen to it off-line usually at my office in the lab where I have studio monitors. (I take my audio quality seriously) Anyway I have noticed that the encoding you use for the download version varies from episode to episode, some are 160Kbps, some are 90Kbps, none have been stereo yet. I understand that this is free content, but there is really no excuse for the poor quality in this day and age with 24bit 96k audio recorders that cost less than 200 bucks. I would be happy to send you one if you want to step your game up in production quality. As a side note, separating or panning the mics of you and your guests not only helps with speech intelligibility, it creates spatial awareness and provides for a more natural sounding conversation. Most of the other pod-casts I listen too are typically distributed in 250~320Kbps at 44.1 or 48khz mp3 format. If your concerned about file size you could use 192Kbps with “joint stereo” encoding at 44.1khz and you are still under 100MB for an hour long episode.
This is a great unit and records directly to mp3 format….
https://proaudio.com/catalog/tascam-dr-40-handheld-4-track-digital-recorder.asp
BTW, I really enjoy your show, I just want to help you make it better.
-Ryan
Agree, that Tascam is a great unit. Used often for podcasts and has wonderful quality.
As of this point, we have been in the same room with our guests now…twice. Dave and I have still never been in the same room.
Instead, we send all of our microphones to guests. This is the one we have been sending lately because it’s affordable and comes with a “tripod”. We do not as for the microphone back from guests because of logistical hassles: http://www.amazon.com/gp/product/B001AIQGUO?psc=1&redirect=true&ref_=oh_aui_detailpage_o03_s00
If you would like to help us out, the best way to do so would be to evaluate lower cost microphones (sub $60) and then do a “best practices” video for guests setting them up. The reality is that we have very little control over the environment that our guests use, including mic placement and volume setting. Here’s a video on how we record the show: https://www.youtube.com/watch?v=BP3Z_N8IkQQ
Yes, this explains a lot. My background is audio, film and video production. So my work-flow would have been entirely different. You could still use mumble for your party-line however I would send portable recorders to each guest/host with time-code capability for re-sync in post production. You are already sending them a mic, why not send them a higher quality field unit with TC and just have them send it back after the session? This would allow consistency from episode to episode, plus the field recorder settings can be locked (high-pass filter, AGC, quality, etc…) so the guest only has to start / stop the recording. I know it sounds like more work but it will be worth it in the long run. You and Dave could just keep using mumble as it seems like you have decent mics, but editing in stereo since you already have the iso tracks from mumble for you and Dave is trivial.
BTW, a “compressor” decreases dynamic range, for dialog recording its best to use limiters rather than compression unless you want to sound like a DJ on a shock-jock radio show.
Definitely an option. We had talked about some kind of portable setup in a suitcase or something. The real thing is logistics. We’re just not prepared enough nor do we have enough time to do that. The reason we can send mics in the first place is they ship direct from Amazon using Prime.
Maybe when we take The Amp Hour to a pro level or something, but I’m guessing we’ll keep it simple and live with the audio we get.
Side note: Watch out for podcasts that use Skype as their medium…they can get really really rough on the ears….
I know very little about board layout. But curious about why having a power and ground plane close should not be considered as capacitive and should only be considered as lowering the inductance.