Eric Bogatin on Breaking Bad Habits in PCB Design - AltiumLive Keynote

Eric Bogatin is Dean of Signal Integrity at Teledyne LeCroy. Listen in and see Eric use the Arduino Uno board as an example of how not to design a 2-layer board. He’ll show you the best design practices to use and how the common Arduino boards violate most of them. Watch Eric in action as he does some live measurements and compares a commercial board with the same design done right.
Pre-register for AltiumLive next year at: bit.ly/altiumlive
Watch all the AltiumLive Keynotes at: bit.ly/AL19vids
Follow Altium on Twitter: / altium
Follow Altium on Linkedin: / altium
Follow Altium on Facebook: / altiumofficial
Subscribe to the Altium Podcast on iTunes:bit.ly/altiumpodcast
Please subscribe to our channel and remember to LIKE and SHARE.
#AltiumLive #Altium #OpenSource

[@Gengh13]

The introduction ends at 9:35, if you want to get straight to the talk, it starts at 10:57.

[@DrTune]

I'm nearly 18mins in now and this whole thing is just fluff so far. Where's the meat?

[@THEELECTRICGUY]

This is the kind of professor I want in my university

[@willcollins9470]

Really good teacher and easy voice to listen to. I wish he did a podcast

[@chaselewis8473]

From like 13:53 onward when he is talking about how people aren't adequately trained today and also talks about maxwells equations and how in school they just teach us how to solve differential equations and what not... he hit the nail on the freaking head.

[@alkhazraji7482]

yo Mr. Bogtain lets design a circuit.

[@ZacksLab]

Eric is an amazing teacher... saw him at PCB West this year, wish I could have been at AltiumLive!

[@Kefford666]

This is gold! I’m so glad I stumbled across these talks because I don’t know how else I would learn these things. When I studied at Uni we covered a range of topics (electromagnetism, digital, analogue, mcu’s, control theory etc) and I felt fairly well educated but we weren’t taught board design like this in any of the modules. PCB design is such a staple skill for an EE too! I guess that was saved for MSc or PhD students but really it should be taught from year 2!

[@tommihommi1]

I help with EMI testing for student projects at my university, every time I see a group roll up with a classic Arduino at the core of their design I know what's coming, the signature 32 MHz noise and it's harmonics definitely will be there, and likely over the limits. the ESP32S nodeMCU boards are another nice example, when they're accessing the flash memory via SPI they're so noisy, it's possible to read the data being transferred from across the room

[@edouardmalot51]

Big thanks for this presentation

[@ruwandarshana7881]

thank you , not only for alitim eda but also bring engineer who uncover his or her experience

[@Decco6306]

Most electrical and software engineers would not be where they are today without open source hardware and software. Without this, we raise the barrier to entry for millions of aspiring engineers around the world who are not privileged enough to own anything else due to the increase in cost due to licensing. With out the Arduino, millions would still be stuck paying out the nose for paid for software and hardware with unethical limitations in place to hinder the creativity of our younger generations. Open source is the best thing in tech this world has ever seen.

[@bloguetronica]

The Arduino is a trap, mostly. It is mostly used to do "projects" easily, but doesn't teach much and/or in depth about electronics and programming.

[@Decco6306]

@@bloguetronica What part of "to get started" did you not understand. Im not a dumbass. i know what a datasheet is. its the exposure factor that is what get people started. you go to a high school STEM class and start talking about registers, alus and interrupts then throw the atmega datasheet in their faces and shit, that's going to turn a lot of beginners off of tech. Its the best thing in embedded systems sense the pc/104 system. is it suitable for professional applications? no. but it's a great educational experience. especially for people who want to see programming be a literal thing they can see in front of them. it's also a great gateway for people going into programming.

[@Decco6306]

@@bloguetronica without open source hardware and software we would not have the massive influx of people wanting to get into electrical engineering and other technological branches.

[@bloguetronica]

@@Decco6306 You are referring to someone who contributed with many OSH projects (me). I cannot testify about the massive influx, but I agree with the principles behind OSH, as I agree with the principles behind OSS. However, with many excellent OSH projects, the presenter decided to pick Arduino as being representative. Arduino is badly designed, a bad platform that facilitates too many things without pushing the developer to go further, and also provides a bad IDE that doesn't allow having multiple files on the same project, forcing one to essentially copy and paste. Lets see from those people that started with Arduino, how many had progressed into more serious platforms? Because Arduino doesn't encourage someone to program directly to the MCU, without the boot-loader's assistance. And Arduino is only meant for quick projects, not for serious ones.

[@bloguetronica]

@@Decco6306 If beginners don't like dealing with registers and interrupts (essentially, the core of MCU programming), then you can assume that those beginners don't have the patience and don't want to learn. The "inclusivity" shield is just a way to conform people to a "norm", and provides no flexibility whatsoever. Programming an MCU, Arduino style, limits you to use certain pins for input and certain other pins for output, while dealing with the DIR and IN/OUT registers directly provides you with a much greater flexibility. Mind that many beginners that start with Arduino are already thrown off by the intricacies of direct MCU programming and won't progress beyond using Arduino. As for others, Arduino is not really needed. I only programmed an Arduino once, and didn't like it, because it is really limited and not flexible, and the IDE is disgusting. Yup, limited options. I was thrown off by that!

[@LoveTheFactory]

Fantabulous !

[@VeritasEtAequitas]

43:00 He says use the thinnest trace, but dont! Manufacturing yield aside, it's significantly decreases the coupling to the return path and the layer below, meaning you'll be greatly increasing the relative coupling to any aggressor signals. Roughly 85% of your return is going to be in the plane below. Using a thinner Trace effectively increases H! You'll find this in oyher Altium Academy and Fedevel videos, maybe even in his own.

[@mhe123321]

I think it is implied that this relates only to low frequency/rise time signals. The return current only starts crawling underneath when the trace-groundplane-subtrate acts more like a waveguide which is only valid for higher frequencies - but at that point you are desigining your traces to have a certain characteristical impedance anyway.

[@samchan5251]

The Arduino board that the student design is no longer an Arduino board, it is just a 328 dev board, since the board sharp and header location is not at the standard location.

[@phil85813]

i thought the same, would have been good to see the noise performance of the student board where board size and header position was fixed

[@MordecaiV]

@@phil85813 one of the bad habits is actually in the connector specification, unfortunately. (not providing grounds for digital signals in your connectors) . A good design (in my not-an-expert opinion) reference would be PCIe for a way to have a flexible and high performance interface.

[@Dazza_Doo]

Altium congratulating themselves on Selling Stuff, also Altium "lets charge them like they are Millionaires". Sorry Altium I won't spend 1/2 my rental money on 1 software for the Month. But I do enjoy these talks, helps me as a hobbyist.

[@AirCrash1]

Habit 7 Don't use non standard units of measurement.

[@damny0utoobe]

Bookmark: Using probes to sense board radiation

[@samihawasli7408]

“ experienced electrical engineers to be“ I should’ve put that one on my résumé long ago

[@paulandsabrinaholmes]

So, a little IC like a 14 pin NAND needs a 22uF (1206 package??) instead of a 0.1uF 0603 to get it to be well behaved on the die? And for the microcontroller, you need a 22uF on each and every Vcc/ground pair?

[@mrkv4k]

No. The problem wasn't value, but placement.

[@bpark10001]

The title of this forum is not addressed in the material. What does "open source" have to do with bad board routing? Are Altium-routed boards, "closed source", free of this malady?

[@flisboac]

Altium Designer, etc., is okay, it just needs to be cheaper.

[@vladlv2]

Not using ground Pours?? is quite strange:) how you gonna do it in practice? Altium will not allow you to route ground traces individually, it will change traces based on topology preference. Switching topology from star? not sure it is good idea, it will become mess very quickly. You still need to use pour for ground... but analyze it more after applying. and maybe rearrange power distribution etc. but it is not as easy like presented in video and requires compromises.

[@juliatruchsess1019]

Use a continuous ground _plane_ on its own layer, not pours in signal layers.

[@vladlv2]

@@juliatruchsess1019 i was speeking about 2 layer boards

[@johnyang799]

@@vladlv2 Route on one layer as much as possible and use the other as continuous return paths. Anything routed on the ground layer should be as short as possible. Signals doesn't care as much about inductance. Power and ground do.

[@Dazza_Doo]

If you haven't learnt, use more than 1 software. ;)

[@bloguetronica]

I love the technique he devised to measure the on-die noise. Never though of that, and although simple, it is quite effective. Anyway, it was a great presentation! However, there are great open source designs out there, make by people that knew what they were doing. The Arduino is a very bad example, and not representative. It is a half-assed design that hardly had any revisions before it was first launched. Proof of that is the odd spacing between two of the SIP headers, that remained as a design "feature" in order to perpetuate what was a mistake.

[@vladlv2]

not sure 6 mil traces ruls is a good idea. because thinner traces have also more parasytic inductance. I would go witth 10mil by default, and use 6 mil only in places where I cannot get away with 10mil.

[@dzidmail]

Inductance is good for power distribution to digital circuitry if sufficient decoupling capacitance is used. Exception will be lines that are shared by both analog and digital. For low speed signaling the inductance doesn't matter much either.

[@yum33333]

It's interesting that even this guy got something partially wrong... the "power puddles" mentioned at 45:40 work mostly because of their capacitance with the ground plane - the reduction in inductance by widening the traces is actually pretty minimal. I also have to disagree with point 7a and 7b. He wants us to use 22uF caps as bypasses? Not only is that an insane degree of overkill - it will actually have drastically worse performance due to the physical size of such caps. As for 7b, it is good practice to designate a layer as GND - I assume he means don't pour GND polygons on every layer? That would actually be good advice.

[@dzidmail]

I think you are vastly wrong about the capacitance vs inductance there. You get like 1pF capacitance between power and ground layer. This is tiny compared to the capacitor.

[@yum33333]

@@dzidmail At very high frequencies - like the transition edge of a switching CMOS circuit, which commonly has energy up to 3 GHz, even a few pF makes a huge difference. Most of the very high frequency current is actually looping through the PCB capacitance to GND, not through the bypass capacitor. This is partially because the bypass cap has a certain minimum inductance imposed by it's physical size. The current has to concentrate down into a small area to reach the cap terminals. Past a certain point, making the traces wider doesn't actually have any effect. Everything beyond that (i.e. making a puddle) is just down to the puddle capacitance to GND.

[@dzidmail]

@yum33333 Your explanation makes sense. I guess it really depends on what frequencies people work with.

[@LochNessAnthony]

Breaking bad

[@michaelstevens630]

Most used and expensive $

[@marcan42]

Excellent talk, bad video title. The talk was titled "Breaking Bad..... Habits". The words "open source" do not appear a *single* time in the entire talk. Why did Altium decide they had to diss the entire open source hardware ecosystem in the video title, completely gratuitously? This has nothing to do with "open source designs". I guess Altium couldn't resist the temptation to diss the entire open hardware ecosystem (which largely does not use Altium but other CAD tools), even though being open source has nothing to do with the bad habits the talk is about. Yes, Arduino boards are terrible. So are many designs made using Altium. There are many great open hardware designs built with Kicad, and many great commercial designs built with Altium. Altium, don't put words in your presenters' mouths. You can do better.

[@AltiumOfficial]

Dear Marcan, We will retract the title and come up with something that more clearly describes the video. We apologize for the mistake and take seriously our ability to provide honest information and label it properly. Thank you so much for raising attention to this. Kindly, The Altium Social Media Team

[@Dazza_Doo]

Just Saying - 8000 new users of Altium (don't tell them it's the New Students in Colleges world-wide), I can make your sales go from discounted student subscriptions to 20,000 new fully paid users per Month - how? That's my secrete - but hey Good Luck with that.

[@PHamster]

This Part...Copper Pour: kzbin.info/www/bejne/enqweq2Bh9KspMU Conflicts with this: kzbin.info/www/bejne/r4TYho17n6aFhrs

[@mrkv4k]

It really doesn't. This video talks about arbitrary (i.e. added without much thinking) copper pours, the oher is about fixing a specific problem by adding board capacitance with a well designed copper pours. Those are two completely different things. Copper pour can be harmfull, if it's not designed properly. If you are just adding copper everywhere, without thinking, it's less likely gonna be well designed...

[@bpark10001]

This guy (or the forum) may be great (according to others?), but he destroys all of whatever credibility he has when he, in an "Altium Live Forum", explains why "open source" is bad. Altium is the premier example of "closed source software", and profits heavily from that! This is "conflict of interest" to the extreme! You say "reverse engineer". How does one do that with (Altium) software? The EULA most likely forbids that! This presentation is dripping with Altium money! look at the fancy auditorium, the blue lighting, the Altium logo on the badge lanyards, and Altium header on all the slides. Breadboard interconnects are not transparent! There is less capacitance than on a circuit board. Especially for High Z analog systems, the worst breadboard is better then the best circuit board, especially in wet environments. All of this presentation (except for the noise sensing scheme, which is the only novel idea presented) is biased in favor of expensive board layout software and expensive board houses. It is also biased for high-speed digital circuits, and not analog systems. It also does not address board reliability (5 mil trace can corrode away in a moment, tin whiskers can short, and tiny via can crack, unless you get board from MIL qualified vendor, to include X-ray tomography inspection). There IS a reason for copper pour. It reduces the chemical waste stream, and the cost of fabrication. An Arduino board has 16MHz maximum inside the processor. These measurements are biased toward >100MHz systems. The I/O lines are typically much lower frequencies. Most of the internal die noise is due to the die bond wires and is beyond the control of the layout engineer. It doesn't matter as all of the internal logic is floating at that voltage. Well-designed chips use isolation between the power for I/O and the internals. They also use 5V which gives you a 5dB margin over 3.3V, all things being the same. There is another way to drastically reduce switching noise. Put series 1K resistors on all I/O. A robot I designed this way passed EMI (100V/M) susceptibility and radiated EMI.

[@mikejoyner3051]

Your response smacks of a grasp of bandaids, mitigating symptoms rather than to fundamentally mitigating the actual source of problems in a given layout.

[@bpark10001]

@@mikejoyner3051 It is NOT bandaids! It is sound design (regardless whether it is open source, or "closed source"). First, lowering the top frequency reduces noise. The design focuses on high-speed digital design, and professes that all designs should be done this way, using these philosophies, when in fact this type of circuitry is in the minority. If the presenter chooses to attack design methods, he should do so based on their merits, not whether it is open or "closed" source. With digital circuits, the majority of noise originates from common-mode high-frequency currents. Keeping the high frequency in a "box" and permitting only low frequencies in and out reduces noise. That is the scheme that I used, and it passed muster at the test house on the first try, both for EMI (emitted radiation) and susceptibility (disruption by external fields). It also does not require geometry that is susceptible to environmental corrosion.

[@mikejoyner3051]

@@bpark10001 again you cannot see the forrest through the trees. It is the same failed premise assumed in audio as well. When you design IC’s and pcb’s you learn what are the sources and what are bandaids or just fixes that mask the real problems. When even the simplest of circuits are subjected to an operating environment of dirty public power distribution, nearfield personal electronics, it is a constant issue unless you live in a faraday cage. Open source designs are often viewed as standards for methods when in practice they are a disservice in illustrating “well it still works” rather than best practices...

[@MordecaiV]

He's not saying "open source is bad." He's saying just because you can look at it, doesn't make it a good example. The switching noise is there even if it 'doesn't matter' because it still works.

[@bpark10001]

@@mikejoyner3051 Who are you, a sales man for "closed source" expensive electronics firm? You are confusing "open source" with "bad designs". There are open source projects and there are closed ones, and there are bad designs and good ones. One thing is evident: open source designs are "aired to the public" and subject to scrutiny, AND CLOSED SOURCE IS NOT. You are pointing your gun at the unshielded soldier! My experience (monitoring engineering in military contracts) was to the contrary. "Closed" designs often are made by "someone who happens to be on the team" and is not competent in the area needed. Other cases are simply human oversight which is undiscovered because no one else can scrutinize the design. That was my job, and boy, some of the designs were doozies! In one case, a Vicor power supply was radiating like crazy (60dB over spec). Because supply design is proprietary and potted, a $2 proper fix could not be applied (putting 2nd copper layer between the FET and the heatsink). instead, "bandaid" was required (connecting the case of the supply to DC bus minus) and surrounding the whole thing with insulation for safety.