It’s a simple fact that, in this universe at least, energy is always conserved. For the typical electronic system, this means that the energy put into the system must eventually leave the system. Typically, much of this energy will leave a system as heat, and managing this properly is key to building devices that don’t melt under load. It can be a daunting subject for the uninitiated, but never fear — Adam Zeloof delivered a talk at Supercon 2019 that’s a perfect crash course for beginners in thermodynamics.
Adam’s talk begins by driving home that central rule, that energy in equals energy out. It’s good to keep in the back of one’s mind at all times when designing circuits to avoid nasty, burning surprises. But it’s only the first lesson in a series of many, which serve to give the budding engineer an intuitive understanding of the principles of heat transfer. The aim of the talk is to avoid getting deep into the heavy underlying math, and instead provide simple tools for doing quick, useful approximations.
Conduction and Convection
Conduction is the area first explored, concerning the transfer of heat between solid materials that are touching. Adam explains how this process is dependent on surface area and how this can be affected by surface condition, and the reasons why we use thermal paste when fitting heatsinks to chips. The concept is likened to that of electrical resistance, and comparisons are drawn between heat transfer equations and Ohm’s law. Thermal resistances can be calculated in much the same way, and obey the same parallel and series rules as their electrical counterparts.
With conduction covered, the talk then moves on to discussion of convection — where heat is passed from a solid material to the surrounding fluid, be it a liquid or a gas. Things get a little wilder here, with the heat transfer coefficient h playing a major role. This coefficient depends on the a variety of factors, like the fluid in question, and how much it’s moving. For example, free convection in still air may only have a coefficient of 5, whereas forced air cooling with a fan may have a coefficient of 50, drawing away 10 times as much heat. Adam discusses the other factors involved in convection, and how surface area has a major role to play. There’s a great explanation of why heatsinks use fins and extended surfaces to increase the heat transfer rate to the fluid.
Modelling Thermodynamics
With the basics out of the way, it’s then time to discuss an example. Given the talk is aimed at an electrical engineering audience, Adam chose to cover the example of a single chip in the middle of a printed circuit board. In three dimensions, the math quickly becomes complex, with many differential equations required to cover conduction and all the various surfaces for convection. Instead, the simulation is simplified down to a quasi-1-dimensional system. Some imperfect assumptions are made to simplify the calculations. While these are spurious and don’t apply in many circumstances, chosen properly, they enable the simple solution of otherwise intractable problems — the magic of engineering! After showing the basic methods involved, Adam shows how such an analysis can be used to guide selection of different cooling methods or heatsink choices, or make other design decisions.
The talk is a great primer for anyone wanting to take a proper engineering approach to solving thermal problems in their designs. And, as a final party piece, Adam closed out the talk with a demonstration of a heat transfer simulation running on the conference badge itself. Thermodynamics can be a dry topic to learn, so it’s great to see a straightforward, intuitive, and engineering-focused approach presented for a general technical audience!
Sounds interesting. But too technical to concentrate on at the moment. Best to save it for later when I can give it full attention.
Hey.. there’s no add to “watch later” button.
Oh well. Click on it to view it in Youtube.
The add to lists button is greyed out.
Try clicking it anyway.
“This action is turned off for content made safe for kids”
Seriously?!?! WTF!
I think the “made for kids” thing needs to be boycotted and I’m saying that as the parent of a 9-y/o.
Thank you for making the world a worse place COPPA supporters.
Also forgot to ask if it was intentional to mark this one as “made for kids”.
No, that’s a new feature YouTube has added. I’ve gone in and specifically clicked “not made for kids” (why you have to opt-out is a weird thing to me).
Try it now and let me know if you can add it to your lists?
Now it works. Thanks!
+1
Myself I added a youtube folder to my bookmarks tool bar. For the most part use to bookmark youtube channels. I drag links to videos I want to watch later to a watch later sub folder. I find using the books marks easier and faster, than going to my account to do the same things. I believe if you are old enough,you can default to restricted mode off. Of course most kids probably a a fictitious google account including a fictitious age. Hell they have to gain access to the better educational material to use for their home work.
Did someone’s experiment go up in flames? I haven’t read the article yet.
No, it probably failed due to a combination of allotropic transformation of lead free solder joints, ion migration in the overclocked CPU, surface oxidisation on connectors, and NAND flash failure due to charge migration.
Or you could have just laughed.
But where’s the fun in that?
Crash course in thermodynamics aka plane crash.
Except that plane crashes are primarily transient phenomena, while this lecture only covers steady state.
Why would someone with an EE education need a “”crash course” in thermodynamics?
Believe it or not, I completed a BSEE program and was never required to take thermodynamics. I had to take a certain number of engineering courses outside of electrical engineering, but there were no specific requirements.
If you “earned” a BSEE without having to completely understand Thermodynamics, then you need to either get your money back or better yet sue that incompetent school you attended. You might get away with not knowing thermodynamics and heat transport as an EE if you are in a field with very narrow technical requirements throughout your career. But over a life of general practice as an EE, this hole in your education will eventually come around to bite you.
I didn’t say I don’t understand thermodynamics. I said I was never required to take a course for it. Just the same, I do feel that I was cheated in many ways by the “top tier” institution I got my degree from.
… why would *anyone* with an engineering degree ask this question?
+1
Packaging engineer. :-p
I can’t tell if they meant that as ‘Real Engineers ™ studied thermodynamics in excruciating detail ‘ or ‘what use would they have’
Ignorance is bliss I guess. Why would anyone want to be more knowledgeable outside their comfort zone, especially when the boss or consumers are paying for their lack of knowldege?
two thermo clases were required for our EE BS. they had setup a course for “non mechanical” engineers.. prof was soo shitty, all the EE’s dropped it every quarter… it was scaring.. i think it was a pissing match between engineering departments, cause the material wasn’t that difficult to understand, they just taught it in the most convoluted way.
I believe the technical term for that is a ‘weed-out course’. Somewhere along the line academia decided to be gate keepers instead of educators. Heaven forbid we have to many engineers or doctors.
Or the natural result of higher-ups demanding one department send personnel to service another department, on their budgetary dime, so obviously they send the idiot or the a-hole.
It might not be so negative. Perhaps there was a deliberate decision to have relaxed admission standards, so that every student gets a chance. But just because the standards for admission have been lowered, doesn’t mean the academic rigour required to succeed has also been reduced (and we probably agree that it _shouldn’t_ be reduced.) So, a rigorous course that has low admission requirements will likely have a high dropout rate. It may be “a feature, not a bug.”
Now, there are certainly academics who are not good teachers/explainers, and course materials that are poorly organised. This of course makes learning the material harder. But we _are_ supposed to be hackers here, so we should be able to rise above it.
Perhaps the university is trying to act as a gatekeeper, but for doctors and engineers the barrier tends to be the professional registration bodies, which seem to be run like medieval guilds combined with pyramid scams :-)
I remember writing a Matlab script to do exactly this as part of my Engineering course work. Would be great to offload the number crunching to the FPGA.