SMT Thermal Design Mistakes

[@happyhippr]

wow one of the best documented experiments on heat transfer in pcbs. best I've seen on KZbin when searching for this type of stuff. no whiteboard mumbo jumbo, just hard measurements and demonstration. bravo

[@warpigs330]

Your videos are really some of the most useful electronics videos on youtube. I don't know why you don't have more subs and views.

[@WizardTim]

Glad to hear the videos are useful but I can guarantee there are some much better produced videos on KZbin about electronics, just not on exactly the same topics. I haven’t made many videos yet nor have I been doing it for very long, but the channel seems to be gaining interest as in the last month viewership went up about 800% and subscribers have doubled.

[@warpigs330]

@@WizardTim Perhaps it is just that the things I am thinking about happen to be the things you make videos on.

[@jaro6985]

So much nicer when the tab is connected to Gnd. Would be interesting to compare to an Al core PCB as well.

[@WizardTim]

I agree. I might revisit this in future with some purpose made PCBs and wider scope.

[@kagu]

It's a simple mistake, but I have to say, that video was made quite nicely!

[@solomondg]

Cool! I always wonder why more linregs don't sink heat to ground - even if that's added manufacturing cost (since it's probably a FET drain thing) I'd imagine it would be a big selling point.

[@WizardTim]

Sadly it’s not just a cost thing it’s a complex silicon design thing. Dropout voltage plays the biggest role in why mostly LDOs end up with Vout on the tab. Your jellybean standard regulator will often use an NPN Darlington pair which needs a high dropout voltage, LDOs with GND on the tab often use a PNP pass transistor but those will often suffer from high quiescent currents, the 1117 uses an NPN to achieve its low quiescent current but the trade-off it makes is the silicon substrate is Vout and thus the tab.

[@nhanvu1566]

Excelent video!! what's the model of your thermal camera? Thanks!

[@WizardTim]

It’s a FLIR E8 which is now discontinued but it has been replaced by the FLIR E8-XT (AU$5,200) which can measure higher temperatures with better thermal resolution (same X-Y resolution). The PC software shown in the video is also an additional AU$500 however it has also been discontinued and replaced by an annual subscription version (I hate it). I don’t recommend the lower end FLIR E series models as after trying them I realised that not only do they have a lower X-Y resolution but the temperature resolution is much worse and they add a lot of noise to the image which makes looking at small thermal gradients on PCBs rather difficult (all models use the same sensor but software intentionally degrades image quality in the cheaper ones for market segmentation). Shop around as well, some distributors will throw in additional accessories like a spare battery or soft carry case for MSRP. However I must note I bought the camera at the end of 2018, and the FLIR E series is now about a decade old, because of that I would recommend against buying a FLIR camera as particularly in the last year there are a lot more better options from other companies for vastly less money and also not crippled by pointless US ITAR restrictions.

[@ninethirtyone4264]

Great video sir!

[@FerdAboubakrOnline]

nice one ,love it .

[@sciencoking]

What I learned: I definitely want a thermal camera

[@WizardTim]

I wanted one for a long time and finally got the courage to spend the money on it late 2018 and it’s been super useful for designing electronics but also repairing electronics and mechanical things (saved many hours by instantly finding the issue). I’ve found it’s useful in all sorts of situations and it’s now a tool I couldn’t live without, but at AU$6,000 it’s difficult to justify even with all the use I get out of it.

[@chitotapia]

@@WizardTim Thanks for the video. very useful. what brand/model of thermal camera did you use here? I plan to recommend one for work. thanks.

[@WizardTim]

@@chitotapia It’s a FLIR E8 which is now discontinued but it has been replaced by the FLIR E8-XT (AU$5,200) which can measure higher temperatures with better thermal resolution (same X-Y resolution). The PC software shown in the video is also an additional AU$500 however there is a free version with limited functionality. I don’t recommend the lower end models as after trying them I realised that not only do they have a lower X-Y resolution but the temperature resolution is much worse and they add a lot of noise to the image which makes looking at small thermal gradients on PCBs rather difficult (all models use the same sensor but software intentionally degrades image quality in the cheaper ones). Shop around as well, some distributors will throw in additional accessories like a spare battery or soft carry case for MSRP. There are also thermal cameras available from Seek, Fluke, Testo and others but I haven’t tried any of them.

[@fredbloggs4829]

search for surface mount thermal bridge to make use of the ground plane as a heatsink.

[@WizardTim]

Yea those SMT thermal bridges are nifty ideas but for the price of them (AU$0.50 - AU$3.00) I could buy an SMT folded copper heatsink that could dissipate much more heat :/ They're only really cost effective in very niche RF applications and certain ultra low profile designs.

[@zianzan4536]

Hi ! Good job but ... If you want to dissipate heat => no green varnish in the copper aera (whish has been ENIG finish) ! :)

[@WizardTim]

Yes 100%, I see that regularly on PCBs in high end products and I've been tempted to do that with several PCBs, but ENIG is just such an expensive option when I'm only ordering 5 or 10 prototype PCBs (4-8x the price of a standard PCB) which is less cost effective than just adding a SMT heatsink, plus the gold is often only like 0.02 µm thick so you don't even get that vibrant gold finish :( I will add on the other hand aluminum core PCBs have really come down in price (only ~2x the price), I've had a play around with some designs using that and it's a massive performance increase, but the single sided limitation really limits the application, only used it so far for LED stuff.

[@zianzan4536]

@@WizardTim I wanted to say that copper or ENIG or HAL are all good for heat dissipation but not green solder mask on copper, which is as running with a pull-over. On your video, it seems that you have green solder mask on the dissipation aera ?

[@WizardTim]

@@zianzan4536 Yes both the green and white PCBs have solder mask over the heat dissipating areas. I don't think I've ever done a PCB with an exposed area for a PCB heatsink that was HAL, I did try once with a silver immersion finish, was happy with it at first but it tarnished really quickly, so I might give HAL a try in future for PCB heatsinks. I've seen some PCB designs that expose strips of HAL in a PCB heatsink and the wave soldering process adds extra solder, I can imagine that massively improves in-plane thermal conductivity.

[@zianzan4536]

@@WizardTim :) I'm doing PCB for 20 years now, and yes : no vernish over dissipation aera. Yes : HAL is not a good finishing in that case of dissipation aera (unesthetic and solder is not uniform), but ENIG is perfect : more than 50 boards done with this finishing and with dissipating areas. With vernish : almost no dissipation is possible, it's not mettalic, it's as a cover. A nice trick with ENIG and a transistor with thermal pad SMD : let a litlle band of vernish between SMD pad and dissipation aera to avoid solder going on dissipation aera. I'm using a thermal camera too and I've strange results with ENIG dissipation aera : heat is not the same with angle of the camera (a hot spot of 60°C can be shown only on some angles, otherwise it can be 25°C and the hot spot is not visible), very strange, do you have same ?

[@WizardTim]

@@zianzan4536 ENIG areas will appear as if the temperature changes with the angle of the camera because ENIG and most other metals have very low emissivity and high reflectance, so that 25°C is just the reflected room temperature. You can get around this by putting some Kapton tape over it which has a high emmissivity, but that will change the thermal performance.