I removed the uC from PCB #2 and cleaned it up with some flux and my soldering iron, then I cleaned it with alcohol. I checked for shorts on this bare PCB and there were no shorts.
With current limited to 100mA, I applied power to the bare board and checked for correct voltage readings at VCC.
Then I put a blob of my flux on top of U1 pins 27 and 28. I applied power and nothing happened, there was no shorting…no current draw on the supply. I thought to melt the flux with my iron. So I melted the flux and gave it power again…
This time I got smoke and lightning. I recorded a video as evidence, it’s in the link below. You probably don’t want to watch the whole video since I didn’t trim it at all.
https://drive.google.com/file/d/1YLSlxWR3WGbX4L_Ll-CuinCWbR5uLWGj/view?usp=sharing
0:00 Setting the voltage and current limit
1:00 Checking for shorts before applying power (melted flux on PCB)
2:00 Applying power
2:30 Current rises, smoking
4:00 Smoking
5:20 Illumination and plume of smoke
6:25 Checking for shorts again after power removed
From the video, you can see there are no shorts on my dollar store meter before power is applied. After the test is complete, there is a short between VCC and ground of -888 Ohms with my meter probes in one orientation, and no short when in the other orientation. I don’t remember which orientation was which, and I don’t know what a negative reading means.
Here is another video after the test and after I cleaned the board twice with alcohol. It still had a short after I cleaned it the first time. You can see the short is gone in this second video. Nothing happens in this video so you’ll want to skip through it.
Nice fireworks.
I’m pleased we don’t have to blame Kicad
Maybe those chips aren’t all dead.
Maybe they didn’t respond because their power was shorted.
You will have to lift them all to get rid of all the flux.
Can’t electronics be an enjoyable pastime
Difficult for me to see but, looking at the screenshot suggests checking:
Gap between Pads:
• Solder bridging (though seeing no resistance on meter)
• ‘Gap Width’ and possible arching at 1A
• Pads/solder looks fairly consistent except for the two in question (screenshot).
Quality of the PCB:
• Deformed/distorted traces (resulting in gap issue)
• Expansion of trace under (thermal) load, causing gap issue)
And, for sure, consider @davidsrsb recent post (re voltage regulator)
The list can go on and on and I’ve seen too many issues to list here and it happens in production with Pick&Place and Wave/flow soldering. Lessons learned keeps me from having anything to do with hand-soldered SMD.
Now I need to clean all three PCBs very well. 91% IPA doesn’t seem to work too well on this flux. Can somebody suggest a different method? I did a search online and people were suggesting things like paint thinner and other harsh chemicals.
Also, I need suggestions on how to remove these SSOPs. I ruined the pins on the one I took off the board in the video. I’m afraid the uC on the first board is probably dead since I wasn’t limiting current on that one. I saw a video of somebody using something called “Chip Quik” low temperature solder. Can I purchase this product at a local store? I’d like to get working on these boards this weekend so I don’t want to order online and wait around…
By the way, I processed the return on Amazon for my flux. Today was the last day of the return period, so I just barely beat it.
I started the Thermal FEA and while waiting for an almost endless time, what caught my attention (relating to my question about trace widths):
• I think the trace width is too narrow (marginal, at best).
• Proximity of trace to pad is narrower than trace width, suggesting too close.
Using the calculator (assuming Cu is 18um) and estimating a Max Temp rise of 65º C (150ºF, adding the Ambient of 70ºF = 220ºF). Is perhaps too much for the PCB, and calculated widths of 7.4mil.
And, that’s for External traces. But, with conformal coating, I’m thinking you should reconsider the trace widths and proximity, mentioned above.
Might be a fun test, if you have a Temperature Measurement tool and measure Temp when it smokes…
Thanks for running this analysis on my PCB. I’ll keep the spacing between traces in mind when I’m designing version 2 of this board.
Just want to point out though that this spacing wasn’t what caused the shorting, it was most definitely the flux. Remember, I had the same short problem on an SSOP protoboard.
It would be interesting if you could experiment with a dab of your flux on a pair of adjacent pads that are powered on a sacrificial board. I find it surprising that such a low voltage can cause arcing through this kind of flux, but I can believe it.
One more thing,
Superglue (cyanoacrylate adhesive) will cause the same problem even at 5v. The smoke just smells far worse. Don’t even keep it in the same room as your electronics.
Another lesson I learnt many years ago along with soldering irons and flux and … oh!, the list is far too long to continue.
