A magic smoke incident

[Lee697]

Cooking chips in the garage last night.... well, just a toasty ESP32 Thing Plus board and an ATtiny85 appetizer! I've been working on a little wifi board that controls a pair of relays, finally had it all working how I want, and had 3D printed an enclosure. Soldered in some tails for some relay outputs (NC) and an power inlet/switch, quick test before connecting up the relay output sockets, as after that getting to the copper tracks would be a nuisance. Grabbed a charged SLA batt from the bench, hooked up - pop! poof! and up went the magic smoke....
I'm unsure what happened.
I've checked and double checked the power connections I made, and they seem OK. The two voltage regs on the (now chipless) board are working OK (seemingly). I'm left with the conclusion I must have mixed up the battery terminal polarity (first time in 30yr I think.... it was late last night). Other possibilities, perhaps a short from a stray lead on the bench ( I do check for these), I'm even wondering if one of the live (12V) relay output wires had curled under the board and tickled somewhere fragile?
Apart from the two chips (ESP32 and ATtiny85), which are easy to replace, there are a handful of bypass caps (on two linear VRegs) and some low-side switching NPNs. If the VRegs are producing correct voltages should I assume the caps are OK?
I'm going to run it back up without the chips and probe the sockets carefully and test the transistors before replacing the chips.... Does anyone have any other advice? (apart from reverse polarity protection diodes! 🙂 )

[TopGun]

Do you have a schematic of your circuit? Did you have the relay directly hooked up to the digital pins of the AtTiny? A few pics would be good.

[Lee697]

There are two relays, they are switched from digital pins on the ESP32, through 2N4401's, with flyback diodes. The ATtiny is there to stop the ESP wifi server 'wandering off' as I've noted it can do every few days sometimes, it drives a 2N4401 to ground the ESP's reset pin every 6 hours.
I scoured the board today at lunchtime, probed all of the IC sockets, tested the transistors/relays, exchanged the caps, and reinserted some new ICs, and all is back up again.... I'll see if I can draw a schematic tonight....

[Lee697]

A bit of block diagram/pseudo-schematic....
There are two individually controlled relays, there's also an LED and a voltage divider over an analog input for battery voltage sensing.....