Hi Claire -
Thanks for responding.
Agreed that it's not good practice; for some reason, I had assumed that the Feather isolated the USB pin from the battery pin, which wasn't the case. My error.
The Feather was a 32u4 Basic Proto model; https://www.adafruit.com/product/2771
The connections were:
3.7v LiPo --> [JST connector] Feather [USB pin] --> [VOut] S10V4F5 [VIn] --> 12v power supply (but switched off).
The ground lines of the Feather, regulator and power supply were all connected.
With the power supply on and the battery disconnected, the circuit operated as expected.
With the power supply off, as soon as the battery was connected, the converter smoked.
(I didn't get a chance to test both power supply and battery connected.)
The voltage on the output pin was therefore coming from the LiPo battery, so would have been ~4.1v.
The failed part appears to be the IC marked BH00 (Semtech SC4503?). There's a small volcano-shaped hole above pin 1 on that chip now, and I'm pretty sure that was where the smoke came from.
I rebuilt the circuit, but can reconstruct it if a photo would be useful. Let me know?