Still getting into the rhythm of weeknotes, not helped by a busy weekend.

I started the week finishing off the PCB work I was doing for Museum in a Box. As I mentioned last week, the circuit design I did last year for our first batch-of-100 worked well, but we found that the headphone volume was a bit on the low side. The amp chip we're using has support for automatically switching to headphones when they're plugged in and as part of that can also reduce the level of amplification.

The amount of reduction is chosen with a simple two-resistor voltage divider and we'd gone with the mid-point initially. I desoldered those resistors on a test PCB last week, and this week replaced that with a small potentiometer on some long wires. That will let George and I adjust the levels on an assembled test box; and once we're happy with it we can specify new resistor values for the next batch. I think a just-changing-component-values is the electronics equivalent of a one-line fix.

Switching projects, but staying with electronics, I did some debugging on the new circuit boards that arrived recently. They're for a fewer-batteries-going-to-landfill version of the LFC LED cufflinks, which I'm switching to rechargable batteries rather than the original coin-cells (which is why they've not really been on sale since the initial launch). My new charger circuit works, but the "charging" indicator LED fails to turn off when the battery is charged. Cue more poring over the charger datasheet, and some test cycles with the circuit hooked up to a multimeter to watch the charge current. I haven't gotten to the bottom of it yet, because I'm mostly working at home and don't have a soldering iron (or my collection of 0603 surface-mount resistors) here. That was parked till I was next in the office and could swap out a resistor for the next test. That got done when I was in for the #BuildBackBetter hackathon on Saturday, so is now ready for when I've next got time for some more testing.

There was more current measuring in the week too, this time looking to minimise power usage during a sensor's wake-from-sleep-measure-and-go-back-to-sleep cycle. The Otii is a great bit of kit to easily capture graphs of power usage. It turns out adding "import json" to an ESP32 micropython program (even without using it) consumes 3uWh of power (not doing that is a 1% gain on this code...)

The week wrapped up with me attending a Liverpool outpost of the Build Back Better Hackathon. A handful of us gathered on Saturday and bounced around some ideas to try to help homeless people with no recourse to public funds. We chose that because Michelle, who was part of the group, runs the excellent The Paper Cup Project, a charity that supports rough sleepers and homeless people. Hopefully that will be a small contribution towards helping her work, and she can take things forwards.