September 26, 2024

Biscotti firmware -- trouble, part 4

The ongoing saga of "trouble".

Still thinking about and working on light 2. I had a sort of brainstorm. I found another Convoy S3 in my stash. These lights are hard to work on, but this light works. So why not steal a board from one of these and swap it into light 2. I can use the S3 (without board) for another project, namely my microscope light where I intend to use some kind of external control for the light.

So I take the S3 apart, unsolder the two wires to the LED, then for no particular reason make myself crazy trying to thread the brass retainer ring back into the light. This is exactly why I never intend to work on (or purchase) any more S3 lights in the future.

Now I have a working and tested driver board in my hand. I cut the trace to pin 6 and then flash new firmware and fuses. Then solder it into the pill for light 2, reassemble the flashlight and it does not work!.

I wonder if it is important what order fuses and flash get programmed? Anyway, this will be more work for another day.

Next day

We disassemble this light yet again. I dig up a 5.6 ohm resistor and use it in series with the LED to test it using a 5 volt supply. It works!

I measure a 2.63 volt drop across the LED and 1.88 volts across the resistor. So we are running 336 mA through the LED. We are losing 0.48 volts someplace, probably through my wires and alligator clips.

I put the PCB in a vise. I solder wires to the two points that should connect to the LED. I connect everything with clip leads. It works! Why didn't it work assembled in the flashlight? I will note that the spring on the PCB seems shorter on the S3 lights. Maybe the dimensions on the head are different? I make careful measurements. They are not.

I reassemble things. I test after soldering the pill into the board and again it doesn't work. I pull it and test -- it is OK. I solder one side and it is OK. I solder the other and it works. Back in the light it goes. I am done. Maybe. I suspect this erratic business is due to the PCB shorting something to the sides of the pill. It is probably important to carefully center the PCB. I half expect this light to flake out at the worst possible time.

Branding the cows

I want to mark these lights without doing something ugly. The important thing to mark is the head with the pill. My idea is to file tiny notches inside the head. I do this and indeed it can be done so that they are invisible from the outside, yet permanent.
Feedback? Questions? Drop me a line!

Tom's Light Info / [email protected]