September 15, 2024

Biscotti firmware -- getting inside a Convoy S2+ flashlight

To start with, I have a disassembled Convoy S2+ that I took apart back in 2019. It is a special version with an illuminated tailcap. It turns out this yields unexpected tricky issues with detecting half way button pushes -- so it was a bad candidate to start with 5 years ago, as this requires some special code in the firmware.

But here is what I did back then to take the light apart. Removing the tail cap, battery and middle tube are trivial. Then you are faced with the head which has the big copper "pill" screwed down inside of it. You have to unscrew this. I have always managed with a screwdriver and persistence, but a spanner wrench might be required.

Once you have the pill out you have an even greater challenge. The ATtiny13 is on the other side of a circuit board that is soldered in place. Unsoldering it is virtually hopeless because you would have to heat up the big copper pill. I used a dremel tool to carefully grind away the solder, then a needle for final picking away of solder remnants. It was actually pretty easy.

Then you have a further challenge. Pin 5 on the ATtiny13 needs to be disconnected from the circuit board before you can reflash it. What I did back in 2019 was to cut the trace in question, expecting to solder a connection back in when I finished.

Now I have a hot air rework tool, and I'll be tempted to just unsolder the entire ATtiny13, flash it, then solder it back in place. We will see. I have a good supply of surface mount ATtiny13 chips, so as long as I don't damage the PCB, this might be the best plan.

NANJG 105D

This refers to the entire circular PCB with the ATtiny13, the eight 7135 chips, and other components.

The other components are a capacitor, two resistors, and a diode.
Here is the schematic for a 105C:

I check this against one of my PCB. Indeed I see pin 6 driving the gang of 7135 devices. The Biscotti code calls PB1 "PWM_PIN" and this is pin 6. An ascii-art diagram in the header comment in biscotti.c is misleading.

I have a 4.7K resistor as shown, and another resistor marked "28C" which measures 11K in circuit. The "28C" marking is 19.1K as per the EIA-96 scheme, so this schematic seems correct.

This resistor divider applies a 0.197 factor to the voltage, but also note that there is a diode drop involved also.

7135 chips

These are a 350 mA current regulator. If you have 8 of them, they will deliver 2800 mA of current working in parallel. Some are selected that deliver 380 mA, and 8 of these will deliver 3040 mA. This is marketing, plain and simple. There is something magic about getting past 3000 that impresses people -- and many people think only about biggest numbers and having the brightest light.
Feedback? Questions? Drop me a line!

Tom's Light Info / [email protected]