So - I made a calibrated voltage readout routine in my attiny85 firmware. Calibration of voltage calculation was done to match readout on a calibrated fluke DMM and verified throughout entire 2.5V-4.4V operational range.
Among other things it reads voltage every 15 seconds and blinks it out.
So this is actually reading the voltage the driver sees live, while under load.
At 3A load A fully charged ncr18650pf was being read with only 3.6V, which is quite less that it should be according to discharge tests.
I have yet to measure where the voltage drop occured, could be one of these things
1. the battery itself
2. switch/switch spring/switch connection to tailcap
3. tailcap-vs tube connection
4. tube vs head connection
5. somewhere on the nanjg105c PCB itself
I will update further based on my findings.