Contrary to the article I am simply using one port of an Attiny85 to output several types of pulse trains. The port is switching the load (R=60 ohms is what I have proven to keep the power bank alive when driven continuously) by means of a MOSFET (BS170). The circuit as such works as expected (proven with a DSO), but the conditions under which it keeps the power bank alive are tricky.
- My first attempt - ordinary PWM at 1 kHz with duty cycle up to ~20% failed. I didn't drive it to higher duty cycles because my goal is saving power.
- My second attempt - manually switching the port at low frequency (<< 1 Hz) with on-time ~1 second, off-time ~4 seconds was successful
- Third attempt - reducing on-time to 0.5 seconds or increasing off-time to more than 4 seconds failed (1 second "on" seems like the minimum for the power bank to recognise intermitting load)
- Fourth attempt - on/off = 1 s/8 s with intermittent 10 ms pulses works for some time (longer than the auto-shutoff time of ~15 s), but eventually fails (shuts off)
What I conclude from these findings:
- the power bank's controller somehow integrates current (otherwise the high load peaks of the ordinary 1 kHz peaks would be enough to keep the unit awake)
- 20%-duty cycle at 1kHz (not working) is not the same as 20% duty cycle at 0.2 Hz (working), so the power bank integrating current cannot be the whole story
- the sensitivity to peak load seems to get lower for longer pauses (8 seconds), but can be reactivated by very short maximum load peaks
Does anyone know something about commonly used strategies for implementing the auto-shutoff in commercial power banks?
Is there any easy explanation for the behavior that I have observed?
Any ideas on how I could reduce the duty cycle substantially further while keeping the unit alive?