Pijuice Zero + Pi Zero 2W -- No RTC/Watchdog wakeup on battery

[DasVinch]

Hello,

I'm working with a PiJuice Zero over a Pi 2W for an autonomous camera projects.

I'm struggling with unreliable wakeups from the RTC alarm... pointing to a possible electrical / firmware problem? Electrical measurement, plus a cursory look at the firmware code in power_source.c seems to point to insufficient capacity at raising the 5V bus all the way to induce a boot when the system is in deep sleep.

Help appreciated !

Configuration

• Pi 2W -- running bookworm lite
• PiJuice software: 1.8
• Pijuice firmware: 1.6
• Battery: 100 Ah LiFePO4 (fully charged, reading 3.22V)

Behavior

Initiating a full poweroff:

• get RTC time
• add 2 minutes and setAlarm at that point in time
• PJ.power.SetPowerOff(60)
• shutdown now

The LED sequence (D1 red when shutdown timer completes) + power consumption monitoring confirms that the pijuice shuts down (~90 mA residual on battery's 3.2V), and then the 5V bus shuts down (<~few mA residual).

Observed behavior

• After one more minute, D1 starts very faint red then goes slightly brighter, stays a few seconds, then turns off.
• Monitoring with a voltmeter shows the 5V rail pushing to ~1.5V, then dropping to 0.4V.
• Pi Zero does NOT turn on.

Expected behavior

• Pi turns on, 5V rail goes up to 5V

Additional info

• Procedure above works fine and expected if a 2.5A power supply is connected to pijuice microUSB
• A single press on button is generally NOT turning the Pi on. Quick repeated presses show the 5V bus voltage ramping up from 1.5 to 3V a little at each press, and the voltage typically clamps to 5V and pi boots after 4-5 fast repeated presses.
• I read 90 uF of capacitance between the 5V rail and ground, seems like a lot. It's just a pijuice zero and pi zero 2W
• Tested and reproduced with 2 (brand new) pijuice zero.
• Tried both LDO and DCDC for SetPowerRegulatorMode, no difference

Any thoughts on improving reliability ? Can any of the configuration flags (battery profile, extended profile, etc) influence this behavior and we got that wrong ?