Menu sleeps at 167µA@3.6V, long-pressing the reset-button will sleep at 187µA@3.6V

[bablokb]

As the title says: going to sleep with a long press of the reset-button ends up in a state that draws 20µA more than "normal" sleep.

From the code, that is not obvious, maybe it has to do with the additional PWMs setup for the fading backlight. But that is only speculating and probably wrong.

[ZodiusInfuser]

Or it could be to do with the pins set up for waking the board with the front buttons 🤔 .

What was your overall current draw measure in the two sleeps?
Edit: I should have read the numbers in the title 🤦

[bablokb]

I could not figure out the code path for the menu - how does it start sleep?

[Gadgetoid]

Under the hood it calls powman.sleep() if there's no user input for a while. That happens here:

badger2350/modules/common/badgeware.py

Line 491 in 183db43

powman.sleep()

Which calls:

badger2350/modules/c/powman/bindings.c

Line 105 in 183db43

mp_obj_t _sleep_sleep(size_t n_args, const mp_obj_t *args) {

Versus long press sleep which calls:

badger2350/modules/c/powman/powman.c

Line 399 in 183db43

void long_press_sleep() {

In both cases they init the buttons and set some pulls, but they are roughly line for line the same code.

These functions are so similar that your speculation is probably the right answer- something about setting up the PWM is causing that extra power consumption.

ie this is insufficient:

badger2350/modules/c/powman/powman.c

Lines 456 to 461 in 183db43

	for(unsigned i = 0u; i < 4; i++) {
	pwm_set_enabled(pwm_gpio_to_slice_num(led_gpios[i]), false);
	}
	gpio_init_mask(0b1111);
	gpio_set_dir_out_masked(0b1111);
	gpio_put_masked(0b1111, 0);

Because 🤦 it's never actually called when sleeping:

badger2350/modules/c/powman/powman.c

Lines 445 to 451 in 183db43

	clear_double_tap_flag();
	if(!gpio_get(BW_SWITCH_UP) && !gpio_get(BW_SWITCH_DOWN)) {
	shipping_mode();
	} else {
	long_press_sleep();
	}
	break; // unreachable

[Gadgetoid]

I've pushed a test build that changes nothing but this, which should pop up in actions soon: https://github.com/pimoroni/badger2350/actions/runs/21743953259

I've also added a commit to our dev branch to - hopefully - fix this. I'm not sure it's the problem but my current measuring setup is ... suboptimal. I'd appreciate you repeating your test with this build if you can!

[bablokb]

Thanks, I will test it. Although I currently have the first very preliminary CircuitPython version running on the board... There is no powman support in CP, but I have a fork that implements it and I am eager to see if this works with the Badger2350.

[bablokb]

The change does not make any difference 😭️

My setup: Nordic PPK2 with a custom made picoblade-to-DuPont-socket cable:

[bablokb]

The only difference in both code paths I found is a call to i2c_disable(), which undos some settings done by powman_init().

[Gadgetoid]

Oh wow, that's a better setup than mine - which is a Tufty with an INA3221 breakout connected to the QWST connector 😆

i2c_disable() could well be the culprit, I'll chuck another commit on to make sure that's the same in both paths.

[bablokb]

I also have an INA3221 setup:

If I would do it again, I wouldn't use the breakout, but put the INA directly onto a single PCB with all the connectors. I also have a firmware that logs the data to serial, UDP or to a SD-card for later analysis. But for everything below 1mA the PPK2 is the better solution (but it is current only, no voltage).

[Gadgetoid]

put the INA directly onto a single PCB with all the connectors

We really ought to do this as an add-on to Tufty/Badger/Blinky and have a current measurement app. If for no other reason than I would use it all the time 😆

Update pushed!

[bablokb]

I'll test this afternoon and report back.

[bablokb]

The newest version fixed it. I am now at 167µA in both cases.

I don't understand why. The POWER_EN pin is explicitly set to floating in powman_init() and i2c_disable() then was just calling gpio_init() which should not tinker with the pulls?!