HWFly Core Power Drain

[james194zt2]

Hi All,

So after suffering with extreme battery drain on my switch whether in Emunand Atmosphere OR just the Sysnand I did some digging, it appears that the HWFly modchip to me is not falling "asleep" but is staying awake the entire time the switch is either "sleeping", or awake and in use.

I am seeing a power drain of about 6-8% per hour in sleep, if I want to take it out with me I basically have to undock and fully turn it off to ensure that it does not loose all it's power.

I have done some digging by taking some measurements off HWFly whilst in both states and it appears the chip is possibly wide awake consuming power the whole time that the switch is in a powered on or asleep state.

I took test measurements off the GD32 chip itself and I am seeing it is powered up with 3.3V the whole time, I even notice sometimes that the switch is warm to the touch when in standby which again is not normal.

Can anybody shed any light on this? Is the chip genuinely asleep even though it is being supplied with 3.3V? I am not really a coder but did a little digging and there is a section in main.c that looks like the power handling, pmu_to_standbymode is in the SDK as well for the STM32/GD32 so think this is the right place.

Also looks like it is what controls the LED so is obviously triggered when the HWFly is booted up as the LED goes off.

void enter_sleep()
{
leds_set_color(0);
fpga_power_off();
while ( 1 )
pmu_to_standbymode(WFI_CMD);
}

I am assume that would be what would put the chip to sleep once it has done it's job, I can only look at it from a hardware perspective and from there it looks powered up, and no debug means I cannot see if it works.

Anybody got any ideas?

 

[thesjaakspoiler]

I am seeing a power drain of about 6-8% per hour in sleep, if I want to take it out with me I basically have to undock and fully turn it off to ensure that it does not loose all it's power.

That would be about 200mAH.
An Atmega324 uses about 100mAH.
The HWFly has a similar mpu and an fpga so this sound reasonable when it is supposed to be running.

But you're correct that the SpacecraftNX code should put the HWFly in a low power mode after it's finished.
So 2 things I can think of :
- they put some modified version of the SpacecraftNX in the HWFly that doesn't put the HWFly in the low power mode
- there is something wrong with the installation or the modchip itself, causing an unintended power drain.

 

[james194zt2]

That would be about 200mAH.
An Atmega324 uses about 100mAH.
The HWFly has a similar mpu and an fpga so this sound reasonable when it is supposed to be running.

But you're correct that the SpacecraftNX code should put the HWFly in a low power mode after it's finished.
So 2 things I can think of :
- they put some modified version of the SpacecraftNX in the HWFly that doesn't put the HWFly in the low power mode
- there is something wrong with the installation or the modchip itself, causing an unintended power drain.

My thinking is it is something to do with the firmware or the most likely cloned GD32!

What could I get wrong on the install, I tested all the continuity etc... Of the glitch flex on the CPU, it was fine and no bridges either, I have put Kapton tape under the exposed heat shield metal that is under the HWFly, and cut the back heat shielding to take pressure off it when the lid is closed, I then secured the HWFly with kapton tape to prevent movement.

I assume if it is glitching the CPU install is correct, so not sure what else I could do with the mod itself to cause it.