Are lightning bolt and overheat indicators enabled in CM3?

[jpers]

Hello everyone,

I've just assembled the prototype device, consisting of Computing Module 3 Lite, CM3 development kit, 2 cameras, LiPo battery pack with china-made regulator (set to 5.15V).

During the testing using HDMI monitor II was kinda surprised that there was not one instance where lightning bolt would appear, although we had constant problems with this when using Raspberry3 devices (we power them from industrial 5V brick PSUs, and for some time no wire gauge was enough to prevent voltage drops during high load). So:

1) Are the lightning bolt and overheat indicators enabled in Compute Module 3? My firmware is apparently quite recent.

Code: Select all

pi@raspberrypi:~ $ /opt/vc/bin/vcgencmd version
Jun  7 2018 15:37:30
Copyright (c) 2012 Broadcom
version 4800f08a139d6ca1c5ecbee345ea6682e2160881 (clean) (release)

2) If not, what voltage exactly do I need to observe to get the same funcionality?

Code: Select all

vcgencmd

can display many voltages, which one and at which threshold causes the lightning bolt?

[6by9]

Overheat should be enabled by default.

Lightning bolt: https://github.com/raspberrypi/firmware ... .dts#L1531

Code: Select all

            pin_define@POWER_LOW {
               type = "absent";
};

So it's not been configured to have a GPIO that is connected to the appropriate external low voltage detection circuit (eg U4 on https://www.raspberrypi.org/documentati ... educed.pdf. The LED and FET is optional).
Note that the Pi3B+ polls the PMIC over I2C, so it too has no GPIO assigned. There is no option for doing that on the CM3.

vcgencmd can only measure the voltages within the SoC, which are all generated by your carrier board. The low voltage LED and lightning bolt are detecting the 5V rail dropping below USB specs, and that rail never actually enters the SoC.

[jpers]

vcgencmd can only measure the voltages within the SoC, which are all generated by your carrier board. The low voltage LED and lightning bolt are detecting the 5V rail dropping below USB specs, and that rail never actually enters the SoC.

Ok, fair enough (thanks for the explanation). Given that lightning bolt starts appearing somewhere between 4.5 and 5V, and that SoC runs on 3.3V, then the voltage drop to e.g. 4.5 V should not be a concern for compute module itself, right? If it is not detectable, then 3.3V regulator does its job, I assume.

That also means that the voltage drop that would cause problem with SD card in ordinary Rpi3, does not affect the integrity of the built in eMMC in CM3, right?

But, on the other hand, if using CM3 lite with development board (or any kind of extension board that is powered by 5V), and using externally mounted SD card for boot and storage, the same problems my appear with SD card due to 5V undervoltage/brownout, just the compute module has no way of detecting this, right?

[6by9]

Ok, fair enough (thanks for the explanation). Given that lightning bolt starts appearing somewhere between 4.5 and 5V, and that SoC runs on 3.3V, then the voltage drop to e.g. 4.5 V should not be a concern for compute module itself, right? If it is not detectable, then 3.3V regulator does its job, I assume.

That also means that the voltage drop that would cause problem with SD card in ordinary Rpi3, does not affect the integrity of the built in eMMC in CM3, right?

But, on the other hand, if using CM3 lite with development board (or any kind of extension board that is powered by 5V), and using externally mounted SD card for boot and storage, the same problems my appear with SD card due to 5V undervoltage/brownout, just the compute module has no way of detecting this, right?

The main issue on the Pi is that USB peripherals (including the USB hub and ethernet) brown out and give random results. There were people on here seeing how long they could take the voltage and keep the Pi running.
AFAIK The SD card runs at 3.3V or lower as it's driven from GPIO bank 2 which is not 5V tolerant (same as GPIO bank 0 or 1).