Pwm-controlled FAN goes out of control

[feudenfer]

Hi everyone,
just have bought a Jetson Nano (B01) kit and experiencing the following issue:
I have a case with power button and the appropriate jumper is placed on the board. The fan (Waveshare 4010-pwm-5v) is attached to the 4-wired connector. When I power on the case but the board is off, the fan starts working at the maximum speed and when I press the power button (so to power on the board) the fan lessens its speed to the minimal (the board takes over the control and the temperature is minimal, I guess. That’s ok).
I checked all the stuff with an oscilloscope and found out that:

• the fan being powered on separately (without pwm signal) provides pullup to 3v on the pwm wire and works at the maximum speed.
• if the pwm wire is pulled down, the fan works at the minimum speed (6Hz).
• the board when unpowered seems to keep this wire in a Z-state and that’s why the fan is going mad.

I assume that certain state of the pwm line should stop the fan completely, but I don’t know if the problem is with the fan or the board in my case…

Please if anyone could check on his/her correctly working Nano, what are the correct levels of the pwm wire when:

• the case is powered on, the board is not?
• the case is powered on, the board too?
• the fan is powered on separately (detached from the board)?

[kayccc]

Hope someone can share info for your reference ith same Waveshare 4010-pwm-5v fan.

[feudenfer]

Finally I have come to a decision that this problem is due to a poor schematics of the Nano devkit carrier board (Nvidia, please comment if not agree). The difference between this board and a regular ATX PC is that PC don’t provide 12V supply for pwm fans in the standby mode. The carrier board does. Its 5V rail is always on in every mode and a pwn fan has no possibility to stop (try to find a fan with no pull up on pwn line… I have failed).
The easiest solution is to pull down the pwm line with 5KOhm resistor (the board has its own 100KOhm resistor but it’s not enough) so that in the standby mode the fan would think that the controlling signal is at minimum.
The resister sinks ~280uA and overall the board consumes ~15mA in the standby mode.

This is not a discovery, I just confirmed existing solutions (ex. PWM Fan Is there a missing pull down resistor on pin4 of J15 on the B01 reference design?).