ATX turning on at boot with setting disabled

[terabyte12]

I have both the "Turn on ATX on Boot" and "Follow USB Power" settings disabled yet when power is first attached, and the controller boots the ATX Supply is turned on.

Controller version 1.2.0

Power Toggle via the Board Button, Case Power Switch, and Web Management works as expected.

[TheGuyDanish]

Thanks for the report, I've been looking into reproducing this behaviour but haven't been able to. Code-wise there's only four places where the power supply can be turned on:

• The USB follow handler
• The power button handler
• The on-boot function
• The web-based /power_toggle endpoint

With both the on-boot and follow options turned off, that'd suggest to me the power button is the most likely trigger. Is the power button maybe being shorted? Or are the power switch pins tied together otherwise?

[terabyte12]

I checked and I do not think the power sw is being shorted. The switch works perfectly to turn on/off after the boot up. I'm using the official SuperMicro Chassis front panel ribbon to break out cable. I did notice that simply plugging/un-plugging the PS SW header triggers on/off. I also am not able to get any of the front panel leds to illuminate with the PS LED header. I briefly got one led to light (the Over Heat LED) using the header but no longer. Perhaps my chassis Front Panel is suspect.

I'll try probing with a meter tomorrow.

[rubylaser]

I checked and I do not think the power sw is being shorted. The switch works perfectly to turn on/off after the boot up. I'm using the official SuperMicro Chassis front panel ribbon to break out cable. I did notice that simply plugging/un-plugging the PS SW header triggers on/off. I also am not able to get any of the front panel leds to illuminate with the PS LED header. I briefly got one led to light (the Over Heat LED) using the header but no longer. Perhaps my chassis Front Panel is suspect.

I'll try probing with a meter tomorrow.

I have the same behavior in my Supermicro 826 case with the LEDs not working and removing the switch header causing a power off.

[terabyte12]

Did more troubleshooting.

• The Power SW header leads are definitely not shorted (continuity test on multi-meter).
• The Power LED Header on the OpenJBOD is providing 3.3v when powered on.
• Trying to test Front Panel LEDs with Multi-Meter in Diode Mode showed open.
  • Either all LEDs are bad, or this is a faulty way to test them

Then I discovered that even with the Power SW and LED Headers disconnected the Board Still Powered on when AC was supplied!

I dismounted the OpenJBOD from the case (i.e. free floating in air) and the problem went away. As long as the OpenJBOD is not screwed down it works as expected. Power remains off when AC is supplied and the setting "Turn on ATX on Boot" is disabled.

Still no luck with LEDs while the board is not screwed down.

I tried taller standoffs just in case the pre-installed ones were too short with no change.

Testing between the chassis and the Power SW header shows continuity and triggers a power on.
Not sure if this is just a SuperMicro Chassis issue or not.

Should I source Nylon/Plastic Non-Conductive standoffs?

[natankeddem]

Going to post this here just to make sure everyone is on the same page about pinouts.
OpenJBOD:

Supermicro 20Pin (I think):

[TheGuyDanish]

@terabyte12 @rubylaser I'd love if you would be able to get pictures of how your boards are connected when the power on happens. I'm assuming you're either not using the big bulky Supermicro header or breaking it out into the base pins? Just so I can try and get a better understanding.

Should I source Nylon/Plastic Non-Conductive standoffs?

While this would make it non-conductive and thus be a workaround, I think it'd be better to figure out what's causing it to exhibit the behaviour. At least for now.

[terabyte12]

I am using the official SuperMicro breakout/split cable which has each connector labeled.

https://store.supermicro.com/supermicro-15cm-16-pin-front-control-panel-split-extension-cable-cbl-0084l.html

[TheGuyDanish]

Interesting! I'd never seen that cable before. In that case, I'd think just hooking up the Power SW header to the PSW row on the OpenJBOD should work just fine unless the power switch is doing something different to just being a switch that closes the signal between the two pins.

[natankeddem]

Testing between the chassis and the Power SW header shows continuity

Are you saying the pwr sw on your chassis has continuity to your chassis at all times? I looked at an 815 with a 20 pin and an 846 with a 16pin connector. Neither showed either leg of pin1 or pin2 connected to the chassis.

[terabyte12]

Testing between the chassis and the Power SW header shows continuity

Are you saying the pwr sw on your chassis has continuity to your chassis at all times? I looked at an 815 with a 20 pin and an 846 with a 16pin connector. Neither showed either leg of pin1 or pin2 connected to the chassis.

No. From the base of the chassis to the OpenJBOD PWR SW Header (No front panel connected).

[terabyte12]

Interesting! I'd never seen that cable before. In that case, I'd think just hooking up the Power SW header to the PSW row on the OpenJBOD should work just fine unless the power switch is doing something different to just being a switch that closes the signal between the two pins.

My thoughts as well.

[natankeddem]

Testing between the chassis and the Power SW header shows continuity

Are you saying the pwr sw on your chassis has continuity to your chassis at all times? I looked at an 815 with a 20 pin and an 846 with a 16pin connector. Neither showed either leg of pin1 or pin2 connected to the chassis.

No. From the base of the chassis to the OpenJBOD PWR SW Header (No front panel connected).

The pin near the edge of the board is tied to ground which should run to the chassis. The pin on the text side is the signal input into the proceesor, this should not tie to ground.

[terabyte12]

Testing between the chassis and the Power SW header shows continuity

Are you saying the pwr sw on your chassis has continuity to your chassis at all times? I looked at an 815 with a 20 pin and an 846 with a 16pin connector. Neither showed either leg of pin1 or pin2 connected to the chassis.

No. From the base of the chassis to the OpenJBOD PWR SW Header (No front panel connected).

The pin near the edge of the board is tied to ground which should run to the chassis. The pin on the text side is the signal input into the proceesor, this should not tie to ground.

Yes, that is what I am seeing.

But even with no front panel leads connected at all the system powers on as soon as AC is supplied. I'll work on getting photos/video.

[terabyte12]

Here are some photos/video:

PWR SW Connected:
https://github.com/user-attachments/assets/151d2d02-10a3-421c-8994-e2a0e8a98336

PWR SW Disconnected:
https://github.com/user-attachments/assets/fa9d599c-3c56-4441-ab91-d579fa7f6524

If I unmount the board and hold it up then the power on when AC is supplied does not happen. Not enough hands / didn't feel like unmounting the board to do a video (Let me know if you do indeed want that).

[natankeddem]

I was able to recreate the issue on my 846 chassis. Could I bother either of you to test out this fix I developed?
https://github.com/natankeddem/software/actions/runs/12151204836/artifacts/2270862699