[sunray]

Hi

After many years of having a broken Indigo2 PSU I fixed it this weekend and the machine boots fine

But, once I connect the AC the machine boots without waiting for me to push the power in button, and also I can not poweroff the machine from Unix or via the buttom!!!

What could be the reason for that, could a faulty NVRAM cause this?

[bjames]

I have had a similar issue with my Octane. If the system fails to boot and I disconnect the power, the octanes shuts off. But as soon as I recnonnect the power cord the system reboots but fails to boot correctly. What I do to fix this is push the reset button. Most times this resets the system to boot correctly.

[sunray]

I can reset the Indigo but I can't poweroff from Unix or using the button and as soon as I connect AC the machine boots.

[weblacky]

I'm a little unclear. When you say you fixed it, do you mean you opened the power supply and fixed it, or you fixed it buy getting another power supply?

Sounds weird, but I'd start by checking your grounding wiring. I've seen some other equipment power supplies auto-turn on if the the ground is wrong. Some circuits compare signals/voltages to ground (reference). If your ground is floating at your outlet (or you use a 2-3 prong adapter), it may think you've press buttons that normally signal by grounding a voltage to 0V.

This would possibly make the PSU think the power button is ALWAYS pressed?

Let us know.

[kirikoo]

I have an O2 and next to the DALLAS chip timekeeper is a normally open jumper. If it is closed, the machine starts as soon as the AC is connected. Maybe it's something similar on the Indigo 2 ?

[sunray]

I'm a little unclear. When you say you fixed it, do you mean you opened the power supply and fixed it, or you fixed it buy getting another power supply?

Sounds weird, but I'd start by checking your grounding wiring. I've seen some other equipment power supplies auto-turn on if the the ground is wrong. Some circuits compare signals/voltages to ground (reference). If your ground is floating at your outlet (or you use a 2-3 prong adapter), it may think you've press buttons that normally signal by grounding a voltage to 0V.

This would possibly make the PSU think the power button is ALWAYS pressed?

Let us know.

I replaced a couple capacitors in the PSU so yes it was opened for 5 years I think

Will double check if anything not connected

Thanks

---

I have an O2 and next to the DALLAS chip timekeeper is a normally open jumper. If it is closed, the machine starts as soon as the AC is connected. Maybe it's something similar on the Indigo 2 ?

Will check that to

Thanks

[Intuition]

Isn't there a terminal command to shut down the system? I remember guessing this with my first O2 when my button was wonky and it worked.

shutdown? or something?

[Jacques]

Isn't there a terminal command to shut down the system? I remember guessing this with my first O2 when my button was wonky and it worked.

shutdown? or something?

try 'shutdown -y'

or if you want a short grace period

'shutdown -y -g5'

for 5 seconds grace. Or to reboot

try 'shutdown -y -i6 -g0'

[weblacky]

Since you say you replaced components, I'm going to kind of go in a different way here and say caps are not the only thing you need to check/replace in a switch mode power supply, they are the first thing, but unfortunately...not the last.

Since you're not getting random shutdowns and such, let's assume your voltage is stable and within tolerance. That leaves the HUGE amount of passive components in an Indigo2 PSU that still need to be checked. I would say you have an out of tolerance resistor (likely in your PSU startup circuit) or some other voltage divider-style circuit used as a comparator, in the middle daughter board that attaches the HV and LV side of the PSU (removable board with pin headers).

While normally resistors tend to increase resistance with age, that's not always the case, they can decrease resistance too under circumstances like moisture exposure (poor storage environment). I don't know if you have a desoldering gun, but if you do, I'd take the PSU apart enough to get the metal case off and stretch the transformer cables a little and desolder one end of every resistor you can find, then test it with a multimeter for accuracy as marked.

If you can locate what looks like a startup area you should begin your check there. The Indigo2 PSU is one of the denses SGI PSUs I've seen and it has minimal semiconductors and a lot of passive components. I think you need to check those passives. My guess would be your startup circuit is malfunctioning on the PSU board, and the management circuit likely is shorted into start-up due to resistor tolerance being off.

[sunray]

Since you say you replaced components, I'm going to kind of go in a different way here and say caps are not the only thing you need to check/replace in a switch mode power supply, they are the first thing, but unfortunately...not the last.

Since you're not getting random shutdowns and such, let's assume your voltage is stable and within tolerance. That leaves the HUGE amount of passive components in an Indigo2 PSU that still need to be checked. I would say you have an out of tolerance resistor (likely in your PSU startup circuit) or some other voltage divider-style circuit used as a comparator, in the middle daughter board that attaches the HV and LV side of the PSU (removable board with pin headers).

While normally resistors tend to increase resistance with age, that's not always the case, they can decrease resistance too under circumstances like moisture exposure (poor storage environment). I don't know if you have a desoldering gun, but if you do, I'd take the PSU apart enough to get the metal case off and stretch the transformer cables a little and desolder one end of every resistor you can find, then test it with a multimeter for accuracy as marked.

If you can locate what looks like a startup area you should begin your check there. The Indigo2 PSU is one of the denses SGI PSUs I've seen and it has minimal semiconductors and a lot of passive components. I think you need to check those passives. My guess would be your startup circuit is malfunctioning on the PSU board, and the management circuit likely is shorted into start-up due to resistor tolerance being off.

It could ofcourse be as you say but I don't thinks so, passive components usually don't behave like that I would say but I could be wrong I am no analog guru but a digital one I think

I think there is something else going on.

Is there no one that knows how the switching off is done?

---

Isn't there a terminal command to shut down the system? I remember guessing this with my first O2 when my button was wonky and it worked.

shutdown? or something?

try 'shutdown -y'

or if you want a short grace period

'shutdown -y -g5'

for 5 seconds grace. Or to reboot

try 'shutdown -y -i6 -g0'

Actually since its a SGI I have only tried via the GUI so far