PC doesn't power down after holding power button for 4 seconds

[HTC]

The only time I've experienced the power button not working after a hard crash is when DRAM is unstable.

Memtest+ 5.01 ran for 2 full passes (that took over 5 hours): zero errors.

I wanted to take out 1 of the kit's sticks to get it's specs for post #48 and, instead of shutting down i restarted by mistake. Tried to shut down before starting windows but the power button didn't work so i had to shut down the PSU to do it.

It did this with previous board too, and now does with this one: the problem is not the board, then. So it's either the button or the cord leading the button's signal to the board, right?

 

[Bill_Bright]

The damn power button still doesn't shut down sometimes after holding it for 4+ seconds: i'm guessing it's an issue with the button or cable leading to the board since it's the 2nd board it does this with.

Understand that "4 second" delay is set in the BIOS. It can be changed to Instant off or Reset.

But since this is the 2nd motherboard, I would look at the switch. Assuming the two wires are connected to the proper 2 pins on the motherboard, it is likely the switch. You need to gain access to the back side of the front panel and inspect it. The switches can break or the clamp holding the switch in place can break or come loose. If the clamp/bracket holding the switch in place is loose, it can sometime be bent back in place. But if broken, what we have done many times here is pull the two wires from the motherboard and connect the two wires from the case's Reset button to the motherboard pins. Then use the Reset button as the power switch from then on.

 

[HTC]

Understand that "4 second" delay is set in the BIOS. It can be changed to Instant off or Reset.

But since this is the 2nd motherboard, I would look at the switch. Assuming the two wires are connected to the proper 2 pins on the motherboard, it is likely the switch. You need to gain access to the back side of the front panel and inspect it. The switches can break or the clamp holding the switch in place can break or come loose. If the clamp/bracket holding the switch in place is loose, it can sometime be bent back in place. But if broken, what we have done many times here is pull the two wires from the motherboard and connect the two wires from the case's Reset button to the motherboard pins. Then use the Reset button as the power switch from then on.

The odd thing is that the power button itself works just fine as it turns on the PC every time: it's just when holding for 4+ seconds that doesn't shut down.

Will check the BIOS to make sure which setting that 4 second thingy is @.

 

[OSdevr]

Anyone heard of this?

Anyone ever heard of this one not working?

I have seen this before on a rather old computer. I don't remember the CPU or chipset (it was one of the first with PCI) but I found out online that the chipset had a bug that affected PCI busmastering which was exactly what I was attempting to do with it. Holding down the power button did not shut it down (but the PC beeper started wailing after a few seconds). I could only turn it off by cutting power. Lucky for me that was more a curiosity than a concern since I was working with a PC booter and didn't have any operating system or hard drives to corrupt. Without going into the technical details of what I think happened, I believe the firmware was too busy dealing with PCI errors that the message to power off was drowned out.

 

[OSdevr]

Memtest+ 5.01 ran for 2 full passes (that took over 5 hours): zero errors.

I'd recommend running Memtest+ for a bit longer than that!

 

[Bill_Bright]

More passes is better, but 2 is the generally the minimum and normally plenty.

That said, you should use MemTest86. Memtest86+ 5.01 is old (2013) and obsolete and it has been years since last updated. MemTest86 is under constant development with regular updates (most recently, July 26, 2017 - my sister's birthday!) to stay current with the latest tests, RAM types (like DDR4) and protocols.

 

[OSdevr]

More passes is better, but 2 is the generally the minimum and normally plenty.

Normally but not always. I always run Memtest for a good 24 hrs, it took just under that to find a problem for me once (good stick, wrong voltage ).

 

[HTC]

More passes is better, but 2 is the generally the minimum and normally plenty.

That said, you should use MemTest86. Memtest86+ 5.01 is old (2013) and obsolete and it has been years since last updated. MemTest86 is under constant development with regular updates (most recently, July 26, 2017 - my sister's birthday!) to stay current with the latest tests, RAM types (like DDR4) and protocols.

I made an Ubuntu installation USB 16.04 LTS stick: the memtest version i used was the one that was in the booting options of that USB stick. Perhaps i mistook it's name but i believe it was memtest+ 5.01 but it could have been memtest86+ 5.01: not sure. Just downloaded the latest USB version to make later.

There seems to be another problem, though. Tried to hook up another HDD and the PC didn't boot: the 4 second power button thing worked and the PC shut down. Removed the newly added HDD and tried to boot again, and again it didn't boot but, this time, the 4 second power button thing didn't work.

Also: can't seem to find that BIOS 4 second setting. Can't test atm since her monitor is also the living room TV, and is being used. Can't hook up my monitor to her PC because her APU doesn't support neither display port nor HDMI.

 

[Bill_Bright]

Normally but not always. I always run Memtest for a good 24 hrs

Yeah, I generally recommend several passes or even over night. But two passes ensures one full pass with the full battery tests at maximum iterations. One pass is not enough (unless it finds problems then the RAM is bad) because it is not as complete as subsequent passes as it is designed to check for and identify the most common problems.

. Perhaps i mistook it's name but i believe it was memtest+ 5.01 but it could have been memtest86+ 5.01: not sure.

There are several with similar names that all based on much of the old core technology. But only the one from PassMark I linked to supports the latest RAM and protocols. So if you did not download the PassMark version, you are using an obsolete program.

That said, no software based memory tester is conclusive. If they report errors, the RAM is bad. But RAM can test good and still fail when installed, or when paired with other RAM. To conclusively test your RAM, you need to use sophisticated and very expensive test equipment, like this $4,400 RAMCheck LX Memory Tester - swap in known good RAM and see what happens.

 

[cornemuse]

This has happened to me several times w/ diff computers/os's. With XP (been a while, tho I still use it) when attempting to install (dont remember which) programs, comp would freeze totally, had to unplug it. With UBUNTU, when clicking on some links (again, dont remember where), a total freeze, unplug comp.
Strange;y enough, the mouse pointer still moved on the screen, the buttons just didnt work, , ,

But they allus worked fine on re-powering, , , ,

 

[HTC]

Yeah, I generally recommend several passes or even over night. But two passes ensures one full pass with the full battery tests at maximum iterations. One pass is not enough (unless it finds problems then the RAM is bad) because it is not as complete as subsequent passes as it is designed to check for and identify the most common problems.
There are several with similar names that all based on much of the old core technology. But only the one from PassMark I linked to supports the latest RAM and protocols. So if you did not download the PassMark version, you are using an obsolete program.

That said, no software based memory tester is conclusive. If they report errors, the RAM is bad. But RAM can test good and still fail when installed, or when paired with other RAM. To conclusively test your RAM, you need to use sophisticated and very expensive test equipment, like this $4,400 RAMCheck LX Memory Tester - swap in known good RAM and see what happens.

Figured out why i had ran version 5.01 (it was indeed 5.01): it was the non UEFI version. Now running the latest 7.4 UEFI version and i plan to leave it running all night.

 

[OSdevr]

That shouldn't make a difference, but let us know if it finds any problems.

 

[Bill_Bright]

it was the non UEFI version. Now running the latest 7.4 UEFI version

Okay but again, MemTest86 and MemTest86+ are still two different programs by two different companies even though their names are very similar.

In any case, hopefully your RAM tests good, though not sure that will get you any closer to the uncovering the actual problem.

 

[Bones]

In reference to shutdown problems:
See this all the time with mine and I know certain boards will act like this, others will power down either by holding the switch or simply pressing the button as some have said.
I'll also say sometimes even with boards that will normally shut off with a press of the switch they can hang as originally described if I have it cranked up hard for OC'ing, switching off the PSU is the only way they will cutoff at that point. All I do is simply restart the system, tell it to start Windows normally and it goes.
The one I'm using as a DD (Stock settings) wants to hang at times but overall it's never been anything to worry about, still works fine as is.

A weak CMOS battery can cause weird stuff to happen, looked back and didn't see anything about the battery being a possible suspect (And I apologize if I missed it) but yes, a weak battery can cause issues with bootup, shutdowns, and just overall freakyness - This is a common issue with certain board models such as an Abit AN7 as an example.
I know that's a really older board but it's still relevant - With an AN7 the Uguru chip on the board eats batteries like a starving animal and these are prone to the weird side of behaviour when a battery starts giving up. Popping in a fresh one makes the weirdness go away and that's good in it's case.
Also know if the battery in any board gets too weak it's possible to corrupt the BIOS in the board regardless of model, make sure if you haven't already to see if the battery is either good or dying. Using a common voltmeter you'd want to see about 3.0v's or a little more for a good battery but if it drops below 2.9v's or so, then it's best to replace it instead of risking a borked BIOS making the board itself unuseable.

 

[Bill_Bright]

The problem with troubleshooting shutdown problems is many different causes give the same symptom. That makes it that much harder to troubleshoot - especially remotely via forums. These are cases I wish I had on my bench.

 

[Bones]

The problem with troubleshooting shutdown problems is many different causes give the same symptom. That makes it that much harder to troubleshoot - especially remotely via forums. These are cases I wish I had on my bench.

Know what you mean - I know my hardware because it's mine and I run it all the time but trying to figure out someone else's problems can be a challenge, esp if you're not there with it.
Had this happen with my REX the other night and although it normally cuts off with a press of the power button it hung and didn't want to shutdown on it's own when telling Windows to restart (Had it cranked way up there ), so after trying the button with no luck I killed the PSU and restarted it, no prob with a restart afterwards.

Still wanted to hang if I tried a standard shutdown or the power button itself but in my case it's no biggie, not a DD machine by any means - Clocked it back down and all was well.

 

[brandonwh64]

UPDATE:

The RAM didn't give any errors @ 1600.

Currently installing Windows 7's updates.

The damn power button still doesn't shut down sometimes after holding it for 4+ seconds: i'm guessing it's an issue with the button or cable leading to the board since it's the 2nd board it does this with.

Since her RAM doesn't have any problems, i'll leave it in her PC instead of having to swap coolers to be able to change the RAM.

My NZXT case I have came from the factory with a faulty power button. I took it apart and solders a lead that was poorly soldered and it works like a champ now.

 

[HTC]

Okay but again, MemTest86 and MemTest86+ are still two different programs by two different companies even though their names are very similar.

In any case, hopefully your RAM tests good, though not sure that will get you any closer to the uncovering the actual problem.

Her RAM completed the test in a bit under 11 hours: zero errors.

 

[HTC]

UPDATE

She sent me a text yesterday, while i was @ work, stating that the PC wouldn't turn on. I arrived from work way too late to do anything, but tried turning on the PC a short while ago and it booted just fine.

Since both the board and CPU were changed, wouldn't this indicate a possible PSU issue?

 

[blobster21]

Time to buy a new power button i guess ?

 

[Shambles1980]

swap the power on for the reset button and see if that changes things.

 

[Bill_Bright]

I agree with Shambles but first, follow the two wires from the back of the case's front panel power button to the two pins on the motherboard's front panel I/O header and pull those two wires from the motherboard pins. Then, with a very steady hand, carefully reach in with a very small screwdriver and short those two motherboard pins together. There is, at most, just a small +5Vsb (standby DC) voltage there so don't worry about getting shocked. You just don't want to short the +5V to the wrong pin, or scratch the motherboard with the hard, sharp screwdriver tip.

If everything is okay, when the two pins are shorted, the system boots just as if you pressed the power button. If it boots, then it is likely the power button is bad (or loose) and swapping the wires from the reset button (making the reset button the power button from now on) will get you going.

If it does not boot, then you have other problems and I would suspect the PSU since the motherboard was already replaced - this assumes that all the power connectors are securely fastened.

If it does appear to be the power button switch itself, as noted, you/she can just use the reset button from now on. But I would try to gain access to the back of the button switch for a visual inspection. Sometimes, the button switch itself just comes loose from its mounting bracket and, depending on how it mounts, can be snapped back in place, or metal prongs holding it in place can be bent back into place, or the button can be hot-glued or epoxy-glued back into place. However, in many cases, the repaired mechanism holding it back in place is never as strong as it was originally - so it may (or likely will) come loose again.

 

[eidairaman1]

Hopefully nothing is chafed either chafing causes some very weird issues