I think it's a failing boot drive, but . . .

[britechguy]

. . . if anyone else has other ideas . . .

Symptoms: Sometimes this PC doesn't boot up. Sometimes, the mouse and fans power on and all the keyboard will light up but beyond that, nothing. There is nothing on the monitor. Other times when it does decide to boot up, it just stays at the at the boot menu (MSI machine) and then the progress indicator keeps spinning for around 10 seconds and then it just freezes.

Because there's always power, it doesn't strike me as PSU related. Because it seems to get stuck "at random points" when booting, that strikes me as the system boot drive in the process of failing.

Other options I may be overlooking?

 

[gadgetfixup]

Boot off Win PE stick and check HD from there. Usually just a SMART check will give you evidence if the drive is failing.

 

[britechguy]

@gadgetfixup

Thanks. Does it strike you that there are any other obvious (but clearly not obvious to me) things I should also be suspecting even if they're secondary, tertiary, etc.?

 

[Markverhyden]

In my experience a failing hard drive very rarely, and I really do mean RARELY, prevents a machine from passing POST or getting past the initial BIOS splash screen. I'd unplug anything, including addon cards, pull the CMOS battery for 30 minutes. Then power up. If you get to the boot device failure screen then I'd only plug in the drive. My experience this type of symptom is usually a failing MB.

 

[Larry Sabo]

BIOS up-to-date? CMOS battery depleted?

 

[frase]

Have you tested your RAM, or used different modules just to rule out. Insert one only - then two etc as well.

 

[nlinecomputers]

. . . if anyone else has other ideas . . .

Symptoms: Sometimes this PC doesn't boot up. Sometimes, the mouse and fans power on and all the keyboard will light up but beyond that, nothing. There is nothing on the monitor. Other times when it does decide to boot up, it just stays at the at the boot menu (MSI machine) and then the progress indicator keeps spinning for around 10 seconds and then it just freezes.

Because there's always power, it doesn't strike me as PSU related. Because it seems to get stuck "at random points" when booting, that strikes me as the system boot drive in the process of failing.

Other options I may be overlooking?

Because PSUs have multiple rails at different voltages you certainly can have power but not proper voltages and that can cause all the symptoms here. Replace the PSU. It’s a quick easy test. Otherwise like @Markverhyden I’m leaning toward bad mobo. HDD should be pulled and tested with manufacturer diagnostics on a known good system.

 

[nlinecomputers]

BIOS up-to-date? CMOS battery depleted?

This. Must be something in the air as I have had a wave of cmos battery failures of late. I almost never see them and suddenly in past two months I’ve had 5. All Dell optiplexs.

 

[britechguy]

Replace the PSU. It’s a quick easy test.

I do not consider replacing a PSU as an initial diagnostic either quick or easy. If it comes to that, it'll happen, but it's certainly not anything I would do straight out of the shoot in a situation like this.

My most common experience with PSU failures is just that - complete failures.

 

[nlinecomputers]

I do not consider replacing a PSU as an initial diagnostic either quick or easy. If it comes to that, it'll happen, but it's certainly not anything I would do straight out of the shoot in a situation like this.

My most common experience with PSU failures is just that - complete failures.

30 years of doing this says you are flat wrong. I’ve several times replaced PSUs that tested good using PSU testers and DVM and fixed PCs doing. Enough that swapping out the PSU is SOP in situations like this.

 

[GTP]

Because PSUs have multiple rails at different voltages you certainly can have power but not proper voltages and that can cause all the symptoms here. Replace the PSU

This ^.
Or just image the drive and replace it. They are so cheap now.

30 years of doing this says you are flat wrong. I’ve several times replaced PSUs that tested good using PSU testers and DVM and fixed PCs doing. Enough that swapping out the PSU is SOP in situations like this.

I have to agree 100% with @nlinecomputers here.
I've had so many situations where the PSU was the problem.
As mentioned - it's quick and easy to replace for test.

 

[NJW]

I do not consider replacing a PSU as an initial diagnostic either quick or easy. If it comes to that, it'll happen, but it's certainly not anything I would do straight out of the shoot in a situation like this.

It's both quick and easy. You don't have to install it, just resting on the open case is fine. Connect motherboard, CPU and boot drive. Switch on.

 

[Sky-Knight]

If a machine ever fails to post, while lights are on and fans are spinning. That's what I refer to as the "lights are on but noone is home" problem. Which clearly indicates a PSU that has a 3.3v rail that's dipping below 3v. Because that's what happens. This rail is notorious for softening over time, and very much does cause intermittent fault. Replace the PSU, if that doesn't sort it out the mainboard is suspect. I wouldn't suspect drive in these conditions at all.

 

[gadgetfixup]

We've recently seen a few drives keep a computer from posting. Plugging the suspect drive into our test rig only to have our test rig hang on post. Unplug the suspect drive and test rig boots again.

 

[Sky-Knight]

We've recently seen a few drives keep a computer from posting. Plugging the suspect drive into our test rig only to have our test rig hang on post. Unplug the suspect drive and test rig boots again.

If you look closer you'll find it's not faulting on POST. It's faulting on BOOT in these cases. Fast boot in the EFI makes this a VERY hard line to diagnose sometimes though, because we don't have that nice BEEP in most rigs to tell us POST is complete.

Though if you're tossing the disk into another unit and having it hang... that's... rare. It does happen! I've seen it too, but dang that's rare. My bench unit has a PCIe SATA card that I use, not to mention the NVME/M.2 SATA card for this reason. The expansion cards fire up later in the boot cycle so I wind up with a bench machine that's hung up on device detection in an obvious way. Helps shorten that specific part of the process.

 

[britechguy]

Thanks to all. I guess once I have the machine in hand I'll test the drive first, and if that indicates no issue, try a "PSU on the side" diagnostic next. That will be a 2-stepper, though, as I'll have to order the PSU as I don't keep anything in stock.

 

[Sky-Knight]

@britechguy Literally the only stock I keep is a single 500w eVGA gold PSU and a collection of ATX to whatever adapters so I have that thing there for testing / quick replacement.

But this desert home of mine... it consumes power supplies! I used to go through a supply a week, but over time that's dwindled down to about 2 a year. Modern power supplies are usually 120v / 240v autoswitching, which means the components can handle the voltages of our grid even when it spikes. So supply faults are vastly less common than they used to be, but... they do still happen.

In my experience the 3.3v going below 3v on a PSU is the 2nd most common machine hardware fault, 2nd only to platter disk fault.

 

[britechguy]

I'm not trying to argue with anyone's experience(s) here, and value them being shared, but that doesn't change the fact that those who've had frequent issues related to PSUs have personal experiences that are diametrically opposed to my own. And I've been in this business (though not directly as a repair tech) since 1985. PSU failures have been relatively rare, and when they've occurred on machines I've dealt with, they've always been "deader 'un a doornail."

 

[Larry Sabo]

It takes maybe 10-15 minutes to open the case, disconnect the PSU cables and connect up the cables from a test PSU. Collectively, we've probably spent 10 times that long providing suggestions.

 

[britechguy]

OK, I'm perfectly aware that doing a PSU diagnostic (if you have the PSU) is quick. But the original statement was replace it, which to me means full pull of the old and in with a new, and that's not particularly quick. It's also, in my opinion (if we mean replace), a "throw parts at it" approach. Using a different PSU "on the side" as a diagnostic is a different thing entirely where that's being used to determine whether an actual replacement is warranted.

But, far more information has come to light subsequent to a "probing conversation" with the client. The machine is less than 2 years old and he replaced (upgraded) the PSU around 8 months ago (with a high-end Corsair unit), so I'm disinclined to believe we have a PSU problem. I'm even more disinclined now that he's admitted that he's been fooling with BIOS and playing with XMP profiles (including messing with RAM speed).

This is now, in my opinion, almost certain to be an issue with UEFI/BIOS, and a complete reset to factory the best way forward, at least initially.

I haven't had a client, recently anyway, who has been one of these tech strip-tease artists where the full picture is never presented and you get small flashes of information that you have to put together. And this is really the kind of client I hate dealing with because they tend to hose things by playing in areas they had ought never enter.