Computer turns on then off repeatedly (NOT new build)

ddawson · Sep 23, 2016

I know there have been similar posts along these lines. But the ones I saw involved new builds/upgrades. This is a computer I've been using for a couple of years. It was working fine when I shut it down. Also, I had not done anything to the internal hardware for quite a while. All of a sudden, when I tried to turn it on a few hours later, it started this. Turns on, then about half a second later, turns off; then 5 seconds after it turns off, it turns on again. Repeats the process until I hold the power button for 4 seconds, or cut power.

Motherboard: Gigabyte GA-B75M-D3V microATX.
PSU (in case it matters): Rosewill RD500-2SB

Testing I did:

■Disconnect all peripherals and the CPU (ATX 12V 2x2 connector, right?). No expansion cards to mess with. Same problem. All tests conducted w/o peripherals.
■Tried with another power cable. Same.
■Tried another, known working PSU. Same. (Note, however, it's an older one with a 20-pin connector, so might not be valid test? I have one I think is better, and also should work, but I need to dig it out. Should I try it?)
■Tried the "paperclip test": disconnect motherboard connector, short pin 16 to ground. Runs fine. Multimeter shows all rails within specs.
■Checked with multimeter while plugged into motherboard. Pin 16 high, but goes low when main power switches on. Pin 8 ("PWR_OK") goes high at this time, then drops when it switches back off. (This is how it's expected to work in relation to main power, right?)

The obvious conclusion from all this is the motherboard is at fault. But some say the PSU might still be the cause even though the voltages look fine. (It sure would be nice if it is, huh?) So, I'm not sure. I'd like some other opinions/advice before I start spending money on replacements.

One other thing: can an absent/disconnected CPU cause this symptom? That's the condition I created when pulling the 12V connector, isn't it? I ask because I want to know whether the testing I did that way is valid.

I appreciate any help you can provide.

amtseung · Sep 23, 2016

A missing CPU will cause post errors, not strange power delivery. And since the post process is within the motherboard itself and has nothing much to do with whether a CPU is present or not, I'd assume at least the motherboard is dead.

I was dealing with this myself a few weeks ago, and I tried removing every component off the motherboard one by one until I got to just the motherboard itself, and power still wouldn't go through it. My tale was rather more unfortunate, as my motherboard dying took my CPU with it. Intel confirmed it, and an autopsy showed its internal voltage regulator (damn haswell chips) took the full blow of a surge and fried itself.

Your PSU seems to be ok, but without an oscilloscope, we can't really tell if it's truly dead or not, especially when testing was not carried out while it was under significant load. You say you tried testing with another known working PSU and you got the same result, which pretty much rules out the PSU as a direct cause, but may still be indirectly related to a power delivery failure. The paperclip test will only tell you if your PSU has catastrophically failed and refuses to put any power through whatsoever.

Have you tried clearing CMOS? Have you checked the rest of your system for live/kinked wires dangling behind the motherboard? Have you tried breadboarding your system (removing the whole system from your case and running it on a cardboard box)?

Watch Jayztwocents struggle with a similar boot looping problem.

ddawson · Sep 24, 2016

My tale was rather more unfortunate, as my motherboard dying took my CPU with it. Intel confirmed it, and an autopsy showed its internal voltage regulator (damn haswell chips) took the full blow of a surge and fried itself.

Ouch. I hope that hasn't happened in my case.

Your PSU seems to be ok, but without an oscilloscope, we can't really tell if it's truly dead or not, especially when testing was not carried out while it was under significant load.

What's considered a significant load? Is having a motherboard connected not enough for that?

but may still be indirectly related to a power delivery failure.

Because it may have done something to the motherboard?

Have you tried clearing CMOS?

Ah, yes. I thought I may have left something out of my OP. I did indeed try clearing CMOS. Twice, in fact, just in case I didn't do it right the first time, or something. It doesn't seem to have helped.

Have you checked the rest of your system for live/kinked wires dangling behind the motherboard?

Yup. Nothing under it except the standoffs. The people who built this were at least able to corral cables.

Have you tried breadboarding your system (removing the whole system from your case and running it on a cardboard box)?

Now, this I had not done. I just tried it. Nope, same behavior.

I did take the opportunity to check voltages, now that I have easier access. It's a little hard to tell with a multimeter, when it samples at about 2.5 Hz, but they looked reasonable.

Watch Jayztwocents struggle with a similar boot looping problem.

Interesting. I didn't miss the fact that guy had a board from the same family as mine (EDIT: or, well, the same chipset, I guess). I really hope it's not a corrupt BIOS and it only decides to recover it after 2,573 power cycles or something. :fou:

However, that one stayed on for longer than half a second, so I'm doubtful.

amtseung · Sep 24, 2016

Just having a motherboard stuffed full of power connections doesn't mean a power supply is being loaded. In fact, there's almost no load there, because the empty motherboard cannot push power to something that does not exist. The most significant load a system faces is on boot, when the post process pretty much demands 100% power to every component connected to the power supply simultaneously, which I call the "initial surge" on boot. If your PSU can't supply that power, or the wall circuit fails to feed your PSU the power it needs, it'll repeatedly fail to boot within the first few seconds or so, which could in turn corrupt a bios and/or fry stuff in your system.

As an experiment, I once tried running my old Antec VP450 450W PSU with dual 12v/18A rails (roughly 430W combined 12v power) to an overclocked system that supposedly needed 420W at full load. It booted every time for months on end. After some fiddling, I tried running all the components off the same 12v/18A rail, and it boot looped, refusing to post, until I pulled the GPU out, and then it went back to repeatedly booting without issue. With the GPU back in, redistributing the load across both rails and replacing the PSU with a unit with a stronger single rail fixed that issue, which is why, for ease of building, I now like recommending power supplies with a strong single 12v rail.

Then again, an HD7950 bios flashed to a 7970 GHZ edition and overclocked to damn near 1.2ghz on the core and 6.8ghz memory is a black hole for power (325W average power draw, nearly 400W peak). Since then, it has been replaced by an r9 380x, but the damn Sapphire bios won't let me push the core past 1.2ghz, so now I'm looking for an unlocked bios for that too.

Back on track, your PSU is listed as having two 12v rails, one 15A and one 16A, which is even weaker than my old VP450, so as you're plugging power into components, you have to make sure the load is balanced between the two rails, for example, motherboard and cpu power off of one rail, and GPU off the other, or else you'll be pushing all that load onto one of two rails, leading to a power delivery failure. You may have fried one of your rails, but not the other, giving the illusion that you're still getting power when you stuff a multimeter into it.

My thoughts are now that your PSU and motherboard are both potentially faulty, and that you should try replacing one at a time until you find your culprit.

Oh, and also, a motherboard will refuse to post without the CPU power plugged in, with or without an actual CPU in the socket.

ddawson · Sep 25, 2016

Points taken about load. I agree that could be an issue in theory. But the thing is... when you say GPU, you're talking about a graphics card, right? Yeah, this system doesn't have one (should I have mentioned that specifically? Well, I did say no expansion cards, and I really meant it). Yeah yeah, I know. Integrated graphics sucks. You don't have to tell me that. Point is, I do know this system normally draws less than 100W (readout on UPS). So it's strange to think the PSU got overloaded in this case, unless it just decided to start dying. (I've considered adding a graphics card, though, and it's helpful to be made aware of the issue.)

That said, I went ahead and tested with that other PSU I mentioned, much better than the one I tried before. Same results. So I'm definitely going to replace the motherboard. Hopefully, that's all I'll need to replace. I'll keep the alternate PSU in case I need it. Thanks for all your advice.

I'm new to this forum. Does it make sense to mark this solved now?

amtseung · Sep 25, 2016

If you want to mark this as solved, no one will fault you, but I would personally hold off closing this thread until you can confirm the problem is fixed. If there is a future problem, like the new motherboard doesn't completely fix the problem, someone else more knowledgeable than me helping you would find it far easier to have all the information in one place.

ddawson · Oct 4, 2016

Good news! Got a replacement board today. Works just fine. I think that proves where the immediate problem lay. Thankfully, everything else, including the CPU and RAM, seems to be intact. Well, that makes sense; since it was on both a surge protector and UPS, it seems unlikely a power surge or cycling killed it.

Anyway, thanks for your support.

amtseung · Oct 5, 2016

Glad to see you've fixed it.

ddawson · Oct 5, 2016

Oh no! It's not fixed after all!

It was running fine for a while. However, I had not hooked up the optical drive, because I didn't have a usable SATA cable for it (because one of the ones on the old board is a piece of **** that absolutely refuses to come loose, but that's another issue). I got ahold of another one and hooked up the drive in question, applying power to it for the first time since this whole saga started. The very next time I tried to turn it on, it did a similar thing to before: when I press the power button, it turns on and then off immediately. Only this time, it stays off until I press power again (but stays off for at least 5 seconds—if I press power within those 5 seconds, it waits until the 5 seconds are up, then tries to turn on again).

And, of course, it persisted even after disconnecting all drives. I did get a proper powerup once immediately after disconnecting and reconnecting the 12V connector (with suspect drive not powered), but that may have been a fluke. Again, tried the spare PSU with same results. Also, later, it did stay on for a few seconds once; all other times, it turns off in about 0.3 sec (it varies just slightly).

I'm not sure what to think, except for a couple of things maybe. Could the CPU be dying after all? Would that cause this? What about the drive in question? This happened with the new board exactly when I connected it. The only other time I saw this behavior was when I forgot to plug in the 12V connector. Silly me, I thought, and plugged it in, and everything seemed fine. Then I proceeded to start using the computer, and reported all okay above. But does a disconnected 12V also cause it to behave like this? (I know you said it would refuse to POST, but I'm unclear if you meant immediate power-down or just staying on and doing nothing.)

My biggest fear with this is that the drive may have killed two boards in a row. But it could have just been a coincidence, and the CPU is the bad part (but then why only failing at power-on and never turning off after the system is booted?). I really don't know. Please, help! Thanks.

amtseung · Oct 5, 2016

Oh wow.

Maybe your disk reader is killing your motherboards. I've certainly heard of it happening, usually with cheaper units, although it's quite uncommon. Personally, I've never built a PC with a disk drive in it (they're obsolete in my book, and USB2.0 has a fast enough data rate that a USB disk reader is practical enough, and a $10 flash drive is more reusable and stores more data than a CD, DVD, or bluray).

I'm not sure what you mean by the 12V connector, since there are a few that go into various components including the CPU 4 or 8 pin connector, 20+4 pin motherboard power, 6 or 8 pin graphics card power, and 4 pin molex. If the 20+4 pin motherboard power (that fat loom that goes into the side of the motherboard) isn't fully connected, absolutely nothing will happen. If the CPU power connector is not in, you will experience the immediate shut down upon trying to start. If the graphics card connector isn't in, you'll get symptoms similar to the previous. If something requires molex but isn't powered, the computer will boot like normal, and the device will just be not detected.

If the CPU were dying, it wouldn't work period. You would forever be getting a symptom that looks like that of the missing power connector. At least, when I had dead CPU's in the past, they all behaved the same way, trying to turn on for a fraction of a second, then shutting off again.

When a motherboard refuses to POST, it'll usually do the immediate power down thing. Staying on while doing nothing usually means something died or kicked in right after post. If there is power but no post, then something else is dead, usually the bios chip (dead motherboard).

It's so hard to diagnose weird, unusual problems like this without seeing it in person. At this point, I can only think of a few things.
1.) Your optical drive is somehow faulty and internally shorted, killing your motherboards.
2.) Your power supply is dying, and is frying everything in sight when it tries to feed power to your optical drive (overload on a single rail).
3.) Your Gigabyte B75 motherboard simply isn't built well enough to handle everything you're trying to plug into it.

I want to blame your optical drive. If your computer worked with it not plugged in, don't plug it in anymore.

Edit: Bad memory will also cause boot issues, but since your computer worked fine until you plugged in that optical drive, I'm ruling it out this time.

ddawson · Oct 5, 2016

Oh wow.

Maybe your disk reader is killing your motherboards. I've certainly heard of it happening, usually with cheaper units, although it's quite uncommon. Personally, I've never built a PC with a disk drive in it (they're obsolete in my book, and USB2.0 has a fast enough data rate that a USB disk reader is practical enough, and a $10 flash drive is more reusable and stores more data than a CD, DVD, or bluray).

Well, sure, but some of us have disc collections. I do like to rip my CDs, but there are cases where I need to use (or burn) an actual disc. I guess I could start using an external drive.

I'm not sure what you mean by the 12V connector

The 2x2. That's the one that feeds CPU power, right?

If the CPU power connector is not in, you will experience the immediate shut down upon trying to start.

Which is what I was describing, right? So, it's acting as if the CPU isn't getting power (or like it's not there, maybe??). Though that's not the only possibility, is it? Also, it's unlikely to be a bad plug or socket, since I've now used two PSUs each with two boards. If anything, it would have to be something in the CPU itself, which I moved from one board to the other.

If the CPU were dying, it wouldn't work period. You would forever be getting a symptom that looks like that of the missing power connector. At least, when I had dead CPU's in the past, they all behaved the same way, trying to turn on for a fraction of a second, then shutting off again.

That matches what I'm seeing exactly. The way you write, though, it sounds as if you think I'm describing different symptoms. Maybe my post was confusing. To me, though, it sounds like you've confirmed a dead CPU as a possibility. What am I missing?

If there is power but no post, then something else is dead, usually the bios chip (dead motherboard).

I.e. staying on and doing nothing? Never experienced that.

1.) Your optical drive is somehow faulty and internally shorted, killing your motherboards.
2.) Your power supply is dying, and is frying everything in sight when it tries to feed power to your optical drive (overload on a single rail).

Both unpleasant possibilities to consider. But I should be able to deal with them from now on.

3.) Your Gigabyte B75 motherboard simply isn't built well enough to handle everything you're trying to plug into it.

Doubtful. The old one worked for years with a hard drive (later two) and the optical, without even a graphics card. Even if it were slowly dying under the strain, that wouldn't account for the replacement dying within hours (or almost instantly, depending on how you look at it).

I want to blame your optical drive. If your computer worked with it not plugged in, don't plug it in anymore.

Yeah, I hear you. Too risky at this point. I wonder if there's a way to test it safely (and not too expensively, of course

). But it might be cheaper or at least easier to replace it and forget about the old one.

Edit: Bad memory will also cause boot issues, but since your computer worked fine until you plugged in that optical drive, I'm ruling it out this time.

Agreed. I've had bad RAM before, so I know what it's like—random POST errors, applications crashing. Well, at least if it were causing complete failure now, you'd think it would have before as well.

BTW, just what is the meaning of the "homebuilt" tag? I didn't add that, did I? FYI, I didn't originally build this. It was built by a company I bought from online. (I suppose you would have liked to know earlier, though it's probably not terribly important.) That said, I'm not averse to messing around inside when I need to, as you can probably tell. (I also have a little experience actually building, just not always the inclination.) So, you could say it applies partially.

amtseung · Oct 5, 2016

There are many components that take 12v power. I listed them previously. To call that a 12v plug is like trying to explain a car engine's internals by saying "that metal thing".

CPU's don't usually just up and die when you plug an optical drive in. In fact, CPU's don't up and die at all. They're the least likely thing to fail in the entire system besides a solid metal heat sink. Something has to be pretty drastically wrong to cause a CPU to fail, and is usually associated with multiple points of failure within the system.

I don't think your CPU is dead. Your computer worked fine, and you okay'd it, and it only stopped functioning when you reconnected that optical drive. You seem rather attached to your optical drive, but I really think it's the culprit here.

To eliminate the chance of it causing issues again, I'd get a USB optical drive over an internal one (I have one myself). If anything does go really wrong, you'd fry a USB port or cable, not the entire computer in one go. I'm seeing USB 3.0 external optical drives on amazon for like $35, even though there's no way in hell an optical drive can benefit from speeds over what's available with USB 2.0. They're not that much more expensive than an internal drive, if at all. Mine was like $18 and can only read/write CD's, but it's also 6-8 years old.

If you didn't build the system from scratch, the homebuilt tag makes it misleading. It doesn't change the diagnosis at all, but it's not nice.

ddawson · Oct 5, 2016

CPU's don't usually just up and die when you plug an optical drive in. In fact, CPU's don't up and die at all. They're the least likely thing to fail in the entire system besides a solid metal heat sink. Something has to be pretty drastically wrong to cause a CPU to fail, and is usually associated with multiple points of failure within the system.

Okay, when you explain it like that, it gives me confidence. Thanks. Admittedly, I realized after writing, if it were somehow dying, there probably should have been noticeable signs for a while, when in fact there were none. And if the drive killed it, things wouldn't have been working perfectly for those few hours.

You seem rather attached to your optical drive

More like I'm attached to things USB drives can't take the place of, like occasional movies, music CDs (yeah, I'm weird like that, though I do rip them for convenience), and burning CDs for others. But I use them infrequently enough that I agree an external drive is a good idea.

but I really think it's the culprit here.

Yes, I do too, now.

If you didn't build the system from scratch, the homebuilt tag makes it misleading. It doesn't change the diagnosis at all, but it's not nice.

I removed it.

ddawson · Oct 9, 2016

Okay. Working again with motherboard #3. That optical drive will not be used anymore, and I'm using the alternate PSU just to be on the safe side. I gave it a day this time before saying anything.

Again, thanks for your advice.

amtseung · Oct 9, 2016

Let's hope it keeps on working this time.

Search

Computer turns on then off repeatedly (NOT new build)

ddawson

Honorable

amtseung

amtseung

Distinguished

ddawson

Honorable

amtseung

Distinguished

ddawson

Honorable

amtseung

Distinguished

ddawson

Honorable

amtseung

Distinguished

ddawson

Honorable

amtseung

Distinguished

ddawson

Honorable

amtseung

Distinguished

ddawson

Honorable

ddawson

Honorable

amtseung

Distinguished

TRENDING THREADS

Latest posts

Moderators online

Share this page