Want to take a stab at my failure to POST (computer) problem?

Factual Questions
1

This is probably too technical to be appropriate for GQ… I’ve also posted to enthusiast forums without too much result so I figured I’d give it a stab.

Hardware is:
Corsair 520hx psu, antec 900 case, DFI tr2s dark (p35) motherboard, 8800 GTS video, 2x2gb ddr2 ram, core 2 duo.

I had a running, fully functional computer when I accidentally pulled the power cord out from the PSU when I was untangling wires.

I plugged it back in, at which point the fans/lights all powered up (I didn’t hit the power button at this point, it just went on by itself), but there was no POST. I held the power button down to shut it off, then turned it back on - the fans/lights all came on (including the loud graphics card fan at 100%), except the 200mm fan on top which seemed like it was struggling to move (movingly slightly back and forth).

So I turned it off and left it alone for a few minutes, then tried to turn the power on again. All lights/fans came on, including the 200mm fan for about 6 sceonds. Then the whole thing shut down for about 3 seconds, and then came back on again on its own and stayed on until I turned it off.

Since then, when I turn it on it either does the 6 seconds on, 3 seconds off, back on indefintiely thing, or it comes on and stays on the first time

When it does this there’s no post beep or any sort of noise except the fans and I think the hard drives spin up but I’m not positive. The monitors won’t even click out of standby mode.

I cleared the CMOS, tried unplugging parts 1 by 1 until I had the bare minimum with no change. I unconnected and reconnected everything, I put the motherboard outside the case in case there was a short, reseated the CPU, every little thing I could think of. Nothing.

I plugged my PSU into my old computer - the new computer uses a 24 pin ATX power connector and 8 pin 12v, and the old one uses a 20 pin and 4 pin 12v, so I’m not using exactly the same connectors on the PSU, but I booted the old computer off the new PSU successfully, went into the voltage monitors and they were all normal. This suggests the PSU is okay - but there’s a chance it’s failing in a way that powers the old computer but not the new one (that requires extra pins, a PCI-E power connector, etc).

The motherboard seemed the most likely culprit so I replaced it with a GA-EP45-UD3R. It still doesn’t work, but it has some new weird symptoms and now I get beep codes.

First, the CPU fan, in the “CPUFAN” socket, does not power up when the system powers up. It starts either 12 or 25 seconds after boot, depending on the hardware configuration. The GPU fan and PSU fans start right as the power switch is pushed.

And sometimes the system will instantly power off when I hit the power button, and sometimes it will require me to hold it for 5 seconds. What does it mean, that sometimes it’s reaching different stages of booting?

If I run with nothing but PSU and CPU, no video, no ram, no HDs, etc, it powers on, continous short beeps at the same interval for 11 seconds (long enough for 8.5 beeps), then it powers off for 2 seconds. At which point it powers itself back on, runs for 11 seconds (with 8.5 beeps), and powers itself on, then restarts itself, etc.

“Continuous beeping” in the motherboard manual indicates a power failure.

If I plug the video card in (along with the PCI-E power connector) but still nothing else, it powers up, beeps 3 times, powers off, powers itself back up, beeps 3 times, repeat. The motherboard manual does not indicate what three short beeps means. I don’t think it’s the “continuous beeping” above, because it distinctly beeps 3 times, stops, and then powers off.

When I then stick in 1 ram stick in addition to the video card, it powered on for 5 seconds, powdered off, powered on again, and then roughly 15 seconds later gave off 1 long beep, and 2 short beeps (I think, it actually sounded like 1 long beep, and then 1 short beep where the volume dipped in the middle of the short beep, I think it’s meant to be 2 beeps).

The manual says 1 long, 2 short beeps (What it probably is) is a gpu/monitor error. 1 long, 1 short (which it maybe is) is a memory or motherboard error. It then powered off as soon as I hit the power button.

I rebooted with the same hardware configuration and this time it didn’t power off first, it just beeped at 15 seconds in. Even with the same configuration as last attempt, now it took 5 seconds of holding the power button to turn it off, it didn’t do it instantly.

Then I tried removing the video card again and switching the ram to another slot. This time it powered up, 1 long 2 short beeps after about 8 seconds (which makes sense since the video card is missing), the CPU fan came on a few seconds after that at about 12 seconds, and it instnatly powered off when I hit the power button.

I put the video card back on, with the ram the same as above, and now after about 22 seconds it gave 1 long, 2 short beeps. 3 seconds later, at 25 seconds after power on, the CPU fan came on. It instantly powered off when I hit the button.

I then moved the ram to the third slot over, and it powered on - but no beeps this time. After 25 seconds, the CPU fan kicked on.

Whenever I had the video card in, I had it plugged into a monitor - and it never clicked out of standby mode.

I am now completely perplexed. So I came here hoping that someone would give me some idea, any idea, what’s going on, or at least what I can try next.

TO add: On a normal boot back when it worked, my video card would spin up to 100% fan when I hit the power button for about 2-3 seconds until it booted up its own bios and got its fan profile and lowered it to like 40%, so you’d hit the power button and get LOUD…2 seconds… quiet. Now it just stays on full blast, which I assume means the video card isn’t getting to the point where it consults its bios. I’m not sure what that means, but I thought it was worth adding.

I’m working on getting good replacement parts to test it with, but anyone I know locally has different hardware that’s not compatable… I’ve got a friend mailing me an old video card to test that. I can’t afford at the moment to just keep buying new parts and hope I hit the right one. So does anyone have any idea what to do?

Another question I can’t get a straight answer to: Does the presence of beep codes indicate that my CPU is functional? That is - if you had a functional system and removed the CPU and tried to boot, would it beep? Or would it be silent because the CPU is necesary to get to the stage of POSTing where it would beep?

2

Have you tried the other memory stick?

3

Dead CPU.

4

Yeah, I’ve been trying one memory stick at a time, while alternating sticks and memory slots. Sometimes it changes the behavior in weird ways (like hitting the system power button will be instant off one time, but hold-5-seconds off the next time).

As far as dead CPU - would the board even beep if the CPU were dead? Does it have enough processing capability on its own to get to that stage of diagnostic?

5

I see this problem is actually over two weeks old and on multiple sites. Time for some fresh blood. Have you overclocked anything?

6

Yeah, everything was overclocked but stable. Resetting the CMOS, and then going to a new motherboard would’ve reset all that. I don’t think that’s related - the problem clearly seems electrical.

7

Overclocking stresses components so it is relevant. I’m not lecturing you about the evils of overclocking either. I do it too.

Have you currently set the BIOS to be not overclocked?

Is this the motherboard? LANPARTY UT P35-T2R I want to download the manual.

When you say pulled the power are you saying the 110v plug to the PSU or a connector to the motherboard?

8

None of the components were stressed - the GPU was at default voltage and the CPU was overvolted by 5 millivolts. I had very good cooling so they were under less stress than stock parts, and they were a healthy margin under what I could push them to. The ram was above the default FSB for the type of cpu but under the rated speed, stock voltage. The northbridge was also stock voltage.

The power cable between the wall and the PSU was pulled out of the back of the PSU. When I plugged it back in, the computer came on by itself immediately - which I suspect may have been some sort of surge or something.

The original motherboard (when the system first failed) was a DFI P35 DK T2RS and the replacement motherboard that I got is a Gigabyte GA-EP45-UD3R. Both boards have similar (but slightly different) problems. Notably, the DFI board wouldn’t put out beep codes, but it did spin up the CPU fan on power up, whereas the Gigabyte board has been giving beep codes at random times and has the weird delayed fan spin up issue.

9

I’m going to ask about the Gigabyte board. Nothing will be about the other.

Does it make it to the point where any text comes up for the BIOS boot?

10

It powered on when you plugged it in without pressing the power button, much of the odd behavior seems to revolve around the power button. My first guess would be a sticky button causing the momentary switch to sometimes stay closed.

11

The board turning on without pressing the power button could be an ACPI function. The behavior after AC power loss can be set to go to S0 or S5 (ON or OFF, respectively).

As for your problem, it seems like the 12V rail to the CPU is flaky. Under certain loads, it’s dropping out. I would try a new PS.

>>And sometimes the system will instantly power off when I hit the power button, and sometimes it will require me to hold it for 5 seconds. What does it mean, that sometimes it’s reaching different stages of booting?

Yes, that means it has booted to a different stage. The “Southbridge” or possibly the Super I/O controller handles that 4 second shut down feature.

>>Another question I can’t get a straight answer to: Does the presence of beep codes indicate that my CPU is functional? That is - if you had a functional system and removed the CPU and tried to boot, would it beep? Or would it be silent because the CPU is necesary to get to the stage of POSTing where it would beep?

Generally, a functional CPU is required to get beep codes, unless the board has some sort of independent microcontroller.

I think that your 12V rail is not stable - sometimes it works longer than other times, so you are seeing strange symptoms.

(I’ll come to Las Vegas for on-site trouble shooting for an airplane ticket… :slight_smile: )

12

No, the monitor never drops out of standby mode - never any boot stage that results in video.

The power button functions normally most of the time, it just powered on by itself the first time I plugged it back in. I’ve run it outside the case without the case power switch (using the motherboard power button) so that’s not likely an issue.

I thought this may be a possibility. That the components that power my old computer (seperate 4 pin 12v rail, no pci-e connector) may be fine but there may be a problem with the components that run the new system that I can’t test now.

Interesting. I’ll try to make more note of when it does and doesn’t power down instantly to find the pattern.

Is this common among mid end consumer boards?

If I could afford it, I’d probably just buy my way out of this mess :stuck_out_tongue:

My friend is mailing me his old PCI-E video card, so that’s what I’ll try next. The fact that I sometimes get the 1 long, 2 short beep sequence may mean it’s fried. If that doesn’t work, I’ll probably work on replacing the PSU.

13

8 yr BIOS engineer, here.

I second everything that Iggins said and only wish to add to it.

Beep Codes: I’ve worked with DFI boards and I can’t imagine that this board would have a special microcontroller to handle the beep codes. It’s the processor running BIOS code and making beeps with the piezoelectric speaker (probably looks like a fat black cylinder).

The “4 second” power button press is part of a feature that’s an industry standard. It’s meant to allow you to press the button for a short time and allow the system to turn itself off, or if it’s really hosed still have the ability to cut its power completely. The feature gets turned on partway through boot (usually near the end of POST), so an immediate turn-off means POST didn’t get very far. That’s all.

“Flaky” and inconsistent behavior with the same hardware setup is the classic “power supply” problem. Whether this means the PSU itself or a Voltage Regulator Module (VRM) on the board frying, or even a cap or trace, it generally means that something isn’t getting the power it needs in the quantity or quality it needs.

You tested the PSU on another board, but the board and CPU were older and didn’t require as much from it. A “broken” PSU may simply be marginal… it’s working and able to handle a significant amount of load, but not enough to do what you’re trying to do. In this case, it may well be that 12V rail.

PSUs are very common things to have fail. Sometimes with no special event precipitating said failure. Yanking a power cord shouldn’t cause a problem like this, but it’s also not the “accepted” or “normal” shut-down procedure, so may not have been heavily stress-tested. Consider also what may have happened if it were only “partially” or inconsistently powered (off & on) for even just a few milliseconds.

An alternate possibility is memory error. Memory is also notoriously fickle (in fact, it must be configured differently on EVERY BOOT because of possible changes in humidity, ambient (electrical) noise, temperature, etc). It also requires a lot of power and power regulation (VRMs). And if it weren’t working correctly, you’d see the kind of errors you’re seeing (it seems to get through memory tests sometimes, not others, but even when it does nothing works quite right). If your memory is flaky, it may pass initial tests and then crap out when actually needed.

I’d say it could also be the memory controller for the same reasons as the memory itself, but you replaced the Mobo, so you replaced the Northbridge/MCH already (unlike AMD processors, this is not a part of the CPU). So this is out.

The fact that you’re not getting video isn’t surprising… video is actually a somewhat difficult thing that requires many things be done before it. I usually comes halfway through the BIOS POST sequence. All this means is that you’re not getting that far. Among other things, it requires that memory be working properly.

If you want better certainty in understanding what the BIOS is telling you, you’ll need to get yourself a port 80 reader. This is typically a PCI card, typically sold in computer parts stores, and typically pretty cheap (it’s just a pair of hex-display LEDs and PCI interpretation stuff, after all). Used is more than fine. If you do this, you’ll never have to worry about “is that 1 long and 2 short or 1 long and 1 short and a hiccup” ever again. Instead, you’ll see a code, such as “EF” displayed, and you can look up what this code means based on whose BIOS it is and who made the memory initialization code (which sometimes isn’t the BIOS vendor, but the memory controller maker!). All this said, you may not want to pay even for a used Port 80 card unless you plan on experiencing more problems than just this – it’s not useful for anything other than diagnosis.

Lastly, regarding you hitting the button to turn it on, and it does something then turns itself off and then back on: this is actually somewhat expected behavior. It’s only SUPPOSED to happen when you change your hardware around, but it’s obvious that the system is confused at this point. The reason it does this is that there are times when the BIOS is detecting components (chiefly memory), realizes that the current system is set to parameters that do not work with those components (chiefly FSB speed), and then realizes that it cannot change the settings without powering down (changing FSB speed on the fly = bad). So it saves the settings that should work with all components in some sort of NVRAM (like CMOS, or more commonly a part of the bus controller itself), reboots itself (with an “off” period to make sure caps clear), and then it “should work”. Since the state is already saved in NVRAM, it shouldn’t need to do this again until you change hardware to something incompatible with these settings.

14

That’s a lot of good info, thanks. I got a video card mailed to me that should be here monday, so that’s the option I’m exploring first. If that fails, PSU will likely be the next thing I try.

It’s a pain in the ass - none of the people I know locally are PC gamers and have ancient, non-gaming, and/or AMD computers so I can’t come by much in the way of spare parts. I’ve thought about trying to find a local computer shop who’d let me test my stuff on their hardware, but I’m not even sure if those sorts of things exist anymore, how I’d find one, or what they’d charge for that sort of service.

15

The PSU is the first component I would say is likely bad. There are other possible bad components, like a bad switch.

Like I asked do you get to to see any text before it dies?

16

Nope, no video at all - the monitor doesn’t get a feed to switch out of standby mode.

17

This is a Bios boot order.

  1. CPU must be good or the board does nothing but instantly hang.
  2. Early chip set initiation.
  3. Memory detection. Bad memory will hang here.
  4. Bios loads into RAM.
  5. Keyboard controller initialized.
  6. VGA Bios start. Text will start appearing.

There is no need to go farther with the sequence for you.

On the issue of a possible bad switch:

  1. Disconnect the reset button from the motherboard.
  2. Try to power up.

If that failed:

  1. Remove the power switch connector from the motherboard.
  2. Connect the reset switch to the motherboard power up pins using it as the power switch.
18

I’ve done both those things. My old motherboard has power and reset switches built onto the board, and I completely disconnected it from the case (including the power and reset switches) for testing for a short.

19

That’s a good POST description. And it’s also a good idea about the switch. I, too, have had switches fail. Obviously, switch that flickers = PSU that flickers = Bad Stuff ™.

I’d guess it’s not as likely as the PSU itself being the culprit, but it’s both far cheaper to check and far easier to fix if so.

EDIT: Whoops, you beat me in answering him. :slight_smile:

20

Now that the the cost nothing option is done, I say try the new card when it arrives and then a new PSU.