Please register or login. There are 0 registered and 1245 anonymous users currently online. Current bandwidth usage: 326.30 kbit/s December 03 - 07:37pm EST 
Hardware Analysis
      
Forums Product Prices
  Contents 
 
 

  Latest Topics 
 

More >>
 

    
 
 

  You Are Here: 
 
/ Forums / Motherboards /
 

  my computer won't boot up or post 
 
 Author 
 Date Written 
 Tools 
Micah Vervoort Aug 11, 2012, 05:45pm EDT Report Abuse
The power seems to be running okay. Fans are all spinning and I can hear the computer working. But it will not post or boot up. It was running just fine and has been for weeks. Then suddenly during reboot it would not boot up or post. Also, not able to turn off the computer now using the power button. Have to unplug it to get it to turn off. I am desperate for help!!


Want to enjoy fewer advertisements and more features? Click here to become a Hardware Analysis registered user.
john albrich Aug 11, 2012, 08:14pm EDT Report Abuse
>> Re: my computer won't boot up or post
.
Enter
system debug
in the upper right hand corner search bar of any HWA page.

Micah Vervoort Aug 11, 2012, 09:30pm EDT Report Abuse
>> Re: my computer won't boot up or post
Thanx for the suggestion John!

So I stripped it down to the basics (PSU, MB, CPU, Vid Card, RAM, display cable and AC cable) and now the computer won't sustain power for more than a few seconds. It starts to power up, the fans start to spin but within 3 or 4 seconds it powers down again, not getting even remotely close to full speed. I'm guessing this means the problem lies with my PSU?!?

john albrich Aug 11, 2012, 11:06pm EDT Report Abuse
>> Re: my computer won't boot up or post
.
Whenever working on exposed electronics, even if everything is "known-good" and working, ALWAYS wear safety glasses and exercise appropriate electrical safety protocols.


A bad PSU is certainly possible. But, since you're not even getting to the point where it would POST if it could, there may be a problem with other part(s). At this point, we're not even looking for a display to initialize, but the ability of the PSU to power up and stay powered-up.

So (as always following power removal and ESD handling practices) remove everything but the motherboard, CPU+heatsink, the motherboard power-switch (front-panel "power" switch), and the PSU.

Connect AC power and switch on the motherboard.

If the PSU still fails to power-up and stay powered-up, then the primary problem is isolated to the PSU, the motherboard, or the CPU+heatsink. Note: a bad CPU, bad CPU+heatsink, or bad motherboard could overload the PSU forcing a shutdown, or cause the motherboard to shutdown the PSU because the CPU is overheating*.

If however, that powered-up and stayed powered-up, then the problem is most likely one of those remaining components you removed. In which case you can try testing them one at a time until/if the failure re-appears. Secondarily, it may be a bad connector/slot on the motherboard OR a short-circuit that's somehow caused when you install a given component into the motherboard (e.g. sometimes a connector or wire can be flexed "just enough" to cause a short-circuit when an item is installed, and it's not visually obvious. Unfortunately, sometimes the failure becomes intermittent because of the changes introduced by the physical actions of installing/uninstalling parts, making it even harder to diagnose.)

To further isolate the CPU+heatsink as bad...

Removing the CPU from the motherboard may require first very carefully removing the heatsink. It depends on the accessibility of the CPU socket lever. But, ideally you can remove the entire CPU+heatsink by using the CPU socket lever. On some motherboards it's very easy while on others it's almost impossible to access the lever with the heatsink still attached to the CPU. You may wish to do that now and retest with the CPU+heatsink removed from the motherboard. If the PSU+motherboard continues to fail the CPU may be good, but you also now know the PSU and/or the motherboard are/is bad.

I won't go into the details and cautions if you have to deal with in-situ CPU heatsink removal (if you decide to go that route). It is risky and could result in catastrophic physical damage to the CPU and/or the motherboard. You can research that on the net.


Note: if you have or can obtain a "go/no-go" PSU tester, that can help ID a catastrophic problem with the PSU, and may eliminate the need to remove the CPU as part of the isolation method. Simple "go/no-go" PSU testers run about US$15. If you buy one, make sure it will handle all the connectors on your PSU. Read the user reviews. Such simple testers will NOT identify all possible problems with a PSU, but if it does reveal a "bad" PSU condition (no voltage/overvoltage/undervoltage), then it's a definite fail. This can be a cheap and relatively easy way to reduce the need to "easter-egg" to locate faulty components, and of course it can be used in the future as well.

Otherwise to test a PSU you need a voltmeter, some way to put a load on the PSU, and a way to emulate the front-panel "power" switch (this is commonly called the "paper-clip" method).

I would NOT try testing a possibly bad PSU on a different, known-good motherboard unless I considered that motherboard "expendable".

However, while it generally won't damage a known-good modern high-quality PSU to test a possibly bad motherboard+CPU, a short-circuit or overload caused by a bad motherboard or CPU should simply cause the PSU to safely shutdown without damage to the PSU. But here again, rather than risk a known-good PSU (and exploding motherboard parts) I prefer to use a simple PSU tester and/or a voltmeter to test (both DC and for AC ripple levels).

*In usually rare cases, it's possible in a previously working CPU+heatsink assembly, that the heatsink is no longer properly thermally "connected" to the CPU. When that happens the CPU heats-up within seconds and can cause temporary or permanent failures. With some CPUs and motherboards, this kind of failure can result in permanent damage and from that point on also can cause the system to shutdown within seconds...similar to the failure you reported.


edit to add:
more info re: CPU/heatsink failures and interactions
edit to add:
more info re: nature of possible failures due to non-mobo/CPU/PSU components/intermittents

Micah Vervoort Aug 12, 2012, 11:39am EDT Report Abuse
>> Re: my computer won't boot up or post
I have done as suggested. Stripped down to CPU, MB and PSU. Turned on, still would not sustain power. Removed CPU and Hetsink and the computer now powers up and sustains power. Does this mean it is the CPU that is "bad" and needs replacing? Or are there further tests I need to do?

john albrich Aug 12, 2012, 03:07pm EDT Report Abuse
>> Re: my computer won't boot up or post
.
Sorry, it's still not a yes/no answer at this point, but at least it's down to these 3 components (hoping no other components were damaged if it was a PSU failure).

Yes, it could be the CPU. However it's by no means absolute...

It could still be a bad motherboard CPU socket or a motherboard short that only occurs when the CPU is installed.

There have been reports of socket connector damage due to overheating of CPU pin-connectors of a faulty socket (not an overheating CPU). Socket damage might result in unpredictable responses, including system shut-down. Visually inspect the socket connectors under magnification, and if possible inspect the bottom side of the motherboard under the socket looking for obvious signs of overheating or other damage.

Also motherboard flexing due to latching the heatsink assembly in place can put a lot of stress on the motherboard, which could aggravate an underlying motherboard problem. The problem may be inside the motherboard and not visible to any inspection. Even the weight of some heatsinks+fans is enough to cause significant flexing of cheap or poorly supported motherboards, which could cause internal problems or create a short between a point on the back of the motherboard and the computer case.

And it could even still be the PSU, as the shutdown might be caused by the added, but proper, power load of a "good" CPU. The faulty PSU shuts-down because it can't supply enough power to the CPU and the PSU voltage on that rail drops more than it should...triggering a PSU self shut-down.

Depending on finances and the risks you're willing to take, there are different ways you can go at this point. If you can get your hands on a really cheap CPU (usually the "lowest" CPU the motherboard supports) that would help isolate whether the CPU is the problem without too much wasted money if it turns out to NOT be a bad CPU. Similarly, you can obtain a low-cost PSU (but NEVER use a dirt cheap, low-quality PSU), and/or you may have to try a cheap motherboard that supports your CPU.

"Easter-egging" is not without risk:
Remember that any time you test a suspected bad component by using it with known-good "test" components, there is a finite risk of damaging the known-good components. Generally, a bad or failing PSU (especially a dirt-cheap low-quality PSU) is more likely to damage a known-good CPU or motherboard than a bad CPU or bad motherboard is likely to damage a known-good PSU. A bad CPU or a bad motherboard have a mutual chance to damage each other if one of them isn't already damaged.

A relatively inexpensive PSU tester might yet uncover a major PSU problem without risking more expensive "test" components. For example, if you test a possibly bad PSU with a known-good CPU, it might not only have damaged the original CPU, but damage the CPU you just bought to test it. But if the PSU problem is load-sensitive, a simple tester may not reveal the problem with a simple tester.

Micah Vervoort Aug 12, 2012, 04:48pm EDT Report Abuse
>> Re: my computer won't boot up or post
Thanx for all your suggestions John! Unfortunately for now I don't have the financial resources to try and figure this out. Feel free to mark this topic as closed.

john albrich Aug 12, 2012, 06:46pm EDT Report Abuse
>> Re: my computer won't boot up or post
.
If you can borrow a decent voltmeter, you may be able to at least do a quick and dirty check on most of the PSU D.C. voltages, without any added cost. This would at least provide some indication of how risky it would be to connect a known-good motherboard and/or CPU to your current PSU.

To put at least some load on the PSU (which is needed to keep it running and provide correctly stabilized output voltages if it's working properly) you can plug it into the motherboard without the CPU. You've already reported that configuration keeps the PSU running. Again, be sure to follow the 100% power removal and ESD protocols when handling the hardware to keep it from being even more damaged...and again, follow appropriate safety protocols including wearing safety glasses. The PSU can easily output a LOT of power, and an accidental short-circuit can melt connections and tools (like voltmeter probe tips) which can blast molten metal into your eyes. One can't over-emphasize hardware and personal safety to the neophyte who has just started learning how to debug computers.

Arm yourself with pin-out diagrams of the connectors, and check the voltages on various connectors.

For example, a limited and easy check of the main +5 and +12 voltages is obtained by connecting the ground lead of the meter to one of the two Black wires on one of the 4-pin power connectors. The Yellow wire is +12, and the Red is +5. But, as the PSU may have multiple power "rails" for a given voltage, you may want to check more connectors...hence the pin-out diagrams.

If the voltages "check out" in this configuration, it's a good indication the PSU is probably operating properly.

You can search for simply
psu pin-outs
and will get a list of a number of sites that provide the pin-out information for the various connectors of a PSU.

If you learn a bit more about PSU voltages and voltage tolerances, you can also use the AC setting on the meter to look for "ripple" that could take a given PSU output voltage out-of-spec and also cause problems, including damage to components. For example, let's say you measure a +12V pin with a DC reading of +12.3vdc. That is within spec (12.0V +/- 5%). You could leave it there, but if you want to honest, the meter also has a measurement accuracy tolerance which needs to be considered as well (meter calibration and measurement error are entire lessons in themselves). Many people forget that part, and on some voltmeters measurement error can be a rather large value. For example, the meter might have an accuracy of +/- 5% of the "scale" on which it is set. Let's assume this is a good quality meter with +/- 1% on the "20VDC" scale. On that scale setting, that would be an error range of +/- 0.2v. That means if (on that scale setting) you measure X volts, the real voltage may be somewhere from X-0.2v to X+0.2v. So, to be absolutely "safe" in considering the maximum value, you have to add 0.2V to that 12.3V reading. At +12.5v you're still within that 5% spec. Then you measure the AC ripple and find that it reads 0.8vac peak-to-peak (for simplicity we assume it is symmetrical about the DC component (0.4v above and 0.4v below 12.5v). Thus, the absolute maximum voltage should be considered to be 12.3(dc reading)+0.2(dc scale max error)+0.4ac ripple component = 12.9V max peak value (we'll also assume the ac scale measurement tolerance for such a low ac component is negligible compared to the other values...and ignore it here). So it's reasonably possible the +12V pin maximum value is exceeded by 0.3V at least part of the time. I've actually come across cheap PSUs that had AC "ripple" values of several volts on top of one or more DC values. A similar calculus is used to determine if the lowest measured value is higher than 12.0v - 5%=11.4v. PSU Voltage Tolerances:
http://pcsupport.about.com/od/insidethepc/a/power-supply-volta...erance.htm


(side note: the digital DC voltages reported by system monitoring software (and even digital multimeters) are sometimes quite inaccurate at the absolute level. In other words, a measured value reported as 1.234 volts can't be reasonably claimed to be actually 1.234 volts. Many people suffer from a form of "digital mania", wherein simply because a digital number is provided to 3 or 4 decimal places, that it not only must be an accurate reading, but that it must be that accurate...even though there are no data to support such an (almost always invalid) assumption.)



 

    
 
 

  Topic Tools 
 
RSS UpdatesRSS Updates
 

  Related Articles 
 
 

  Newsletter 
 
A weekly newsletter featuring an editorial and a roundup of the latest articles, news and other interesting topics.

Please enter your email address below and click Subscribe.