My PC is intermittently not starting when I press the start button. Before I take it to the repair shop, I thought I would solicit ideas.
My thoughts:
Any thoughts? All wisdom welcome.
Could you clarify what you mean by "not starting"?
i.e. do you men you press the button and absolutely nothing happens, no LEDs, no fans spin no signs of life at all?
Or do you mean it starts booting but then gets stuck at some point in the process? Anything displayed on screen? Any error messages etc?
Does it beep? If so check out PC beep codes, they report certain errors. For instance one long followed by two short beeps signals a video card error. Mike
Is the PC permanently connected to the mains - ie have you turned it off at the wall or had a power cut recently? I've come across some PSUs which get into a funny state which is cleared by removing the mains power for a few minutes (and pressing the PC's power-on button to discharge the PSU's internal capacitors).
As I understand it, a PC's power switch connects straight through to the line on the rectangular Molex power connector (from the power-switch header connector via the motherboard to the Molex). So if the power switch is working, you *should* always be able to turn on the PSU. The PSU may turn itself off immediately if there is a fault (eg short to ground on the +5,
+12 or -12 V rails). It may be worth listening for the PSU fan starting and stopping - or look at it.You could check the connection from the power switch to the motherboard header. It's usually got a multi-coloured twisted cable. See
The PSU can start even if the CMOS battery is dead or there's no HDD (let alone an HDD with a non-booting Windows). So if it;s not getting that far, it's a fairly fundamental problem. I would suspect poor contacts in the power switch, which normally make contact as you press the button but occasionally fail to. The switch only needs to make momentary contact to
*start* the PSU; there's no requirement for the switch to remain connected all the time that the PC is turned on - indeed that would probably make the PC turn off again!Scott snipped-for-privacy@gefion.myzen.co.uk> posted
When this happened to mine, I swapped in a spare PSU and that cured it. I bought the spare from a PC repair shop for a tenner (it was obviously second-hand).
Before touching anything in the PC, earth your hands by touching the PC case (assuming it's plugged into the mains) to discharge any static which can damage the electronics of the PC.
While you've got the back off the PC, check also the case fans and the CPU fan to see if they start spinning - or failing that do they even twitch.
And, yes, listen for beep codes which can indicate a fault with the memory, the display etc. Unfortunately the codes vary from one motherboard to another, but anything other than a single short beep (everything OK) is cause for concern.
If the PSU/case/CPU fans spin OK, look for any messages on the screen - you usually get white-on-black messages from the BIOS (the code that is executed even before Windows or Linux starts to boot) and then the normal images for Windows/Linux as booting proceeds.
Yes, if you can borrow a PSU, it's worth a try. Make sure the replacement PSU has the same additional power connectors (in addition to the Molex): some have extra power leads terminating in square connectors, often plugged into the motherboard close to the CPU and maybe directly onto the graphics card if it's plug-in rather than motherboard circuitry.
A sign of PSU capacitors being on the way out is it being grumpy to start, only starting after it's been turned off or unplugged for a while.
It isn't. The line on the Molex power connector is a simple 'ground to switch on', whereas the power button tells software (ACPI) that the PC should do a safe shutdown. If you hold the power button down eventually the embedded controller on the motherboard will decide you mean 'force power removal' and deassert the Molex PWRON line. But it's not a straight through connection.
You can remove the Molex connector from the motherboard and short the PWRON line to ground, if you want to test just the PSU, and probe the voltages. The PSU fan may or may not run - it may be on a temperature control loop and only run when the PSU is hot.
Both the connector and wires can be black, or other colours :-) But it's the biggest connector coming out the PSU (a 24 way Mini-Fit Jr).
Theo
The following is a useful link:
You do often need a minimum load attached for a PSU to start up, I usually attach a HDD to a 4 pin molex connector or to a known good mobo.
PS, the multimeter video while useful, doesn't test the PSU under a full electrical load so attaching a full load or using a dummy load and then measuring voltages is often worthwhile.
DO be careful in shorting out the PS-ON or PWRON wire to GND as you dont want to have a short across a high current capable voltage rail. The bestthat will happen is a PSU shutdown and teh worst is either a loud bang or a sudden release of magic smoke!
No.
The first observation to guide your analysis, is what happens when the PC is running, and you push the button on the front.
If the OS is running, the button must be depressed for a full four seconds, before the ATX supply goes off. This is to prevent a running PC from receiving a dirty shutdown when the cat bumps against the switch. It takes a dedicated push on the switch, to activate emergency shutdown. 4 seconds worth.
When you sit in the BIOS, the switch response is not armed yet. In the BIOS, pushing the button for 20 milliseconds is enough to turn off the PC. That's a "real" emergency response, suited to actual emergencies.
Obviously the switch is not magical. Let's draw a hypothetical diagram for it. The switch is conditioned by a logic block. The logic block is powered by the supervisory rail, which is +5VSB.
+5VSB is running, when the fans are not spinning. It's a separate PSU block doing that, making +5VSB. +5VSB | +--------+-------+ PS_ON#, 20/24 pin cable +---------+ Front ------| Logic block |------------------------->| ATX PSU | switch | Runs off +5VSB | | | momentary ON +----------------+ +---------+The front switch is momentary action. Whereas PS_ON# stays asserted the whole time the PC is on.
__________ ________________ 4+ sec emerg off __________ Front# |___| |_____________________| __________ __________ PS_ON# |________________________________________|
The "#" symbol means active low.
You need +5VSB to be running, to start the PC. No logic block power, equals nothing to pull down the open collector PS_ON# signal.
Shorting PS_ON# to ground, will cause the ATX fans to spin and the PC to start (bypassing the logic block). Again, the PSU will not start, if for any reason the supervisory rail is not available. The reason you can ground PS_ON# is because the logic type is Open Collector and not a regular totem pole output.
(Typically, we ground PS_ON# when the PSU is out of the PC, and we're testing it on the bench. I have a switch for this, that plugs right into the 24 pin connector. The wires have connectors on the end with the right shape for the Molex.)
Now, let's plug two Apple iPads into the PC, to start charging the iPads (USB ports run off +5VSB). Enough current can be draw while charging them, to cause the +5VSB to switch off (OCP) inside the ATX supply. And this will prevent the front button from turning on the PC.
To "reset" an ATX supply, switch OFF at the back, wait 60 seconds, switch ON at the back. Now press the front switch and try again.
And while it does not happen on all models, a few models seem to be sensitive to VBAT. If the CR2032 is completely flat (exactly zero volts), a few PCs will not start when that happens. My two PCs in the room here, both run with a flat CR2032. There is *no* info in a datasheet, admitting to the existence of the "zero volt no start" logic. Probably less than 1% of PC designs, do stupid stuff like that.
For more timing diagrams, try here. There are three versions of this, and this is a relatively recent one:
Some info on PC power is here. This shows the connectors, where you can find PS_ON# and so on. Maybe some day, the author of this site will start adding the new PSU type details. Now that they made a lot of the wires in PC black in colour, we've lost the convenience of colour references for signal type. This means you have to be a good "pin number counter", to find stuff. I liked the old way, with the properly coloured wire in it.
Mine did this a few months ago. I unplugged the USB hub I was using and all was back to normal. The fault was repeatable so I threw the hub away.
This is a picture of a dummy load (on a RED and BLACK) and the switch for PS_ON# on GREEN and BLACK.
The supply in this picture, is an older 20 pin unit.
The playtool site (and others) will have 20 pin and 24 pin pinouts.
[Picture]The red package, is a package of male pins that are soldered to the ends of the power resistors. They fit the Molex shape.
You can use ATX extension cables instead, and connect your power resistors to an ATX extension you've cut in half, to make a wire harness. The only problem with ATX extension cables, is you can no longer buy them with the right wire colours (for easy verification while soldering).
So that's an example of preparing to bench test a supply. The bat handle switch is my ON and OFF.
Paul
It's the first of these conditions - the blue light inside the switch did not even come on.
Now that has been an issue, as I disconnected the mains to install a UPS. It was after a delay that the PC returned to life (on pressing the switch).
Thanks. I'll look inside.
It may be total nonsense but I always remove my shoes and socks as well.
There's also this kind of thing for making connections:
In message snipped-for-privacy@4ax.com, Scott snipped-for-privacy@gefion.myzen.co.uk> writes
Do this even if the PC is NOT connected to the mains. The important thing is to ensure that there is zero voltage potential between you and the PC. To be really sure to be sure, you can get a conductive strap - one end to your wrist, and the other to the PC metalwork. Failing that, try to keep some bare part of your body in contact with the PC.
A sweaty foot on the PC is perfect!
That is a very neat adapter. I might just buy one!
John
Yes I tend to have one hand on the metal chassis, and re-earth myself to it if I move away from the PC and return to it.
I once fell foul of the fact that my PC was earthed. I was unplugging the TV aerial from a USB digital TV tuner, and I had one hand on the PC and the other on the metal aerial plug. When they were separated, I received a nasty jolt - not a full-blown mains shock but a shock that was enough to make me not want to repeat it as part of tracking down where the voltage was coming from. My TV was connected to the aerial amplifier and then to the aerial cable which fed other aerial sockets around teh house via a splitter. The TV was also connected to the VCR by aerial and phono audio, and by audio to my hifi. So there were a lot of culprits. I tracked it down to the TV which applied about 80 V (with no load) between aerial ground and mains earth; this reduced to about 50 V when I added a large resistance to simulate hand-to-hand resistance of my body as if I was touching TV earth to mains earth.
And when I complained to the manufacturer, they said it was to be expected. This was a CRT TV bought in 2000, so not an ancient live-chassis TV.
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.