Baffling electrical connection issue

[Dr Wu]

TL;DR Summary

Mysterious electrical connection issue (domestic).

I'm faced with a puzzle. Lately my desktop PC requires me to first disconnect, and then reconnect its mains cable, before it will start. By this I mean the cable has to be physically disconnected from the mains socket, as well as disconnected from the male plug that goes into the back of the PC, and then reconnected again, before it will operate.

Simply removing and reconnecting the latter plug won't do. It has to be a full disconnection/reconnection at both ends, otherwise zilch. Neither can the problem be exclusively due to the cable itself, for I've swapped it with three others in my possession, including a brand new one that I had kept as a spare.

Obviously there's a connection issue taking place, but I can't even begin to see what the cause of it might be. Also, the PC continues to function flawlessly once switched on - no power failures, brownouts. . . whatever. In addition I have no similar issues with other electrical equipment running from the same power supply: just the PC, nothing else. Any answers?

 

[Guineafowl]

Could be a cracked solder joint inside the power supply, but it also sounds like the power supply or downstream circuitry might have developed a fault and is going into shutdown. Your unplugging resets it, and the extra time to unplug both ends allows internal capacitance to discharge and reset the controller.

I don’t know your skill level, but the best solution would be a new power supply unit - the cost of diagnostic/repair work on the existing one may well exceed the price of a new ATX supply.

 

[Baluncore]

I would shake the cable near each end to identify possible mechanical failure of the end connections.

 

[Dr Wu]

All I can add to the above is that the PC's on/off switch briefly lights up when I try to start it, but goes off again when the PC fails to fire up at all. Otherwise the switch remains lit when the PC is running normally.

Guineafowl: when you say power supply unit, I take it you mean the PC's power unit? PS. I'm afraid my electrical capabilities don't extend much beyond changing a fuse.

 

[Baluncore]

What AC mains voltage do you have, 115 Vac or 230 Vac ?
How old is the power supply, when was it purchased ?

 

[Guineafowl]

I would shake the cable near each end to identify possible mechanical failure of the end connections.

I think the OP has tried different cables with the same result. It sounds like the ATX power supply is at fault.

 

[Guineafowl]

All I can add to the above is that the PC's on/off switch briefly lights up when I try to start it, but goes off again when the PC fails to fire up at all. Otherwise the switch remains lit when the PC is running normally.

Guineafowl: when you say power supply unit, I take it you mean the PC's power unit? PS. I'm afraid my electrical capabilities don't extend much beyond changing a fuse.

Yes, the main power supply unit inside. Given your experience, I would recommend you take it into a repair shop to a) establish that the power supply is indeed the root cause, and b) replace items as needed.

 

[Vanadium 50]

(domestic).

Which is where, exactly? (And how should we guess?)

If it truly requires both ends of the power cable to be disconnected, it can't possibly be electrical. Try another power cable. If you're like me, you've accumulated a bag full.

If it's just the time it takes to do a double unplug, it's more complicated, but points to a power supply. Do you have a power supply tester? These are around $10 and will keep everyone from guessing. Also, when you open the case to test, clean the case with compressed air. It might not help, but it sure won't hurt.

 

[Keith_McClary]

clean the case with compressed air

In my experience, outdoors.

 

[Dr Wu]

If it truly requires both ends of the power cable to be disconnected, it can't possibly be electrical. Try another power cable. If you're like me, you've accumulated a bag full.

Yes, as I mentioned in my first post, I've already tried that option, and (again) no joy there.

So, all in all, it looks as if the PC's power supply unit is at fault here. Probably it needs replacing, which (for me at least) means a visit to my local computer menders. Even so, assuming the villain of the piece is the PSU, as seems likely, I still find it strange how I've had to go through the fan-dance of unplugging everything first in order to get the PC started. That I don't understand.

 

[Vanadium 50]

I still find it strange how I've had to go through the fan-dance of unplugging everything first in order to get the PC started. That I don't understand.

If it's not the cable, it's the time.

 

[DaveE]

PC power supplies never turn 100% off. There is a small power supply that operates the on/off switch. They also may have a different turn on procedure for when power is applied vs. the normal turn on command.

I would replace the power supply. If that's too expensive then add a switchable outlet strip between the computer and the wall.

 

[Baluncore]

I think the OP has tried different cables with the same result. It sounds like the ATX power supply is at fault.

When you dismiss my test suggestion you reduce the field based on your statistical guess, rather than by any rational process. The OP observation that both ends of the cable must be unplugged appears irrational. When you shake a cable you test the integrity of the connectors at both ends of the cable. It probably does not matter which cable you use during the test. Gathering available evidence costs less than replacing the supply immediately.

What AC mains voltage do you have, 115 Vac or 230 Vac ?
How old is the power supply, when was it purchased ?

The age of the supply is useful information to those who understand supplies and the ways that they fail. Has the power supply been replaced before?

 

[yungman]

It is a desktop, I assume you said cable is a straight power cable (no power supply like the laptop type) from wall to the desktop tower. Unplug both end requires more time.
One thing to try first. Disconnect from one end, wait for a minute, then plug it back in. This is to make sure you let the desktop totally power down before reconnecting.

Also, try jiggling the power cord to see whether you have intermittent connection.

 

[DaveE]

The OP observation that both ends of the cable must be unplugged appears irrational.

Yes.

I've seen this several times when troubleshooting difficult problems*. Experiments that give contradictory or nonsensical results. or measurements misinterpreted or done wrong. One of the hardest things for people to do when troubleshooting is to be skeptical of their own results. There are many times I have said something like "let's start over", "I need to see that myself", or "let try that test a different way". More often than not, difficult troubleshooting problems involve the human thought process as much as the hardware being tested.

I learned that as soon as you are confused, or think "that doesn't make sense", you need to go back and reassess the quality of your data and all of your assumptions.

*Maybe I'm wrong, but I don't think this one qualifies as difficult. Just get a new PSU.

 

[Guineafowl]

When you dismiss my test suggestion you reduce the field based on your statistical guess, rather than by any rational process. The OP observation that both ends of the cable must be unplugged appears irrational. When you shake a cable you test the integrity of the connectors at both ends of the cable. It probably does not matter which cable you use during the test. Gathering available evidence costs less than replacing the supply immediately.

No dismissal or general rudeness intended, Baluncore. Apologies if it came across that way. The perils of group remote troubleshooting.

Since swapping 3 cables in, with associated wiggling, hadn’t changed the fault, I had guessed it wasn’t the connectors. On the other hand, if either of us got called into investigate the fault, I bet the first thing we would do is have a shake of the cables.

 

[Dr Wu]

The age of the supply is useful information to those who understand supplies and the ways that they fail.

I've had the desktop for about seven years now, and it's been used almost on a daily basis, and often for long stretches of time too. So, yes, it now looks as if the power supply is on its way out. Solution: new power supply. All the same, even if this had been apparent to me from the start (which it wasn't), the weirdness of how the problem presented itself would still have prompted me to starting this thread. Thanks for all the expert advice. It's much appreciated.

 

[Guineafowl]

I've had the desktop for about seven years now, and it's been used almost on a daily basis, and often for long stretches of time too. So, yes, it now looks as if the power supply is on its way out. Solution: new power supply. All the same, even if this had been apparent to me from the start (which it wasn't), the weirdness of how the problem presented itself would still have prompted me to starting this thread. Thanks for all the expert advice. It's much appreciated.

Good luck, and let us know how you get on. I have an actual physical itch to open up the supply and fault find.

 

[Keith_McClary]

By this I mean the cable has to be physically disconnected from the mains socket, as well as disconnected from the male plug that goes into the back of the PC, and then reconnected again, before it will operate.

If you disconnect from the back of the PC first, does the PC "know" whether you disconnected from the mains?

 

[wirenut]

If you disconnect the pc end, does it still require removing the other end? If not could it be your moving a connection in the male part of the pc connection, and "fixing" it until
it either heats up and re-breaks the connection, or one of the pins in the male part may have overheated or worn away from arcing and moving the cord by unplugging and reinserting temporarily fixes it?

 

[wirenut]

Or a bad solder joint on the board in the ps at the back of the male plug.

 

[sophiecentaur]

clean the case with compressed air.

. . . . or use a vacuum cleaner.
Did you try a different mains lead?

 

[Dr Wu]

I've tried three mains leads in all, including a brand new one. Result: no difference. Meanwhile I've "solved" the problem by very slowly pressing the start button to fire up the PC, instead of giving it the usual quick dab. This seems to give the starting mechanism more time to do its stuff - and it works. I've had a 100% success rate throughout this last week. So could the fault be simply due to dirty contacts inside the start button? (Still doesn't address the leads issue, though).

 

[256bits]

If it's not the cable, it's the time.

Had a similar problem once.
If I let the PC be off too long, like over a day or so IIRC, the next time it would not start.
Forget what exactly, but I had to unplug the PC socket end and replug a few times before it would start up.
Probably a weak capacitor or other component in the PSU resulting in an incorrect voltage for other some part of the circuit.