Can a power supply only work sometimes?

[hondablaster]

My hypothesis comes from experience with my own hardware. A small few of my builds would reboot constantly without a battery, or without the jumper, or a jumper in the clear spot. I always believed that it was in an auto configuration loop. When you power it on for the first time the BIOS detects settings and saves them to CMOS this makes the boot up process faster. Some of these settings cannot boot up the first time without being configured properly so they are recorded in the CMOS and the PC is rebooted up freshly by BIOS.(like my P5QC when booted for the first time or battery is missing uses DDR2 or DDR3 so the first boot sees which slots are used and second boot actually fully runs, its about a 3 second interval)

CMOS holds and retains several settings (ram, size, timings, voltages)(HDDs and types SATA, IDE and boot order) If I left the battery out or stuck it in clear mode that process would happen back to back because upon each reboot it would forget, detect, restart, forget, detect, restart, forget. When it should have been boot, detect, restart, boot, and run the correct configurations. Keep in mind every PC is different. This is only for a PC that has a short CMOS charge time.

I also have an old Compaq Pentium 3 that if you didn't have a BIOS battery the screen would come on saying it detected RAM(does a small test) and HDD and will now reboot...... to infinity This would happen every 30 seconds because there was a RAM quick test in between. And it hold didn't hold past the 30 seconds.

CMOS simply holds the most basic configuration settings your BIOS needs to get the hardware up and going. Any changes you make to BIOS are stored in CMOS. The battery simply keeps CMOS settings alive. If your standby power is failing then that can also contribute to some of your problem.

My hypothesis simply revolves around that. But I think you found the real culprit. A CPU barley contacting can create this situation. Heat changes make it expand just enough to where the loose connection doesn't connect at all.

 

[B Trevathan]

I'm not sure how many amps it takes to kill one, but 5v certainly isn't going to drive it. Out of curiosity, I just stuck my DVM probes in my mouth and got around 200K ohms, which means 5v would drive all of 30 microamps. So even if he used his tongue to test that 5v line, it's certainly not going to kill him.

Speaking of sticking things in your mouth, have you ever touched a 9 Volt battery to your tongue, it a shocking experience. I don't recommend you do this.


I've always thought that you have a great chance of dying from low voltage shocks than the higher voltage ones. I also think that AC is more dangerous than DC, I think AC is so dangerous because of the frequency (60 Hz here in US).


From the Ohio state website:

It's The Current That Kills

Offhand it would seem that a shock of 10,000 volts would be more deadly than 100 volts. But this is not so! Individuals have been electrocuted by appliances using ordinary house currents of 110 volts and by electrical apparatus in industry using as little as 42 volts direct current. The real measure of shock's intensity lies in the amount of current (amperes) forced though the body, and not the voltage. Any electrical device used on a house wiring circuit can, under certain conditions, transmit a fatal current.

While any amount of current over 10 milliamps (0.01 amp) is capable of producing painful to severe shock, currents between 100 and 200 mA (0.1 to 0.2 amp) are lethal. Currents above 200 milliamps (0.2 amp), while producing severe burns and unconsciousness, do not usually cause death if the victim is given immediate attention. Resuscitation, consisting of artificial respiration, will usually revive the victim.



"As shown in the chart, shock is relatively more severe as the current rises. For currents above 10 milliamps, muscular contractions are so strong that the victim cannot let go of the wire that is shocking him. At values as low as 20 milliamps, breathing becomes labored, finally ceasing completely even at values below 75 milliamps."


Danger - Low Voltage

It is common knowledge that victims of high-voltage shock usually respond to artificial respiration more readily that the victims of low-voltage shock. The reason may be the merciful clamping of the heart, owing to the high current densities associated with high voltages. However, lest these details be misinterpreted, the only reasonable conclusion that can be drawn is that 75 volts are just as lethal as 750 volts.

More shocking info to read:

OSHA http://www.osha.gov/SLTC/etools/construction/electrical_incidents/eleccurrent.html
Georgia State http://hyperphysics.phy-astr.gsu.edu/hbase/electric/shock.html
Illinois University http://pchem.scs.uiuc.edu/pchemlab/electric.htm
Wikipedia http://en.wikipedia.org/wiki/Electric_shock​

 

[loavesfishescomp]

Honda, thanks for taking the time explain further. It is clearer now.

Trevathan, I will make sure I wear my bright yellow dish gloves to check the voltage. These are usually used for those laptops that are disgustingly full of organic items I wish not to touch before Windexing. Cheers.

Will post back.

 

[red12049]

You are wrong and Im not sure what your point is. And you didnt post a link or state why I was wrong. He asked why it would reboot faster. I gave an accurate hypothesis. Have you ever stuck a CMOS jumper in clear CMOS mode. Has it not rebooted repeatedly? (depending on mobo) What are you talking about.

http://www.tomshardware.com/forum/251082-30-battery-motherboard-matter

Is this ACG?

No Honda, you didn't give an accurate hypothesis. And no, this isn't ACG. And again, the page you linked to doesn't support you.

Here is the deal. A computer at my workplace would not boot up after shutting down. It would boot up if I waited at least ten minutes or more

In the old days (before BIOS's had "auto" in their settings), if you booted a machine with no or a dead CMOS battery, it would give a series of error messages, and then sit and wait. On those machines, you could go into the BIOS, enter the correct info, save it to CMOS. The machine would reboot and run fine, until you shut it down. At the next start, you'd have to do it all over again. In those days, there were quite a few different CMOS batteries, and some of them were soldered to the motherboard, so this happened with some frequency.

Nowadays, most machines with a dead/missing CMOS battery will give a date/time warning. If you bypass that, they'll continue booting, but will take a little longer because they have to "rediscover" all the peripherals attached to them.

Try it for yourself on a few different machines.

The original poster stated that the machine was rebooting every one to one and a half seconds. I'll ask you again; if the machine were rebooting that fast because of a CMOS battery, how would you ever get into CMOS to reset it?

Here's something more than just a simple link. Grab a coffee or coke, and sit down to read a bit.

http://www.google.com/#hl=en&source...ty1CgAAAKoEBU_QC-kX&pbx=1&fp=689666b5549dfcbc

Rick

ETA: PS: Setting the CMOS jumper to "clear mode" is an entirely different operation to the CMOS than having a dead or missing battery.

 

[Jay85]

My experience is when a PSU starts acting up, Just replace it, Its not that expensive and will probably save the system in the long run,

Sometimes i try taking them apart and seeing if i can find a short or something, But i always like replacing the unit altogether.

I fit 500 Watt Psus as standard now. And a good quality too, A cheap PSU can be the death of a comp. I dont think the general public realise how important this component actually is.

On PSU's ive tested especially like the Dell Factory Fitted ones, The output is only about 70% of what it says on the outside, So like a 400 Watt is actually putting out about 280 Watts, Which I think is scary.

 

[Jay85]

Also as a side note. I would change the PSU before I start trouble shooting the Motherboard or any other component, 8 out of 10 times this will fix system errors due to insufficent power.

As an example, A client came in with a Dell Emachine. The comp would sometimes boot and sometimes would just have an amber light blinking on the front.

I tested the HD Drive, Memory and everything checked out fine. I swapped out the PSU and presto.

An underpowered system can also bring about loads of errors, or especially a faulty PSU

If a system is acting up or you have a problem diagnosing or cannot pinpoint the problem exactly, just pop in a new psu. I always have a spare on my test bench. And a Bought a PSU tester on ebay for $10...It just tells you if sufficent power is being fed to all the rails. Testing with an ammetere is not as accurate coz that just tests if current is flowing. but how much is more important and you need to know that.

 

[loavesfishescomp]

Update

Update

Client has informed me that the computer is still experiencing the power on/off cycle after sitting overnight off. Power supply has arrived, new, and will install it, and let sit overnight to test when they bring it back in.

Will post back.

Great J85. I will buy a power supply tester. I didn't realize they were so inexpensive. In regards to replacing power supplies, I agree with my limited experience: For just a little more, once can get a high quality PS. I really like Corsair, and computers love clean power.

Honestly, if this power supply doesn't fix the problem, I can only attribute it to a bad Mobo, though the capacitors don't look bad on it. I haven't any other idea what else to test.

I always like to look on slickdeals.net everyday. When I find a good deal on Corsair/other, I snatch up a few.

Thank you.

 

[B Trevathan]

I was shopping at Office Depot so I got this one: Antec® ATX12V Power Supply Tester it was 19.95

Honestly, if this power supply doesn't fix the problem, I can only attribute it to a bad Mobo, though the capacitors don't look bad on it. I haven't any other idea what else to test.

Did you test the power switch? Its real easy.

Did you re-seat all the PSU connectors and check them for any damage?

Did you test the PSU by shorting the green wire to a black wire?

I'm guessing this PSU has a voltage select switch under the plastic have you test it?

PS: are you wearing your bright yellow dish gloves? LOL

 

[ATTech]

Power supply testers are virtually worthless. They don't detect slight variances in voltage that can cause issues, the only thing that they can really tell is that the power supply is completely dead. A known good power supply can tell you the same thing, as well as help diagnose issues that can be caused by a power supply that is defective, but is still providing power

 

[loavesfishescomp]

Update

Update

I tested the power supply purple wire, and it is 5.12 volts.

I wore my yellow gloves.

I don't know how to further test a power button other than pressing it and things fire up. Am I missing something? Should I be testing it when the problem has occurred?

I didn't short the power supply because it works if it's not plugged in. Am I missing something? Should I be testing it when the problem has occurred? of course!

I checked all connections to the mobo and they look fine.

I will replace the power supply and leave it plugged in over night and prey on my knees by my bed.

Running everest stability test and the CPU is running at 91C after 4 minutes?! uhhh... will shut down and check furthur


Will post back.

 

[loavesfishescomp]

Don't forget to put the plastic CPU shroud back on or you will see your chip at 99 C. Put it on while the computer is running and you'll see the temperature go down to 40 C in 3 minutes!

 

[B Trevathan]

Since this is a intermittent problem, yes you should be testing only when the problem is actually occurring.

Sorry for this being so long.

I tested the power supply purple wire, and it is 5.12 volts.

The purple wire should be within 5% of +5.0 volts.

I don't know how to further test a power button other than pressing it and things fire up. Am I missing something?

Its real easy to test the power switch, just unplug the power switch from the motherboard and connect the power switch to a multimeter (stick one probe from the multimeter to one wire of the switch and stick the other probe to the other switch wire), set the multimeter to read continuity (on the multimeter it will say OHM or have a little Omega symbol) and press the power switch, if you're using a analog multimeter you will see the needle move way over to the right when you press the power switch.

I didn't short the power supply because it works if it's not plugged in.

What?
The reason you would want to use a PSU tester or short the PSU is because it isolates it, when you do it you unplug every power supply connector, when you short it if the power supply fans come on then the odds are your PSU is good.
You can see from the picture on this 24 pin connector they are shorting pin 16 to pin 19

I checked all connections to the mobo and they look fine.

You re-seated all of the PSU connector on the motherboard and the drives?
Pics of bad connectors: http://www.technibble.com/why-you-should-avoid-cheap-power-supplies/



Some Dell power supply units are proprietary, they use a proprietary pinout arrangement on a standard ATX plug. If you try to use a regular non-Dell psu on it, you will fry the mobo, psu or both.

The main motherboard power connector (P1) looks complicated but once you learn the wire color code it's really not.

Color Code Rail
Yellow +12V
Red +5V
Orange +3.3V
Black Ground / COM
White -5V
Blue -12V

Green Power On / PS_ON#
When this green wire is grounded to a black wire it tells the the PSU that power should be delivered to the power rails (Turns your computer on, it is like pressing the power switch). When it is not grounded it tells the PSU not to deliver current to the rails (Your computer is off).

Gray Power Good / PWR_OK
The gray wire is called PWR_OK it’s a signal from the PSU to the motherboard stating the DC voltages are within normal ranges. It indicates that the +5VDC and +3.3VDC outputs are above the undervoltage thresholds of the power supply.

Purple +5V Standby / 5VSB
This purple wire gives 5 volts to the motherboard when it’s shut down, yet connected to a wall power outlet. This is why it is a good idea to unplug your computer when you work on the inside of it because it has 5 volts running into it and this 5V line is capable of delivering 2 Amps (Less than 1 Amp can kill you).

Standard 24 Pin connector
1. +3.3V
2. +3.3V
3. Ground
4. +5V
5. Ground
6. +5V
7. Ground
8. PWR_OK
9. +5VSB
10. +12V
11. +12V
12. +3.3V
13. +3.3V
14. -12V
15. Ground
16. PS_ON#
17. Ground
18. Ground
19. Ground
20. NC (Removed) -5V
21. +5V
22. +5V
23. +5V
24. Ground

Helpful Links:
http://www.pcpower.com/support/ATX_troubleshoot.htm
http://www.technibble.com/computer-technicians-103-knowing-your-power-supply/
http://www.technibble.com/computer-technicians-103-knowing-your-power-supply/2/

 

[Jay85]

If the CPU is running that hot. Maybe try removing the heatsink, Scrub off any trace of thermal grease and apply a fresh coat, also make sure its a good quality like arctic silver, just a pea size blob in the centre, Attach the heatsink and make sure its firm and making full contact,

Also maybe try change the heatsink for a better one perhaps. Make sure the fins and fan are all clean and dust them out if you need to

 

[loavesfishescomp]

Update

Update

Trevathan, Thank you for the time to educate me in full detail. I owe you a beer. EDIT: I WILL TEST THE POWER SWITCH.

After letting the computer sit over night and half the next day, the cycle occurs after I press the power button. When it is in this cycle, the power button has no effect on system. Only way to remedy is to unplug and plug back in- after that, it runs fine.

Please see the video of what is happening here.

Next on my list to do. Start unplugging cards while this is occurring, and the power button-to-mobo connection. Errrrrr...... I almost give up.

 

[loavesfishescomp]

Update

Update

Disconnected everything (cards, fans, memory, all drives) except P1 & P2 and problem still occurs.

Took out the CMOS battery, and the cycle sped up to twice as frequent for 5 seconds, then it booted up on it's own? The battery is 3.14V. I'm going to update BIOS.

EDIT: The power switch isn't that easy to test. It has a ribbon cable running from it, to another small board near the CD drives, then another ribbon cable from that board, to the mobo. So.... I'm almost ready to give it up and give it back and say, sorry, you need a new computer.

 

[B Trevathan]

I just watched the video and I was going to ask you if that was a ribbon cable running to the front panel that I saw in the video next to the blinking standby light on the motherboard? If you have a ribbon cable instead of the individual connectors that I talked about using to test the power switch then that makes it harder to test the power switch, Dell front panel connectors are mostly "Proprietary and Undocumented". Some Dell users have posted the pin outs on some models but they are not standard across all Dell models. Dell has never officially documented any of them that I know about.
You can still test the switch if you have a power switch board that connects to the motherboard by a ribbon cable, there are a few ways you can do this, you can go through the pins on the cable with a multimeter and find out which one is on the switch, or you can test where the ribbon cable connects to the power switch board and find out which pin is on the switch, or you can test the switch where it is solder on the power switch board.
On a lot of these power switch boards that the ribbon cables connect to have three or four capacitors, one of these caps may be bad, check these caps and double check the caps on the motherboard.

Have you put in the new PSU and are still having the problem?
What model Dell is this?
How many wires are on the ribbon cable? 34?
When the computer is running right, have you tried taking out the CMOS battery and see how it effects the start up process?

 

[loavesfishescomp]

Sorry for the delay

Right Trevathan. Yes, there is a ribbon cable that runs from the power switch, to another board behind it... then from that small board to the mobo. I looked at the (1) capacitor on the board in the middle, and it looked fine to me. It looked like a Pain to remove, so I just inspected in place with a flashlight.

I gave to computer back to the client because they really need the computer working for their home business. For a temporary fix (which I don't like), I recommended they hook up a timer to the outlet, therefore cutting the power to the computer for the duration of the night, alleviating the problem. Best thing I could think of, as I don't know what else to do. I don't have much of a problem telling them I can't figure it out and they should bring it to another repair shop that has seen many, many more computers than I.

I think there were 34 strands, yes. I did NOT remove the CMOS battery when the computer was running properly. I read your post after I gave it back.

Client is cool though... he says "No problem. One step at a time"

Thanks.

 

[PcTek9]

http://www.fonerbooks.com/power.htm

 

[gazza]

In order to solve these types of problems one of the best ways is to swap out components to see if the problem goes away. Unfortunately as techs we generally only carry ATX/BTX power supply units for testing and not proprietary model psu units. Some proprietary models, you have to use the exact same power supply or you will burn out the motherboard. If you replaced the existing psu with a exact same DELL psu and the power issue still remains then that just leaves the motherboard. By now you have spent a lot of time and money on a repair that is not going to be in the customers interest. Backup the clients data for them, tell them the motherboard is failing and that they would be better off to put the cost of new motherboard towards a new computer, preferably not a Dell.