Author Topic: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)  (Read 9693 times)

0 Members and 1 Guest are viewing this topic.

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
PC repair question...

Is it safe to use a DMM in "ohm" setting connected between power pins (yellow, red, orange wires) and ground pins (black wires) on the motherboard 24-pin (or 20-pin) power connector and 8-pin (or 4-pin) CPU power connector to check for a shorted supply voltage rail?

I'm talking about probing the connectors on the MOTHERBOARD with the power supply disconnected and the meter set to "ohms". I'm NOT talking about checking the PSU output with the meter set to "VDC".

It seems like it should be a straightforward operation, but I'm not sure if it's safe for the MB. All the warnings about static electricity and improper handling potentially damaging a board have me nervous about trying this without some feedback first.

I know that a DMM produces a very small voltage and measures the current through the probes to determine the resistance. I wouldn't think that a few millivolts from the DMM applied to a rail that normally gets 3.3, 5, or 12 volts would cause any problems, but I wanted to see if anyone here has firsthand knowledge on the matter.

I don't think it should matter for such a general question, but my MB is an Asus 990FX rev 1.01, and the CPU is an AMD FX-8150.

Specifically, I suspect a short to ground on the 12 volt CPU power rail, so I'd like to check resistance between power and ground on the 8-pin CPU power connector. I want to first do this with the CPU installed, and then remove the CPU and check resistance again. If there is a short in both cases, then the problem would be in the MB. If there is a short only when the CPU is installed, then I'll know the CPU is fried.

Also, if these measurements are safe to do and can provide meaningful readings, can anyone tell me what a "good" reading and a "bad" reading would be? Should a good board read in the kilo-ohms range? mega-ohms?

Lastly, should I only perform this with the meter in the lowest range? I don't know if the voltage a DMM produces increases as the range is increased.

There seems to be plenty of info on checking voltages on a MB, but I can't find much about resistance checks. Thanks in advance!
 

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #1 on: September 06, 2017, 09:08:20 pm »
What leads you to suspecting a short on the 12V rail?

Testing the power inputs between them and ground is safe. Use the Diode Check function, red lead to ground/negative.

Make sure you are on working as static-free as you can by using a wrist-strap.  Also, touch the ground plane when you are handling the board.  Typically any shielding will be ground.

The CPU would not be active on the 12V rail.  On the 3.3V rail, yes.

Have you swapped PSUs?

ASUS mobos carry a 3 year warranty.

T
veritas odium parit
 
The following users thanked this post: door king

Offline helius

  • Super Contributor
  • ***
  • Posts: 3689
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #2 on: September 06, 2017, 09:18:36 pm »
The test voltage of the DMM should be printed in its specifications. That will also tell you if it uses different voltages on higher ranges.
For example, the Fluke 8060A applies an open-circuit test voltage of under 4.8V on its lowest range, but higher ranges apply lower voltages. The safest range (lowest supplied current at the lowest voltage) is 200k ohm. It is unusual in having a continuity indicator on every range; most other DMMs do not have this.

4.8V is not exactly a "very small voltage" when you are talking about semiconductors. The available current on the lowest (200 ohm) range is only a milliamp, but that is still a lot of energy to a nanoscale transistor. The higher ranges on the 8060A use lower currents.

As to whether you can get any useful information from this test, I do not have high hopes. The very high power consumption at low voltage of a modern processor means that resistance between Vcc and ground must be negligible. You have a better chance of detecting something with the processor removed. 12V doesn't go to the processor, anyway, it is converted to around 1V by POL DC-DC modules.
 

Offline Monkeh

  • Super Contributor
  • ***
  • Posts: 8158
  • Country: gb
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #3 on: September 07, 2017, 01:55:23 am »
The CPU would not be active on the 12V rail.  On the 3.3V rail, yes.

Er?
 
The following users thanked this post: vk6zgo

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #4 on: September 07, 2017, 02:33:18 am »
Yeah, my bad.  |O  Stuck in the old daze...  :o

12V for CPU, 3.3V for RAM, North/South bridges.

Regardless, having the CPU in or out would not be read through the DC-DC converters with an ohmmeter as helius pointed out.

T
veritas odium parit
 

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #5 on: September 07, 2017, 10:31:46 pm »
What leads you to suspecting a short on the 12V rail?

Power supply works fine when disconnected from all loads and "jumped" to turn it on.  Also works fine when only the 24-pin connector is attached to the MB (8-pin CPU power disconnected).  Does not power up when both 24-pin and 8-pin connectors are installed:  When I press the power switch, briefly the HSF will twitch and the CPU LED on the MB with flash and then nothing.  No voltage from the PSU except for the 5V standby power.  So it seems a fault in the motherboard pertaining to the circuit on the 8-pin connector is causing the PSU to go into protection mode. 

Have you swapped PSUs?

Yes. Same behavior noted above with both PSUs. 

ASUS mobos carry a 3 year warranty.

The Sabertooth actually carries a 5-year warranty.  Unfortunately, I'm about 5-6 months past that.  (ordered the board on 19/3/2012)  :'(
 

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #6 on: September 07, 2017, 11:06:06 pm »
The test voltage of the DMM should be printed in its specifications. That will also tell you if it uses different voltages on higher ranges.

My Fluke 81438 manual just says "less than 1.5V" for "open-circuit test voltage" on the specs for the resistance range.  Not as large as your example, but I always assumed most meters put out a fraction of a volt. 
 

Offline pigrew

  • Frequent Contributor
  • **
  • Posts: 680
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #7 on: September 07, 2017, 11:18:26 pm »
Most of my meters output between 7 and 12 V when measuring resistances.

I ended up getting an in-line banana plug adapter thing and putting two 1n4148s across it (in parallel, with polarities opposite). This limits the output to less than 0.8 V or so, which is safe for most transistors (some Ge devices may complain). The only downside is that it reads about 600 ohms in continuity mode, so any resistance more than 100 ohms or so will be significantly wrong.

I asked this exact question last year.
« Last Edit: September 07, 2017, 11:53:28 pm by pigrew »
 

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #8 on: September 08, 2017, 12:35:11 am »
For curiosity's sake, what are the PSUs you tested with?

With your response, I would suspect the DC-DC converter(s) of a failure.  Do the DMM diode check on the mobo and see.  If you do the check with the mobo mounted in the case, you greatly reduce the chance of any static related problems.

Were there any indications of "something" going on with the system prior to the failure?  Sudden shutdowns or reboots?

Were you overclocking?

Any recent hardware additions/upgrades to the system?

Try testing with no CPU & RAM or cards installed.  Then just RAM.  Then add CPU.  Be sure to discharge the PSU in between each test.

I think you need to have the CPU fan connected.

T
« Last Edit: September 08, 2017, 12:44:14 am by Toasty »
veritas odium parit
 
The following users thanked this post: door king

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #9 on: September 08, 2017, 03:33:12 am »
For curiosity's sake, what are the PSUs you tested with?

The computer had a Seasonic X-760 to start with, then I swapped in a Fortron/FSP SAGA+ 450R. 

Were there any indications of "something" going on with the system prior to the failure?  Sudden shutdowns or reboots?

No. 

However...

Where I live we seem to lose power quite often.  I haven't had a battery backup/surge protector for some time.  Also, a few times when we've lost power, our LED bulbs remained dimly lit and I measured like 50V at the outlets.  I think even when the power is up, the quality is pretty s**t.  Even though there was no power issue immediately before the failure, I wonder if over time the flaky power situation reduced the lifespan of my rig.  I would assume though that bad power would be more likely to hurt the PSU than anything else, and I would think that a nice Seasonic like mine would keep the rest of the system protected. 

I also have a tendency to keep an obscene amount of firefox tabs open, which can work the CPU and RAM pretty hard at times. 

Were you overclocking?

No.

Any recent hardware additions/upgrades to the system?

No.

Be sure to discharge the PSU in between each test.

My MB has a standby power LED.  I assume that if I wait for it to go out after I unplug the PSU, it is considered discharged?

I think you need to have the CPU fan connected.

CPU fan has been connected for all of my tinkering thus far.  I'll keep it connected for all of your suggested checks. 
 

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #10 on: September 08, 2017, 03:43:25 am »
Testing the power inputs between them and ground is safe. Use the Diode Check function, red lead to ground/negative.
Do the DMM diode check on the mobo and see. 

I'm curious to know what the reasoning is behind using diode mode and the polarity you suggest.  Is it a measure to protect the circuitry being tested, or is it to prevent forward-biasing a diode or p/n junction in the circuit and misinterpreting it as a short?  Or something else?
 

Offline stj

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #11 on: September 08, 2017, 04:24:28 am »
saving you some time,
if it's shutting the psu down, check the fets around the cpu socket.
 

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #12 on: September 08, 2017, 04:47:44 am »
The use of the diode check supplies a voltage that is constant. The results are quick and show the voltage drop across the circuit.  Faster picking up shorts in my experience.

When you use the other ohms functions, the voltage can vary widely from meter to meter depending on the range setting.  Plus, it can take a while to "settle" on a reading, especially where capacitors are involved.

Example: I have a handheld DMM that puts out 0.5V on every range except for diode check. On diode check, it gives a solid 2.5V, which is enough to forward bias a junction.

My bench DMM puts out 3.0V

Also, some meters have been known to have the polarity of the leads reversed on the ohms functions, but fine on the diode check.  Weird  :o

T
veritas odium parit
 

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #13 on: September 08, 2017, 04:57:11 am »
My MB has a standby power LED.  I assume that if I wait for it to go out after I unplug the PSU, it is considered discharged?

Also press the power button.  You can do that first to speed the process along.  Usually the LED on the mobo will flash and the fans might twitch.


saving you some time,
if it's shutting the psu down, check the fets around the cpu socket.

From Badcaps?

Would have been my next step once he got a reading.  There's only 20 or so under that heat-pipe heatsink.  lol  :-DD


T
veritas odium parit
 

Offline stj

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #14 on: September 08, 2017, 11:36:12 am »
 

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #15 on: September 14, 2017, 04:27:31 am »
Sorry about the lapse.  I've also been having some automotive woes that I had to address...

Do the DMM diode check on the mobo and see.
CPU installed:  Reading settled to 0V instantly. 
CPU removed:  Reading rose to about 0.44V over a couple seconds. 
(For comparison, doing the same test on the 12 pins of the 24-pin connector, the DMM reading quickly settled to about 0.503V.)

Try testing with no CPU & RAM or cards installed.  Then just RAM.  Then add CPU. 
The only thing that affected the results was whether or not the 8-pin CPU plug was connected or not. 
If the 8-pin plug was connected, the computer had the same behavior as I described previously (CPU LED flashes, fans twitch, then nothing). 
If the 8-pin plug was removed, the PSU comes to life, fans spin up, and the CPU LED remains permanently illuminated. 
This was the behavior regardless of whether there was no CPU or RAM, just RAM, or both CPU and RAM. 

The only other things installed for the above tests were the CPU heatsink fan, the power switch, and the PC speaker. 

I was disappointed that I didn't get any beep codes for any of the above tests.  Have some MoBo companies been doing away with system speaker error code output?

It seems that there's definitely an issue with the MB.  Agree?
One nagging question in my mind:  How likely is it that the MB took the CPU with it as it went down?
 

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #16 on: September 14, 2017, 08:13:01 am »
Yes. It seems like the VRM is bad.  Perhaps the heatpipe failed...?

I'd give Asus support a shout, and see if they'll cover it.  Fudging the purchase date "I got it as a gift at Christmas, I don't have the receipt" might help.

Failing that, it's time to take the heatsink off the VRM and have a go at identifying the failure(s).

"ASUS Sabertooth 990FX features a 10-phase VRM, with 2 highly efficient Low RDS(on) MOSFETs per phase. There are 11 x 820µF and 3 x 270 —F capacitors, all solid, made in Japan, with up to 500,000 MTBF at 65°C. 'Military grade' even. Coils feature composite cores. The device is controlled by the familiar Digi+ VRM that offers great leeway for overclocking. The motherboard supports all AM3/AM3+ processors with up to 140W TDP." -iXBT Labs, 12/2011



T
veritas odium parit
 

Offline door kingTopic starter

  • Contributor
  • Posts: 14
  • Country: us
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #17 on: September 14, 2017, 09:56:06 pm »
I'd give Asus support a shout, and see if they'll cover it.  Fudging the purchase date "I got it as a gift at Christmas, I don't have the receipt" might help.
Well, it looks like getting ASUS to cover anything is hopeless.  The guy on the phone told me it was out of warranty without even asking for proof of purchase.  I can send it in for them to fix it on my dime, but I don't want to mess with that.  I've read too many horror stories about people who've gone that route. 

Failing that, it's time to take the heatsink off the VRM and have a go at identifying the failure(s).
So, if I pinpoint the failure(s), do I actually stand a chance at fixing this thing?  Are the FETs in the VRM really just plain old MOSFETs or are they ICs with integrated driver/controller circuitry?  Are the parts going to be something I can buy, or are they proprietary ASUS stuff?

Perhaps the heatpipe failed...?
How can I tell for sure?  Is that something that can be repaired?  I assume if the heatsink has an issue that can't be resolved then there's no point in repairing the electronics. 
 

Offline Toasty

  • Regular Contributor
  • *
  • Posts: 98
  • Country: us
  • Component Level Repair Tech
Re: Testing PC motherboard for supply rail shorts w/ ohmmeter (is it safe?)
« Reply #18 on: September 14, 2017, 10:37:23 pm »
>>The guy on the phone told me it was out of warranty without even asking for proof of purchase<<

How?  Did you provide a serial number?  If it were 5 months ago, how would that change things and how would they know?  Did you register the board 5 years ago?

>>Are the FETs in the VRM really just plain old MOSFETs or are they ICs with integrated driver/controller circuitry?  <<

Hard to say without a part number.  :)

>>How can I tell for sure?  Is that something that can be repaired?  I assume if the heatsink has an issue that can't be resolved then there's no point in repairing the electronics. <<

1) Remove the heatsink. Apply heat at the very far end that's over the VRM and feel the opposite end for warming. You can use one of those infrared laser devices.

2) No

3) Maybe.  Check the overclockers forums. Some 30mm or 40mm fans mounted direct to the heatsink might work.

T

PS: I'd be more confident if you could get that CPU tested.



« Last Edit: September 14, 2017, 10:43:02 pm by Toasty »
veritas odium parit
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf