CElliott2
asked on
Why are good, expensive PSUs failing?
In the last two to three years I have had to replace the power supplies (PSUs) for my main computer with units of continully increasing wattage (650, 750, 850, and 1050 watts) at least 6 times. This are good, expensive name brand PSUs that the manufacturer has replaced under warranty. The brand new 1050 watt unit recently began causing BSODs, so I had to put the 850 W unit back in. I operate these PSUs at 1/2 to 3/4 of rated wattage; e.g., the present 850 W unit is supplying 620 to 665 wildly varying watts. Last night I was watching the GPU-Z output of the power demands of two Nvidia Geforce 570 GTX display cards, which both erratically varied between 22 and 55 Amps (at only 1 volt) as they processed Seti@Home work units. Is it possible that the switching unit of these switching PSUs cannot keep up with these wildly varying power demands and so is failing? Is it a good, bad, or neutral sign that the PSUs 12-volt rail is outputting at present 18.82 volts measured by a VOM at a Molex hard disk connector?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
John
Another thing that could cause issue is a mis-formed bus socket or mis-formed card plug that is causing a false current path. ... Thinkpads_User
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Four different supplies have been used. That would seem to rule out a supply-only issue. The way the supply is connected in is more likely, I think.
... Thinkpads_User
... Thinkpads_User
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Clarification of ID: 38379732:
Have you checked the "mains" AC power supply is stable & of constant voltage?
Have you checked the "mains" AC power supply is stable & of constant voltage?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thank you GaryCase for yur answer. In response to your questions:
"First, there's no 1v rail in the PSU, so your comment r.e. "... varied between 22 and 55 Amps (at only 1 volt) ..." doesn't make sense. How did you determine this power draw was at 1v?"
GPU-Z 0.6.4 says "VDDC" is 0.943-0.951 V. It also says that VDDC Current varies now between 28.0 and 57.5 A (not ma).
"Second, a 12v rail should be within 5% of 12v -- NOT anywhere near 18v. That's so far out of spec it could easily damage your PC. Are you sure you're measuring this correctly??"
NO! Those measurements were taken with a digital VOM I received as part of a computer electronics course. The company has since gone out of business. Using an expensive Radio Shack "Micronta" digital VOM the 12V hard disk supply is generally solid at 12.41 V, but it dips down to 10.5 to 11.5 V occasionally, and down to 8.56 V much more rarely.
Your suggestion to purchase a decent PSU tester is a good one; thanks again.
The TX 850 has only one 12V rail, but that is true of all of Corsair's 850 W PSUs, even the most expensive.
The voltage at the wall socket varies between 114.8 and 117.1, but it is pretty steady between 116 and 117.
"First, there's no 1v rail in the PSU, so your comment r.e. "... varied between 22 and 55 Amps (at only 1 volt) ..." doesn't make sense. How did you determine this power draw was at 1v?"
GPU-Z 0.6.4 says "VDDC" is 0.943-0.951 V. It also says that VDDC Current varies now between 28.0 and 57.5 A (not ma).
"Second, a 12v rail should be within 5% of 12v -- NOT anywhere near 18v. That's so far out of spec it could easily damage your PC. Are you sure you're measuring this correctly??"
NO! Those measurements were taken with a digital VOM I received as part of a computer electronics course. The company has since gone out of business. Using an expensive Radio Shack "Micronta" digital VOM the 12V hard disk supply is generally solid at 12.41 V, but it dips down to 10.5 to 11.5 V occasionally, and down to 8.56 V much more rarely.
Your suggestion to purchase a decent PSU tester is a good one; thanks again.
The TX 850 has only one 12V rail, but that is true of all of Corsair's 850 W PSUs, even the most expensive.
The voltage at the wall socket varies between 114.8 and 117.1, but it is pretty steady between 116 and 117.
ASKER
Thank you Michael-Best for your answer.
"You my have a supply issue.
Have you checked the AC power supply is stable & of constant voltage?"
Yes, the voltage at the wall socket varies between 114.8 and 117.1, but it is pretty steady between 116 and 117.
"You my have a supply issue.
Have you checked the AC power supply is stable & of constant voltage?"
Yes, the voltage at the wall socket varies between 114.8 and 117.1, but it is pretty steady between 116 and 117.
ASKER
Thanks Dave Baldwin for your answer.
"Heat and a lack of airflow are the things that usually cause power supply failures."
So true! But here I have a humongous Honeywell floor fan (purchased at www.kaz.com at a great price, BTW) that blows across the whole rig, which sits w/o case on my work table. Plenty of hot air comes out of everything, the CPU cooler, the display cards, and the PSU.
"Heat and a lack of airflow are the things that usually cause power supply failures."
So true! But here I have a humongous Honeywell floor fan (purchased at www.kaz.com at a great price, BTW) that blows across the whole rig, which sits w/o case on my work table. Plenty of hot air comes out of everything, the CPU cooler, the display cards, and the PSU.
If the power supply is decent and is properly filtered, then the particular DVM should not matter. I use B&K Precision and ExTech DVM's for electronic work. Neither are cheap, but they are both very reliable if you are looking for something better.
I assumed your AC supply was fine and you have verified that. You have used 4 PSU's, so the motherboard and/or connecters are definitely looking bad.
.... Thinkapds_User
I assumed your AC supply was fine and you have verified that. You have used 4 PSU's, so the motherboard and/or connecters are definitely looking bad.
.... Thinkapds_User
ASKER
Reply to nobus and thinkpad_user: Thanks for your comments.
"Four different supplies have been used. That would seem to rule out a supply-only issue. The way the supply is connected in is more likely, I think."
The house current is OK. It is correctly gounded and the voltage at the wall socket only varies between 114.8 and 117.1, but it is pretty steady between 116 and 117.
However, all the current goes thru a UPS, which claims to output a pseudo-sine wave when on battery power, which almost never happens in Center City Philadelphia. I checked the UPS output when not on battery power with an oscilloscope just now, and the output wave form is no different between when the leads are connected to the UPS and when they are connected to the wall socket. Moreover, the frequency is correct: about 17 ms between wave peaks. However, I have read that some UPS's can wreak a PSU with automatic power factor correction, which mine all have had, if the UPS does not output a true sine wave. I told the PSU manufacturer about the UPS, but it never commented in reply. Could this be my problem?
"Four different supplies have been used. That would seem to rule out a supply-only issue. The way the supply is connected in is more likely, I think."
The house current is OK. It is correctly gounded and the voltage at the wall socket only varies between 114.8 and 117.1, but it is pretty steady between 116 and 117.
However, all the current goes thru a UPS, which claims to output a pseudo-sine wave when on battery power, which almost never happens in Center City Philadelphia. I checked the UPS output when not on battery power with an oscilloscope just now, and the output wave form is no different between when the leads are connected to the UPS and when they are connected to the wall socket. Moreover, the frequency is correct: about 17 ms between wave peaks. However, I have read that some UPS's can wreak a PSU with automatic power factor correction, which mine all have had, if the UPS does not output a true sine wave. I told the PSU manufacturer about the UPS, but it never commented in reply. Could this be my problem?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Here for reference is the output of an APC Smart UPS 750XL. You can see a small level of harmonic distortion top and bottom (at least I can), but it is nothing at all significant and does not cause any issue with my gear.
.... Thinkpads_User
.... Thinkpads_User
well - if all PSU are failing, the first suspect is the UPS
you can also test by connecting the UPS to another AC rail in the house (to verify that is not the cause - eg, bad contacts)
also, for checking AC waveforms, and UPS, you shoud have a power monitor, since the scope only records a moments value
see : http://www.fluke.com/fluke/uken/solutions/pq/default.htm
you can also test by connecting the UPS to another AC rail in the house (to verify that is not the cause - eg, bad contacts)
also, for checking AC waveforms, and UPS, you shoud have a power monitor, since the scope only records a moments value
see : http://www.fluke.com/fluke/uken/solutions/pq/default.htm
"The house current is OK. It is correctly gounded and the voltage at the wall socket only varies between 114.8 and 117.1, but it is pretty steady between 116 and 117."
This may not be the case over a long period...you may have spikes or lows under certain conditions.
If this is the case it will be hard to trace, but your UPS Voltage regulator is probably the cause.
It may be simpler to just replace the UPS and see if the problem is solved by that.
This may not be the case over a long period...you may have spikes or lows under certain conditions.
If this is the case it will be hard to trace, but your UPS Voltage regulator is probably the cause.
It may be simpler to just replace the UPS and see if the problem is solved by that.
@CElliott2 - I thnk it might have been more reasonable for you to split points on this answer.
.... Thinkpads_User
.... Thinkpads_User
ASKER
That is what I thought I did. I clicked on accept multiple solutions, and once I had done that, the system would not let me change anythng. If I didn't give everyone points, I screwed up. Sorry.
In any case, thanks for your help!
In any case, thanks for your help!
ASKER
So, how do I distribute points to multiple experts?
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
I marked the best soln as the one that suggested that I check all the voltages coming out of and running into the PSU. That was the most practical and time-consumming suggestion for failing PSUs. I only run PSUs between 1/2 and 3/4 of full load, as that is their most efficient operating range (see the PSU efficiency chart on Corsair's website, which I cannot find now), so overloading should not be the answer. That heat is the nemesis of all electronic equipment was an excellant reminder.
I marked the soln as good rather than excellant because no one said, "Here are the causes of PSU failure, pick yours!" Just to be pedantic, a know-it-all, and a pain in the A**, here is what I think is a good solution:
The major causes of PSU failure, somewhat in order of occurance, are:
Overheating
Overloading due to insufficient capacity for equipment connected
Power surges
Dust & dirt
Faulty connections or equipment (e.g., mobo or hard disk failure)
Physical shock
Improper electrical supply current
(sources: results of a Bing search on "Major causes of power supply failure")
I marked the soln as good rather than excellant because no one said, "Here are the causes of PSU failure, pick yours!" Just to be pedantic, a know-it-all, and a pain in the A**, here is what I think is a good solution:
The major causes of PSU failure, somewhat in order of occurance, are:
Overheating
Overloading due to insufficient capacity for equipment connected
Power surges
Dust & dirt
Faulty connections or equipment (e.g., mobo or hard disk failure)
Physical shock
Improper electrical supply current
(sources: results of a Bing search on "Major causes of power supply failure")
imo - it's not quite correct
while overheating iS a major issue (if trespassing the max temp of components) it is the heating up/cooling down cycle that causes most problems, by crimping expanding the solderings, and different materials in a different way
also - tyou never answered my question about the 17V measured : are you sure you measure the 12 V? between the YELLOW and black wire on the molex?
while overheating iS a major issue (if trespassing the max temp of components) it is the heating up/cooling down cycle that causes most problems, by crimping expanding the solderings, and different materials in a different way
also - tyou never answered my question about the 17V measured : are you sure you measure the 12 V? between the YELLOW and black wire on the molex?