Sign in with
Sign up | Sign in
Your question

My power supply acting strange?

Last response: in Components
Share
November 14, 2007 1:20:20 AM

Hi everyone, I have just noticed that in my bios that my CPU Vcore voltage moves up & down from 1.416 - 1.424 volts, with OC 1.544 - 1.552 volts. Also it seems to move from Vcore voltage to my +3.3 rail, but doesn't show any of this in Speed Fan. My bios lately has seemed slow and flickering as the RPMS are changing. I guess the good thing is my voltage isn't dropping below my setting in bios for Vcore or any of my rails, but rather moves up in voltage rather than down.

This all started when I got my 8800GTX after getting rid of my 7800GTX that only needed one 6Pin PCI-Express connector. I don't have another graphics card to try and all my friends are still using AGP. My PSU has two PCI-E plug-ins in red color, one says master the other says slave. I havent had any stability problems yet so that might be a good sign.

Do you guys think that my PSU is detecting that its needs to up the voltages because of my 8800GTX power sucking card that it is needs more wattage? I am running Orthos Blend priority 8 on my OC in my sig, so far its ok. My spec's are in my sig.



By systemlord at 2007-11-13

More about : power supply acting strange

November 14, 2007 8:44:05 AM

Strange that the BIOS is slow and flickering as the RPMs change. Like you said, it's a good sign that you are not experiencing any stability issues. It might be that the power supply is compensating for the extra power requirements of your beast of a graphics card. :-)
November 14, 2007 8:57:18 AM

You think the motherboard's voltage regulators might have anything to do with it? Just a guess. I might be completely on the wrong track.
Related resources
November 14, 2007 9:08:47 AM

cfvh600 said:
You think the motherboard's voltage regulators might have anything to do with it? Just a guess. I might be completely on the wrong track.


I just failed Orthos after 6 hours, plus my +3.3v rail is 3.280 which I don't know if its ok being that low. Wonder if my mobo or PSU thats bad, one thing is for sure is there is some voltage regulation problem. Theres only two things it could be and thats my mobo or power supply.
November 14, 2007 9:16:34 AM

****UPDATE****

I just remembered something, when I began my overclocking experience everytime I had to flick that on/off switch on back of the PSU it would beep loud four times everytime like its suppose to and it doesn't do that anymore. Damn Enermax is supposed to be the best, I guess I'll have to RMA it the day I get Crysis. :fou:  Anyone know if the +3.3 volt rail is supposed to drop below 3.3 volts?
November 14, 2007 9:41:03 AM

I've found a table for voltage tolerances on ATX power supplies. And there is an + or - 5% tolerance on the 3.3V rail. Your reading is within that limit.
November 14, 2007 10:02:27 AM

cfvh600 said:
I've found a table for voltage tolerances on ATX power supplies. And there is an + or - 5% tolerance on the 3.3V rail. Your reading is within that limit.


Now my 5V rail is fluctuating as well, so I went into the bios and upped the Vcore, Northbridge Vcore, FSB volts and then the 3.3v rail, 5v rail and my Vcore were acting up again. The fact that this all started when I upgraded to a more power hungry graphics card tells us what? It seems the more voltage I ask for the more I see the fluctuations.
November 14, 2007 10:15:53 AM

Poor PSU can't keep up!
November 14, 2007 10:45:04 AM

That's a top flight PSU, and you certainly aren't loading it up. Before I blamed it, I would get a good DVM and measure a spare MOLEX while it is under load. I would be more inclined to suspect the mobo as cfvh600 said early on.
November 14, 2007 6:34:15 PM

You think because I untroduced a new component (from 7800GTX to 8800GTX) to my system that that might make my preveus overclock no longer stable? When I got my 8800GTX thats when this all started. But my CPU Vcore shouldn't move up and down should it? Maybe the rails but not CPU Vcore fluctuation, right?
November 14, 2007 8:03:14 PM

Maybe try a default BIOS and see if your problems go away. I wouldn't think that would cause your voltages to fluctuate. That is a pretty high OC maybe your board got tired. I still think you should measure the power at the MOLEX connector. That would rule out the PSU
November 14, 2007 8:20:52 PM

I currently have 10 hours on Orthos blend everything at stock clock with no overclocks at all. I looked at that link you gave me DellUser1 about the Enermax 720W that I have and is well within spec. Is it normal for any Vcore voltage to flucuate between 1.544v to 1.552, maybe its upping the volts to compensate for the added power draw from my 8800GTX.

Because of this extra power draw (8800GTX) do you think that when you start adding more power hungry parts to you PC that might make an otherwise stable OC unstable?
a c 243 ) Power supply
November 14, 2007 8:51:43 PM

I don't know if it's normal, but after reading this thread I have been watching my Vcore with more interest today. Seems mine fluctuates between 1.28 and 1.296 ( twice the amount of yours ).
I suppose adding power hungry hardware could have an effect on the oc, but with a quality psu such as yours, I would think there would be no problems due to it.
November 15, 2007 2:33:03 AM

delluser1 said:
I don't know if it's normal, but after reading this thread I have been watching my Vcore with more interest today. Seems mine fluctuates between 1.28 and 1.296 ( twice the amount of yours ).
I suppose adding power hungry hardware could have an effect on the oc, but with a quality psu such as yours, I would think there would be no problems due to it.


There must be an effect on my OC because I'm still running Orthos for 16 hours now with no errors at stock clocks. Anybody else have any opinion on how this might be possible, losing an OC because of new hardware?
November 15, 2007 2:52:47 AM

ALso, like cfvh600 said, All PSU's have a tolerance rating. Most are +/- 5%. I calculated your 3.3V tolerance and it was -0.6%. Thats a good tolerance IMO.

To calculate tolerance (same calculation as % error)= ((New - Original)/ Original) X 100

% Error= ((Measured - Theoretical )/Theoretical) X 100

Whenever you OC, you are stressing the system, so by adding that card, you stress it more because the card requires more power. You will probaly lose the OC, but you do not have to be at stock. If it is not stable at 3.7Ghz, step it down a few Mhz until it is stable.

Also, check the caps on your motherboard to make sure they are still fine. If your PSU is out of warranty, open it up to check the caps (or if you don't care about the warranty, open it up).

For pics about what a "bad" cap looks like go here:
http://www.badcaps.net/
November 15, 2007 4:15:53 AM

soloman02 said:
Whenever you OC, you are stressing the system, so by adding that card, you stress it more because the card requires more power. You will probaly lose the OC, but you do not have to be at stock. If it is not stable at 3.7Ghz, step it down a few Mhz until it is stable.


I have already checked all mobo components and nothing looks burned or discolored. Would you recommend any good monitoring software for reading voltage? I tend to agree that whenever you enter something new to a perfectly stable system that its very possible that the added hardware might change things and overclocks a tad because now more power is being fed to the system which can change things ever so slightly, enough to give just an error but not a stability problem which isn't the case.

One error in 6 hours is very small, I'll just back off a little on my OC and see what happens. If I had any component to guess is to blame for this I'd have to say it has something to do with my mobo voltage regulaters, they have tolerances to.

Anyone care to add your view on this?
November 15, 2007 5:07:04 AM

I tend to agree with you on the mobo regulators being the culprit, you have stressed them a bit. The DVM would rule out the PSU. They are not that expensive for a cheap one, and they are good to have around the house.
November 15, 2007 6:35:42 AM

Zorg said:
Maybe try a default BIOS and see if your problems go away. I wouldn't think that would cause your voltages to fluctuate. That is a pretty high OC maybe your board got tired. I still think you should measure the power at the MOLEX connector. That would rule out the PSU


A digital volt metter? How do I know where to point those sharp pins? Also would a DVM show me all the rails and not just the 12 volt rails? Hay guys 21 hours without any errors, think I'll wait till 24 hours.
November 15, 2007 7:31:01 AM

systemlord, I use SpeedFan Charts to observe Vcore as well as other supply voltages while stress testing. I also have an Enermax. Although it's only a Noisetaker 495, it's always run my rig event free, which is astonishing, because I know the little guy is being pushed to within an inch of going up in a mushroom cloud when I simultaneously stress test the Q6600 at 3.6 with Prime95 Small FFT's, the twin RAID 0 Raptors with a virus scan, and the overclocked 8800GTS with ATI Tool Show 3D View.

Even during such an extreme total system stress test while watching voltage and temperature Charts with an unflinching eye, I have never seen Vcore go UP! My smallest incremental Vcore sag (droop is an improper electrical term) on this P5k Deluxe is 8 millivolts, which is typical, but never exceeds 16 millivolts, thanks to the Vcore Damper feature. My former P5W DH Deluxe was 24 to 36 millivolts, and my Opty 170 on the MSI K8N Neo 4 Platinum was really sloppy at 80 millivolts!

Here's a few ideas:

  • Since your P5B Deluxe has a secondary 4X PCI Express slot, I suggest that you temporarily move your GTX to the other slot to see whether the problem is slot related.

  • If your BIOS has settings for PEG Slot Power, you can experiment with different combinations to see whether the problem is affected.

  • Molex connector female pins can become sloppy, which can cause intermittent sags, and may result in compensative supply surges. I would check and / or tighten every power connector's female pin in the computer.

  • Since the motherboard uses a +12 rail to create Vcore, I would observe +12 to see whether it fluctuates with Vcore.

  • As your supply is modular, you can try reconfiguring output cables to see whether the problem is cable related.

  • If the problem consitently present immediately on power up, then power off at the plug. Spray "circuit chill" on the regulators, and re-test. If the time interval to recurrence changes, then the problem is a heat sensitive component in the motherboard's voltage regulation circuitry.

    That's about all I can think of for the moment. That should be enough to keep you busy for awhile.

    Comp :sol: 
    November 15, 2007 7:56:06 AM

    Just thought I'd let you guys know that with three different P5B Deluxe mobo's (my two other friends each have one) I'll set Vcore in bios to 1.5875 Vcore then go to "Hardware Monitor' in bios it shows 1.544 Vcore and when I boot up in Windows it shows up in Speed Fan as 1.55. I know thats a Lot of Vcore sag, but theres not much I can do about that. Thanks for all your suggestion everyone, I'll settle for a smaller OC cause I'm already at max Vcore for air cooling.

    I will do all those things that you suggested Comp. :sol:  Thanks!
    November 15, 2007 8:37:59 AM

    Good luck systemlord. :-)
    November 15, 2007 1:11:15 PM

    systemlord said:
    A digital volt metter? How do I know where to point those sharp pins? Also would a DVM show me all the rails and not just the 12 volt rails?
    On the 4 pin Molex you can only measure the 12v and 5v. You can wrap a paper clip tightly around the positive probe tip and measure the 24 pin connector from the back while the machine is running as well. Touch the ground (black) probe to the case and use the positive probe (red) to check one of each voltage, just be careful. Read the manual for the meter and make sure you have the leads plugged into the voltage testing plug on the meter. If you have them in the current testing plug (Amps) then you will create a direct short which is not good to say the least. I'm not trying to scare you just saying be careful. Here are pics and links, should you decide to test your PSU voltages.


    With the 4 pin molex you can put the ground on a black wire in the molex, you can leave it there to test the 24 pin connector, if you want.

    4 Pin Molex Connector



    24-pin ATX power supply connector






    November 15, 2007 1:17:14 PM

    Nicely done Zorg. :-)
    November 15, 2007 1:29:01 PM

    Thanks, taking a meter to anything, especially a running PSU/mobo, can be a daunting task. I tried to be as complete as possible, without writing a novel. If anyone has additional pointers feel free to post them.
    a c 243 ) Power supply
    November 15, 2007 5:56:22 PM

    As cfvh600 said, nicely done !
    November 15, 2007 7:34:41 PM

    Zorg said:
    On the 4 pin Molex you can only measure the 12v and 5v. You can wrap a paper clip tightly around the positive probe tip and measure the 24 pin connector from the back while the machine is running as well, touch the ground (black) probe to the case and use the positive probe (red) to check one of each voltage, just be careful. Read the manual for the meter and make sure you have the leads plugged into the voltage testing plug on the meter. If you have them in the current testing plug then you will create a direct short which is not good to say the least. I'm not trying to scare you just saying be careful. Here are pics and links, should you decide to test your PSU voltages.


    With the 4 pin molex you can put the ground on a black wire in the molex, you can leave it there to test the 24 pin connector, if you want.


    I looks simple enough for me, I just finished Orthos 32 hours at stock setting Vcore stock no OC. So I have to find a new happy OC for my processer, and with that in mind our summer temps have returned. :ouch:  I'll be moving my PC components to a SilverStone TJ09 next month so thats going to help my temps a lot.

    Thats so much for giving me very helpful ideas everyone!
    November 15, 2007 8:22:53 PM

    One last thing that I notice when I disabled my OC, when exit the bios I noticed that my bios screen froze for a very slight second and displayed lots of different colors over bios display then booted to Windows. Is this something to worry about?
    November 16, 2007 3:44:53 AM

    Zorg said:
    Thanks, taking a meter to anything, especially a running PSU/mobo, can be a daunting task. I tried to be as complete as possible, without writing a novel. If anyone has additional pointers feel free to post them.


    One more question, is 2.42 volts to low for a +3.3v rail? And what percentage is that above 5% -?
    November 16, 2007 3:46:29 AM

    WAY out of tolerance by 31%.
    November 16, 2007 4:02:45 AM

    CompuTronix said:
    WAY out of tolerance by 31%.


    I use Speed Fan to get that setting.
    November 16, 2007 4:09:07 AM

    Your rig probably wouldn't run if the 3.3 volt rail measured 2.42 with a DVM. It would be best if you were to perform the voltage measurements using the excellent information Zorg provided.
    November 16, 2007 4:17:51 AM

    CompuTronix said:
    Your rig probably wouldn't run if the 3.3 volt rail measured 2.42 with a DVM. It would be best if you were to perform the voltage measurements using the excellent information Zorg provided.


    I'm running Orthos now for 7 hours, durring this time is where I detected the +3.3v rail drop to 2.62v. Is this a bullshit reading from Speed Fan?
    November 16, 2007 4:21:10 AM

    The only way to confirm the actual value of the 3.3 volt rail is to perform a DVM measurement.
    November 16, 2007 5:04:48 AM

    CompuTronix said:
    The only way to confirm the actual value of the 3.3 volt rail is to perform a DVM measurement.


    I'm getting one tomorrow.
    November 16, 2007 6:37:42 AM

    Zorg said:
    On the 4 pin Molex you can only measure the 12v and 5v. You can wrap a paper clip tightly around the positive probe tip and measure the 24 pin connector from the back while the machine is running as well, touch the ground (black) probe to the case and use the positive probe (red) to check one of each voltage, just be careful. Read the manual for the meter and make sure you have the leads plugged into the voltage testing plug on the meter. If you have them in the current testing plug then you will create a direct short which is not good to say the least. I'm not trying to scare you just saying be careful. Here are pics and links, should you decide to test your PSU voltages.


    With the 4 pin molex you can put the ground on a black wire in the molex, you can leave it there to test the 24 pin connector, if you want.

    4 Pin Molex Connector

    http://i15.tinypic.com/6x6v87p.jpg

    24-pin ATX power supply connector

    http://i3.tinypic.com/87dabdz.jpg


    Ok I'm a bit confused here, there are three 1/2/12 +3.3v rails pins, so which one is the one I want? I can use any of the black wires for grounding?
    November 16, 2007 6:45:58 AM

    Pins 1,3 & 12 to any black. Carefully measure all Orange connectors, then make a connector chart, and measure all voltages at Idle as well as Load.

    Comp :sol: 
    November 16, 2007 6:51:37 AM

    But not 1,3 and 12 at the same time? 1 and 3=ground, then 3 and 12? Sorry just want to be very clear here.
    November 16, 2007 6:57:13 AM

    Sorry...1,2 & 12 is correct. Measure 1 to ground, then check 2 to ground, then 12 to ground, etc.
    November 16, 2007 7:09:49 AM

    Its going to be a pain in the neck to just stand there holding everything (probes) in place. If the voltage drop is only happenning 1-3 time an hour this might prove to be hopeless. I wish I could just plug something in and leave it there all day.
    November 16, 2007 7:20:00 AM

    If you have an unobstructed area above the motherboard power connector, you can actually push the needle probes into the individual connections, provided that they're small enough. Otherwise, as Zorg suggested, you can fabricate a pair of small paper clip segments using needle nose pliers to create snug loops in the ends of each segment, then insert the probes into the loops, and the paper clip segments into the motherboard connections.
    a c 144 ) Power supply
    November 16, 2007 9:21:08 AM

    How rapidly is CPU vcore fluctuating? You are reporting an 8 mv. fluctuation. That could be one of three things: a normal result of a varying CPU load, a side effect of the way A/D converters work, or a sign of a problem.

    I am a little more troubled about the reported 3.3 volt values. But if you are operating with no stability problems, you probably do not have anything to worry about. I would take everything back to stock speeds and just watch the system for awhile.

    Yes, indeed, Zorg. Nice work.
    November 16, 2007 10:03:36 AM

    jsc said:
    How rapidly is CPU vcore fluctuating? You are reporting an 8 mv. fluctuation. That could be one of three things: a normal result of a varying CPU load, a side effect of the way A/D converters work, or a sign of a problem.

    I am a little more troubled about the reported 3.3 volt values. But if you are operating with no stability problems, you probably do not have anything to worry about. I would take everything back to stock speeds and just watch the system for awhile.

    Yes, indeed, Zorg. Nice work.



    The higher the CPU Vcore is 1.544v the more rapidly the flucuations are, the flucuations then go from the Vcore then down to the other two rails (+3.3v & +5v) one at a time then begin flucuating like a snake climbing down the vine. This only happens one rail at a time though, so after I saw all the flucuation at 1.544 Vcore I desided to perform a little test.

    Next I reloaded back into bios and began to lower all Vcores (Northbridge/FSB voltage/CPU Vcore) values to there lowest and the rapid voltage flucuation almost stopped completely. The other issue was right after POST I would very very quickly see lots of vertical colorful lines not moving, kind of like when a movie projector freezes up on one frame of film. This happens about 1 in every 10 reboots, and looks the same everytime. The white letters in bios almost wink at me every few seconds.

    I have had no instability whatsoever, I probibly would have never found these little problems if I hadn't wanted to try a lower voltage for my RAM. I ran Orthos at stock everything for 32 hours with no errors, passed with flying colors. My OC that I am running now at 400x8 @ 3.2GHz 1.32 Vcore has 13 hours on it with not one error.


    By systemlord at 2007-11-15
    a c 243 ) Power supply
    November 16, 2007 10:34:46 AM

    systemlord said:
    Its going to be a pain in the neck to just stand there holding everything (probes) in place. If the voltage drop is only happenning 1-3 time an hour this might prove to be hopeless. I wish I could just plug something in and leave it there all day.


    Tape the probes to the connector.
    November 16, 2007 11:07:51 AM

    1.55 Vcore sounds a little extreme to me. My system is not much different than yours in terms of power consumtion and i'm running an Enermax Infinity 620WT PSU. My Vcore is set to 1.335 I think (P5-B Vanilla BTW) and I've gone 8 hours in Prime95 with no errors on max heat/power settings. Asus has a few new Bios' out for the P5B, you may want to try and flash it with an upgraded bios (forgive me if you said you already have the latest as there were way to many posts for me to read this early in the morning).
    November 16, 2007 11:41:56 AM

    Ibanezrg570 said:
    1.55 Vcore sounds a little extreme to me. My system is not much different than yours in terms of power consumtion and i'm running an Enermax Infinity 620WT PSU. My Vcore is set to 1.335 I think (P5-B Vanilla BTW) and I've gone 8 hours in Prime95 with no errors on max heat/power settings. Asus has a few new Bios' out for the P5B, you may want to try and flash it with an upgraded bios (forgive me if you said you already have the latest as there were way to many posts for me to read this early in the morning).


    I have use 1.55 Vcore for my CPU since last May, but this summer I had big heat issues so I am settling for around 1.33 Vcore running @ 3.2GHz 400x8. I want my mobo to last so I'd rather not heat it up to much and stress the voltage regulators out, this was a wake-up call for me.


    I have been told that you shouldn't update your bios unless you are having hardware compatibility problems, guys what is your view on this updating bios thing. Most of the updates are for the 1333 FSB processers, I look at the bug fixes and if they don't apply to me I don't update the bios.

    O yea, why does everyone use Prime95 when Orthos is simpler to use? Is one better than the other?
    November 16, 2007 11:53:55 AM

    This is what the Asus bios update said when I tried to download new bios, "Please note, BIOS update is only recommended when experiencing technical difficulties with your system, And is not recommended to be performed regularly.

    Moreover, due to the nature of BIOS update, there is certain level of dangers involved. BIOS update must be performed with extreme caution . During BIOS update process, your system must be maintained without interference or power loss to prevent unexpected damage.

    In case of BIOS update failure, please follow instructions in your User's Manual for guidelines on BIOS recovery via CrashFree BIOS. In the event that BIOS recovery is not recoverable via CrashFree BIOS, please contact your place of purchase for further assistance on BIOS recovery".

    November 16, 2007 12:03:01 PM

    systemlord said:
    Its going to be a pain in the neck to just stand there holding everything (probes) in place. If the voltage drop is only happenning 1-3 time an hour this might prove to be hopeless. I wish I could just plug something in and leave it there all day.
    That power supply probably only has one regulator output per voltage, so if the voltage is sagging it is probably doing it on all of the pins of the same voltage.

    If you want to measure a voltage for a extended period of time, then you can remove the connector an carefully jam a piece if wire in the back of the pin and one to a ground on the Molex or chassis, and extend them to your desk. Tape the end, make sure it is secure and handle the wire gingerly to avoid pulling it out. Use the biggest gauge wire that will fit and keep the length as short as possible, within reason, to minimize any voltage drop created by the wire resistance, 18 to 16 gauge should be OK. Check the voltage at the end of the wire and at another pin on the connector to see any difference in voltage and adjust the readings accordingly.

    When you get a meter, scan the face of it, assuming you have a flatbed scanner, and post the pic. We will give you assistance with the proper inputs for the probes.

    Edit: I imagine that you have already seen this review, but I figured I would post it anyway. You can compare your voltages with those in the article. Only compare voltages measured with a DVM, the numbers acquired with Speedfan, or any other program for that matter, from the mobo will not accurately reflect the PSU output.
    November 16, 2007 9:41:07 PM

    My friend will be bring a DVM over tonight, but just to recap the modes I shouldn't have it in is current mode because that would complete the circuit right? What modes should I have the DVM in for the 3.3v rail and if you guys have any do's and do nots share them with me please.

    Theres no way I'm going to trust Speed Fan now that it showed the +3.3v rail as 3.7v, thats a bs reading.
    November 17, 2007 12:09:41 AM

    systemlord said:
    My friend will be bring a DVM over tonight, but just to recap the modes I shouldn't have it in is current mode because that would complete the circuit right? What modes should I have the DVM in for the 3.3v rail and if you guys have any do's and do nots share them with me please.

    Theres no way I'm going to trust Speed Fan now that it showed the +3.3v rail as 3.7v, thats a bs reading.
    NO NO NO!!! I hope your friend knows how to use his meter. Set the DVM for DC Voltage and insert the test probes into the meter as per follows:
    Black lead - GND/COM (common)
    Red Lead - Volts/OHM

    DO NOT plug the red lead into the 10A

    The following pic denotes the location in red. This is for illustration purposes, your meter will be different.


    !