Low 5v rail, psu or mobo?

muddocktormuddocktor
edited July 2003 in Hardware
OK, I have a Turbo Link 420 psu in my Chieftec case that has the A7N8X mobo in it and both MBM and bios are showing my +5v rail at 4.56v or so. I opened the case up and put my voltmeter(Amprobe AM-22 digital) on one of the molex connectors and I'm showing the +5v rail at 4.86v while loaded, running F@H and 4.91v idle. I've been watching the meter for the last 20 minutes or so and it is holding rock steady at 4.86v and the hi/low log on MBM is showing from 4.68-4.52, with an average of 4.56v.

What do you think is happening? Do you think that the mobo is starting to go tits-up or do you think that the psu is croaking? I have a new NF7-S 2.0 waiting to drop in this rig but I don't want to install it if this psu is about to go. The rig seems stable enough, but I had to back my overclock down a couple of days ago to 2305 from 2337 to stop unexplainable lockups, which haven't recurred since backing down the fsb 2 MHz.

Any help that you can give would be much appreciated.:)

Comments

  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited July 2003
    grab a multimeter, grab a molex connector hanging off the PS. Measure the 5v rail, and see what it actually is. That's your only way of knowing for sure.
  • muddocktormuddocktor
    edited July 2003
    UH, Geeky, did you read my post through? I've already done this and I'm showing 4.86v on the molex connector, but MBM and bios show 4.56v or so.;)

    Any other ideas?:scratch:
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited July 2003
    :rolleyes2 :banghead: :banghead:

    Sorry... I skimmed thru it... missed that part...

    Sounds to me like the PS. If the voltage @ the molex connector is low, the PS is simply not putting out the right voltage. I'd replace the PS. I'd go for the Antec TrueControl 550... the fan control circuit on mine just died, so I have to RMA it, but the front-panel adjustable voltages are very, very convenient...

    btw... what do you think, should I adjust those so the molex connectors are @ 12v, 5v, & 3.3v (or a bit over) or so the board is reading +12v, 5v, & 3.3v? I'm inclined to think that it should be adjusted by the actual voltage, but I thought I'd double check...
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited July 2003
    Actually, the best way is a recordingVOM. BUT, lacking that the lows actually could be minor voltage drops on the incoming voltage leading to drops in the mohterboard connector 5Volt line-- can you back-probe that???

    Backprobe is done by sticking a probe into the back of the connector at right wire so you probe into the line at connector, and other probe gets stuck into the back of a black (not dark gray).

    Another way is to stick harddrives on all the connectors until it is at rated draw for PSU. IF at that point the PSU dips on 5 volt rail you have probably a problem.

    Unless you can log the voltages with time stamps or use a voltage recording device, you cannot be sure you have a superbig issue INSIDE the computer. What you are saying would make me want to try the other mobo or swap out a PSU temporarily with existing one. It looks like the CPU(did you drop voltage and MHz or just MHz, which would mean fewer watts of draw probably), mobo, or PSU or incoming power are flaking out.

    MY strategy would be to swap out, in this order, PSU, mobo, CPU. If swapping all three out do nothing then the problem is outside the case, possibly a damaged Surge Suppressor or UPS.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited July 2003
    Board closest to perfect, but nothing more than 7% out of range from true. Keep within 5% of ideal all round if you can.
  • TheSmJTheSmJ
    edited July 2003
    Actually my nForce2 mobo also shows the 5v and the Vcore a bit on the low side. I've heard that this is just a sensor glitch with some mobo voltage sensors.

    In short, it's a small problem linked to nForce2 mobos. As long as the VOM is within an acceptable range, it's nothing to worry about.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited July 2003
    TheSmJ; but 4.8v on the 5v line is kinda low... I'd rather have it @ like 4.95...
  • muddocktormuddocktor
    edited July 2003
    Yeah, it's a little low there Geeky, but well within the 5% tolerance range. I'm a lot more concerned about the super low readings I'm getting from the board's sensors.

    I found a real good thread over at overclockers.com talking about low 5v rails and what you can do to correct them. I'm going to back probe the ATX connector like Ageek said and I'm also going to measure the 5v rail at the power mosfets, like Hoot says in the referenced thread, and see what difference there is. If there is very little difference, then I won't worry about it and will change the mobo out when I get the chance but if there is a big difference I will carefully inspect the ATX connector to check for corrosion or bad contacts on it before swapping out the boards.

    BTW Geeky, that Antec truecontrol 550 was the very psu I was thinking of getting if this psu is crapping out. I really like the idea of being able to individually set each rail from the front panel of my rig and I've had good luck with the other 2 Antec psu's I have.
  • edcentricedcentric near Milwaukee, Wisconsin Icrontian
    edited July 2003
    So MudD, do you have an old analog meter around?
    Check the 5V in both DC and AC. If there is an AC signal then you are loosing a diode and/or cap in the PSU.
    If the no-load and load were the same and low I wouldn't worry. But the drop under load is not a good sign.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited July 2003
    why do you need an analog meter to do that? My digital multimeter supports AC & DC...
  • TheSmJTheSmJ
    edited July 2003
    An analog meter can show very fast dips in voltage where a digital VOM cannot.

    Unless you're using a $200 Fluke... I suppose...

    Basically, your voltage may drop too low for a hundredth of a second... a digital cannot see this, while an analog meter could show this "blink of an eye" activity.
  • jj Sterling Heights, MI Icrontian
    edited July 2003
    I'd take the reading from the MBM as the correct reading. 4.56 is very low and would explain you rig locking. Your multimetter could be out of tolence and give you a higher reading. So use the MBM. and since most microprocessors have a min. operating voltage of 4.5, AHH i'd replace that PSU right quick. could fry anything in your system that reqires 5v.
  • muddocktormuddocktor
    edited July 2003
    Finally got around to doing some probing with my VOM and it looks like I'm probably having 2 different problems with the 5v rail.

    I back probed the ATX connector and all the +5v connections are reading 4.75v, which is just barely within the 10% tolerance range. However, when I checked the middle leg of the 5v power mosfets, they were reading 4.55v, which is entirely too low.:banghead: So it looks like MBM is giving me correct reading on what is actually getting to the mobo.:banghead: It looks as if there is some extra resistance in the 5v circuit, most probably in the ATX connector itself.

    I'm going to swap out the Asus board right now for the NF7-S board and see if the voltage improves on the 5v rail. If it does, then I will know that the ATX connector is bad on the mobo itself most probably and if it doesn't improve, then the ATX connector on the psu itself is probably the culprit. If it turns out that the ATX connector on the mobo is bad, then I guess that I will RMA the Asus board back as I just got it last November or December.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited July 2003
    Yes, unless socket contacts are somehow corroded or the connector on mobo is cold-soldered for one or more 5V connectors or a connector in PSU that clamps onto prong is loose (female connector spread at its split), that is HARD to achieve unless the board is just bad.
Sign In or Register to comment.