Testing my power supply

[entropy]

Well, mbm5 has been reading some low results ... and since i realized I had a multimeter laying around, I figured - wth! So where do i exactly put the red and black leads, to test the 3.3, 5, 12, and is there a way to test the -12 (right now THAT one reads about -6v :zombie: )?

[Black Hawk]

Your motherboard's manual should say which pin is which. Can't find an image on Google :-/

[entropy]

Wouldn't you just jump them into a molex? I have no idea which, but I do know that for the 3.3 it's on the big atx power, and I'm not fond of messing with that, really .. and it's only for memory, which is working perfectly fine. (again, please correct me if i'm wrong since i don't know a lot about this stuff :))

[Black Hawk]

You don't just "jump them into a molex". You gotta find out which is the positive and negative for each one. The manual says that. You stick the black cable in the negative pin and the red in the positive. As for the -12v, I have no idea since I didn't pay attention in digital electronics class too much :-/

[entropy]

well, when i said that, i didn't mean in a totally random fashion ;D
this is from 4-n-zics over at EOC forums (I'm a traitor :eek2: )

yes.. to measure the +12 volt rail, take the red lead from the multimeter into the yellow wire on the molex. For the +5 volt rail, plug the Red lead from the multimeter into the red wire on the molex. For the +3.3 volt rail, plug the Red lead on the multimeter into an orange wire on the 20pin ATX connector. Black lead on the multimeter plugs into any black wire on the molex/ATX connector.

set the multimeter to 20V..
I would like a backup opinion or three, from people who have done this. Also, when I test, should i start with the psu OFF and then stick them in, and turn it on? or should I do it while it's already running?

[Missileman]

4 - n - zics is right on his wire colors. I have done it several times. Easiest way is to go spend 15 bucks and buy an Antec power supply tester at Best Buy/CompUSA. Take the big ATX connector off the MB and plug it into the tester. Power on and stick your leads into the marked holes to measure.

If you want to do it manually I usually put in my leads and then turn on. Turn off and move to the next. I do this just to be safe, but a multimeter shouldn't cause any issues. They are an isolated device except if they are set to measure amps. Then they will cause a BAD short so just make sure it's set for Vdc and you'll be fine.

[CaffeineMe]

I've always had good luck sticking the black into a molex connector...black to black....the black is ALWAYS negative, and any negative line will do. Then, connect the red side to any colored wire, and you can test the voltage across it. In fact, I usually just stuff the black wire from the meter into the black connector on the molex, and probe to my hearts content! :)

Just NEVER let the red probe touch a black wire/connector, and you shouldn't have a problem.

I am not responsible for any errors that result from the above information!!! :)

[entropy]

lol CF, nice disclaimer :p
Thanks much, guys. I have an Auto-Range Digital Multimeter...and Missileman, you say vdc .. and I assume DC V is the same but hell, i'd like to not blow this power supply, cuz i might be able to sell it for an antec. so DC V then? and i should probably test idle and load, right?

[entropy]

sry for the double post ...
multimeter to 20v? what does that mean?

[mmonnin]

No modern computer uses the negative rails like the -12V so dont worry about that one.

[entropy]

alright thanks :) then all appears to be in order. +12v is 12.08-12.1 ... +5v is 5.07-5.08 ... those are good, right? not perfect, but nothing that's limiting my overclock?

[muddocktor]

Those are real good and shouldn't give you any problems with your overclock.

[mmonnin]

Ditto that!

[entropy]

Curses!
/me shakes fist at the world
That means my chip is topping out :-/ I can go from 1.6v to 1.68 lol ... bit too high of a jump for my tastes, seeing as how SNDS occurs at 1.7-1.75 :(

[MediaMan]

Getting a higher voltage on any of the rails isn't the crucial determining factor for overclocking. What does affect system stability is abnormally low voltage (11 v on the 12 v rail) or constant fluctuation.

When you do overclock you are putting more of a load on the PSU...it is the drop/fluctuation that can produce problems. Unless you are asking for 20v off a 12v line.

[Straight_Man]

Getting a higher voltage on any of the rails isn't the crucial determining factor for overclocking. What does affect system stability is abnormally low voltage (11 v on the 12 v rail) or constant fluctuation.

When you do overclock you are putting more of a load on the PSU...it is the drop/fluctuation that can produce problems. Unless you are asking for 20v off a 12v line.

Perfect, but I think 4_N_Ziks (SP?) was saying to set DMM\VOM(Meter) to 20 VDC RANGE for testing. What I want is a Fluke graphing\recording DMM\MiniScope (about $1,800.00 USD wholesale with the PC software (from Fluke) to interface an analyzing PC or laptop to the miniscope), myself-- because it has a much smaller error factor than the RadioShack I do have and I can record flux patterns to the analyzing PC or laptop, but I make do with a little 40 range RadioShack higher-end DMM\VOM (about $235.00 USD retail as of 4.5 years ago)-- and on that the 20 VDC range gives more accuracy than the 40 VDC range does (accuracy on the DMM I use is +\-.1% of RANGE for the voltages at 20 and 40 VDC ranges, the 20 VDC yields an average error of .02 V, the 40 VDC an average of .04 VDC).

Rule for DMM VDC (Voltage\DC) testing is: set DMM to range just above the value the voltage SHOULD be in for first test, then go to next higher range if you get an out of range error (on my DMM it shows a numeric 3 integer as out of range error)in order to try to get an absolute value-- if its out of range for the next higher range, on any test, trash the thing being tested unless you can replace the power ciruitry in it.