How To Read Your Power Supply Voltages

[theAnimal]

In order to correctly determine your power supply voltages, you will need a DMM or digital multimeter (example).

Open your computer case and locate a Molex connector. (Note: The computer must be operating while taking readings, preferably with a load on the CPU and GPU.)

Make sure your DMM is set to DC Volts and to the appropriate range (the lowest available value which is larger than 12V).

IMPORTANT REMINDER: ONLY TOUCH THE DMM PROBES TO THE POINTS MENTIONED BELOW AS THERE IS THE POSSIBILITY OF A SHORT CIRCUIT WHICH COULD POTENTIALLY DAMAGE YOUR COMPUTER AS WELL AS THE RISK OF A SHOCK. IF YOU ARE UNSURE OF WHAT TO DO, ASK!

Insert the black probe from the DMM into a socket on the molex connected to a black wire, and leave it there until all three readings have been completed.

Insert the red probe into the socket connected to the yellow wire to obtain the 12V reading.

Insert the red probe into the socket connected to the red wire to obtain the 5V reading.

Care must be taken when obtaining the 3.3V reading, as it will be taken at the motherboard power connector. Gently insert the red probe into the back of a socket with an orange wire (adjacent to the wire).

According to the ATX specification the readings should be between the following values:

For 3.3V, 3.125 to 3.465
For 5V, 4.75 to 5.25
For 12V, 11.4 to 12.6

 

[zerocool84]

Can this be stickied? Would save a lot of threads.

 

[BTRY B 529th FA BN]

+1 on sticky

and what about other PSU & mobo voltage reading tools

how about it jonnyguru or anyone!

subscribed!

 

[Gillbot]

Lets keep the good info coming, we can update the OP with details as we go.

BTW, stickied!

 

[Zap]

:thumbsup:

Just a caveat for anyone wanting to try this. You are playing with live voltages, so take care to not accidentally short stuff out. Translation: Make sure your DMM probes touch what you WANT them to touch, and don't touch what you DON'T want them to touch. Live long and overclock!

Originally posted by: theAnimal
According to the ATX specification the readings should be between the following values:

For 3.3V, 3.125 to 3.465
For 5V, 4.75 to 5.25
For 12V, 11.4 to 12.6

Thanks for adding this in. Some people don't quite understand that it is a +/- range. For instance I've seen posts saying that PSU-A tested at 12.6v while PSU-B tested at 11.8v and thus PSU-A is the "better" unit because it gave more voltage, or because it didn't "droop" as much. Strange, because I'd think that PSU-B was the "better" unit because it can better regulate power at +/- 0.2v while PSU-A was at +/- 0.6v. Anyways, not 100% sure I'm right, but that's just my thought.

 

[HOOfan 1]

right, you might want to add that caveat to the first post...this can be potentially dangerous to both the tester and the tester's computer.

Might also want to post some pictures

 

[BTRY B 529th FA BN]

What plug would I test for fluctuations in vcore?

 

[theAnimal]

Originally posted by: HOOfan 1
right, you might want to add that caveat to the first post...this can be potentially dangerous to both the tester and the tester's computer.

Added.

 

[mindless1]

Open your computer case and locate one connector per each voltage rail, that connector being the one that goes to the most power hungry part powered by that rail. (It is not enough to measure an unused connector, that would ignore the loss in the wire which can be significant in higher current circuits, particularly with cheap PSU using higher gauge wiring).

Make sure your multimeter's probes are plugged into the right jacks, it will short out the PSU and potentially blow a fuse in the DMM or damage it if the probes are in the current measurement jacks regardless of what the dial is set to.

There is no risk of a short if touching the multimeter probes elsewhere IF each probe only touches one thing. For example, if you put one probe on the 5V socket in the connector and stick the other probe on the pin of a mosfet, that will successfully measure the difference in voltage between these two points.

 

[mindless1]

BTRY,
there is no plug for vcore, you'd put one probe on a ground wire and the other location depends on what areas you have access to. Best would be a wire going into the socket. 2nd best would be the EDIT: power plane pads on the back of the board. 3rd best (and most reasonable/easy way for most people) would be a leg on the last stage of inductors just prior to the socket in the VRM subcircuit, or with access to the rear of the board, the positive leg of the last set of capacitors immediately after the aforementioned inductor(s).

 

[Quiksilver]

Originally posted by: theAnimal
Gently insert the red probe into the back of a socket with an orange wire (adjacent to the wire).

Which wire?

 

[theAnimal]

Originally posted by: Quiksilver

Originally posted by: theAnimal
Gently insert the red probe into the back of a socket with an orange wire (adjacent to the wire).

Which wire?

The adjacent one, of course!

 

[ibex333]

When you get these readings, are they supposed to be steady? Is it ok if they fluctuate like crazy on the screen of the multimeter?
For example, in my case, when measuring the 5v reading, what I get is numbers jumping from 4.69 to 4.77 and anything in between. They do not settle at any one value for more than a second.

 

[theAnimal]

Originally posted by: ibex333
When you get these readings, are they supposed to be steady? Is it ok if they fluctuate like crazy on the screen of the multimeter?
For example, in my case, when measuring the 5v reading, what I get is numbers jumping from 4.69 to 4.77 and anything in between. They do not settle at any one value for more than a second.

The numbers may move some. I'd be more concerned that the voltage is out of spec, and would suggest a new PSU. If it's for the rig in your sig, CorsairVX450 or Seasonic S12II 430 or Antec Earthwatts 430 would be my recommendations.

 

[ibex333]

Thanks

 

[brblx]

i just wanted to clarify that there is no way to short something with a voltmeter as mentioned in the OP. you could physically short two parts that are close together by touching a the metal part of one meter lead across them, but that's it.

all a voltmeter does is measure the difference in voltage between two points. nothing is going to 'jump across' and harm anything.

edit- nm i'm stupid, mindless pointed that out above.

it's still worth pointing out that this will only uncover large problems, as quick voltage variatons won't be displayed by a meter- you're going to see an average. the only real way to uncover rails that are unstable but hover around the correct voltage is with a scope.

 

[HOOfan 1]

Originally posted by: brblx

it's still worth pointing out that this will only uncover large problems, as quick voltage variatons won't be displayed by a meter- you're going to see an average. the only real way to uncover rails that are unstable but hover around the correct voltage is with a scope.

Well the quck voltage variations (ripple) can be a huge problem. Of course the entire point of checking your voltage with a multimeter though is to check if the PSU is defective overall.

The ripple can kill your components pretty quick. The ATX spec is no more than 120mV or .12V anyway, so even if a multimeter could read high frequency ripple, it wouldn't likely stand out to the user.

 

[Modelworks]

Couple tips:
Replace the leads that come with your multimeter. They are usually poor quality, especially with cheap meters. Get some that have a very sharp point. That makes connecting to the right spot easier and less likely for the probe to slide off the point.
These are what I recommend: http://www.probemaster.com/sho..._24_26&products_id=198

Get some simple attachments like alligator clips also. It is always best to only have one hand near anything you are measuring with live voltage. With the clip you can attach the ground to one point, and measure with the other. Put the free hand in your pocket.

Measuring the Vcore using points like coils on the board is not a good idea. The supplies on the boards use a switching design. When you contact the coil solder point with a probe to the meter you are adding a long length of wire to a high frequency switching point It is like adding an antenna to a highly tuned circuit. Not a good idea unless you know what you are doing and use the proper techniques.

If you see voltages that are far off from the normal voltages, disconnect the meter from the pc and switch it to AC. Now measure again. Regardless of what wire you measure you should not get a reading over 20mv. Lower the better. That is the amount of AC in the DC. Usually caused by poor filtering in the power supply.

If you want to see quick changes with voltages, go analog. Analog meters are great for seeing voltage change over time and patterns like oscillating voltages, especially really quick times that a digital meter cannot display.

 

[buddatech]

Very good info here! I think I'll have to give it a try on my Cooler Master 850w PSU, but what my chances of getting shocked?

 

[Sixtyfivebravo]

Thanks for this post it really helped me out a lot. I used to have my programs tell me the voltage but often it was inaccurite. Anyone can help me with my 5 years old Dell 9800? The supply went out last month and I need help finding a replacement.

 

[theAnimal]

Originally posted by: Sixtyfivebravo
Anyone can help me with my 5 years old Dell 9800? The supply went out last month and I need help finding a replacement.

Start a new topic for this.

 

[shirleydeng]

Really helpful post, I'm going to buy a new laptop ac adapter.

 

[ceejay2005]

try to see if you can set it higher in bios itself. If it doesn't work then i would recommend you to install a software

 

[shirleydeng520]

Really informative, thanks.

 

[theAnimal]

Bump for re-sticky.