Power Supply check

[Tiamat]

Hi guys! Can u help me with a doubt?

Is it possible to evaluate the actual conditions of my XFX 550W PRO SERIES by checking the voltage offered in each connector using for this, the sensors detection showed in some software (like aida 64)


For example, , AIDA 64 shows

Voltage Values
CPU Core 1.260 V
CPU VID 1.076 V
+12 V 12.221 V
VTT 1.044 V
DIMM 1.512 V

If yes, what are the expected values (and variation) in each case. Do u know any other program to do this?

Ty!!!

 

[Mr.Scott]

Software cannot be trusted to test PSU voltages. Only a DMM can.

 

[erocker]

If you can find a "Kil-A-Watt"/power meter, etc. ( you can get them at a hardware store) plug your system into that and see how much it's pulling under full load. Balance that with the age of the PSU and you can decide if you need a new one or not.

 

[Bill_Bright]

If yes, what are the expected values (and variation) in each case.

The only conclusive way to test a PSU is with an oscilloscope or dedicated power supply analyzer - expensive and sophisticated test equipment. So for most users swapping in a known good spare is the best alternative.

To answer your question, PC power supplies are required to output 3 primary voltages; +3.3VDC, +5VDC and +12VDC. The ATX Form Factor standard stipulates a maximum allowed ±5% tolerance. So,

Acceptable Tolerances:

12VDC ±5% = 11.4 to 12.6VDC
5VDC ±5% = 4.75 to 5.25VDC
3.3VDC ±5% = 3.14 to 3.47VDC
​

Note using a multimeter is not conclusive either. For one, to test properly, the supply must be tested under a full range of expected loads, from an idle state all the way up to 100% load, at the supply's rated operating temperature (typically 40° - 50°C). This is difficult to do with a multimeter. But also, a PSU must be tested and analyzed for excessive ripple and other anomalies that affect computer stability. Most multimeters cannot test for ripple.

 

[peche]

there are some decently priced items...

 

[Bill_Bright]

Those are better than nothing. I have one like the one to the right of the X model I keep in my tool bag for house calls. But note those only have a small 10Ω load. They don't provide a full range of loads. And they don't test for ripple either. They can definitely tell you if a PSU is dead or missing a voltage, but they cannot conclusively tell you a supply is good.

 

[EarthDog]

Again though, its better than nothing, or software. We cant all be EE (not saying you are, just saying) and have truly proper testing equipment. So, a simple dmm or those testers which essentially do the same thing, perhaps with a bit of resistance, will have to do....if its still wonky, send it in. Its just one of those boxes where we cant do much about it.

Balance that with the age of the PSU and you can decide if you need a new one or not.

What does this tell you?

 

[Bill_Bright]

I no longer have a scope. And my shop is not that big that I need or can justify spending the money on a dedicated power supply analyzer.

Nope - swapping in known good spares is a tried and true troubleshooting method used by amateur and professional technicians alike for years. And you sure don't have to spend big money on a Titanium certified PSU as a spare. Any old PSU will easily verify if your old PSU is the problem or not - as long as it can handle the load. If you get the exact same symptoms with the spare PSU, the original PSU is conclusively not the problem. You can't say the same thing with a tester or multimeter.

And while I do have a bunch of certs and a couple degrees in electronics, I am not an EE. Technicians make things work in the real world what EE's "claim" works on paper and in theory! lol

 

[EarthDog]

Weird... felt like most of that post was a repeat of a repeat...so here it is again...lets see if what i think will happen happens......

So in the end...

1. We know software isnt reliable for voltage readings, period.
2. A dmm is accurate for voltages and will tell you if voltages are out of whack at idle and load if you are able to put one on the unit.
3. We cannot test for ripple (derp).
4. Psu testers, those listed above, except for perhaps that coolermaster, are no better than a dmm.

Essentially there is little we can do. If people are lucky enough to have a second known good psu, is really a good way to iso test the unit in question. Otherwise, we use the tools that we have and understand their limitations. Its the best weve got.

 

[dirtyferret]

How I test my PSU ...and yes never trust PSU software

 

[jsfitz54]

Use "Speccy" here: https://www.ccleaner.com/download

When open, go to motherboard, values for voltage there.

CPU CORE 1.240 V
MEMORY CONTROLLER 1.736 V
+3.3V 3.312 V
+5V 4.944 V
+12V 11.598 V
VIN6 1.072 V
CPU 1.237 V
DRAM 1.654 V
ICH Core 1.111 V
ICH PCIE +1.5V 1.508 V
CPU PLL 0.026 V
IOH PCIE +1.5V 1.812 V
IOH Core 1.098 V
QPI 1.230 V

 

[EarthDog]

Thats software. See half a dozen posts above you going over its concerns.

 

[jsfitz54]

Any hardware solution that only has a pass/fail light is insufficient.
You need raw numbers, so I'll take a software solution as a first clue, then a DMM.

The gratuitous comment about being software was not needed as you can't download hardware.

 

[EarthDog]

We agree.. and said as much multiple times earlier in thread.

Another software suggestion wasnt needed either, especially considering all the bunking that method garnered in this same thread...the user has software already. Speccy reads from the same registers as others. Is there a reason its better? Please share.

 

[dirtyferret]

Software is completely inaccurate, they all get their info from the same Mobo IC. I have an Antec PSU on a gigabyte Mobo that states my 12v rail is only at 50% of spec. Yet if I swap out PSU with another PC then the Antec is well within spec. All software tells you is how accurate or inaccurate your mobo is, not your PSU.

 

[jsfitz54]

That's funny , because the first post doesn't show 3 and 5v rails, in the software solution. Only the 12v was displayed.

So perhaps a secondary software solution that's more expansive, is a help.

Software is completely inaccurate, they all get their info from the same Mobo IC. I have an Antec PSU on a gigabyte Mobo that states my 12v rail is only at 50% of spec. Yet if I swap out PSU with another PC then the Antec is well within spec. All software tells you is how accurate or inaccurate your mobo is, not your PSU.

You can also look at the voltages directly in the BIOS, or was that mentioned already?

 

[dirtyferret]

BIOS gets it from the same IC

 

[EarthDog]

That's funny , because the first post doesn't show 3 and 5v rails, in the software solution. Only the 12v was displayed.

So perhaps a secondary software solution that's more expansive, is a help.



You can also look at the voltages directly in the BIOS, or was that mentioned already?

Sorry, Fitz, the OP likely just didn't post the 3.3V readings as AIDA64 does read that. I am staring at it now as a matter of fact. If its readable, AIDA (and Speccy) will likely catch it.

As DF just said, and I said above, they (software) get their information from the same register, that includes the BIOS.

 

[jsfitz54]

BIOS gets it from the same IC

Without the contended inaccurate secondary software solution. Cut to the chase. If the software numbers appear wrong, as Bill said in post 4, then look to a DMM to back that up. The software is not as bad as claimed by some. And it shows numbers under load.

 

[EarthDog]

Isn't the BIOS the same 'intermediary' as software is to the register?

If we are cutting to the chase... Software can be accurate, but the only way to confirm is through a DMM, not through an additional piece of software (Speccy) which reads off the same registers.

 

[dirtyferret]

Without the contended inaccurate secondary software solution. Cut to the chase. If the software numbers appear wrong, as Bill said in post 4, then look to a DMM to back that up. The software is not as bad as claimed by some. And it shows numbers under load.

The software is not as good as claimed either and can easily state information of a PSU in spec when it's not. It's the equivalent of using web md to diagnose the flu. Just because you have flu like systems does not mean web md is an accurate replacement for a flu test at the doctor's office.

 

[jsfitz54]

You keep arguing over minutiae. Are you really proving anything? The software is a first good indicator that something is wrong. The first post only showed the 12v rail. Tiamat may have omitted the other rails because he thinks the 12v is the only important V rail, when in fact, if the others are bad, he can have a bad PSU.

 

[dirtyferret]

I am not arguing over anything. I'm stating a fact as others have already said that all software that reports PSU from the Mobo IC is not an accurate measure. That fact seems to bother you because you feel at times it can be accurate. So I will make you feel better and state yes sometimes speccy can be accurate in the same way a broken clock can be accurate twice a day. I'm done with the thread as now it's completely off topic.

 

[jsfitz54]

So what exactly is your internet armchair diagnosis? AND What pill should the OP take?

So far, you haven't offered a cure.

 

[EarthDog]

Is Speccy, a piece of software which we all know can be inaccurate (like others), really more helpful? Ive had software read fine but with a MM attached was out of spec.

What the OP may have omitted or why, who knows. Your initial point was that AIDA didnt read it. I corrected that telling you it does and he just left it out. Now you've moved on to something else. You are spot on in this different discussion about all the rails being important. My point was that we bunked the use of software which he already used to get 'an idea' of the voltages. Then, a post comes in without context saying to use different software (which reads from the same registers). I noted it was software and referenced earlier posts going over the efficacy of software in general. So, a "secondary software that is more expensive" isn't really a help as you stated? DId I miss something?

Yeah, rehashed, just like any other software suggestion... I think the point has been made ad nauseum without a paid solution showing the same thing being injected. God Bless TPU........

So what exactly is your internet armchair diagnosis? AND What pill should the OP take?

So far, you haven't offered a cure.

And your solution was a "more expensive piece of software" that shows the same exact information. And here we are spinning our tires over it rehashing points already made both about testing and that software isn't a great barometer....

Oye Vey. Unsubscribed.