Screeching noise, probably from powersupply

Todilo

Distinguished
Sep 20, 2010
11
0
18,510
Hello community

I just put together a new computer and I am experiencing an annoying sound. It is a screeching sound, sometimes even a scratching sound(almost like a working harddrive). It seems to be comming from the powersupply. I can create it easily by opening firefox and some big site and there it is. I also got it(sometimes) from moving windows around, scrolling and just pressing down my mouse. I have ran occt and sometimes during the test(powersupply test) I got higher sound(especially when moving the windows).

Also to note is that I sometimes get a screeching sound sent through my mic(annoying for people listening to me), this is easily fixed by unplugging and plugging in the mic, though it re-appears after some random time.

I have run memtester, occt and no obvious errors have turned up. All games/application seem to be working fine.

My specs:
Windows 7 x64
Formula IV motherboard
Athlon Phenom x6 1090 be CPU
750w AX Corsair POWERSUPPLY
120GB Corsair ssd drive
2TB WD 5400 RPM harddrive
Corsair 1600MHZ DDR3 2x2gb
Fractal R3 CASE

Some occt pics:


6704tx.png

103siut.png

ju7608.png

16kw03c.png
 

d1rtyju1c3

Distinguished
Sep 1, 2010
874
0
19,060
It may not be a big problem but if it is new just RMA it and get another one. It may be coil oscillation, capacitors leaking or a fan and it may go away. My father in law has had this sound since he built his rig about a year and a half ago and hasnt had a problem. Make sure it is the PSU it could be coming from something else MOBO or GPU.
 

1haplo

Distinguished
Feb 24, 2007
541
0
19,060
If the values are correct on the graphs, Your computer should not run. Your 12v is at 7v, 5v is at 3.7-3.8 and your 3.3v is at 1.8-1.9v

Check the voltage with a voltmeter if you have one.

 

Todilo

Distinguished
Sep 20, 2010
11
0
18,510


Where could that problem come from? Have I used the wrong positions when using the sectionated-PSU? I will get a voltmeter and see what I can see. Have these small mesasurepoints on the Formula IV motherboard.
 
^ I think the graph has incorrect labels, The "Blue" line looks like the voltage, Not sure about what the red line is. As you said the computer should have shut itself off if red was the voltage. 2nd verifing with a DVM - Yes. Might also try a diff software package to look at voltage. I generally use HDMonitor - look at Idle voltage, then run Prime 95 (or furmark) and look at the deviation (Min to max).

PSU noise is generally (ie coil/Xformer) a high pitch whin (around 20 Khz) or a low freq ie 60 Hz which is line freq not being filtered . Screeching, I would guess, Fan bearings. But what some call screeching others might call whin.

Noise in Mike, or speakers, is generally: (1) connection or (2) grounding problem (3) noise on the supply Line ( but this is generally at a given freq).
 

Todilo

Distinguished
Sep 20, 2010
11
0
18,510


Got a hand on a voltmeter but well it was broken :(. Have to get a new one to test. But true, it just looks like the graph is miss-labeled. The Microphone problem, how can I most easily ground the case? My electric skills are very low.

Is there a way to test the PSU for the noise, meaning a way that it SHOULD appear. So I know if I have to return it. I will soon test with another PSU just to rule out that the noise is not coming from somewhere else(like the GPU).
 
While an O'scope is normally used to look for "noise" it is possible to to do a LIMITED check with a DVM. For checking the +12 V & +5 V, you can normally use a Molex connector. Set the DVM to DC, Black meter lead to Black on the molex, Red meter lead to Red lead on Molex for +5 V, Yellow (or orange) lead on molex is for +12 V. When measuring switch meter to AC, should be in the millivolt range. You can also check the +12 by leaving the black lead and placing the red meter lead to the colored wire on the pci-e connector and/or at the colored lead on the 4/8 Pin MB connect. If you can not get the red meter lead to make contack, I use a straight pin (Just be carefull).

For grounding. With ac plug disconnected ohm check the ground (Computer case, or MB)to the Ground wire on the AC plug (round prong). Also do a AC voltage reading at your outlet. Black lead to Round hole, Read lead to the other to prongs one should be around 120 VAC, and ideally the other should be Zero, but no higher than 1 - 2 volts.
 

Todilo

Distinguished
Sep 20, 2010
11
0
18,510
Hi and thanks for all the dedicated replies so far. I am still waiting to get my hands on a working DVM but in the meantime I downloaded everest ultimate as suggested and the sensor results are shown below. All seem to be in order by the looks of it:

My next step is to swap out the powersupply and see if it is really the one that is causing the problem, I notice opening photoshop is a sure way of creating the noise :D.

sqpj50.png
 
Temps look great, Possibly to grear.

- Temps - what is your room temp. Unless your room temp is slightly below 68 F temps are probably low. Depending on HSF I would expect Core temps to be from 5 to 8 C above your room temps. Don't have AMD proc (Have a E6400 @3.2) and a I5-750 @ 3.2). For a full load on CPU (For an average HSF) I would expect somewheres in the ballpark of a 15 -> 19 C increase over abient. Also Not familar with CPU loading using Occt, I use, INTEL TAT with E6400, and Prime 95 for my I5-750.


On Plus 12V - Looks rock solid, While this is quite possible, most PSU will drop a very small amount when going from Idloe to Max load (Max load is max for an indiviual system, which is less than max for PSU).
For load on +12v try furmark. GPU will generally load the +12 V down more than a CPU.

Maybe some one more fomilar with your Proc can chime in.

Added. If furmark does not cause ANY decrease in +12V then the +12 V should be checked with a DVM to Verify it is infact correct.
I would also try a sencond program such as CPUIDs HWMonitor just to see if they both jive with one another.
 

Todilo

Distinguished
Sep 20, 2010
11
0
18,510


Using CPUID's HWMonitor and a OCCT CPU test(LINPACK): I got values :

Voltages
Type Min Max
+12V 11.73 11.78
+5V 4.89 4.91
CPU VCORE 1.24 1.43

Temperature ranged from 34(93F) to 41(105)
I do have a big fan :D, room temp is about 20C

 
That looks more realistic, and look good on voltages and great on temps.

I would still run furmark to verify your +12 does not drop much more. The min spec for it is 11.4 V, my own limit is to not go below 11.6.
+5 Your in good shape, I dought it would drop much more and its limit is 4.75V
http://www.ozone3d.net/benchmarks/fur/
 

d1rtyju1c3

Distinguished
Sep 1, 2010
874
0
19,060
I agree with RetiredChief, the tolerance is +/- 5% I would also run futuremark as it will load the gpu more and pull the 12v rail down more than the cpu will. I think RetiredChief also said that in an erlier post (not trying to steal the glory RC) [:bohleyk:1] . Let us know how it goes.
 

d1rtyju1c3

Distinguished
Sep 1, 2010
874
0
19,060
I am pretty sure that is the northbridge from what I have seen though the temp is most likely a false reading if you can touch the heatsink and it aint extremely hot you are probably fine. 128c(262f) would burn you. GPU looks ok may be a little high but not bad, try turning the fan up to 80% when in benchmarks and games.
 

Todilo

Distinguished
Sep 20, 2010
11
0
18,510
It sais 41C in BIOS, so should be ok. But aint that a bit too high? I know that the Formula IV have had problems with, at least I think it was, the northbridge. I am going to buy those sort of plastic plugs you put in the poweroutlet with only grounds connected to them and attach that to my computer to remove the voltage build-up. How do I best control the fanspeed for the graphics card? At the moment I am letting the BIOS control the case fans,(will add an extra fan), so perhaps I should adjust that somehow to help the GPU fan(without increasing the noise too much).