i7 PC electrical noise leaking into audio recordings

skbanandtech

Junior Member
Jun 24, 2009
4
0
0
Hello,
I have a very challenging and annoying electrical noise problem that I believe is traceable to either my PC's power supply, motherboard, or processor. I hope some knowledgeable and helpful experts out there can help me figure out what to do next. Unfortunately, I don't have other power supplies, motherboards, or unlimited funds to drop and swap until the noise is gone. I need to carefully consider my best option(s) and go one step at a time. Here's the story:

Last month I built a new Intel i7 PC for music composition and audio recording. Everything was going great until I tried recording line-level analog audio from my hardware synthesizers. My i7 PC has an M-Audio Delta 1010 sound card (external A/D converter box) that was previously used in a Pentium III PC and made pristine recordings in that role. To my horror, analog recordings into this new i7 PC are accompanied by an obnoxious, wide-spectrum buzzing noise centered near 2.5kHz. This electrical noise is very obviously audible in .wav files recorded on this i7 PC.

Fortunately, I was able to reduce the noise by 8dB by disabling Speedstep in my bios as noted in this thread (ironically, these people cite acoustic noise whereas this helped reduce my electrical/audio noise). After this solution, however, the noise is still audible and needs further reduction/elimination for these recordings to be up to par.
http://forum.corsair.com/forums/showthread.php?t=75330.

A further solution is having my Delta 1010 A/D breakout box powered by a rackmount Furman power conditioner (RP-8). Supposedly these units have AC filtering circuitry within, and sure enough, placing this unit between the A/D box and the AC wall plug reduced the recorded electrical noise by another 2dB. But that's still not good enough - it's still audible, albeit at a low level.

So I have a few older PCs that have been reliable for pristine audio recordings for many years - one of which is a humble Toshiba 1.4GHz Pentium M laptop with an RME Hammerfall PCMCIA sound card (Multiface external A/D box). I repeated these recording tests using this old laptop with the following two conditions: new i7 powered OFF and i7 PC powered ON.

With the i7 PC merely powered ON (keep in mind, there are no connections between these 2 PCs.. and the synthesizer is being recorded directly into the Hammerfall A/D inputs.. so the only common link is that they both are drawing AC power..), the laptop records the SAME ELECTRICAL NOISE as originally found/noted in the i7 recordings (at a slightly lower level)!

With the i7 PC powered off, however, the laptop audio recordings were pristine as expected - no noise whatsover.

Therefore, this electrical noise is making its way from the i7 PC and into my other audio gear via my home's AC power path.

My question for the experts that know PCs, power supplies, motherboards, and AC: how should I go about eliminating this noise? What's my next best step?

(FWIW, Antec customer support recommended that I update my motherboard bios, which I promptly did, and found no improvement/change in my recording tests).

If the root cause is the i7 CPU, would switching to a different CPU be the best solution? For example, would I find solace in an Intel Core 2 Quad or AMD Phenom II CPU?

Or, is it the Gigabyte motherboard that needs to be swapped out? Isn't it the motherboard's job to filter and control the emissions of noisy chips? But isn't Gigabyte well-known as the best motherboard manufacturer out there? What could I try that would make a difference - Asus? Foxconn?

Or, is the PC power supply at fault? If I simply bought a much more expensive power supply, would I also buy better DC -> AC filtering? Is there a specific power supply model out there designed to handle the i7 and keep its noise within the PC?

Sorry for the long post, but there is a bit to consider. Your thoughts, comments, and questions would be most welcome. Here are my PC specs:

CPU: Intel i7 920 2.6 GHz
Motherboard: Gigabyte GA-EX58-UD4P
RAM: Super Talent 6GB DDR3
GPU: Sapphire Radeon HD4770
HD: Western Digital WD10000LSRTL
DVD: Sony AD-7240S-0B
PS: Antec Earthwatts 430W
Case: Antec Sonata Elite
Sound: M-Audio Delta 1010


Thanks,
Scott
 

skbanandtech

Junior Member
Jun 24, 2009
4
0
0
Thanks for the article link. It indicates that ground loops manifest as 60Hz hum. However my electrical noise problem is centered around 2.5kHz and is more of a high-pitched buzzing than a hum.

BTW, I apologize if I posted this in the wrong forum.
 

doubleOseven

Member
Jun 23, 2008
113
0
0
Could you try to swap in another brand name power supply with more wattage, to see if that helps? That's my top guess as to the source of the problem.

I'm generally not a fan of Antec (I've had multiple bad experiences with them, due to leaking capacitors and questionable designs)

Your Antec 430 watt may be underpowered for the job, such that it might be causing signal distortion (harmonics etc), that feedbacks into your line. I appreciate you're talking about a 2.5 kHz pitched hum, but who knows if it's emanating as a harmonic of another frequency etc.

Also, perhaps in your motherboard's BIOS, there is a "Spread Spectrum" setting... try enabling it, as this might help reduce spurious noise.

Also, in your video card setting (CCC), try turning on/off the vsync. I know that this, believe it or not, can sometimes help with electrical noise.

If all else fails, try another soundcard?!?

 

dualsmp

Golden Member
Aug 16, 2003
1,627
45
91
Have you ever tried overclocking the processor to see if the noise changes or stays the same?
 

Rubycon

Madame President
Aug 10, 2005
17,768
485
126
Ground loops can carry a plethora of broadband noise not just a 60Hz hum.

Filtering and isolation does NOTHING if you allow noise to ride in through a back door known as a ground loop. Using a wiring topology based on balanced power (vs. L1/CT/L2) cancels the effects of hash from ground loops as well.
 

WaitingForNehalem

Platinum Member
Aug 24, 2008
2,497
0
71
It's your motherboard. Gigabyte X58 boards produce an annoying squeal which is what made me return my UD5. Google "UD5 squeal" or "UD5 noise". The only way I could make it stop was by disabling C1E.
 

skbanandtech

Junior Member
Jun 24, 2009
4
0
0
Just wanted to write back and say thanks for all of your help and suggestions!

My initial nasty -70 dB Delta 1010 noise floor is now -84 to -86 dB. The Delta 1010 still has a dB or two of pitched AC noise riding on DC, whereas the RME (still on the Pentium M laptop...) has negligible noise on its DC around -89dB.

Another significant related finding is that analog output #1 of the Delta 1010 is MUCH more noisy (+15dB) than outputs 2-8. The DC power input happens to be right next to output #1...

Here is a summary of my initial efforts:

- relocated i7 AC cable from the same power strip as the Delta 1010 to an "upstream" path closer to the AC plug. result: at least 1 dB electrical noise reduction.

- added Furman power conditioner between Delta 1010 AD box DC adapter and AC outlet. result: 1-2 dB electrical noise reduction.

- disabled C1E and E1ST ("speedstep") parameters in the Gigabyte EX58-UD4P bios. result: 6-7 dB electrical noise reduction.

and my follow up efforts:

- disabled Delta 1010 MIDI out driver in Sonar. result: 5-6 dB electrical noise reduction.

- disabled Sonar's recording metronome audio output. result: ~1 dB electrical noise reduction.

- swapped Antec Earthwatts 430 PSU for BFG GS-550. this effected the frequency components of the AC noise (i.e. normalized AC noise "sounds" somewhat different), but did not effect the electrical noise amplitude.

- physically routed the Delta 1010 parallel cable and AD converter box to a new isolated location in the room. also tried moving power, source analog, and data cables around while recording. no effect on electrical noise.

- strip down the PC down one component at a time. decided not to do this because the Delta 1010 AD converters are disabled prior to Windows XP loading the M-Audio drivers. therefore, any measurements of Delta 1010 performance with the PC in bios (or DOS) would be unrealistic and misleading.

- disconnect DC case fans. decided not to do this because this is an unrealistic and possibly unsafe operating condition.


In summary, it was primarily the combination of the Gigabyte bios "speedstep" settings and Delta 1010 MIDI output that made up the majority of the recorded electrical noise. Among my recording equipment, this electrical noise is unique to my new i7 PC and concerns me because I have to be careful about dynamics processing applied to audio recorded through its analog inputs.

"Waiting for Nehalem" suggested that the Gigabyte EX58 i7 motherboards create an audible whining/buzzing noise. It is possible - although unproven - that the source of that problem also causes the electrical noise I experienced. Personally, I would suggest that anyone who needs professional quality audio recording to be careful when choosing a motherboard and consider the issues presented in this thread and elsewhere. Good luck.

Thanks,
Scott
 

skbanandtech

Junior Member
Jun 24, 2009
4
0
0
Originally posted by: beray
Have you turned off DES (Dynamic Energy Saver) yet? If not go here -> http://www.gigabyte.com.tw/Sup...oard/Utility_List.aspx to get the utility to do it.

DES is your mobo's regulator equivalent version of Intel speedstep implementation, it also created broad spectrum noise when dynamicly kicking in and out while activated.

Yes. Disabling the C1E and E1ST parameters in the bios accomplishes the same thing.
 

beray

Member
May 30, 2008
194
0
0
Originally posted by: skbanandtech
Originally posted by: beray
Have you turned off DES (Dynamic Energy Saver) yet? If not go here -> http://www.gigabyte.com.tw/Sup...oard/Utility_List.aspx to get the utility to do it.

DES is your mobo's regulator equivalent version of Intel speedstep implementation, it also created broad spectrum noise when dynamicly kicking in and out while activated.

Yes. Disabling the C1E and E1ST parameters in the bios accomplishes the same thing.

No, they do not do the same thing. DES turned on and off the power phases of your processor voltage regulator while speedstep changed voltage, frequency, and internal components demand of the processor (indirect control of the power phases).

As far as I know you can only turn DES off with the GIGABYTE utility, it can not be done in the BIOS as of current. DES default is on for all my GIGABYTE mobos.



"how should I go about eliminating this noise? What's my next best step?"

The best is eliminating all wide-band noise source origin such as DES and speedstep.

Read this thread --> http://forums.anandtech.com/me...id=28&threadid=2282402 then turn on LLC in your BIOS to eliminate the next wide-band noise source. GIGABYTE's LLC setting default is off.