Please identify pc audio noise in my rig!

Injuis

Member
2015-02-01 8:56 pm
So I have an Asus Essence STX II that puts out great, noiseless audio when by itself but begins to produce a lot noise of when coupled with a H6 daugther board. I'm talking the mouse movement sounds, webpage scrolling noise - lots of high frequency stuff.

I go from a almost inaudible tweeter hiss when the STX II is by itself to a bunch of usb/mobo induced noise with the H6. All the channels from the H6 also produce this noise.

The STX II gets it's power from a separate 4-pin molex cable, which is coming from a Seasonic x-750 modular cable (no other devices attached).
The H6 gets it's power from a 26-pin io ribbon cable from the STX II. It is not connected to the PCI-E directly.

My amp setup consists of a high wattage SMPS powering 3X TPA3116 amplifier modules, which I'm pretty sure is in a ground loop. Might this somehow be triggering this PC noise?

What do you guys think might be causing the H6 board to induce noise into the STX II?
 
Last edited:

Pano

Administrator
Paid Member
2004-10-07 6:05 am
Panama
Just FWIW, my system was noisy in a similar way, mostly related to what was on the screen. My system is not at all the boards you are using - it's a laptop with USB sound card. - but I knew it was a power supply problem because running the laptop in battery was noiseless. Figured it was just a cheap, noisy PSU.

Then one day I tripped over the power cord and broke the ground pin on the mains cord. Ah-ha! No more noise. The cheap PSU wasn't noisy, it was a ground loop. But no 60Hz mains noise, all whiny computer noises like yours. Surprising. I don't understand why the ground loop made those sort of noises, not 60Hz and its harmonics.

You might be having a ground loop problem. Worth checking.
 

Injuis

Member
2015-02-01 8:56 pm
Sounds like you're experiencing common-mode noise issues. What's unclear to me is whether when your H6 is in operation you are also using the outputs from the main STX II?

One way to significantly reduce common-mode noise effects is using input transformers in your amps.

Yes, I am still using the STX II while the H6 is in operation. The whole setup is meant to provide 5.1 and the STX II provides the front speaker audio. It's just surprising on how much worse it gets just by powering the H6 board.

One interesting to note is that I use an Edcor isolation transformer (only on the STX II output) and the noise still gets across, albeit slightly dampened. I've read a lot of documention and posts about ground loops but it seems I truly don't understand it. Could the unisolated inputs from the H6 be affecting the isolated input from the STX II?
 
Looks like then through adding the H6 you've created a ground loop within your PC. That ribbon cable running between the two is part of the loop and its carrying noisy ground currents to the add-on board, meaning a voltage drop exists across it. Then when you run single-ended RCA cables from both the STX II and the H6 you have a loop, so the induced voltage across the ribbon will force currents through the screens of your cables. Yucky!
 

Injuis

Member
2015-02-01 8:56 pm
At one point in testing, I had the H6 out of the case and measured a .02-.03 volt difference on the backplates. That made we wonder if the the STX II was ever designed to work with the H6. I believe the H6 was initially released with the ST I and I just figured it'd work with the STX II, since they are almost the same.
 

wwenze

Member
2008-03-07 12:46 pm
So with the H6 out of the case, what is the voltage between RCA ground vs signal,

And then with H6 in the case, what is the voltage between RCA ground vs signal?

Also check voltage between STX signal and H6 signal... this will tell us whether the issue is just thin wires

Well if the noise disappears with the H6 out of the case then the voltage between RCA ground and signal is presumably nil.
 
Power supply ground --> STX --> ribbon cable --> H6 --> H6 power ground --> Power supply ground

Yep, looks like a first-rate ground loop to me. Question is, why is it even relevant, as stuff on the ribbon cable ought to be digital. You can have as many ground loops as you want, as long as they are not sharing paths with audio ground. Well, jitter might not exactly improve even on a digital signal, but still.

Hang on a microsecond. What if...
Front output on STX --> front audio cable --> stuff that ties input grounds together --> surround audio cables --> H6 --> H6 power ground --> power supply --> STX power ground
Eww. I think this could be the problem here. Referencing audio ground to slot bracket ground on both cards would largely eliminate this (inside the case anyway), but maybe the slot bracket doesn't make good contact here (heck, they're making cases that are painted where the power supply goes these days, so I wouldn't be surprised by anything) or they didn't do it like this on the H6 or whatever. This grounding stuff is tricky business, potentially even the routing on the main board figures in.

Sounds like a job for a line isolation transformer, either from STX output or H6 output to stuff. Depends on your priorities I guess, one stereo output needs less of 'em than three, but quality requirements might differ...
 
Last edited:

Injuis

Member
2015-02-01 8:56 pm
As far as I can tell the brackets are properly grounded with the case, and the case is grounded with the amp. The potential difference should be minimal. I've looked at my PSU and it is painted, so the only contact it makes is with the 4 steel screws securing it. Is this adequate?

Also, I didn't to Pano about lifting the ground pin for my PSU, but I did inadvertently do this with a bad cable. For as long as I had that bad cable my computer was completely lifted! Dangerous! Caught this while testing continuity between the amp and my pc.

It's really looking like I'm going to have to abandon the STX II and just go with HDMI audio on a receiver. Throw more $$$ at the problem, I guess :(
 

wwenze

Member
2008-03-07 12:46 pm
Why common mode noise happens with single-ended connection

The output generates a voltage between the output pin and its own ground pin

The input measures a voltage between the input pin and its own ground pin.

Usually the input pin voltage = output pin voltage when connected. (A fair assumption, because load impedance is much higher than output impedance.)

But the input's ground may not be at the same voltage as the output's ground.

Alternative explanation: Imaging swinging a fork (the cutlery) up and down. The height of the fork keeps changing greatly, but the distance between the pins is always constant. That's your -120dB low noise soundcard, when measuring between the source's signal and ground pins. And the fork + your swinging hands is the computer. The movement is the noise generated by the computer.

Now poke that fork into something. If you're lucky, the something is a piece of potato, and it gets stuck onto the fork and follow its movement. But most of the time, you poke a wall, and only the bottom pin on the fork gets stuck while the upper pin is not stuck. You move the fork. The fork's lower pin bends. The distance between the pins is now bigger at the tips (and constantly changing as you keep moving the fork), but at the base the difference is unchanged.

Next try two forks into the same hole in the wall and move your hands at a different speed. That's two sound cards connected to one load. Although in your case the second fork isn't on your second hand, but duct-taped onto the first.

Oh, and ***safety warning*** do not disconnect the mains earth pin of a computer PSU. The earth pin is necessary for the common mode EMI noise filter, and disconnecting it will send the ground voltage (in other words, the chassis, the secondary 0V, etc) to AC voltage divide by 2.
 
Last edited:
The IO cable is pretty long, roughly 35 cm. Perhaps long enough to act as antenna?

You have at least 2 ground loops, the one above and Mobo ground -> STX -> Ribbon -> H6-> mobo ground.

As I understand it, it isn't really how long the cable is but the area of the ground loop and how much light at what frequencies goes through the loop, causing current in the loop wire. Straight wires can be an antenna a la rabbit ears but I haven't learned anything about that with regard to audio.

Also, there doesn't have to be a difference in ground potential to form a ground loop. However the loop can can cause momentary signal on ground.

Running a signal without a ground near it causes more electric field than with a ground, which is why people use twisted pairs. So that radiation in a ribbon cable can I believe add to the size of the ground loop excitation signal.

If possible you might try to get an extra cable and cut its ground, or if you are comfortable break the correct pin out. That would destroy the loops.

Please correct me if I am mistaken.
 
No. We're talking about stuff that happens at audio frequencies. A ferrite bead would do precious little about that.

There's a few unknowns involved here, unfortunately, including routing on the STX and H6. And I can certainly imagine greater fun than poking around on multilayer boards.

I don't think you can just cut the grounds on the ribbon cable. Signal integrity may turn out a bit peeved. Plus it appears to carry the power supply as well, definitely a bit hard to cut that. ;)

I suspect the ground loop may be running from STX to H6 over the ribbon cable and back to your multichannel input, with audio ground being referred to the ground that comes over the ribbon cable rather than the slot bracket (I think the Asus cards have jacks isolated from the slot bracket, but a multimeter might tell more). In such a case the only thing you can do to contain it is reduce loop area - install H6 immediately next to STX, keep ribbon cable neat and tidy, combine all audio cables as soon as possible. And even that may not help if it is supply current flowing back from the H6's digital section (in which case ground return impedance would have to be lowered). That being said, it would be dumb to combine all the "grounds" on the ribbon cable into one (I would expect separate AGND and DGND at least), but you never know!
 

Injuis

Member
2015-02-01 8:56 pm
From my multimeter, the STX II output grounds are tied to the slot bracket. Will test H6 when it returns.

In a different thread, member goodguys has got his STX II 7.1 setup working without any noise so that gives me hope that I can replicate his results without using a transformer.

sgrossklass, when you say 'combine all audio cables' do you mean combining their grounds?