Computer noise carrying through USB Cable

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 1:34 am

X-Trade wrote:This won't help because the noise is getting into the synth through the USB port.

Hi X-Trade! I regard your posts highly and would not like to argue with you, but in this case I think you're wrong. USB carries binary data and no analogue signal. Based on the drawing noise can get in only via the audio lines - or if the Microstation is faulty. I think you can trust me on that, had way too much experience with ground loops. It is true that USB is involved, but the final poblem is the audio signal not being referenced to the same ground on both ends. A balanced line might solve or attenuate this problem, but - like most synths - I assume that the Microstation is unbalanced too (unlike the Mbox), in which case cutting the GND line on the signal cable is not an option (and counter-advised anyways, even on a symmetrical signal path).

By the way, thinking of it a special cable might also be a cheapo solution: on a cable TS jack on one end and TRS on the other. TS should be connected to TR on the Mbox side, S (sleeve) left unconnected. This will reduce the volume but should eliminate noises within the Mbox's maximum input voltage range (around 25V I believe, probably more than enough to compensate for the noise). I'd try this first if I had some cables, connectors and a soldering iron handy...

Edit: something like cable 14 on this picture: (edit2 to better specify the drawing)

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 1:38 am

Oh, and one more remark: indeed, a new PSU _might_ solve the problem, but a transformer isolated DI box _will_ solve it for sure.

xmlguy · Post by **xmlguy** » Thu Jan 20, 2011 1:51 am

synthjoe wrote:Oh, and one more remark: indeed, a new PSU _might_ solve the problem, but a transformer isolated DI box _will_ solve it for sure.

Actually, no, a transformer isolated DI box will not solve a noisy ground from a PSU. A noisy ground is not the same as a ground loop. The synth is a DC device with the signal referenced to a ground that can float, so noise can still be carried even if the ground is later isolated through a transformer or even optoisolation. In other words, the noise technically gets added to the signal itself through the ground in this situation.

In a ground loop situation, there is a circuit that occurs due to the voltage difference between grounds that allows current to flow, which has secondary effects such as hum and noise. A DI box with audio transformer can solve this problem. USB does have analog connections, +V and ground. Normally we think of ground as being nicely stable and clean, but many computer switching power supplies lack the filtering necessary to keep the ground clean of noise transients.

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 2:05 am

xmlguy wrote:Actually, no, a transformer isolated DI box will not solve a noisy ground from a PSU.

True. But in that case the noise would be present on the Microstation's output (e.g. on the headphone out - that's why I mentioned headphones in my earlier post) even when the audio connector is not plugged into the Mbox, only the USB cable to the computer. As I understood such is not the case - sorry if my bad. I'll re-read.

BTW: if the noise is present over the HP output as above, it is as much the fault of the Microstations' PSU (or its internal construction) as the PC's PSU. Digital and audio power supplies and grounding should always be isolated in circuitry. It is true that PC PSU's are very likely to carry noise - but sometimes motherboards are just as prone to interference, so without proper diagnosis I'd not advise changing it. I'd suggest giving the cable a go, first - balanced end plugged into the Mbox and unbalanced into the Microstation's out. Only in the case, of course, if the noise is not present on the HP out with the USB cable connected.

Good luck!

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 2:39 am

I re-read the OP and indeed, we don't know for sure whether the noise is present on the Microstation's output when it is connected to the mains and the PC via USB only. So this is something for the OP to try first.

On the other hand:

xmlguy wrote:The synth is a DC device with the signal referenced to a ground that can float, so noise can still be carried even if the ground is later isolated through a transformer or even optoisolation. In other words, the noise technically gets added to the signal itself through the ground in this situation.

Yes, but the signal is referenced to ground very close to the output (i.e. the last amplifier stage), so even in a half-decent construction this should not be a problem (otherwise the device's S/N ratio would be an unacceptable figure). A floating ground should not be a problem as long as the current flowing through the resistance of the GND (or shield) lead(resulting in error voltage) is insignificant.

In fact the problem most often with such USB connections is, that even though the device's PSU is properly floated, the USB input is not. Meaning that the audio ground is practically tied to the USB ground. Now, if only the PC's ground is noisy but everything else is properly floated, then there should be no noise (error current) on ground conductors (or at least not in relation to the musical signal). The problem is, that 'proper floating' is quite difficult to achieve and probably the Protools' input is also not perfectly isolated from the USB ground. It is also likely that the the PSU of the Microstation leaks some ground. And true that the PC PSU is the most likely source of ground leak/noise.

So, assuming that both the Microstation's and the Mbox's S/N figures are believeable (with USB connected - which I think should be a basic requirement), the noise can be borne solely out of a ground loop problem.

BTW, I think we should demand manufacturers of USB equiped gear to publish two new figures in the specs: symmetrical and asymetrical USB power rejection (probably in dB). This should tell us everything about noise from a USB connection, or PC PSU, for that matter. Just my 2 cents...

xmlguy · Post by **xmlguy** » Thu Jan 20, 2011 2:50 am

synthjoe wrote:
xmlguy wrote:Actually, no, a transformer isolated DI box will not solve a noisy ground from a PSU.
True. But in that case the noise would be present on the Microstation's output (e.g. on the headphone out - that's why I mentioned headphones in my earlier post) even when the audio connector is not plugged into the Mbox, only the USB cable to the computer. As I understood such is not the case - sorry if my bad. I'll re-read.

BTW: if the noise is present over the HP output as above, it is as much the fault of the Microstations' PSU (or its internal construction) as the PC's PSU. Digital and audio power supplies and grounding should always be isolated in circuitry. It is true that PC PSU's are very likely to carry noise - but sometimes motherboards are just as prone to interference, so without proper diagnosis I'd not advise changing it. I'd suggest giving the cable a go, first - balanced end plugged into the Mbox and unbalanced into the Microstation's out. Only in the case, of course, if the noise is not present on the HP out with the USB cable connected.

Good luck!

The effect of a noisy ground can vary based on the specifics of the circuit design down the chain. It's possible for the main output to pick up the ground noise while the headphone driver amp doesn't: different amplifiers with different characteristics act differently in different situations. My advice about replacing the PSU first was an educated guess based on the experience I've gained seeing this specific problem repeatedly on other Korg gear, the R3, MKXL, & X50. A bad PSU on the computer has always been the root problem, and the best solution is solving the root problem rather than dealing with symptoms (which is what using midi cables and DI box address). Of course history doesn't guarantee that other problems may not be at fault, but it's a good starting point that will save a lot of time and trouble if the probabilities follows the trend.

Besides, most PSUs that come with PCs are usually garbage unless hand selected, so getting a good one with a quiet, large fan is worth doing regardless of this problem.

I do appreciate your thoughts and opinion on the issue, synthjoe. In this case I have some advantage of empirical evidence to guide my advice, but I guess we'll have to wait to see how it turns out. My four Aces is a strong hand, but there's still a possibility of a straight-flush in the deck.

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 3:14 am

xmlguy wrote:It's possible for the main output to pick up the ground noise while the headphone driver amp doesn't...

Absolutely right, that's why I always suggest (just as I did here, too) to connect a battery powered headphone amp to the subject output - of course not everyone has one at hand and so the HP out (if there's one) is the next best educated guess...

xmlguy wrote:Besides, most PSUs that come with PCs are usually garbage unless hand selected, so getting a good one with a quiet, large fan is worth doing regardless of this problem.

Couldn't agree more. For a proper audio workstation you must get a proper PSU. The reason I'd take the other path is that however good a PSU I have, in the case of two Yamaha devices of the many I own (CVP-92 and MM6) data and audio grounds are not properly separated. USB/'to host' ground was noisy compared to the 'auido in' ground on the same PC (not a PSU issue, unfortunately). Due to the relatively high currents involved, error voltage between the two grounds was significant. There was no other way to overcome the problem than to isolate the audio GND from data GND by using a DI box. Of course it does not help if the output of the device is not clean in the first place...

sproyd · Post by **sproyd** » Thu Jan 20, 2011 7:57 am

Hi All

I really appreciate the robust discussion around this unfortunately problem I am facing. In fact I am actually fairly surprised this problem hasn't reared its head with other users.

I have new

1) I purchased a 700W ThermalTake PSU from my local shop on the recommendation from the tech that it was a good quiet PSU within my price range. Have just installed it and while the PSU is audibly quiet.... the bad news is the pitchy signal noise remains.

2) My ATX Tower has a multitude of USBs. The noise varies between them, some worse than others but all unusable and none which eliminates the noise

3) Plugging my Sennheisers straight into the 3.5mm jack on the front of the microSTATION eliminates ALL erroneous line noise. The only noise left is a nice clear sound from the synth. However, obviously when recording through the MBox 2 the line noise is still present.

So to summarise what I have tried
i) Using MIDI cables (or just unplugging USB) fixes the problem but then I can't use the plugin in my DAW which defeats the purpose to me of having the mS - so we know the USB is causing the noise
ii) a new quiet PSU made negligible or no difference
iii) the headphone jack on the from of the mS has no noise so the noise is definitely travelling through the USB and down the 1/4 inch cables to the MBox

xmlguy · Post by **xmlguy** » Thu Jan 20, 2011 9:24 am

You can use the headphone output to connect to the MBOX2 line input, so long as you adjust the level to not overload the input, then use the MBOX headphone output for headphones. See if that works without noise.

This is the first time I've seen on this forum that replacing the PSU didn't fix this kind of problem, but not many people here use the MBOX2 (since it's more of Pro Tools related I/O box). Set the inputs to LINE, not mic, on the MBOX. You might also try the DI jack inputs on the MBOX, just to see if it has any effect. You may indeed have a ground loop problem, and a stereo DI box with ground lift should solve that problem, as synthjoe suggested. Good stereo DI boxes are rather expensive though. If the phones output on the microStation works without noise, then that's what I would use because it's a free workaround with no real disadvantages, as your MBOX has phones output anyways.

I would also try another computer, just to see how it affects the scenario. If the noise goes away with another computer, then the blame clearly lies with the computer's USB port. The PSU is the only relatively easy change vs the USB on the motherboard. You could try a third party USB card using a PCI slot.

Unfortunately, we have to make educated guesses when we don't have the gear on our test benches with all of our tools. So that means trying different elements along the signal path to see how it affects the problem.

If your MicroStation is still within the store return period, I would personally exchange it with a different model. There could be something wrong with the main outputs that make it overly sensitive to the USB power. Trying a different MicroStation will also be informative.

sproyd · Post by **sproyd** » Thu Jan 20, 2011 9:32 am

You can use the headphone output to connect to the MBOX2 line input, so long as you adjust the level to not overload the input, then use the MBOX headphone output for headphones. See if that works without noise.

Stereo 3.5mm jack vs 2 x 1/4 inch though... won't that potentially reduce quality?

Could it potentially be to do with the fact I imported the MBox from the US but am using it on NZ/AU power system (even though its plugged in via USB?)

Any suggestions on a 3rd party USB card?

xmlguy · Post by **xmlguy** » Thu Jan 20, 2011 9:52 am

No problem with the quality of the audio on the phone jacks. They just have a slightly more powerful output amplifier to drive headphones. The phones output comes from the same signal source as the main outputs use. The size of the jack makes no real difference, so long as you have a good electrical contact between the jack and plug. You'll need a different cable though, a Y cable with 1/4" stereo TRS jack to two 1/4" mono TS jacks (one labelled Tip and the other Ring).

Are you using the correct power adapter for 220-240VAC on the microStation?

As far as 3rd party USB PCI, whatever you find that's cheapest could be worthwhile to test. I'm trying to suggest the cheapest alternatives before getting into the more expensive ones.

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 10:36 am

xmlguy wrote:I'm trying to suggest the cheapest alternatives before getting into the more expensive ones.

On the same token I'd still hint to try that cable (No. 14) on the top of the page. It has a rather slim chance to cure the problem, but a cheap one to try. However, that solution works - if it does, at all - with symmetrical inputs only, which the Mbox has, but not many other computer audio inputs are like that (pro quality gear usually are).

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 10:47 am

sproyd wrote:iii) the headphone jack on the from of the mS has no noise so the noise is definitely travelling through the USB and down the 1/4 inch cables to the MBox

sproyd, you have tried the headphones with the mS USB cable connected to the PC, right? Just the signal cables were disconnected from the Mbox, correct?

X-Trade · Post by **X-Trade** » Thu Jan 20, 2011 11:45 am

It is possible that it is the MBox which has the interference.

Perhaps if you were using its dedicated PSU (if it has one?) this problem would go away. I doubt it but worth trying if you have one that will work on your power grid.

Yeah I was worried you were going to run into that problem. Popular amongst gamers and musicians running massive tower PCs are 'ultra low noise' PSU components, which are acoustically quiet. At least you have a nice quiet PC now, right?

But maybe it is not directly the problem.

Finally, have you checked that the sound is actually in your recordings, and not just on your outputs?

synthjoe · Post by **synthjoe** » Thu Jan 20, 2011 1:38 pm

X-Trade wrote:Perhaps if you were using its dedicated PSU (if it has one?) this problem would go away. I doubt it but worth trying if you have one that will work on your power grid.

Mbox does not support an external power supply. Unless you hack the USB cable which might be complicated and certainly not adviseable. IDK