USB ground loop mystery with synths and Pulse 16

[unusualcadence]

Hello! Hoping some experienced minds can help me with this age-old issue. I’ve googled this extensively and scoured these forums for help, but it seems that everyone’s circumstances are different when trying to fix a USB ground loop issue. I'm also a little bit confused as to why mine is happening, so I might be able to learn something in the process! I'm very sorry for the long post ahead, but it helps to be specific in matters of weird audio stuff!

I’ve just re-cabled my home setup with a new interface for recording synths in to my PC, and I’m getting grotty digital noise at the A/D stage if I have an audio cable and a USB cable connected to the same synth. Every synth has unbalanced outs, so I’m using unbalanced cables to a patchbay, which then goes in to the 16 channels of the interface (Ferrofish Pulse 16 in to an RME Digiface). The only device that uses any bus power is the Digiface, everything else has a power brick in to a power strip in to the UK mains.

A good problem example is with the MS-20 Reissue - I don’t have any noise or hum whatsoever anywhere in my signal path to the speakers UNTIL I plug in a USB cable from the PC to the MS-20 for midi purposes. When I do this, all the horrible computer noise appears in the Pulse 16 input that audio cable is patched to. Gross!

It doesn’t even matter if I unplug the MS-20 power from the mains, the noise still appears when the USB cable makes a connection between a synth and the PC, whether from the PC directly or from a powered USB hub. I also get this digital noise in the output from a vintage Pro-1 if I connect the CV/Gate to a Roland SBX-1 USB interface! It seems all the usb connections on my PC are cursed!

The PC, USB hub, Pulse 16, SBX-1 are all plugged in to a belkin surge protected power strip together (along with two active monitors and two computer screens), so they should all be grounded together at the same outlet, which makes me very confused about the whole thing!
If I plug the audio cable form the MS-20 in to a lil' Behringer Xenyx mixer on the same power strip, there is magically no noise... unless I then plug the output of the mixer in to the Pulse 16, which brings it all back. Argh. It feels like the Pulse 16 is the problem point here??

Synths are all on a power strip going to a separate wall socket - and even if I unplug them and remove them from the equation and have just the PC, USB Hub and Pulse 16 powered on, the ground loop noise occurs if the PC is switched on, and a USB cable and an audio cable is connected to the same synth.
Even if I remove any connection from the Pulse 16 to the PC (through the ADAT connectors on the Digiface), the noise is still there. I admit that electronics is not my strong suit, but this is the first time that I have had no idea how to break a ground loop!

I have tried shuffling all the power around to run off of the same two dual wall outlets, putting the usb hub and pulse 16 on a different wall socket to the PC, putting the PC on its own wall socket, putting ferrite clips on the usb cables and power cables, swapping usb cables out, swapping usb ports, removing everything from my PC other than the USB cable going to the synth, using balanced cables _just in case_...
...pretty much everything except using one of those usb ground lift devices, which I don’t hugely trust after a youtube review stated that one 'burst in to flames as soon as it was plugged in'. Not keen on setting fire to my flat over this!

What did work was bodging together a connection to a stereo RCA ground loop isolator on the MS-20 output - that completely cleared up the noise, so it suggests that it is a ground loop issue. If there’s no better solution than ‘ground loop isolator on the synth output’ (although I can't seem to find any 1/4" jack isolators, only stereo RCA or 3.5mm ones, if there are any suggestions for a reputable one I'd gladly take the recommendation) then I will go with that, but I’m wary that those are a plaster over some bigger problem, AND I don't really want to spend a fortune on buying sticking plasters for all my synths! Ready to get spenny when I have properly diagnosed the issue, however, just want to make sure I'm going to be buying the right thing!!

If anyone has any more good investigative ideas to help me solve this, and maybe even be rid of it entirely, I’d be very grateful for the help!

[ef37a]

https://www.amazon.co.uk/AV-Link-Ground ... th=1&psc=1

Those ^ are cheap enough that you could get a couple? Yes, RCAs but RCA to TS jack adaptors are common and very reliable in my experience. You can easily fit jacks if you can solder? The transformers are of course not of the best quality but at the -10dBV levels or lower that synths seem to put out I think they will be fine, except perhaps below about 80Hz. However, if you look upon them as a diagnostic tool and they fix the noise then you can invest in better from the likes of Art or Orchid.

Do you own a digital multimeter? If not get one and check if the power supply for the Pulse 16 carries the mains earth through to the DC side. If it does, there I think is the root of your problem...Now: YOU MUST NOT REMOVE THAT EARTH! What you CAN do however is look for a 12V 3A supply with a fig 8 bi-pin mains connector, i.e. SAFELY earth free.

You say "PC"? But do you mean laptop? If so do you get the hum on battery operation? Again an earth free charger should solve the problem. I have such a universal supply of Duracell brand.

Hope this helps.

Dave.

[unusualcadence]

Thank you for reading all of that and replying!

Those ^ are cheap enough that you could get a couple?

I have a similar looking isolator from doing laptop gigs, which _did_ work to quell the noise when I hooked it up to the MS-20. However, I had to use the stereo Phones output rather than the mono Line output, as the latter through the isolator seemed to act as a huge high-pass filter.

Definitely wouldn't hurt to have one of those Orchid isolators around the studio, so worst case I'll invest in a few of those!

Do you own a digital multimeter?

Yes I do! It's about time I learned how to use it as well. I'll check the earth on the Pulse 16. The plug on it is fully metal pronged, but I will measure this as it's one of those that comes off the device down to a 'brick' that has a removable fig 8 bi-pin socket on it (looks like this https://www.acadaptorsrus.co.uk/12v-2-0 ... r-fig-8-c7). So maybe it's the source of the problem after all!

The PC is a full-fat tower workstation, unfortunately. I might whip out a laptop for a test though!

[Hugh Robjohns]

Ground loops are mysterious and confusing, and tower PCs can be a complete PITA because so many are badly are badly grounded internally.

Reading this might be helpful:

https://www.soundonsound.com/sound-advi ... ound-loops

In your case, the USB grounds seem to be causing most of the problems, but isolating them all would be expensive.

So too would be buying a load of transformer audio isolators....

However, since the Pulse has balanced inputs, my solution would be to make up (or have made) pseudo-balanced audio cables for all the problematic synths. These will break the ground loops and, hopefully, cure the noise problems.

In essence, what you need to do is connect the synths' unbalanced output signals to the Ferrofish's balanced inputs without linking their grounds via the cable screen... ideally using balanced cables.

So, the Ferrofish end is a TRS plug wired as normal — hot on tip, cold on ring, and screen on sleeve.

The synth end is a TS plug, and the easiest wiring arrangement is with the hot on the tip and the cold on the sleeve. The cable screen is cut back and insulated so it can't touch the sleeve at all — and thus there is no direct ground path and so no ground loop.

This is the cheapest and easiest solution if you can solder your own cables.

This simple 'fix' normally works fine... but you can tart it up with a small resistor and capacitor (in parallel) between the cable screen and TS sleeve for better interference rejection, if required. However, there isn't much space in a TS plug, so this is tricky to achieve unless you use smd components... which is fiddly!

It's worth noting that ground loops normally involve the equipment's mains earth and audio connection screen. In devices with two-core
Mains cables, or with wall-wart / line-lump power supplies there isn't a mains earth connection, and thus can be no ground loop.

However, connecting a USB cable can provide an alternative ground path and a new loop... which appears to be your problem.

Conventional 5-pin DIN MIDI was designed with opto isolators specifically to avoid direct electrical connections between gear, and thus avoid ground loops... so if you can use DIN MIDI instead of USB MIDI you'll make life much easier for yourself!

As per the article mentioned above, I'd start with powering everything from a single or dual wall outlet. Powering different gear from different wall outlets is guaranteed to cause ground loop issues.

And I'd also disconnect all the synths, establish that the system is quiet, then connect the synths' USB and audio one by one, resolving the ground loop noise as you go until everything is connected.

[ef37a]

Ah, b***r! That PSU is NOT earthed so you won't solve things that way.

Yes, some outputs can be fussy when feeding transformers* this could still be a problem even with a high quality unit from Art or Orchid so make sure you can return whatever you get.

There are USB isolators, 'pulse transformers' I think. They are quite expensive and I have never tried one and they might have a detrimental effect on the USB bandwidth?

*There is actually no such thing as a "universal" audio transformer. Even those with a 1:1 ratio fall into two extreme camps. "Output" transformers which are designed to work at a few mW from a very low, <100 Ohms source and "10k line bridging" transformers which can be used on high (for most things) impedance inputs, "10k line" ins say, These are generally much more expensive than low Z traffs. The stuff sold as "earth isolators" seem to be somewhere in between but I always use them on outputs.

I actually have a pair of very good OEP 10k traffs that I installed in a die caste box many years ago. It is not used any more and I MIGHT be persuaded to donate it to a VERY deserving cause! Such devices are very valuable for stopping serious RFI intrusion.

Ah yes, Hugh's suggestion is well worth a shot.

Dave.

[unusualcadence]

Thank you for your detailed response Hugh, much appreciated!!

It's worth noting that ground loops normally involve the equipment's mains earth and audio connection screen. In devices with two-core
Mains cables, or with wall-wart / line-lump power supplies there isn't a mains earth connection, and thus can be no ground loop.

However, connecting a USB cable can provide an alternative ground path and a new loop... which appears to be your problem.

So if the Pulse 16 is not earthed through it's PSU, the USB cable is making that ground connection back to the PC, and that is where this loop is actually manifesting? That would make sense as to why the conventional method of plugging everything in to the same wall socket isn't working for me here - even if I have ONLY the Pulse 16 and PC plugged in and powered up, that noise is present!

So it seems like my options are:

* Hacked TS-TRS cable for problem synths. Seems like a good idea, and I just so happen to have a TRS cable to experiment on! I haven't used balanced looms going from the patchbay to the Pulse 16 inputs as I didn't want to create any issues with TS and TRS cables messin' with each other, but I should probably just bite the bullet and get balanced looms for all the Pulse 16 inputs.

* Ground loop isolator for problem synths. Also seems like a good idea, if it doesn't have any side effects to the signal.

* Swap to DIN-MIDI. I am probably going to do this anyway with a Kenton THRU-12, so I can get everything grooving off of the SBX-1. This solves pretty much all my problems!

* USB ground lift dongle thingy. Ever since I learned on a fire safety course that dodgy USB chargers are responsible for an alarming number of house fires, I am eyeing this option with extreme suspicion!

I actually have a pair of very good OEP 10k traffs that I installed in a die caste box many years ago. It is not used any more and I MIGHT be persuaded to donate it to a VERY deserving cause!

This is very kind to bring up, although I wouldn't count myself as a worthy cause, haha! Hugely appreciate your time in answering my posts though. Helps keep away the madness that comes with troubleshooting these things!

[Lostgallifreyan]

Ground loops may have various noise on them, but usually with a strong mains hum in there. You say it's mostly digital noise, so try getting a bunch of USB cables with ferrite beads on them. You're mixing unbalanced audio connections and USB in the same system so this is best done anyway. It may be enough to solve this, or maybe not, but it's easy to try and usually does something useful.

You may well have to make special cables for audio but if you do, have a look at this:
https://www.ranecommercial.com/legacy/note110.html
There's a downloadable PDF of that too.

[Hugh Robjohns]

Ground loops may have various noise on them, but usually with a strong mains hum in there.

Yes... if the 'loop' is via the mains safety earth cabling inherently passing mains frequency currents.

In the OP's case, the loops are apparently via the USB screening and the PC's internal grounding arrangements, so assorted digital 'shash' is the predominant noise.

...try getting a bunch of USB cables with ferrite beads on them.

Ferrites are helpful for reducing the effects of radiated RF interference, but do nothing for audio frequency earth loop currents. So while generally a useful feature on high-data rate cables they are unlikely to be much help in this situation.

[Lostgallifreyan]

Ferrites are helpful for reducing the effects of radiated RF interference, but do nothing for audio frequency earth loop currents. So while generally a useful feature on high-data rate cables they are unlikely to be much help in this situation.

He has digital noise. That may be RF based, demodulated in the oxide on some connector so it becomes audible. If so, then stopping the RF at source is useful to try. It's the easiest and cheapest thing to do, so worth trying first. It's not so much the rate that is high, it's the fast edge transitions, so a 'low rate' USB system might still have a lot of strong RF harmonics even on occasional pulses. So why not get some USB leads with filters on general principle, it's what they are for..

Doing something about the unbalanced lines will help, but it's best to reduce the scope of need for it first.

[ef37a]

Ground loops may have various noise on them, but usually with a strong mains hum in there.

Yes... if the 'loop' is via the mains safety earth cabling inherently passing mains frequency currents.

In the OP's case, the loops are apparently via the USB screening and the PC's internal grounding arrangements, so assorted digital 'shash' is the predominant noise.

...try getting a bunch of USB cables with ferrite beads on them.

Ferrites are helpful for reducing the effects of radiated RF interference, but do nothing for audio frequency earth loop currents. So while generally a useful feature on high-data rate cables they are unlikely to be much help in this situation.

Right. I am old enough to remember, as are many here, when every cable that connected to a computer, mouse, kbd, printer etc, had a lump of ferrite each end. The practice seems to have died out and I for one have never noticed any difference in the equipment's performance.

I have advocated the use of ferrites on audio cables, especially guitar cables for cases of RFI as a first line solution but it rarely does much good. If the peeps that built the kit did not put proper RFI stops inside about all you can do is retro-fit them.

Dave.

[Lostgallifreyan]

I have advocated the use of ferrites on audio cables, especially guitar cables for cases of RFI as a first line solution but it rarely does much good. If the peeps that built the kit did not put proper RFI stops inside about all you can do is retro-fit them.

I agree it often fails to help, but it's an easy first try. I'd do it on the USB lines before I went into a rats-nest of unbalanced audio lines. The big help in this case may be that unusualcadence didn't have trouble before plugging in something specific. In which case that Rane document I linked to might be a good place to look next.

(About phasing out use of ferrites on loads of gear, I think firms realised it was cheaper to buy cables with precise twisted pairs instead of relying on expensive ferrites. Reducing RF from cables that way works better. It still amounts to reduction at source where possible.)

[Hugh Robjohns]

To reiterate... ferrite chokes can indeed be useful for reducing RFI issues.

However, the OP's description of symptoms and experimental results give no indication that RFI is the principle cause of unwanted noise here.

The evidence all points to being ground-loop related. Indeed, his experiment with an audio isolation transformer effectively proved both the ground loop theory and evidenced a classic solution which fully resolved the problem.

The Rane document explains how and why audio isolation transformers offer the ideal answer to issues of this sort... but good ones able to pass high levels of deep synth bass with negligible distortion are expensive, bulky and, if not positioned carefully, can introduce more unwanted noise.

Altering the wiring as I suggested above, to take advantage of the interface's balanced inputs, is simple, cheap, quick and usually very effective in the OP's situation.

I have employed this technique very successfully in my own studio on some of my own synths to avoid USB MIDI noise problems. No ferrites have been installed on any USB lines.....

[Folderol]

If push comes to shove, you can get USB isolators - don't know if there are any that can do the full USB3 negotiation

[Sam Spoons]

That probably doesn't matter as none of his devices are USB3, even the Digiface with 32/32 channel I/O uses USB2.

[unusualcadence]

I did clamp ferrite sleeves on to every cable I could imagine as my first port of call - much to my disappointment, it didn't do much! I've referred to it as 'digital noise' as it's all pulses, whines, static and other detritus that's obviously originating in the PC, but I'm probably using that term as a bit of a misnomer. In my head, ground noise is the constant angry mains hum I'm used to hearing in guitar amps, but I think what I'm experiencing is just a different flavour of ground noise!

[ajay_m]

The USB1.1 isolators are cheap and work well in many scenarios. However they can only supply about 400mA downstream as their little isolated power convertor has limited power. If the target device is bus powered this isnt of course a problem and otherwise it is possible to overcome this limit by splicing in a phone charger downstream of the isolator with a bit of cable hackery.
The more recent usb2 compatible devices would only be needed for more than 4 channels of audio and I purchased one but returned it as it did not work in a number of situations where the usb1.1 unit worked perfectly.

[Lostgallifreyan]

Maybe one good powered hub (could even be a USB3 type on the basis that it's compatible with older variants, and may have better filtering), connected to the PC via a short cable that may or may not include an isolator, might help stop noise getting out of the PC as much as it does now.

It may even be worth going inside the PC if there is a USB chassis port attached via a pin header on a mainboard. That pin header might have a weak ground contact, and the wiping action of replugging it might help. If the USB port is attached directly to the mainboard there's not much you can do except try to find a powered hub that does some filtering of noise from the PC end.

About the Rane doc, I think its early emphasis on a transformer might be on the basis that no other intervention is done. Those special cable arrangements may do more, at the expense of more work to determine which fix to apply, and where..

If I had some nice old gear with unbalanced feeds, I'd consider retrofitting a balanced line driver for TRS or XLR output. It's possible that small bulk supplies might make this economical, but I'd only do it if finding room for internal fitting and wiring was easy.

[ajay_m]

I have posted on this topic before but what is happening here is that inside the PC very large switched transient currents flow through the power distribution system. The voltages are low, at the CPU less than a couple of volts, but the currents can exceed 50A or even 100A for more powerful chips.

Now those currents flow through the motherboard and between two nominal ground points on different connectors and also they flow from.a connector ground on the PC through attached cables to any other ground point on external gear.
As an audio signal source this then can develop several millivolts across a pretty solid piece of wire because the source impedance is probably 0.01 ohms or less.
You can often see that if you bridge a cable with a stout piece of thick speaker wire grounded where the cable ends plug in the spurious signal decreases because now the target impedance isn't as high. Most usb cables have pretty thin wire so could easily look like 0.1 ohm and that's far from a short circuit for a source impedance 10 times lower.

So the cure is not to give that low impedance noise source a ground to ground connection to act as a target. This is where breaking grounds almost always resolved the issue.

In some extreme cases the noise spectrum goes into the RF region and in this case ferrite beads can prevent noise coupling via cable capacitance but this is a separate conduit to the primary issue.

[Lostgallifreyan]

As an audio signal source this then can develop several millivolts across a pretty solid piece of wire because the source impedance is probably 0.01 ohms or less.
You can often see that if you bridge a cable with a stout piece of thick speaker wire grounded where the cable ends plug in the spurious signal decreases because now the target impedance isn't as high. Most usb cables have pretty thin wire so could easily look like 0.1 ohm and that's far from a short circuit for a source impedance 10 times lower.

So the cure is not to give that low impedance noise source a ground to ground connection to act as a target. This is where breaking grounds almost always resolved the issue.

I like your explanation of the high currents as a source, and of a low impedance source driving noise currents to remote grounds via cables. It suggests that if the source were shunted to ground locally, with a very low resistance, it would reduce the voltage that drives ground currents beyond the case. This seems to be what the 'backplates' in computers are trying to do (in addition to RF shielding). (unusualcadence said it's a full tower PC so I guess it uses one).

On the better backplates I have used, there is a thin leaf-type layer of metal formed to get a very tight fit with the aperture behind a mainboard, and several curved spring-loading contacts that touch all the metal parts of the socketry on the mainboard. Sometimes these things are fitted well, sometimes they aren't, so it might be worth looking at whether the backplate is doing its job. If it's not, the voltage in the USB connector ground may be high enough to drive noise currents. Checking tightness of standoff screws on the mainboard might help too.

Unusualcadence, you mentioned trying with a laptop... if you did, did it make a difference? If it was better, then looking at the PC backplate might show whether a better fit might be had. It was a definite weakness in some systems I have seen (curved contacts lifted high during a difficult fit, sometimes the whole plate fitting loosely).

[unusualcadence]

Thanks to everyone bringing their wealth of knowledge here! I am definitely learning some things along the way.

I have hooked up my laptop and connected all the offending gear to it, and magically there are no ground loops to be heard at all. So that's something! Although is this because I now have a situation where there is no ground path at all?

Regardless, it seems like poking around in the PC Tower might be an idea. It's a bit of a Ship Of Theseus build, so wouldn't surprise me if there's something in there that isn't fitted in the optimal way. At one point I added a PCI-E USB 3.0 expansion card, and also fitted two extra USB ports to a header on the motherboard (https://www.amazon.co.uk/Bracket-Two%E2 ... B09C8HBWDC) because I expected to be using a lot of USB devices, but half of them are going un-used now. A bit of triage inside the case might yield some results.

I've got DIN MIDI up and running for everything that needs it now thanks to a splitter box, but I'm in to deep to just call it quits without trying _everything_ I possibly can!

[Hugh Robjohns]

! Although is this because I now have a situation where there is no ground path at all?

I don't know which (if any) other devices are class-I (earthed). Desktop PCs are normally class-I, so you've removed one potential ground-loop path.... but you've also replaced the motherboard and thus changed its internal grounding arrangements, and that could well be the reason all that unwanted noise has reduced.

In systems with multiple ground-loop paths, the level of unwanted noise can increase or decrease when paths are added or removed due to differing current phase relationships on different loops.

That's why trying to find problems in a large system can be a hair-pulling nightmare, and why it's so important to add one device at a time to the rig and resolve any noise issue before adding the next.

A bit of triage inside the case might yield some results.

Indeed it might!

[ef37a]

Whilst you are inside the PC check all the MOBO scews, usually M3, are present and tight but also check that the pad they fix to is actually being bonded to the chassis. Sometimes peeps go a bit nuts with the red locking paint and insulate the pad from the screw!

Dave.

[Lostgallifreyan]

And another thing to check, if you're going into the PC...
Look for bulging or leaking electrolytic capacitors, especially near the power supply connector and near the CPU. A failure of decoupling there might account for a lot of switching noise spreading where it shouldn't.

If you do see capacitors in this state, then eventually, after noise, you can expect data corruption, so it's an important thing to check. A lot of mainboards got fitted with caps that failed early because of a poor quality component supply. Some of those boards may still be in use even though newer ones are less likely to be at risk.

[unusualcadence]

I gave all of those things a look-over while scrambling around inside the case. Thankfully no freaky capacitors, as far as I can tell!

I'm not sure specifically what was most effective, but pulling out all of the unused internal power cables, taking out the USB PCI-E card, tightening the motherboard m3 screws, generally neatening up the routing inside the case, and swapping the power lead for a newer one has significantly reduced the volume of the ground noise. Not eliminated it entirely, but it's definitely had a very noticeable effect!
I suspect a fancier power supply will help me route some of the cabling in an even more practical way (annoyingly right now I have to route a motherboard power cable very close to the GPU and CPU, which I'm sure is not ideal), but I'm pleased that I've actually had some success in reducing the noise at the source. It really goes to show just how all these little things can add up!

I've got a date with a soldering iron coming up, and I'm very interested to see what a difference a pseudo-balanced cable can do on top of this. I have several practical ways to chip away at the issues from every angle now, hooray!

[ajay_m]

The "bad caps" story (which plagued a lot of gear about a decade or so ago) comes from a piece of industrial espionage that went wrong. Low ESR (equivalent series resistance) capacitors are critically important in switching power supplies and the composition of the electrolyte is a vital component.
A rival company stole the formula for the electrolyte from a competitor and it then became widely disseminated but unknown to the thief, they had only part of the formula. This omitted an essential stabiliser component which meant that the bogus caps failed prematurely.
Behringer were hit particularly hard which is why almost all of their older gear is probably going to need recapping, but many other manufacturers were also affected.
Moral of story, choose reputable manufacturers like Rubycon for your caps, you get what you pay for.