• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

USB ground loop isolator


JDS Synapse 3D Printed enclosure version is USD$49

It's too premature for iGPR3 to order a USB isolator. It would be wiser to isolate the source of noise first, per solderdude's and KSTR's suggestions (i.e. by breaking the connection, first at the input of DAC, then at the input of DSP). Chances are the USB isolator won't solve his problem (especially if he uses those awful XLR to RCA "adapters").
 
I still don't get it
Its not just the mains PE that is the culprit but rather leakage currents in those devices from primary to secondary in those power supplies.
Furthermore not all audio devices have proper signal and ground wiring/routing inside gear and could even have inductors or resistors between analog and digital ground planes that can make things worse in practice while working fine in a lab.
Add to that sensitivity/gain of gear (transducers) causing some unwanted signals to reach audible levels where they don't in other systems.

I agree that audio gear should be immune but should and does don't always go hand in hand. :)

When wiring is done properly SE connections work fine. Balanced can surely help.
SE out to balanced in and v.v. is not a good idea either (as you already mentioned)
 
Its not just the mains PE that is the culprit but rather leakage currents in those devices from primary to secondary in those power supplies.
Furthermore not all audio devices have proper signal and ground wiring/routing inside gear and could even have inductors or resistors between analog and digital ground planes that can make things worse in practice while working fine in a lab.
Add to that sensitivity/gain of gear (transducers) causing some unwanted signals to reach audible levels where they don't in other systems.

I agree that audio gear should be immune but should and does don't always go hand in hand. :)

When wiring is done properly SE connections work fine. Balanced can surely help.
SE out to balanced in and v.v. is not a good idea either (as you already mentioned)
I wholeheartedly agree to the above. The concepts of star or hybrid grounding that are used in audio equipment are quite often very difficult to work under all possible scenarios. At work, when I was reviewing audio circuit designs and systems (not hi-fi though) I had often difficulties to explain to the less experienced colleagues that "ground loops" can be predicted (and avoided) if you know the parasitic elements and draw the possible current paths. Even a single-ended system might work then. But it is not a game of arbitrarily adding or removing groundings here and there until noise stops. This is even more difficult to explain to the average user of course.

The "I don't get it" was due to my inability to explain why the OP (original poster) didn't use fully balanced cabling in the first place (since, as he said later, his equipment does have balanced i/o). Of course, a badly designed balanced input or output will not provide enough noise attenuation but it would be a good start anyway.
 
Yep, when ordering from music instrument sellers or even China balanced cables are not that expensive.
In any case they can be cheaper than a high-speed USB isolator but more expensive than those cheap slow speed ones.
And ... those isolators will only help if the USB source is the culprit and might not help if it comes from elsewhere.

Balanced cables would be the safest option, certainly if the gear is designed to work using balanced connections.

I guess its a financial thing...
 
It's a bit frustrating that the specs for the two SMSL DACs I have (and I'm guessing the same is true of most others) are so vague about power draw, simply stating "<5W" which is just saying within USB 3.0 spec (although I recall reading somewhere that the SU-1 actually draws about 2W, but presumably with in-rush a lot higher switching on). That's higher than USB 2.0 which is kinda interesting since pretty much every PC has a couple of 'legacy' USB 2.0 ports which still have the bandwidth required. What happens when you connect them to one of those? (I'm not trying it).
 
It's a bit frustrating that the specs for the two SMSL DACs I have (and I'm guessing the same is true of most others) are so vague about power draw, simply stating "<5W" which is just saying within USB 3.0 spec (although I recall reading somewhere that the SU-1 actually draws about 2W, but presumably with in-rush a lot higher switching on). That's higher than USB 2.0 which is kinda interesting since pretty much every PC has a couple of 'legacy' USB 2.0 ports which still have the bandwidth required. What happens when you connect them to one of those? (I'm not trying it).
The SMSL DACs are all USB 2.0 only, even though the SU-1 has a USB C port.
 
The SMSL DACs are all USB 2.0 only, even though the SU-1 has a USB C port.
Yes, I know, but their power draw is quoted as "less than 5W" (i.e 1A) which is on the limit for USB 3.0, the max for USB 2.0 is 2.5W (0.5A). ETA >> I doubt the 'legacy' USB 2.0 ports on MB's of the last decade or more are actually limited in terms of PCB traces and components, or clamped to that current, but still ....
 
Last edited:
You're 100% sure this will stop a USB ground loop? I don't want another balanced cable situation where people say it'll fix my ground loop, and then for it to not work. Had to cancel my order of balanced cables earlier....
Recommendation from one happy owner using it to isolate USB for measurement:
https://www.audiosciencereview.com/...ds/cesy’s-usb-3-0-isolator.55215/post-2006151
I think MC_RME has also recommended this one as inexpensive and effective.

Ground noise problems need a systematic process to solve them, such as the one presented by Whitlock in Jensen Transformers' AN007. Note that it can include systems using balanced interconnects if one or more components has a 'Pin 1 Problem' say because the manufacturer failed to follow AES48, or because a fault has developed. If the process leads you to the USB connection as the place to break ground then an isolator such as the one above is what you need.
 
Just found this USB 2.0 version of the ADUM3160 for $22 (uses the ADUM3165):
Could this be a good alternative to the more expensive ones?
ADUM3160 is limited to 12Mbps so 192kHz 24 bit stereo is about the max it can do (9.2 Mbps)
ADUM3165 can do 480Mbps.

First it needs to be established that the USB connection is the root cause of the encountered problem for an USB isolator to be helpful.
Using one of these as a 'precaution' to prevent possible problems is pointless in most cases.
 
Last edited:
First it needs to be established that the USB connection is the root cause of the encountered problem for an USB isolator to be helpful.
Using one of these as a 'precaution' to prevent possible problems is pointless in most cases.
I know it's a ground loop, at least from what I can test. Running the DAC on just a wall powered USB has no issues. Unplugged it has no issues. And only when plugging it into my PC with a power only or power + data cable there's noise. Definitely ground loop, don't know for sure if it's USB. But I'd guess USB isolator would be sure to do the trick.

My friend said he can lend me his scarlett 2i2 and pair of balanced cables so that I can swap it into my setup to see what changes. If TRS won't fix it then USB isolator will be the way
 
Last edited:
I know it's a ground loop, at least from what I can test. Running the DAC on just a wall powered USB has no issues. Unplugged it has no issues. And only when plugging it into my PC with a power only or power + data cable there's noise. Definitely ground loop, don't know for sure if it's USB. But I'd guess USB isolator would be sure to do the trick.

My friend said he can lend me his scarlett 2i2 and pair of balanced cables so that I can swap it into my setup to see what changes. If TRS won't fix it then USB isolator will be the way
Yep - Same here with my MiniDSP situation (unbalanced from AVR > MiniDSP, but only had noise when connected to PC for tuning / USB PSU for the +5v power is quiet)... The cheap USB isolator solved it - simple as can be.
 
Last edited:
My friend said he can lend me his scarlett 2i2 and pair of balanced cables so that I can swap it into my setup to see what changes. If TRS won't fix it then USB isolator will be the way
Sounds like a plan.
But at any rate an USB isolator is a good investment as it really works perfectly by design, removing the root cause by breaking the loop rather than only mitigating its effects. It might be useful in future, in other scenarios than your current one.
 
If TRS won't fix it then USB isolator will be the way
yep, probably the isolator will work.
You can buy a cheap one with ADUM3160 if you don't have any files beyond 192/24 (or 96/24) otherwise you need a more expensive one with an ADM3165.
 
Interestingly, just discovered that my SMSL Sanskrit appears to completely switch out the 5V in from USB when the dedicated 'Aux' USB power-in is used. I was using a USB isolation dongle on the assumption that I'd get the same GPU and USB mouse noise as with every other USB interface I've tried, but on removing the dongle, silence. Very good news (and kudos to SMSL for designing this DAC right - I'm surprised they don't make this feature better known).

ETA > Spoke a little too soon - I generally use digital volume control with the amp's volume control left at c. 11.00, just tried turning it to 3.00, ear up to a speaker running a GPU stress test and there's a faint trace of whining/buzzing. Fortunately that's still silent for my purposes (in fact actually silent - I don't generally have the GPU running flat out while listening to music or watching movies).
 
Last edited:
Interestingly, just discovered that my SMSL Sanskrit appears to completely switch out the 5V in from USB when the dedicated 'Aux' USB power-in is used. I was using a USB isolation dongle on the assumption that I'd get the same GPU and USB mouse noise as with every other USB interface I've tried, but on removing the dongle, silence. Very good news (and kudos to SMSL for designing this DAC right - I'm surprised they don't make this feature better known).

ETA > Spoke a little too soon - I generally use digital volume control with the amp's volume control left at c. 11.00, just tried turning it to 3.00, ear up to a speaker running a GPU stress test and there's a faint trace of whining/buzzing. Fortunately that's still silent for my purposes (in fact actually silent - I don't generally have the GPU running flat out while listening to music or watching movies).
Yeah, sadly the noise for me is just unbearable. Even playing music kinda loud you can still easily hear it.

My friend let me borrow his DAC and TRS cables, still no fix. So I'll need an isolator. At least I know what the problem is now and how to fix it.
 
Yeah, sadly the noise for me is just unbearable. Even playing music kinda loud you can still easily hear it.

My friend let me borrow his DAC and TRS cables, still no fix. So I'll need an isolator. At least I know what the problem is now and how to fix it.
What set-up are you using (DAC, Amp, PSU)?

A Class II (two-pin AC) PSU for speaker or headphone amps renders USB completely silent on mine (any of last four DACs/dongles I've used, five if you include the Denon PMA-60 integrated which also has a two-pin AC in).

I wasn't able to find a Class II big enough that I'd trust for the TPA3251 amp I recently bought so using a Class 1 (grounded) Mean Well, the noise is there but with the Sanskrit's 'aux' power in connected it's much lower than with the other DACs.
 
I recently bought so using a Class 1 (grounded) Mean Well, the noise is there but with the Sanskrit's 'aux' power in connected it's much lower than with the other DACs.
Maybe the Sanskrit's USB GND line/receptacle shield is not directly joined to the shield/return line of the output cinches, reducing thus the ground-loop currents. The DAC pictures show the connectors are a bit distant from the back side cover. If you have an ohmmeter, maybe resistance measured between the USB-C shield and the cinch shields would tell more.
 
Maybe the Sanskrit's USB GND line/receptacle shield is not directly joined to the shield/return line of the output cinches, reducing thus the ground-loop currents. The DAC pictures show the connectors are a bit distant from the back side cover. If you have an ohmmeter, maybe resistance measured between the USB-C shield and the cinch shields would tell more.
Well, I've got multimeter, might take a gander. I could also try pulling the aux power cable out and see if it's as bad as previous USB DACs.

That said, I don't seem to be able to help myself lately - since the last post I ordered a medical grade Class II brick (a bit chunkier, 160W vs. the 120W I'm using) which will render it moot.
 
Sorry for the 1000th post on ground loops, but after looking at how expensive regular USB isolators are, would something like this cheap aliexpress isolator work?
I bought this exact variant. With some "welcome" promo it cost me $2. Not only didn't help with my mic's 1kHz noise but increased it from -75 to -66 dB. I've got similar results on differently located USB ports. The red LED kills your eyes in the dark.
 
Last edited:
Back
Top Bottom