USB bus noise

[AnalogSteph]

Most importantly, unplugging the DBX from the wall removes the mains PE connection as well. So now, you have eliminated the ground loop that I was suspecting was there:

I suspect what's effectively a Pin 1 Problem that is uncovering the ground loop between the DBX (an IEC Class I device) and the PC.

Your audio cable is a plain old' balanced 1/4" TRS to TRS cable, not an instrument cable (TS to TS), right?

If so, it may be time for the soldering iron:

Are you able to modify your TRS cables? If I'm right, the problem should go away once the shield connection is disconnected at one side. Likewise when adding a line isolator like a Behringer HD400.

In theory, this sort of thing should not even be happening when using a balanced connection. In practice, it is possible for shield currents to impair the signal if internal ground routing does not follow AES48-2005 guidelines, effectively reducing the valuable common-mode rejection to much less than the input should normally be capable of. As the shield connection is carried by XLR pin 1 (which is the sleeve on a TRS plug), this is known as a "Pin 1 Problem". It is often easier / cheaper to connect pin 1 to local circuit ground compared to giving it an independent ground connection all of its own.

 

[Dilettante]

@AnalogSteph ,

Admittedly, it was a TS cable. Albeit a good and short one: Mogami W2524 wire terminated with Neutrik NP2X-B.

I have found noname, longer TRS-to-XLR at the bottom of the drawer and tested both. TRS/XLR goes first:

As you can see, plain TS picks-up significantly less noise. Also, might be relevant:

• DBX output plug - noise present.
• DBX insert plug - noise present.
• DBX line-in plug - noise is not present.

Unit was connected to mains, but turned off during all testing.

Given the information above, do you think modifying the cable / adding line isolator is the way to go here?

 

[Dilettante]

To avoid the brand-name bias, I ran 1Khz signal through both. Connected as loopback in ID4.

Unbalanced, short branded TS

Balanced, long unbranded XLR-TRS

Latter is obviously better.

 

[Dilettante]

Further, I have noticed that the problem is greatly reduced by connecting DBX to ID4 via balanced XLR/TRS and turning on the power on DBX. Last bit is not optional.

"Computer glitch" sounds are now replaced by the monotonous hum:

Zooming-in, the hum is mostly at frequencies of ~50Hz and ~150Hz with some leftovers fading away towards 700Hz.



To summarise:

• Balanced cable removes "computer noise", but introduces hum.
• If DBX is turned off - computer noise is present regardless of the cable and balanced one is actually worse.
• Adding CPU load reduces the computer-induced noise.
• Adding GPU load increases the computer-induced noise.

I think the "hum" version is the best one yet. How to get rid of it is an open question.

 

[Dilettante]

Hum lines, courtesy of REW:

Peaks are precisely at 50 and 150Hz

 

[Dilettante]

Possible workaround: ID4 II can work as a standalone preamp. Connecting it to a dedicated PSU allows to create a noiseless chain as following:

Mic --[ XLR ]--> ID4 --[ TRS in one of the channels ]--> DBX

Unfortunately, with this setup there is nothing performing ADC on the resulting signal from DBX...

Also, somewhat surprising difference between "Cheutrik" and high-quality XLR cables:

 

[AnalogSteph]

Admittedly, it was a TS cable. Albeit a good and short one: Mogami W2524 wire terminated with Neutrik NP2X-B.

It still makes it an unbalanced connection, so no wonder you were having issues. Instrument cable is not patch cable.

Peaks are precisely at 50 and 150Hz

I'd say that's pure mains leakage, possibly indicating a Pin 1 Problem as previously discussed. You're obviously located in 50 Hz terrain.

TRS and XLR inputs on the iD4 seem to differ in maximum input level, so the TRS should allow you to run even higher levels.

I would look at the TRS to XLR and see whether you can't remove the shield connection (pin 1) at the XLR end, I'd expect that to help. (Space permitting, a small ceramic capacitor of maybe 10 nF between shield and pin 1 would help keeping RF out then.) Also test continuity for the rest and fix any bad solder joints found, the cable may have been at the bottom of the drawer for a reason. Two of the most common issues on cheap cables are shoddy workmanship and strain reliefs of dubious effectiveness.

 

[Dilettante]

TRS and XLR inputs on the iD4 seem to differ in maximum input level, so the TRS should allow you to run even higher levels.

Are you saying that TRS is capable of carrying the higher signal? I was under impression that XLR is generally preferable. Won't be able to quote exactly, but I've seen various specs where device would accept either XLR or TRS and former had "better" parameters.

This is valuable as I can connect ID4 <--> DBX using either TRS-to-TRS or TRS-to-XLR.

This whole exercise makes me willing to pay for a high-quality factory made patch In fact this is what I am going to do once the difference between TRS-TRS and XLR-TRS is cleared.

As for ID4, the manual states a difference between "line-in" and "preamp on" modes which is expected:

I would look at the TRS to XLR and see whether you can't remove the shield connection (pin 1) at the XLR end, I'd expect that to help.
(Space permitting, a small ceramic capacitor of maybe 10 nF between shield and pin 1 would help keeping RF out then.)

Sure, I'll try it!

So... Literally cut the pin #1 on XLR and then solder 10nF polyester 100V DC capacitor between the shield and pin, where there is currently a wire:

 

[somebodyelse]

Are you saying that TRS is capable of carrying the higher signal? I was under impression that XLR is generally preferable. Won't be able to quote exactly, but I've seen various specs where device would accept either XLR or TRS and former had "better" parameters.

It's not about the connector, it's abut the input circuitry that it's connected to. It's not unusual for audio interfaces to have the XLR going to the mic preamp, with the option of phantom power, while the TRS is used for line level input.

 

[Dilettante]

It's not about the connector, it's abut the input circuitry that it's connected to. It's not unusual for audio interfaces to have the XLR going to the mic preamp, with the option of phantom power, while the TRS is used for line level input.

Ah, got it. Not the case with ID4, back connector is combined and serves as both line-in and mic-in. Former is activated by turning the Gain knob all the way to the minimum.

 

[somebodyelse]

Ah, got it. Not the case with ID4, back connector is combined and serves as both line-in and mic-in. Former is activated by turning the Gain knob all the way to the minimum.

I'll take your word for it about the Id4, but not everything with a combined connector works that way. The XLR and TRS parts of the connector may still lead to separate PCB connections so they can still be used for different input circuits.
Neutrik NCJ9FIH

 

[Dilettante]

I'll take your word for it about the Id4, but not everything with a combined connector works that way. The XLR and TRS parts of the connector may still lead to separate PCB connections so they can still be used for different input circuits.

Well, I think it can be tested in a straightforward manner: by adjusting the microphone gain.
If there is no change and connector is TRS, then the device is using the "smart" connector.

 

[Digby]

Is all equipment connected into your computer plugged into the same power outlet through a multi-socket extension. If not, try this and see what the result is.

 

[Dilettante]

Is all equipment connected into your computer plugged into the same power outlet through a multi-socket extension. If not, try this and see what the result is.

Here it is, click on one of the images to zoom-in:

Different sockets	Same socket

As you can see, the noise on 50 and 150Hz stays pretty much the same, but harmonics shift.
Using different sockets also reduces peaks by ~2dB.

small ceramic capacitor of maybe 10 nF between shield and pin 1 would help keeping RF out then

Here is the mockup, hope this is what you've had in mind.

Current state	Desired state

Re. capacitance: https://www.audiosciencereview.com/...citor-between-xlr-shell-and-ground-pin.34888/

 

[Dilettante]

Tested with shorter, high-quality cable made by someone who actually knows how solder. I was told that pin #1 is not connected to the shell.

Situation with the ground loop (?) has improved, especially at 150Hz which now sits 3dB lower.
Also, notice the lack of two peaks that were previously surrounding the 50Hz frequency:

I will cannibalise upgrade the old cable with the capacitor setup once this topic is clear.

Strangely, running 1KHz sine shows slightly worse performance compared to the cheap version with shortened ground/shell.
TD+N has dropped from approximately -92 to -86, however SNR has increased by ~2dB.

 

[Dilettante]

Small breakthrough. Problem goes away if:

1. Interface and channel strip are connected with the balanced cable where pin #1 is not shorted to the shell.
2. "Process bypass" button on DBX is pressed (e.g. no processing is done apart from the preamp)

Two points above shift the noise to even lower frequencies and reduce it by ~30dB which is more than enough.

Now, I wonder if it's time for a new topic with the subject "Debugging processing module in DBX"...

 

[AnalogSteph]

Sorry for missing this topic for so long.

Here is the mockup, hope this is what you've had in mind.

Current state	Desired state

Whoops, small misunderstand there. What I meant was
CABLE shield --> 10n --> pin 1. So that it's basically open at audio frequencies but effective at RF.
(Incidentally, 22 µF is about 3 orders of magnitude more than 10 nF. I was talking a small ceramic capacitor.)

I am generally assuming the shield to be hardwired to pin 1 either inside the device or the connector (or both), if it is even hooked up at all, but you know what they say about "ass-u-me". I mean, in theory all you need is the cable shield --> XLR shield connection (and it is what makes the most sense if you think about the cable and case as forming a continuous Faraday cage), but I wouldn't consider the contact nearly as good and reliable as with one of the regular pins.
I would have to think about when this could pose a problem... Let's say you have XLR shield hooked up to chassis in a particular device but pin 1 goes to regular local circuit ground - then just plugging in a cable that connects both would make pin 1 issues appear. It's probably a very specific case (and I have zero clue how common it would be), but one could make a dummy connector with nothing but a pin 1 to shield connection plus a pin 2 to 3 short to research this.

Few people are thinking of XLR connectors as actually having 4 terminals. That might be good for some gotchas in the field.

Small breakthrough. Problem goes away if:

1. Interface and channel strip are connected with the balanced cable where pin #1 is not shorted to the shell.
2. "Process bypass" button on DBX is pressed (e.g. no processing is done apart from the preamp)

Two points above shift the noise to even lower frequencies and reduce it by ~30dB which is more than enough.

Item #2 may suggest that ground routing in the processing section is problematic, or it's not the cleanest for other reasons. The bypass may have been included for good reason.

 

[Zeke]

I'm having a similar issue. I've got a ID4 MK2. Did you figure something out? I'm so fed up.

 

[shoto]

I have the USB isolator from ali in the first post, it is to avoid groundloop between DAC and ADC.

There is high pitch noise from DACs connected to a speaker amp using isolator.
It is with laptop on battery with usb DAC only connected, no ground loops, the isolator causes the noise.
I tried different DACs, laptop, cable, I even removed DC-DC converter on isolator and added power from USB phone charger, no help...

 

[Ashimaru]

I have the USB isolator from ali in the first post, it is to avoid groundloop between DAC and ADC.

There is high pitch noise from DACs connected to a speaker amp using isolator.
It is with laptop on battery with usb DAC only connected, no ground loops, the isolator causes the noise.
I tried different DACs, laptop, cable, I even removed DC-DC converter on isolator and added power from USB phone charger, no help...

You are using Adum3160 or 4160?