What's wrong with my RME Audio Interface?

[peniku8]

While doing some voltage calibrations today, I found out that the positive and the negative end of a balanced output on my RME Fireface UC do not output the same voltage. I measured 1.9V on one lead and 1.3 on the other (don't remember which was which, but I think the positive had a higher voltage).
I then recorded from both wires and also recorded a differential signal (which should sum to zero if I understand that correctly?).
Negative: -18.4dbFS; Positive: -15,9dbFS; Sum (Positive->Positive, Negative->Ground): -11.6dbFS

Signal generator was REW in ASIO mode, 1khz. Here is the waveform of the measurement across + and -

Am I missing something? If I understand the differential nature of balanced connections correctly, the signal should be mirrored, which should sum to zero, or just be twice the voltage of the individual channels, when measured differentially (since they should output the same voltage?).

 

[MRC01]

... If I understand the differential nature of balanced connections correctly, the signal should be mirrored, which should sum to zero, or just be twice the voltage of the individual channels, when measured differentially (since they should output the same voltage?).

Yep, that's how it should work: sum is zero and difference is twice the + signal. Differencing them doubles the signal and cancels some noise & distortion. But if they aren't mirror images of each other then it wreaks havoc and simply doesn't work.

 

[DVDdoug]

I don't think the differential signals are required to have any particular relationship to ground...

 

[restorer-john]

There is obvious distortion visible too. What is your ground reference where you are measuring from? The 3rd pin (sleeve) of the TRS bal output or have you picked a chassis point?

Many interfaces have a resistor to Gnd/0v to protect the outputs from shorts. This can result in some strange output voltage readings depending on where you connect your negative. Also, the USB shell will connect to a PC ground.

I also had a bunch of trouble with some interfaces in the past where the balanced output is not really balanced. It was psuedo balanced via resistors.

 

[MC_RME]

Output signals not identical is fully normal and no issue - you're holding it wrong, so to say.. You need to measure (or record) the signal as it is intended to be used, differentially, not using one signal to ground then the next to ground and then trying to make them balanced on the computer - that won't work, and it is not the way these outputs are used. You either connect Ring (or pin 3 XLR) to ground for unbalanced, or use TR (pins 2 and 3 XLR) for balanced operation.

Use a simple multimeter to measure ground free, and use a balanced AD input to verify that they just work as expected. They do.

 

[peniku8]

There is obvious distortion visible too. What is your ground reference where you are measuring from? The 3rd pin (sleeve) of the TRS bal output or have you picked a chassis point?

Many interfaces have a resistor to Gnd/0v to protect the outputs from shorts. This can result in some strange output voltage readings depending on where you connect your negative. Also, the USB shell will connect to a PC ground.

I also had a bunch of trouble with some interfaces in the past where the balanced output is not really balanced. It was psuedo balanced via resistors.

I measured against the XLR's ground, not the chassis. I'm pretty sure I have properly balanced outputs in my interface here. Anything else I'd declare as fraud.

Output signals not identical is fully normal and no issue - you're holding it wrong, so to say.. You need to measure (or record) the signal as it is intended to be used, differentially, not using one signal to ground then the next to ground and then trying to make them balanced on the computer - that won't work, and it is not the way these outputs are used. You either connect Ring (or pin 3 XLR) to ground for unbalanced, or use TR (pins 2 and 3 XLR) for balanced operation.

Use a simple multimeter to measure ground free, and use a balanced AD input to verify that they just work as expected. They do.

How would I go at calibrating the output voltage, when I'm testing an external device? I thought I could calibrate it to 2V single ended and then expect to see a 4V input when using a balanced connection, but measuring positive to negative shows something in the 3.5V range in that case.
The device has been RMA'd once before and I'm not sure I trust it fully. It loses connection/resets the driver sometimes when a ceiling light is switched off and I get rather high distortion readings near full output (I can post a measurement in a bit, but maybe that's normal, I'm not so sure).

 

[SIY]

I measured against the XLR's ground, not the chassis. I'm pretty sure I have properly balanced outputs in my interface here. Anything else I'd declare as fraud.


How would I go at calibrating the output voltage, when I'm testing an external device? I thought I could calibrate it to 2V single ended and then expect to see a 4V input when using a balanced connection, but measuring positive to negative shows something in the 3.5V range in that case.
The device has been RMA'd once before and I'm not sure I trust it fully. It loses connection/resets the driver sometimes when a ceiling light is switched off and I get rather high distortion readings near full output (I can post a measurement in a bit, but maybe that's normal, I'm not so sure).

Don't conflate "balanced" with "symmetrical voltage wrt ground." They are not the same thing.

 

[restorer-john]

Don't conflate "balanced" with "symmetrical voltage wrt ground." They are not the same thing.

When it comes to a differential signalling system, would you want equal opposing voltages, or not, to enable the best performance?

Any form of offset will decrease or diminish the performance in such a setup.

 

[pma]

While doing some voltage calibrations today, I found out that the positive and the negative end of a balanced output on my RME Fireface UC do not output the same voltage. I measured 1.9V on one lead and 1.3 on the other (don't remember which was which, but I think the positive had a higher voltage).
I then recorded from both wires and also recorded a differential signal (which should sum to zero if I understand that correctly?).
Negative: -18.4dbFS; Positive: -15,9dbFS; Sum (Positive->Positive, Negative->Ground): -11.6dbFS

Signal generator was REW in ASIO mode, 1khz. Here is the waveform of the measurement across + and -

Am I missing something? If I understand the differential nature of balanced connections correctly, the signal should be mirrored, which should sum to zero, or just be twice the voltage of the individual channels, when measured differentially (since they should output the same voltage?).

Looks like overload with signs of phase inversion. You should measure it with the scope. Soundcard is always prone to mistakes. Decrease generator amplitude and repeat it.

 

[peniku8]

When it comes to a differential signalling system, would you want equal opposing voltages, or not, to enable the best performance?

Any form of offset will decrease or diminish the performance in such a setup.

I was under the same impression, but maybe someone can explain why that is not the case.

 

[SIY]

When it comes to a differential signalling system, would you want equal opposing voltages, or not, to enable the best performance?

It matters not a bit since "ground" is taken out of the picture as far as the receiver is concerned. That's the beauty of differential.

https://www.diyaudio.com/archive/blogs/alexcp/attachments/1926d1460406090-bruno-putzeys-micropre-g-word.pdf

 

[SIY]

I was under the same impression, but maybe someone can explain why that is not the case.

See Bruno's clear explanation in the link above.

 

[restorer-john]

Soundcard is always prone to mistakes.

They should be banned. Scopes only, preferably analog.

 

[peniku8]

It matters not a bit since "ground" is taken out of the picture as far as the receiver is concerned. That's the beauty of differential.

https://www.diyaudio.com/archive/blogs/alexcp/attachments/1926d1460406090-bruno-putzeys-micropre-g-word.pdf

Thank you for this, I love delving through papers like these! Will read through it later

 

[MC_RME]

How would I go at calibrating the output voltage, when I'm testing an external device? I thought I could calibrate it to 2V single ended and then expect to see a 4V input when using a balanced connection, but measuring positive to negative shows something in the 3.5V range in that case.

Well, there is a manual. And it explains the levels at the servo-balanced outputs. This is not a HiFi unit...

 

[restorer-john]

It matters not a bit since "ground" is taken out of the picture as far as the receiver is concerned. That's the beauty of differential.

Is it? The source is never actually differential in the first place. It's psuedo, at best, especially with digital.

 

[MC_RME]

They should be banned. Scopes only, preferably analog.

Scopes won't help a bit here unless the OP also invests into differential probes...

 

[restorer-john]

Scopes won't help a bit here unless the OP also invests into differential probes...

F#ck it , just add and invert on hots only. Does the same thing...

 

[pma]

Scopes won't help a bit here unless the OP also invests into differential probes...

2 probes and X-Y. Nothing is easier.

 

[SIY]

Is it? The source is never actually differential in the first place. It's psuedo, at best, especially with digital.

It absolutely is differential. Get the "g" word out of your thinking, it's irrelevant in differential applications.

Where "ground" comes in is consideration of balanced, and there, the ONLY thing that matters is equal impedances to ground. Equal voltages to ground aren't necessary for balanced transmission and reception.