Measured of RME Fireface UC - Does my RME work properly?

[Robciak]

Hi,

here is my very first review, of the audio interface. The device under test was interface RME Fireface UC and measures were conducted on Audio Precision APx525 analyzer.
Based on the results I started to worry that something may be wrong with the measured interface. Other RME product measurements usually give much lower THD+N distortion. In addition, for comparison, I measured my other Focusrite Scarlett 3gen interface and the distortion was much lower too.

If anyone has had a chance to test this interface model before, I would be happy to compare test results.

Also, as I mentioned, these are my first measurements, so if you feel that some measurements are not relevant and some have been omitted, please let me know what you would change. Unfortunately, I did not have the opportunity to measure Jitter.

In the review, I focused on output channels 1 and 2. I also attached measurement reports for the rest of the channels, if anyone is interested in the details.

Basic parameters

The table below shows the values of each parameter for each channel.

I do not know why the SINAD and RMS levels are so low. These measurements were taken with a balanced connection, but when the connection was unbalanced, the values were very similar.

FFT Spectrum - multitone

The first multitone measurement was taken with a sample rate of 44,1 kHz, second one with SR 192 kHz. Both look fine.

Linearity (-20,000 dBFS)

Perfect linearity

THD+N Ratio vs Measured Level

Surprisingly high distortion values

THD+N Ratio vs Measured Level

Surprisingly high distortion values

SINAD vs Measured Level

High THD+N values result in low SINAD

RMS Level

Low-level value for balanced connection (expected around 4 Vrms). No difference in level between balanced and unbalanced conection.

Balanced:

Unbalanced:


Relative Level (1,00000 kHz)

Linearity within +/- 0.1 dB for the entire range.

THD+N Ratio

Equal but high THD+N over the entire frequency range.

SINAD

As well as THD+N

Intermodulation distortion (IMD Level Sweep ( SMPTE ))

 

[nagster]

I get fired when I download ZIP files uploaded to internet bulletin boards, so I'm guessing from the information displayed.
If the FFUC output reference level setting is "+4dBu", there seems to be no problem with the measurement result of 3.4V output with 0dBFS input.

The THD+N ratio of ch1~2 has a large difference from the specification value. There's probably a problem somewhere. Is it a misconfiguration? Is it a measurement method? Is it the environment? Is it hardware? Is it software?
Could it be that the design makes the THD+N value worse for channels 1 and 2 from the beginning? Of course, it is unthinkable that RME would design something like that.
ch3~6 may be within the acceptable range. It is subtle. Since we do not know the measurement conditions for the specification values, a strict comparison cannot be made.

Unlike simple DACs for consumer use, there are many variables when it comes to audio interfaces for production use.
We recommend that you start by identifying the cause of the large difference in THD+N ratio between channels. After that, we verify the difference with the specification value and the difference with other measurement results.
If it were me, I would reinstall the drivers and software on another computer, initialize FFUC, update the firmware to the latest just in case, and redo the settings. This eliminates misconfigurations and computer problems.
If there is still a difference, it may be a hardware problem. Measure distortion and noise separately and also observe the FFT spectrum.

 

[AnalogSteph]

Based on the results I started to worry that something may be wrong with the measured interface.

Looks like you're stuck at 16 bits per sample. Presumably a settings issue.

 

[peniku8]

We have this interface in the studio and I'm planning on taking a full set of measurements when I get the time to after my presumably broken foot has healed and I can walk again.
From what I can remember, I've gotten about 100dB SINAD from it and definitely over 6Vrms on the output. Are you sure your cable isn't broken? With these figures, it looks like you only get half the signal (e.g. hot or cold disconnected).

 

[Rja4000]

No difference in level between balanced and unbalanced conection.

I read in the specs that the UC has servo balanced outputs, so that explains that the level remains the same for balanced and unbalanced.

I do not know why the SINAD and RMS levels are so low.

I think the very reason your THD+N doesn't match the specs is that you're using +4dBu reference level and not high gain.

In high gain, you'll see the output level increased by 6dB to reach 19dBu, which is 6.9Vrms output.
And you'll probably see SINAD increased similarly.

 

[Robciak]

As you said, the low level was related to the +4dBu setting. Thanks to you guys for pointing this out.

Here is a measurement with different output settings. I reinstalled the RME drivers, which probably had a positive effect. Now the THD+N is around 0.0015% which is close to the specification

It looks like the low SINAD is caused by harmonics. I got the best SINAD for the +4 dBu setting, and the worst for Hi Gain.

Output leve set: -10 dBV

Output leve set: +4 dBu

Output leve set: Hi Gain

 

[AnalogSteph]

The increased H2 level in Hi Gain would be a direct result of the higher analog levels in the output stage and as such not a real surprise. The strong H5 must be a DAC thing.

BTW, spectra in Vrms scaling are a bit hard to interpret, I would set the scale to dBr instead.

The noise floor still looks a bit high - what sort of dynamic range readings do you get? Specs say both A/D and D/A should be good for 110 dB unweighted (20/22 kHz, I assume) or 113 dB(A). Apparently there should be an AK4620A inside? The DAC side of this codec is specified with -100 dB THD+N (or -97 dB for AK4620B) and a dynamic range of 115 dB(A). Slight losses due to surrounding electronics are expected. In these codecs, the ADC side is often a bit stronger - I suppose the idea is that this is the important part for the thing you can't generally repeat, the recording.

 

[MC_RME]

Also please note that the Fireface UC is a design of 2009, with the latest hardware revision from 2014, meanwhile discontinued (but still supported)..