RME Babyface Silver Edition Measurements and Review (after 10 years of use :)

[dominikz]

Just finished a bit of more extensive testing and measurement of my audio interface (RME Babyface Silver Edition, 1st generation bought in 2011.) and though this audio interface is now quite old (3rd generation is the current one) I thought the results might be interesting to some so decided to share

A bit of background first - when I bought the Babyface, it has been my third or fourth audio interface in a span of about 5 years, the others either dying on me or going out of SW support. So I decided I needed something more robust and reliable. RME seemed to fit the bill nicely with good specs, robust build and somewhat legendary SW support and driver stability.
10 years later I have to say I'm pretty impressed - I'm still getting driver and SW updates, and even an occasional firmware update. SW is in general rock-solid, and I don't believe I ever experienced any major bugs or any dropouts related to the interface in all these years. The device itself is also working as well as day-one HW-wise (I did have to replace the breakout cable though at one point - that thing is not overly robust), so really this is A+ on reliability and robustness.
RME Totalmix FX is a great SW mixer/router and DSP that I really love, and is a big reason why I'd probably buy another RME interface if this one ever dies or gets broken/lost. I especially like the per-channel 3-band PEQ, which I use to tame the problem LF area in my nearfield setup, and the HW-button selectable B-speaker option.

Some complaint can be made about the headphone output which has a a very high output impedance (30 ohm) and limited power (max voltage ~1,7 Vrms / +7 dBu) - so I'm using a DIY kit-based O2 headphone amp with mine instead.
[EDIT 2021-11-25] Some examples of how high output impedance of the headphone amplifier causes FR deviations with various headphone loads have been added in post #7.
Using the Babyface for generic computer multimedia, audio measurements as well as home studio recording duties over all these years you can imagine it is powered on and in use a lot of the time, and it has been an amazing workhorse really.

Full manual with manufacturer specifications are available here. Some highlights of manufacturer specs:

• AN1 and AN2 ADC:
  • Balanced
  • Input impedance: 2 kOhm
  • Maximum input level XLR, Gain 0 dB: +12 dBu / i.e. ~3 Vrms
  • Signal to Noise ratio (SNR): 108 dB RMS unweighted, 111 dBA
  • THD: < -100 dB, < 0.001 %
  • THD+N: < -98 dB, < 0.0012 % / i.e. SINAD: >98 dB
• AN2 HiZ mode
  • Unbalanced
  • HiZ Input impedance: 470 kOhm
  • Maximum input level TRS, Gain 9 dB: +12 dBu
• AN1 and AN2 DAC:
  • Balanced
  • Output impedance: 75 Ohm
  • Output level balanced at 0 dBFS: +15 dBu / i.e. ~4,3 Vrms
  • Output level unbalanced at 0 dBFS: +9 dBu / i.e. ~2,2 Vrms
  • Dynamic range (DR): 112 dB, 115 dBA @ 44.1 kHz (unmuted)
  • THD: - 104 dB, 0.00063 %
  • THD+N: -100 dB, 0.001 % / i.e. SINAD: >100 dB
• AN3 and AN4 DAC (i.e HP output):
  • Unbalanced
  • Output impedance: 30 Ohm
  • Output level at 0 dBFS: +7 dBu / i.e. ~1,7 Vrms

Performance as stated is definitely quite a bit below the current SOTA ADC or DAC performance, but I'd say still respectable (even more so for a device designed over a decade ago).

Anyway, to start off, let me share RightMark Audio Analyzer (RMAA) loopback results first - the test was done with balanced out connected to balanced in with output level reduced to -2,5dBFS to avoid clipping the input (input gain set to the minimum 0 dB), done at 48kHz/24bit . Since these are loopback tests, there is typically no easy way to be sure if the limiting factor is the input (ADC) or output (DAC), so I did some additional analysis that I will show further down to try and determine DAC and ADC specific performance.

Regardless of limitations of RMAA, we can see already the numbers seem to correlate quite nicely with the manufacturers specs for the ADC.

As we can see there is a small level mismatch between the two channels (right channel down by ~0,2 dB vs the left one), but here it is not obvious whether this comes from the input or output side (or both). I measured the output level of both channels with 400 Hz signal at -9 dBFS peak with my true RMS multimeter to determine the level matching of the DAC and output stage and saw the output channels are only 0,03 dB off from each other - meaning that most of the loopback level difference comes from the input preamp and/or ADC stages.
We can also see that there is some deviation from flat, mostly in the LF area - where the response is ~2dB down @ 20Hz. Unfortunately I was unable to determine by measurement whether this deviation is caused by the ADC or DAC, but the manufacturer specs seem to suggest it should be the ADC:

ADC:
Frequency response @ 192 kHz, -1 dB: 30 Hz - 90 kHz
DAC:
Frequency response @ 192 kHz, -1 dB: 5 Hz - 80 kHz

As we can see, according to the manufacturer the ADC is down by 1 dB between 30 Hz - 90 kHz, while the DAC is specced from 5 Hz onwards with the same deviation.

Noise floor and dynamic range, while not the best showing, are quite uneventful - as one might hope for.

With a 1kHz tone we can see the third harmonic dominates the distortion components at ~-101 dBFS. Right channel seems worse performing than the left one.

IMD test again shows worse performance in the right channel vs left.

Stereo crosstalk is stable and pretty low across the full frequency spectrum.

Balanced loopback tests

Next let's look at some distortion vs level loopback tests, done with ARTA and REW.

First the balanced loopback, in all 4 combinations:

As we see, the AN1/left input exhibits the 'hump' above 0,01 V that the AN2/right input doesn't have. On the other hand, AN1/left input seems to have better distortion performance above ~0,4 V, where the AN2/right input starts to show early rise in distortion.

I also tested distortion frequency dependency on AN1 and saw reasonably stable performance:

And a sweep test from REW at a single level:

IMD performance:

Multitone tests and -1 dBFS seems to show ~15 bit of clean range:

J-Test:

Just a single side tone at ~12,5kHz at about -125dBFS, so reasonably clean.

Next let's look at the difference in performance between three modes of AN2 input (balanced / unbalanced Lo-Z / unbalanced Hi-Z):

Pretty good match, except the early rise in distortion in Lo-Z unbalanced mode - which appears to have some dependency on impedance matching and gain setting (I assume due to relatively low input impedance in this mode). Hi-Z unbalanced mode is the one I use for most of my measurements, as the high input impedance makes the measurements relatively unaffected by the DUT impedance.

Unbalanced loopback tests

Hi-Z input distortion vs level vs frequency plot:

And a sweep test from REW at a single level:

Let's look at all 4 DAC outputs in unbalanced mode, looped back to the HiZ input:

We see outputs AN2 (right) and AN3 (HP left) show an early rise in distortion vs the other two.
Multitone test shows a few dB better performance with the Hi-Z unbalanced input vs balanced shown previously (but still slightly worse than 16-bit resolution):

Measurements with an external analogue 1kHz oscillator and twin-T notch

To quantify the distortion contributions from ADC and DAC, I pulled out the ultra-low distortion 1kHz analogue audio oscillator I got from Victor and the matching twin-T passive notch filter, described previously here. Since RME don't provide the reference level for their THD(+N) figures it is difficult to say exactly how well the match to the spec is - but as you can see it is definitely in the ballpark.

ADC 1kHz distortion measurements made by oscillator and compared to loopback measurements:

As we can see there is a pretty good match, implying that the ADC is the limiting factor for distortion - so in line with expectation based on manufacturer specs. Note that I couldn't measure the balanced input with levels lower than ~0,4V, as introducing a passive attenuator caused the distortion to shoot-up due to low impedance of the balanced input, and this was as low as I could set the level on the attenuator However we have a good match on the unbalanced input.

DAC 1kHz distortion measurements made through the notch filter:

As expected, we see the DAC distortion levels are much lower than measured by loopback, which is expected since loopback measurements are apparently limited by the ADC. We can see again that outputs AN2 (right) and AN3 (HP left) show an early rise in distortion vs the other two.
We also see pretty good match to the manufacturer declared maximum output level of each output type.

Bonus: Roundtrip latency measurements

The manufacturer specifies the following figures in the manual for the AD and DA converter latency:

In my loopback measurements with the Oblique Audio RTL Utility I got pretty much the same values:

• 44,1kHz: ~1,8 ms / 78 samples
• 48kHz: ~1,7 ms / 78 samples
• 88,2kHz: ~0,9 ms / 79 samples
• 96kHz: ~0,8 ms / 79 samples
• 176,4kHz: ~0,4 ms / 73 samples
• 192kHz: ~0,4 ms / 72 samples

Conclusion

I'd say these measurements pretty much confirm the specs provided by the manufacturer, which is always nice to see.
My motivation for doing these measurements was to determine the baseline performance when using the audio interface to test other devices. For my use it is sufficient, but definitely behind what the latest generations of audio interfaces achieve.

 

[Tks]

One thing I noticed aside from the comprehensive review was how brand spanking new the unit still looks.

 

[dominikz]

One thing I noticed aside from the comprehensive review was how brand spanking new the unit still looks.

What can I say - it is quite a robust device. Body is some kind of metal (it's surprisingly heavy) and the finish is quite durable too.
Also, it has mostly been used at home so hasn't been kicked around (too) much

 

[audiofun]

I would say you can't do jitter test in loopback mode --- the dac and adc share the same clock and jitter does not make any sense at all. If you have 2 units, the jitter test would make much more sense.

 

[dominikz]

I would say you can't do jitter test in loopback mode --- the dac and adc share the same clock and jitter does not make any sense at all. If you have 2 units, the jitter test would make much more sense.

That is a good point - thanks!

 

[wildgooselake]

Have been using one of these for many years as well. My only ADC/DAC/Headphone-out.

The sound is good, the drivers are good, the controls are good. Probably the best bang-for-buck in used market.

I record audio and MIDI, and the in/out latency is amazing, better than anything in this price range. In addition, the zero latency DSP equalizer, reverb, and echo are very very useful for recording. The EQ comes in handy for simple headphone/speaker correction if needed as well.

In terms of headphone sound, I compared listening to Schiit Modi + JDS Atom. Was not some fancy ABX test, but a simple close-to-real-life approach: plug in to output one, listen to song. Now output 2, listen to same song. Repeat for many songs. Could not tell a difference.

It's not the most powerful headphone amp, but for the headphones I use, more than enough loudness even at 30%.

In terms of speaker out, no issues I can speak of.

There are of course better measuring (and considerably more expensive) options, but my ears can't tell and my wallet can.

Another added benefit that's imporant to me: without the chonky breakout cable, the Babyface is quite small and discreet, no bigger than a USB hub or portable hard drive. So I can use it portably for headphones when working in a cafe/train/etc.

Very happy with it and not planning to change it.

 

[dominikz]

Have been using one of these for many years as well. My only ADC/DAC/Headphone-out.

The sound is good, the drivers are good, the controls are good. Probably the best bang-for-buck in used market.

I record audio and MIDI, and the in/out latency is amazing, better than anything in this price range. In addition, the zero latency DSP equalizer, reverb, and echo are very very useful for recording. The EQ comes in handy for simple headphone/speaker correction if needed as well.

Fully agree

In terms of headphone sound, I compared listening to Schiit Modi + JDS Atom. Was not some fancy ABX test, but a simple close-to-real-life approach: plug in to output one, listen to song. Now output 2, listen to same song. Repeat for many songs. Could not tell a difference.

It's not the most powerful headphone amp, but for the headphones I use, more than enough loudness even at 30%.

As I mentioned in the OP, the main issue is the high output impedance and lack of power.

Power requirement of course depend on headphone sensitivity and your SPL requirements - if it plays loud enough for you, then there is obviously no issue
However with some of my headphones I couldn't get enough loudness from the included headphone amp while tracking loud instruments - hence my comment.

Similarly, the high output impedance will only cause significant frequency response issues if both of the following two conditions are satisfied:

• The headphones have nominal impedance much lower than ~250 Ohm
• The impedance vs frequency characteristic of the headphones show large impedance swings

Let me illustrate with some measurements I took a while ago, here's how the frequency response of my RME Babyface headphone output (with its 30 Ohm output impedance) is affected based on the specific headphone connected to it:

Note how very different the FR is depending on the specific headphone. Of course, the deviations in the graph are exaggerated - note the highly zoomed vertical scale. In the worst case vertical swings are within 1 dB - but still, since Q is pretty low that is significant and likely audible in worst cases.

On the other hand, here's how the O2 headphone amp (with its 0,5 Ohm output impedance) frequency response is affected based on the specific headphone connected to it:

The scale is the same as before, but we see no such deviations. In other words, O2 couldn't care less what is connected to it

Of course I'm not trying to say that it is mandatory to use a dedicated headphone amp - the built-in one does the job for sure, and with certain headphones I'm sure works pretty well. It's just that the performance will vary depending on the specific headphone that is used with it.

Hope that is interesting!

Another added benefit that's imporant to me: without the chonky breakout cable, the Babyface is quite small and discreet, no bigger than a USB hub or portable hard drive. So I can use it portably for headphones when working in a cafe/train/etc.

In time I've actually come to like the breakout cable and see it as practical. It allows me to put the connections out of sight and have a cleaner desktop. Also it makes it easier to disconnect everything when I need to take it with me as I just disconnect one D-Sub connector. I have an additional breakout cable for on-the-go, so one of them always stays connected to cables at home.

 

[wildgooselake]

Fully agree

As I mentioned in the OP, the main issue is the high output impedance and lack of power.

Power requirement of course depend on headphone sensitivity and your SPL requirements - if it plays loud enough for you, then there is obviously no issue
However with some of my headphones I couldn't get enough loudness from the included headphone amp while tracking loud instruments - hence my comment.

Similarly, the high output impedance will only cause significant frequency response issues if both of the following two conditions are satisfied:

• The headphones have nominal impedance much lower than ~250 Ohm
• The impedance vs frequency characteristic of the headphones show large impedance swings

Let me illustrate with some measurements I took a while ago, here's how the frequency response of my RME Babyface headphone output (with its 30 Ohm output impedance) is affected based on the specific headphone connected to it:
View attachment 168073
Note how very different the FR is depending on the specific headphone. Of course, the deviations in the graph are exaggerated - note the highly zoomed vertical scale. In the worst case vertical swings are within 1 dB - but still, since Q is pretty low that is significant and likely audible in worst cases.

On the other hand, here's how the O2 headphone amp (with its 0,5 Ohm output impedance) frequency response is affected based on the specific headphone connected to it:
View attachment 168076
The scale is the same as before, but we see no such deviations. In other words, O2 couldn't care less what is connected to it

Of course I'm not trying to say that it is mandatory to use a dedicated headphone amp - the built-in one does the job for sure, and with certain headphones I'm sure works pretty well. It's just that the performance will vary depending on the specific headphone that is used with it.

Hope that is interesting!


In time I've actually come to like the breakout cable and see it as practical. It allows me to put the connections out of sight and have a cleaner desktop. Also it makes it easier to disconnect everything when I need to take it with me as I just disconnect one D-Sub connector. I have an additional breakout cable for on-the-go, so one of them always stays connected to cables at home.

Oh yes, Jacobi’s law at work. Thank you for illustrating nicely that it makes a difference for some headphones, and might even be lightly audible for some cases.

If portability wasn’t a concern and my headphones were very low and varied impedance, would have probably considered other options as well, probably with separate AD and DA chains.

For my headphones it made no discernible difference though, and together with the other convenience factors it made a very good output device for my needs.