Measure and evaluate quality of USB audio interfaces!? (Presonus, Focusrite, Mackie, Behringer)

[Schmidt123]

Hi guys,

I am a musician. At my studio I have an RME-Fireface. Now - because of corona reglementation - I cannot go there and I want to buy a simple cheap USB-Audio-Interface to mix some old recordings, maybe record something new and listen to music.
I have 3 interfaces here to test: Presonus Audiobox 96, Behringer UMC202HD, Mackie Onyx Producer 2-2.

Can I use the measurements from within the same interface (loop output to input - like calibration) to judge it's quality or at least to rate those three interfaces in sound quality to choose one?

Apart from the frequency response REW-measurements show me THD and noise floor. I tried to measure to crosstalk between the two stereo channels, by setting the left output for the sweep but looping the right output to the measuring input.

However I am not sure if this makes sense. What is the right way to setup to make it comparable?
And I don't now what to look for. And I don't know how to read or interpret the THD-charts.

I try to attach the measuring-session.

Hope you can give me some insights.

Btw: I calibrate the interface by looping output to input. Then I want to verify my calibration. But the frequency response is not flat? Why? However it is repeatable always the same curve with every other measurement.

Cheers,
Schmidt

Edit: Uploading the 60mb .mdat did not work. Here are graphics.

 

[John Mulcahy]

Pretty odd result for that Mackie measurement. Main thing to look out for is any monitoring being active, that will mess up the measurements. For accurate measurements of soundcard distortion best using stepped sine measurements, but it is very slow. You can compare the THD results on the Overlay window, Distortion graph. Lower is better, but connections between measurements and subjective perceptions are elusive unless the measurements are very bad.

 

[Schmidt123]

Pretty odd result for that Mackie measurement.

You mean the "Mackie verify.jpg"?
The "Behringer verify" (attached) looks similarly odd.

I calibrate the interface by looping output to input. Then I want to verify my calibration. But the frequency response is not flat? Why? However it is repeatable always the same curve with every other measurement.

Is something wrong with my calibration?

 

[John Mulcahy]

Easier to tell if you post an mdat of a measurement. That is a highly zoomed in measurement, the whole span is only 1.2 dB, but it will be easier to check an mdat. In your Analysis preferences make sure the option to "Truncate IR after 1.7 s" is selected in Impulse Response calculation to help keep the file size manageable, can also select the "Allow 96 PPO log spacing" option for smaller files.

 

[DanDan]

That looks like a C Weighted Curve. I doubt very much if any of these interfaces will have anomalies big enough to be noticeable in numbers.
If it were me, I would lose the Behringer on principle. Listen to the other two, pick the one with the best sounding headphone output.

 

[Bernard]

Hello,
DanDan said : "choose the one with the best headphone output"
This is a good way to determine the quality of the card, but what should you listen to as an input?
What if the headphones aren't great?
An idea came to me (when you don't have the intellectual means to create sometihing, ideas don't cost much!).
What if Master John added a new tool in his toolbox?
For example, a "Test Your Sound Card" button that would send a program in silent mode and tell us at the end of the test "Yes, that's fine, you can continue."
It's up to the boss to determine what is necessary and sufficient!
Isn't that a good idea?
Bon dimanche,
Bernard

 

[John Mulcahy]

That looks like a C Weighted Curve.

Look at the scale Dan, the graph only spans 1 dB top to bottom.

 

[Wayne A. Pflughaupt]

What if Master John added a new tool in his toolbox?
For example, a "Test Your Sound Card" button that would send a program in silent mode and tell us at the end of the test "Yes, that's fine, you can continue."
It's up to the boss to determine what is necessary and sufficient!
Isn't that a good idea?

Probably no good reason for John to do that. As Dan alluded, virtually any soundcard will be suitable for use with a measurement platform like REW, except perhaps for the most low-end pieces.

With the headphone output Dan is probably referring to personal use beyond REW measurement, such as the OP would have as a recording musician.

Regards,
Wayne

 

[Schmidt123]

Thanks for your replies.
Here a the *.mdats.
@DanDan: I know to much about (my) psychology to only rely on my ears.

However I used a small Behringer mixing console for routing for a while. When I changed it to simple Mackie, I was surprised how much better the Mackie sounded. So I wanted to go with the Mackie as the interface.
However the Mackie don't work well. Input and output gain increase very slowly in the beginning and after 3 o'clock they suddenly jump in big steps. A real pain. No fun to work with.

So I was starting mixing a song on the Presonus Audiobox96. However it did not reconnect by itself after sending the PC to sleep mode. I had to disconnect and reconnect the cable each time.

Because of this I switched to the Behringer and thought it sounded not so good - like not so open not so clear. But maybe it is just my bias.
And I cannot A/B the interfaces with the push of a button keeping the same volume to blind test it.

So maybe you can find some hints in the measurements, that the Behringer really sounded a little worse?

I use mainly my small studio monitor and not so often the headphones. But headphone amp is also a good point.

 

[DanDan]

I would have guessed that the Mackie is the better device. Clearly your ears and headphones are capable of appreciating a better sounding path.
There are no simple threshold numbers for good enough or audibility. I keep a bunch of Reference Tracks which I have gathered and used over decades.
When I play, even just one and listen, I can tell if everything is working OK and can easily pick out nasty sounding loudspeakers or rooms.
Differences in electronics are more difficult, they all sound very similar. But even there I have been surprised to immediately hear very obvious differences between CD players.
Hi end ones, all with ruler flat response and 0.00000 Distorting, Noise.
Note that room anomalies are 3.7 Gazillion quantums bigger than any measurement on any working interface.
If you simply ignore the Calibration process entirely you can get on with whatever it is you are really trying to do.

Converter Comparison

ethanwiner.com

 

[John Mulcahy]

The Behringer has a relatively high level of 2nd harmonic distortion at around 0.07%.

The "Presonus verify" and "Mackie verify" 96 kHz measurements are using soundcard cal files called "Presonus 44.1.cal" and "Mackie 44.1.cal" respectively, the cal files must be generated at the sample rate that will be used for measurement. A 44.1k cal file only has data to 22.05 kHz and the soundcard responses are different at different sample rates.

Something is wrong with the Mackie's settings, it is rolling off very early, already -3 dB at 13 kHz.