Audio Interface Suitable for Measurements

[Sergei]

I was interested in a similar DAC/ADC some time ago. After reading about and playing with some, I realized that there are use cases in studios which put competitive pressure on evolution of loop-precise DAC/ADC interfaces. One of them: addressing a situation when there is just one valuable piece of vintage outboard processor in a studio, which needs to be applied to multiple tracks and/or channels.

So, instead of using that vintage piece in real time, mixing or mastering engineer inserts it in a DAC-vintage-ADC loop, applied to several tracks in a DAW, one track at a time. A simple test of faithfulness of the interface is to insert a straight wire (vintage processors tend to have a bypass button for that) and null the input vs output.

Another test is to automate the loop in DAW and keep repeating it until the engineer can hear the difference between the original digitized sample and the outcome of multiple DAC/ADC conversions, which may run into hundreds of repetitions for good interfaces made after 2007.

The list of usual suspects in lower-priced range included top-tier interfaces from RME, Focusrite, and MOTU. One price range up was noticeably more expensive, and included top-tier interfaces from Lynx, Antelope, and Merging. There was even more expensive range, yet those were too expensive for me, and I didn't investigate them any further.

Price-wise, I believe professional gear doesn't have as many undiscovered bargains as audiophile electronics. A ten years old Lynx Aurora costs about the same as a new MOTU interface with comparable characteristics. If you think you found a bargain nobody knows about, I advise you to think again

AES research articles have frequently listed RME DAC/ADC over the past decade in the context of measurements. Lately Focusrite started appearing more often than before. All good brands. I think you'd be OK with pretty much any top-tier rack-mountable interface from the six companies I mentioned, made in the last 10 years or so.

 

[Interference]

@Interference is a second hand E-MU 0404 USB a good choice for measurements from a Macbook? I'm looking for entry-level / beginner measurement instruments and I found one locally for about $48. Does it work out of the box on recent macOS?

I do not own a Mac so I cannot really tell you. Your best chance is googling it up.

 

[hyperknot]

I do not own a Mac so I cannot really tell you. Your best chance is googling it up.

There is an open source driver which makes it work even in 192k mode. https://github.com/Wouter1/EMU-driver

(I wouldn't be super happy about disabling System Integrity Protection for security though.)

So my question was mostly related to the ADC performance, but I've read a lot yesterday and it seems that it's still better than anything under 200 EUR today, or even 300.

Having said that, it seems second hand units go super fast, it seems the one I've found is gone already.

 

[audio_tony]

For noise, just build a 20db-30db gain opamp based preamp for noise measurements. And divided by the gain later.

@JohnYang1997

Can you clarify this method please?

I'm not sure I follow what you meant.

 

[Earfonia]

You need the latest beta version for a Noise+Distortion figure, which should be measured at -60 dBFS and combined with the max level to provide a DR figure. The "DR" figure was based on an assumption of -0.1 dBFS max level, it has been retired. To have the generator allow up to 0 dBFS select the "Full scale sine rms is 0 DBFS" option in the View preferences.

The latest REW beta version is super awesome! Thank you!

RME Babyface Pro IO Analog 3 (3.5mm headphone output left channel) - 1Vrms output @ 32 ohms load:

RME Babyface Pro IO Analog 3 (3.5mm headphone output left channel) - -12 dBFS 10-40k Hz Multitone:

 

[hyperknot]

I do not own a Mac so I cannot really tell you. Your best chance is googling it up.

Thanks for recommending the E-MU line, I picked up a 0202 after carefully checking that their ADC is exactly the same across the line (0202/0204/0404). It measures really well and it uses USB power, which is a big plus for me! Uploaded some RMAA measurements here.

 

[CauliflowerEars]

Thanks for recommending the E-MU line, I picked up a 0202 after carefully checking that their ADC is exactly the same across the line (0202/0204/0404). It measures really well and it uses USB power, which is a big plus for me! Uploaded some RMAA measurements here.

Hey, how has the E-MU been? I might be in a similar boat: just want to do some room measurements, but want to keep the gear as much bang for buck as possible. There’s a 2nd hand 0202 for 48 usd.
1) Just curious, if it would be better than new budget interfaces, such as M-Audio M-Track or the Behringer ones. Or even comparing to Focusrite or Steinberg. @Interference ?
2) Does the wire loopback work well? Apparently with such, an XLR mic can do real time Impulse Response measurements.
3) Has there been any problem with the Mac OS? I have v. 10.15.7 Catalina.

Also, debating if to get a Dayton Audio EMM-6 mic or Sonarworks Xref20. They seem the same as the likes of Behringer ECM8000, just with a calibration file …

 

[Interference]

Hey, how has the E-MU been? I might be in a similar boat: just want to do some room measurements, but want to keep the gear as much bang for buck as possible. There’s a 2nd hand 0202 for 48 usd.
1) Just curious, if it would be better than new budget interfaces, such as M-Audio M-Track or the Behringer ones. Or even comparing to Focusrite or Steinberg. @Interference ?
2) Does the wire loopback work well? Apparently with such, an XLR mic can do real time Impulse Response measurements.
3) Has there been any problem with the Mac OS? I have v. 10.15.7 Catalina.

Also, debating if to get a Dayton Audio EMM-6 mic or Sonarworks Xref20. They seem the same as the likes of Behringer ECM8000, just with a calibration file …

I remember the 0202 being slightly lower spec'd than 0204 and 0404, but please check the data sheets to confirm.

The old E-MU site had all that nicely outlined in text but now is buried in the Internet Wayback Machine and I don't have a link at my fingertips.

ADC performance should still be superior to most entry level interfaces, at least this was the case the last time I researched this and decided to buy a Motu M4 (as I cannot justify the budget for a RME). Beware that the E-MU drivers are pretty old and I would recommend something more recent if you are gonna use something other than Linux (I have no idea what's the status of Mac support).

For acoustic measurements I don't really think extreme ADC performance is that critical, most modern interfaces should be up to the task. If you want to try some electronic measurements with it, then you should care (but then modern DACs will surely outperform the interface's ADC).

I should mention my 0404 has developed a fault to one gain trimmer.

 

[CauliflowerEars]

Thank you for a great response!

ADC performance should still be superior to most entry level interfaces

1. Which specs in particular should I look at, so that I choose a good enough interface for room measurements? I guess this would be most helpful in picking one. Someone has mentioned it might be 24-bit over 16-bit. Apparently 48khz good enough?

If you want to try some electronic measurements with it, then you should care (but then modern DACs will surely outperform the interface's ADC)

2. What do you mean by electronic measurements? I already have a Topping e50 DAC. Would that cover the case?

I should mention my 0404 has developed a fault to one gain trimmer

3. Oh, what sort of a fault? If I go a 2nd hand route, is there a gear check/measurement I can do at home, so that I can say if it's a faulty unit?

I'm just a music lover that wants to create the best quality at home, but happy to expand my capabilities and/or knowledge

 

[Interference]

Thank you for a great response!

1. Which specs in particular should I look at, so that I choose a good enough interface for room measurements? I guess this would be most helpful in picking one. Someone has mentioned it might be 24-bit over 16-bit. Apparently 48khz good enough?


2. What do you mean by electronic measurements? I already have a Topping e50 DAC. Would that cover the case?


3. Oh, what sort of a fault? If I go a 2nd hand route, is there a gear check/measurement I can do at home, so that I can say if it's a faulty unit?

I'm just a music lover that wants to create the best quality at home, but happy to expand my capabilities and/or knowledge

1. Given today's prices, I would not pick anything which is below 24 bit / 96 kHz. 192 kHz are also quite common but it's not a must. As long as the interface is modern enough I would not worry about the usual S/N numbers: they are sure good for the job.
2. I mean measuring S/N and distortion of signal sources, similarly to what is done with expensive equipment in this forum. It is a very tricky business and PC audio interfaces do not have good enough spec to evaluate modern state-of-the-art DACs. Except for something top notch like the RME ADI-2. But still, an audio interface is not a eletronic measurement instrument, although with sufficient effort it could work as one. I would say you don't need any of this to set up a sound system
3. it's been a while since I touched it, but if I remember correctly the gain pot of one of the channels would make the signal go to zero beyond a certain rotation point (could be at the max gain). It's easy to see on the 0404 because this has a VU-meter, but for a 020x you would need to check the signal on the PC. A cable loopback and a REW instance should be sufficient for this.

 

[CauliflowerEars]

Ah, thank you!

1. Given today's prices, I would not pick anything which is below 24 bit / 96 kHz. 192 kHz are also quite common but it's not a must. As long as the interface is modern enough I would not worry about the usual S/N numbers: they are sure good for the job.

1. This Thomann in-house brand is 24-bit/192khz ... Do you think it could be good? Or maybe it's necessary to see the actual measurements? Curious, if it's actually the same OEM as some famous brand ... https://www.thomann.de/intl/swisson...LCJjdXJyZW5jeSI6MiwibGFuZ3VhZ2UiOjJ9&reload=1

I would say you don't need any of this to set up a sound system

2. OK, sounds like a purely engineering measurement kind of case. Probably nothing for me ... yet

A cable loopback and a REW instance should be sufficient for this.

3. Ah, sounds it could be done at home, then.

 

[CauliflowerEars]

 

[hyperknot]

The old E-MU works perfectly, but there is no driver support for it with any modern OS, except for Linux. Basically you need to use it with an old laptop, or with a Raspberry Pi, etc.

 

[Sokel]

The old E-MU works perfectly, but there is no driver support for it with any modern OS, except for Linux. Basically you need to use it with an old laptop, or with a Raspberry Pi, etc.

Works fine with win10.I do a gazillion measurements every day with it.

 

[hyperknot]

Under virtual machine I couldn't make it work, but it's good to know that it works well under real Windows. In macOS it's not supported anymore.

 

[Sokel]

Under virtual machine I couldn't make it work, but it's good to know that it works well under real Windows. In macOS it's not supported anymore.

What you see is as sloppy as it gets,6 meter cable to DUT,powered by the PC alone,newbie operator (me),etc.
It's not that is supported,driver I use is 6.0.1.1 (2010) and the Microsoft one in Device manager I/O is 10.0.19041.1 (2019).
Is just that it works ok for a 13yo interface,I bought it new back then and never had a problem.

 

[arvidb]

I did some measurements of my old Lexicon Alpha audio interface. These are my first ever such measurements; I hope they make sense. Feel free to describe how I could make them better! (And sorry about the height of the images; I used fixed aspect ratio setting in REW which might have been a bad idea.)

Line 1 out was simply connected to Line 1 in with a balanced cable for these tests. Voltage was measured with a Brymen BM235 DMM at 1 kHz. Its ACV range is only specified to 440 Hz, but I did some experiments and it seemed to only be off a few percent at 1 kHz.

For all measurements, output level was calibrated to a specific FS rms voltage using the DMM, and the input level was set so the measured REW RTA signal level corresponded to the FS-relative output level set in REW's signal generator.

First a measurement with no signal. I suppose this is meant to measure self noise - on the other hand I'm told FFT noise floor is meaningless, so not sure how useful this is:


Then a 1 kHz signal at what seemed to be "best case" settings, more or less. Output is -12 dBFS; FS = 1 Vrms:

And at -3 dBFS output:

The input "Peak" indicator lights up at -6 dBFS so it does give some warning that it isn't happy here.

I made a series of measurements at different settings:

FS(rms) (5.00V)        (4.00V)        (2.00V)        (1.00V)        (0.50V)

-0dB    (clips)        (clips)        (clips)        (clips)        (clips)

-3dB     3.54V          2.83V          1.42V          0.71V          0.35V
THD      -45            -48            -51            -51            -51
N        -83            -84            -87            -88            -87

-6dB    (2.50V)        (2.00V)        (1.00V)        (0.50V)        (0.25V)
THD      -50            -55            -64            -73            -76
N        -88            -89            -89            -91            -89

-12dB   (1.26V)        (1.00V)        (0.50V)        (0.25V)        (0.13V)
THD      -60            -65            -75            -81            -87
N        -91            -92            -92            -92            -90

(clips) = visible clipping on REW oscilloscope view; THD > -35 dB
Voltages within () are calculated

 

[Sokel]

I did some measurements of my old Lexicon Alpha audio interface. These are my first ever such measurements; I hope they make sense. Feel free to describe how I could make them better! (And sorry about the height of the images; I used fixed aspect ratio setting in REW which might have been a bad idea.)

Line 1 out was simply connected to Line 1 in with a balanced cable for these tests. Voltage was measured with a Brymen BM235 DMM at 1 kHz. Its ACV range is only specified to 440 Hz, but I did some experiments and it seemed to only be off a few percent at 1 kHz.

For all measurements, output level was calibrated to a specific FS rms voltage using the DMM, and the input level was set so the measured REW RTA signal level corresponded to the FS-relative output level set in REW's signal generator.

First a measurement with no signal. I suppose this is meant to measure self noise - on the other hand I'm told FFT noise floor is meaningless, so not sure how useful this is:
View attachment 288883

Then a 1 kHz signal at what seemed to be "best case" settings, more or less. Output is -12 dBFS; FS = 1 Vrms:
View attachment 288885
And at -3 dBFS output:
View attachment 288887
The input "Peak" indicator lights up at -6 dBFS so it does give some warning that it isn't happy here.

I made a series of measurements at different settings:

FS(rms) (5.00V)        (4.00V)        (2.00V)        (1.00V)        (0.50V)

-0dB    (clips)        (clips)        (clips)        (clips)        (clips)

-3dB     3.54V          2.83V          1.42V          0.71V          0.35V
THD      -45            -48            -51            -51            -51
N        -83            -84            -87            -88            -87

-6dB    (2.50V)        (2.00V)        (1.00V)        (0.50V)        (0.25V)
THD      -50            -55            -64            -73            -76
N        -88            -89            -89            -91            -89

-12dB   (1.26V)        (1.00V)        (0.50V)        (0.25V)        (0.13V)
THD      -60            -65            -75            -81            -87
N        -91            -92            -92            -92            -90

(clips) = visible clipping on REW oscilloscope view; THD > -35 dB
Voltages within () are calculated

A Blackman-Harris 7 window would show the results better than the rectangular.
I never used this interface but I would expect far better results.

 

[arvidb]

Yeah, it's pretty bad, isn't it? It's old though, perhaps its power supply rail caps are bad or something? Or it's measurement error, but I'm not sure how I could have messed it up if that's the case.

I've been trying to read up on FFT windows*. As I understand it, a rectangular window is optimal if an integer number of periods of the signal fits in the FFT sample set. When doing these measurements I had the REW Generator Lock frequency to RTA FFT set, so - if I understand correctly - the measurements should be as good as they can be in that regard. Please correct me if I'm wrong.

Now the first measurement I did - the one with no signal - as I suspected is meaningless. I did another measurement now, this time with the unit set to V/sqrt(Hz) and the input calibrated to 1 Vrms FS (calibrated at 0.253 Vrms, 400 Hz this time). This produces a Linear Spectral Density (rather than a Spectrum) and should be relevant as a measure of noise - again if I understand correctly:

Indeed changing FFT length and windowing had no effect in this mode - REW always calculated a figure of about 15.7 µV 22-22k UNW, and the shape and level of the graph did not change - so that's encouraging!

* Heinzel et al - linked by Mark Tillotson in How to - Distortion Measurements with REW. Also https://www.roomeqwizard.com/help/help_en-GB/html/spectrum.html.

 

[Sokel]

I'm also a newbie in measurements and although I use Multitone analyser I sometimes test with REW just for the fun of measurements.
A good guide how to set-up is this by @mdsimon2 :

Minidsp Flex Review (Audio DSP)

Just so I understand what you were seeing with the Flex, are you saying that if you reduce the Flex volume such that you are no longer clipping you still saw degradation in noise and or distortion? Michael Yes,and severe one,specially noise.Multitones was also not good,at all. All of that in...

www.audiosciencereview.com

There's also a really good thread here about Multitone:

Beta Test: Multitone Loopback Analyzer software

This is a software beta-test for IMD and THD aficionados with an intense desire for multi-tone audio testing :) The start of the documentation and the download website is here: https://distortaudio.org/multitone.html (for 64-bit Windows) Multi-tone Analyzer (MA) was more of an experiment than...

www.audiosciencereview.com