Suggestion: more audio interface measurements

[YoniV]

I've been following this website for a long time, and it is my humble and personal opinion that we kind of reached a saturation in terms of DAC and HP amp measurements.
I think that for quite some time now, ever since say the Topping D50s, we reached a point where anyone that can spend about $200-300 can get an essentially flawless DAC/Amp combo that will last them for life without any reason to upgrade ever.
Speakers are also being measured, which is great and super helpful and important. Likewise for receivers/streamers etc.

What I, personally, am missing is more measurements of other hardware types, and particularly audio interfaces.
These can be measured just like DACs and HP amps, but they also have other aspects that are worth measuring such as latency and ADC.
I think everyone from hobbyist musicians such as myself to semi-pro gear enthusiasts can benefit from these measurements (maybe even actual professionals, IDK).

Anyway, I know it depends on a lot of things (people send Amir their gear, donate, measure themselves, companies send items etc), and I am in no position to be anything but grateful (all of my own purchases were heavily based on this site, as well as a few close friends'), so I'm just expressing my thoughts, and I'd be happy to read what others are thinking in this regard.

Cheers,
Yehonatan

 

[sweetchaos]

Which audio interfaces would you like to see measured? Most of the popular options were already measured by Amir.

Motu, Focusrite, RME.

 

[Blumlein 88]

I'd like to see those too. While many pro companies don't make outrageous claims, it is true in that field too, that more money doesn't dependably mean better performance. So measuring these devices would be very worthwhile. Amir will measure them if you send him one.

Too my knowledge cheap DACs can get close to SOTA performance while I don't know of any ADCs that can match it. Which kind of makes a DAC having a SINAD of 125 db a suspect benefit if there are no ADCs capable of that. Not very many of them can reach 110 db SINAD.

I don't think the problem is Amir, it is down to what forum members, and/or companies will send him to measure.

 

[Blumlein 88]

Which audio interfaces would you like to see measured? Most of the popular options were already measured by Amir.

Motu, Focusright, RME.

SSL, Antelope audio, other models from the three you mention, Universal Audio, and Apogee for starters. I'd really like to see more upscale stuff like Burl, Crane Song etc. But likely a limited audience for that here.

 

[sam_adams]

Too my knowledge cheap DACs can get close to SOTA performance while I don't know of any ADCs that can match it. Which kind of makes a DAC having a SINAD of 125 db a suspect benefit if there are no ADCs capable of that. Not very many of them can reach 110 db SINAD.

This is one area that we have yet to see much progress in terms of the all important SINAD measurement. As-far-as performance goes, there seems to be only our own @IVX and his E1DA Cosmos ADC that is breaking ground here.

 

[YoniV]

SSL, Antelope audio, other models from the three you mention, Universal Audio, and Apogee for starters. I'd really like to see more upscale stuff like Burl, Crane Song etc. But likely a limited audience for that here.

I agree, there are plenty other brands and models not tested (Native Instruments, M Audio, Steinberg...), and many of those that were tested are of outdated versions (MK 1 vs MK 2), which warrant measurements in the same way that e.g. new SMSL models get renewed attention.

Of course Amir is in no way "the problem", just pointing this out in the hope that the community might be interested in putting more focus on this type of hardware.
ADC and mic preamps can really be a huge PITA, I had many frustrations in the past with interfaces that just didn't deliver...

 

[Blumlein 88]

I agree, there are plenty other brands and models not tested (Native Instruments, M Audio, Steinberg...), and many of those that were tested are of outdated versions (MK 1 vs MK 2), which warrant measurements in the same way that e.g. new SMSL models get renewed attention.

Of course Amir is in no way "the problem", just pointing this out in the hope that the community might be interested in putting more focus on this type of hardware.
ADC and mic preamps can really be a huge PITA, I had many frustrations in the past with interfaces that just didn't deliver...

For any ADC you own, one possibility is to get one of the SOTA or near SOTA inexpensive DACs Amir has measured. A clean reference source if you will. Then measure it with your ADC. You'll see where it comes up short vs the AP results. It isn't the reason I purchased it, but for instance the Topping D10 Balanced is useful for this purpose. With REW or pkane's Multitone you can do some good measuring.

 

[amirm]

The best thing to do is contact the companies and ask them to send me gear for testing. Or buy them yourself to send to me for testing. Most interfaces are too expensive for me to purchase unfortunately.

 

[tiramisu]

There hasn't been a lot of movement in audio interfaces in the last few years.
They tend to be a mixed bag. Good inputs shit outputs, crap power supplies, or worse USB powered so the outputs are anemic.
Drivers are usually a crap shoot on the window's side.
Very hard to find inserts and midi as a feature even the expensive gear tends to half-ass the features.

Buying premium doesn't get you any better than buying cheap in most cases.
This space could use a disrupter.

 

[BeerBear]

You can always upgrade playback gear, but a sub par recording is ruined forever.
So yes, it would be nice to have more measurements of "pro audio" gear, including hardware effects and all.

And if you think that people who produce music have it all figured out, you need to spend more time around musicians and recording/mixing guys, or visit a forum like GearSpace.
A lot of these people are just as ignorant and superstitious as your average audiophile. Not just hobbyists, but even professionals working with big name artists.

 

[DownUnderGazza]

Definitely in favour of more pro audio gear being put through the full range of Amir tests!

I had a Behringer X32 based mixer that I mixed with an iPad. The app was pretty good, but when inputs were driven hard it didn’t sound good. More distortion or lower head room and early clipping?

I replaced it with an Allen and Heath Qu-SB mixer. Lighter, more compact, and never had any of the sound degradation issues I had with the X32. Sounded great in the gigs I did.

Would have loved to know how both measured on the bench.

The problem I see is that these digital mixers are not plug and play devices. They need configuration to ensure input is routed to a known output. Not for the faint of heart.

Analogue mixers are however much more straight forward devices. Those with usb inputs are easily configured.

 

[Rja4000]

But then you'll also need to add some specifics tests.

As an example, it makes little sense to just measure a microphone input with 4V signal at minimum gain.

And also a comparison of EIN or dynamic range for different gain values (after aligning the gains on some physical common meaning).

I proposed elsewhere on this website to align all mic preamp gains around the max gain to not saturate 100mV.
Then you begin to see the difference between old, new, expensive and cheap mic preamps.

 

[RayDunzl]

Do the currently available better ADCs not already exceed the (let's call it) SINAD of what is to be recorded?

 

[Rja4000]

when inputs were driven hard it didn’t sound good

I measured a Motu 828mk3 firewire: the mic input is ok, but saturates below 0dBFS.

 

[Blumlein 88]

Do the currently available better ADCs not already exceed the (let's call it) SINAD of what is to be recorded?

That is one of those gray area questions. Actually not many ADCs break the 100 db barrier. For any music not of electronic origin that is probably enough. People should also remember with microphones any gain eats into the available dynamic range because it applies gain to the noise including self noise of the microphone. You probably don't usually end up with more than 70 db DR to work with in terms of overall noise. However, like other noise, the level where we hear sensitively (3-5 khz) is much lower than this. With complex recording and post processing you'd like to have more wiggle room, but physics tends to bottleneck things. I'd say usually environmental noise is the real limit at lower frequencies and all other factors together at mid to upper midrange frequencies. Of course the way most music is compressed for loudness wars you don't need much clearance of noise for that.

 

[RayDunzl]

For any music not of electronic origin that is probably enough.

That's kinda my point...

If of electronic origin, it won't need an ADC - assuming it has a digital out.

Actually not many ADCs break the 100 db barrier.

Found a cheap cable rats nest loopback on my Focusrite from two years ago: at a level of about -7.5dBfs where 0dBfs = 6.15V per documentation

 

[Blumlein 88]

But then you'll also need to add some specifics tests.

As an example, it makes little sense to just measure a microphone input with 4V signal at minimum gain.

And also a comparison of EIN or dynamic range for different gain values (after aligning the gains on some physical common meaning).

I proposed elsewhere on this website to align all mic preamp gains around the max gain to not saturate 100mV.
Then you begin to see the difference between old, new, expensive and cheap mic preamps.

I'd like to see EIN tests. And it isn't that hard to do. I've done them with metal film resistors. I think the AES-17 standard is 150 ohms and 60 db gain (or max if less than 60 db is available). But it is more informative to do the EIN at different gain levels.

Not sure I agree with testing gain at a level for 100 mv. Are you referring to input or output levels (I'm assuming input)? Some microphones on peaks can put out several volts or even over 10 volts (which would be at a lowered gain).

 

[Blumlein 88]

That's kinda my point...

If of electronic origin, it won't need an ADC - assuming it has a digital out.



Found a cheap cable rats nest loopback on my Focusrite from two years ago: at a level of about -7.5dBfs where 0dBfs = 6.15V per documentation

View attachment 227528

Yes, but THD is more than -100 db and that doesn't include noise. So probably a mid to low 90 sinad. Some are worse, I think Amir tested an Arturia Fuse barely at 90 db. The RME Babyface was something like 107 or 108 db and only a four tested were better than that. Now that is likely enough. I think the best ADC he tested is the RME without microphone inputs followed by a two Unversal Audio units. And a couple of Motu units. Some more expensive units are worse and maybe some unknown cheaper units are better.

 

[Rja4000]

I'd like to see EIN tests. And it isn't that hard to do. I've done them with metal film resistors. I think the AES-17 standard is 150 ohms and 60 db gain (or max if less than 60 db is available). But it is more informative to do the EIN at different gain levels.

Not sure I agree with testing gain at a level for 100 mv. Are you referring to input or output levels (I'm assuming input)? Some microphones on peaks can put out several volts or even over 10 volts (which would be at a lowered gain).

Here are some examples
EIN (A weighted) with 150 ohm resistor, 20Hz-20kHz

But I think the Dynamic range plot is more "readable"
As an example you see more clearly what the AD8HR is doing.

The important thing, in my opinion, is to align "gains" in a meaningful way.

And the only meaningful way is: "I have that mic that gives me that Voltage max for my use case. I want to compare all mic preamps with a gain that is the max gain allowing not to saturate my preamp or ADC with that voltage."
(I decided a threshold THD of 0.1%, but saturation is usually very brutal, so this threshold value doesn't matter much)

Here above, my "reference gain" (0dB) is the max gain to allow 100mV rms input.
(That's about the max level you'll get from a Shure SM58 with a singer in real life)
That's 35dB on the RME (max ADC level at 0dB gain marking = 18dBu) and 44.2dB on the Millenna (45dB gain marking - max output before saturation 27dBu)
For the Yamaha, it's -37dB gain marking (max ADC level for 0dB gain marking is 20dBu).

NB: For fun, I added the RME ADI-2 Pro fs R for its 4 input ranges.
Of course, it can't act as a mic preamp. For this kind of level, you'll need a condenser microphone, that requires a phantom power, and the ADI-2 doesn't have one.
Still, you understand here why you can't get better measurements from the UCX II for mid level signals.


You learn from the above what really makes the difference between an high end mic preamp and a cheaper -but still excellent- one:
The gap is big for LOW gains.
So use in priority such a preamp with high level output microphones (Condenser) or sources (drum kit).

You also realize that it is more meaningful to measure EIN (A) at 30dB gain than at 60dB gain...

 

[Rja4000]

I'd like to see EIN tests. And it isn't that hard to do.

Well, it's not that simple.

For my plots above, you have to first measure accurately the gain of your preamp for each gain marking (for an analog preamp) or the ADC level variation from gain to gain.
You also have to check at what level it will saturate (if any) for each gain.
Then you have to measure noise with your resistor (that's the easy part).

But, ideally, you also have to compensate for you actual resistor value vs 150 ohm and for the temperature.

All this is a lot of manipulations.
Worse: On some old mixers I have, you can't set the gain knob to the exact same value twice.
So, you'll have to do all the above for each gain then move to the next value.
Which means a lot more cable changes.
Quite a lengthy process...

Not sure I agree with testing gain at a level for 100 mv.

Actually, it doesn't matter, as long as we agree on ONE input level to align all gains around it.
I plan to do a "dashboard" for mic preamp at 3 gain settings:
Gain for 100mV max, for 1V max and for 10mV max.
Each measurement at an input level to get -10dBFS.

Here is an example
(You may find more in the posts of the thread here)