• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Measuring RME ADI-2 Pro with QuantAsylum QA401

Joined
Feb 21, 2019
Messages
10
Likes
3
#1
Hi All, newbie here looking for some measurement guidance. Hopefully I post this topic in the right place.

I have QuantAsylum QA401 and I borrow my friend's RME ADI-2 Pro. I would like to measure the SNR and SINAD of the line output.
The setup is as the following:

ASUS Laptop USB C with foobar playing 1 kHz tone (96k-24bit) using WASAPI driver >>
RME ADI-2 Pro Unbalanced Line Out (TS socket) with TS to RCA adapter >>
Belden 1505F coaxial cable (50cm - DIY with Canare RCA to BNC) >>
BNC splittler with 10k ohm resistor on the other end >>
QuantAsylum QA401 positive BNC input.

As shown in the pictures:
20190221_233745_tn.jpg


20190221_233633_tn.jpg


20190221_233649_tn.jpg


The QuantAsylum QA401 negative inputs were terminated using 50 ohms BNC termination.

RME ADI-2 Pro Output level was set to +4 dBU reference level, and volume was at 0 dB.
20190221_233848_tn.jpg


20190221_233905_tn.jpg


What I don't understand is, measurement showing only:
SNR: 89.5 dBA (@ ~1Vrms)
SINAD: 88.3 dB (@ ~1Vrms)

A weighted:
RME ADI-2 Pro LO_1-2 TS Ref +4dBu Vol 0dB Out 1.1V 10k dBA 01.png


No weighting:
RME ADI-2 Pro LO_1-2 TS Ref +4dBu Vol 0dB Out 1.1V 10k 01.png


I expect to get much higher SNR and SINAD from a high end audio interface like this RME ADI-2 Pro, at least on the 100 dB range as it is still within the sensitivity range of the QA401. I don't understand why I couldn't get anything better than 90 dB. Tried other settings on the RME, different Ref level, different volume, result were all close to 89-90 dBA SNR.

The RME connected to my ASUS notebook on USB C port, and powered by 3x 18650 battery for cleanest possible power supply.

Hopefully someone could point to me what's wrong with the setup, and why SNR and SINAD measurement result were relatively low in comparison to the equipment specs.

Thanks in advance!
 
Last edited:

KSTR

Active Member
Joined
Sep 6, 2018
Messages
118
Likes
198
Location
Berlin, Germany
#2
I would suggest a balanced connection and ASIO driver. Also upload the latest firmware.

RME's specs can be trusted upon, I wouldn't worry.
 
Joined
Feb 21, 2019
Messages
10
Likes
3
#4
The 50ohms termination is just to avoid open unterminated inputs that cause increase in noise floor. Probably I should just short the negative inputs to the ground. Noted for ASIO driver as well, will check on that. Once I re-test it, I will post an update. Thanks!
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,373
Likes
2,719
Location
Perth Western Australia
#6
Hi All, newbie here looking for some measurement guidance. Hopefully I post this topic in the right place.

I have QuantAsylum QA401 and I borrow my friend's RME ADI-2 Pro. I would like to measure the SNR and SINAD of the line output.
The setup is as the following:

ASUS Laptop USB C with foobar playing 1 kHz tone (96k-24bit) using WASAPI driver >>
RME ADI-2 Pro Unbalanced Line Out (TS socket) with TS to RCA adapter >>
Belden 1505F coaxial cable (50cm - DIY with Canare RCA to BNC) >>
BNC splittler with 10k ohm resistor on the other end >>
QuantAsylum QA401 positive BNC input.

As shown in the pictures:
View attachment 22340

View attachment 22341

View attachment 22342

The QuantAsylum QA401 negative inputs were terminated using 50 ohms BNC termination.

RME ADI-2 Pro Output level was set to +4 dBU reference level, and volume was at 0 dB.
View attachment 22343

View attachment 22344

What I don't understand is, measurement showing only:
SNR: 89.5 dBA (@ ~1Vrms)
SINAD: 88.3 dB (@ ~1Vrms)

A weighted:
View attachment 22345

No weighting:
View attachment 22346

I expect to get much higher SNR and SINAD from a high end audio interface like this RME ADI-2 Pro, at least on the 100 dB range as it is still within the sensitivity range of the QA401. I don't understand why I couldn't get anything better than 90 dB. Tried other settings on the RME, different Ref level, different volume, result were all close to 89-90 dBA SNR.

The RME connected to my ASUS notebook on USB C port, and powered by 3x 18650 battery for cleanest possible power supply.

Hopefully someone could point to me what's wrong with the setup, and why SNR and SINAD measurement result were relatively low in comparison to the equipment specs.

Thanks in advance!
The problem you are seeing there is primarily the 401s problem with noise floor modulation. As the signal level rises so does the noise floor. It's not the ADI. The 401 also contributes to the harmonic distortion.

Put the ADI into loop back with balanced connection and use something like Arta software to do the measurement.

It's actually a test I would very much like to see the result of. You will find the ADI is a better measurement ADC than the 401.
 
Joined
Feb 21, 2019
Messages
10
Likes
3
#7
The problem you are seeing there is primarily the 401s problem with noise floor modulation. As the signal level rises so does the noise floor. It's not the ADI. The 401 also contributes to the harmonic distortion.

Put the ADI into loop back with balanced connection and use something like Arta software to do the measurement.

It's actually a test I would very much like to see the result of. You will find the ADI is a better measurement ADC than the 401.
Oh I see, didn't know that QA401 has noise floor modulation problem. But I do notice that on every measurement the noise floor raises with the signal. Hopefully QuantAsylum could fix it. Thanks for the info! I probably need to get Arta and a good ADC.
 

dreite

Active Member
Joined
Aug 10, 2018
Messages
127
Likes
114
Location
Bremerton, WA
#8
Oh I see, didn't know that QA401 has noise floor modulation problem. But I do notice that on every measurement the noise floor raises with the signal. Hopefully QuantAsylum could fix it. Thanks for the info! I probably need to get Arta and a good ADC.
There's nothing to fix, per se. It's inherent in the way you're implementing your test and the characteristics of the QA401 system.

You don't need the 10k terminators, and either a short or 50 ohm terminator in the unused inputs of the QA401 is fine. Changing to the ASIO driver won't make any difference either.

You might change your interface to a balanced configuration though.

Dave.
 

KSTR

Active Member
Joined
Sep 6, 2018
Messages
118
Likes
198
Location
Berlin, Germany
#9
I probably need to get Arta and a good ADC
Well, just get an Adi-2 Pro FS and you're set. The interface is really excellent as a measurement frontend.
For the software, most things can be done with the REW freeware. For expert level Audiotester is what's recommended by many (incl. RME themselves). ARTA isn't a good choice for general audio measurements, IMHO.
 
Joined
Feb 21, 2019
Messages
10
Likes
3
#10
I made XLR to BNC cable to test the balanced XLR output using small coax cable, Belden 8218. 50cm length.
20190223_085541_s.jpg


Setup as the following, only Left channel tested:
20190222_221420_s.jpg


20190222_221348_s.jpg


For this test I still used the 10k ohm termination.


Unweighted results at 0 dB volume (1.1Vrms), -7dB volume (0.5Vrms), and -13dB volume (0.25Vrms) are as the following:
RME ADI-2 Pro LO_1-2 XLR Ref +4dBu Vol 0dB Out 1.1V 10k 01.png

RME ADI-2 Pro LO_1-2 XLR Ref +4dBu Vol -7dB Out 0.5V 10k 01.png

RME ADI-2 Pro LO_1-2 XLR Ref +4dBu Vol -13dB Out 0.25V 10k 01.png



'A' weighted results at 0 dB volume (1.1Vrms), -7dB volume (0.5Vrms), and -13dB volume (0.25Vrms) are as the following:
RME ADI-2 Pro LO_1-2 XLR Ref +4dBu Vol 0dB Out 1.1V 10k dBA 01.png

RME ADI-2 Pro LO_1-2 XLR Ref +4dBu Vol -7dB Out 0.5V 10k dBA 01.png

RME ADI-2 Pro LO_1-2 XLR Ref +4dBu Vol -13dB Out 0.25V 10k dBA 01.png


Am I right to assume that since ADI-2 Pro seems to be using digital volume control, the noise relatively should remain the same regardless the volume level?

As mentioned by @March Audio the noise floor measurement on QA401 seems to follow the signal level. This makes SNR result hardly be any greater than 90 dBA.
 
Joined
Feb 21, 2019
Messages
10
Likes
3
#11
There's nothing to fix, per se. It's inherent in the way you're implementing your test and the characteristics of the QA401 system.

You don't need the 10k terminators, and either a short or 50 ohm terminator in the unused inputs of the QA401 is fine. Changing to the ASIO driver won't make any difference either.

You might change your interface to a balanced configuration though.

Dave.
Yes, tried ASIO and WASAPI drivers, no differences. I will compare with and without 10k terminators later.
 
Joined
Feb 21, 2019
Messages
10
Likes
3
#12
Well, just get an Adi-2 Pro FS and you're set. The interface is really excellent as a measurement frontend.
For the software, most things can be done with the REW freeware. For expert level Audiotester is what's recommended by many (incl. RME themselves). ARTA isn't a good choice for general audio measurements, IMHO.
Audiotester seems to be end of life. Got the download link, but couldn't purchase the license.
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,373
Likes
2,719
Location
Perth Western Australia
#13
I made XLR to BNC cable to test the balanced XLR output using small coax cable, Belden 8218. 50cm length.
View attachment 22380

Setup as the following, only Left channel tested:
View attachment 22381

View attachment 22382

For this test I still used the 10k ohm termination.


Unweighted results at 0 dB volume (1.1Vrms), -7dB volume (0.5Vrms), and -13dB volume (0.25Vrms) are as the following:
View attachment 22383
View attachment 22384
View attachment 22385


'A' weighted results at 0 dB volume (1.1Vrms), -7dB volume (0.5Vrms), and -13dB volume (0.25Vrms) are as the following:
View attachment 22386
View attachment 22387
View attachment 22388

Am I right to assume that since ADI-2 Pro seems to be using digital volume control, the noise relatively should remain the same regardless the volume level?

As mentioned by @March Audio the noise floor measurement on QA401 seems to follow the signal level. This makes SNR result hardly be any greater than 90 dBA.

This is exactly my experience of the 401 and why I decided to move on from it.

Are you able to do a XLR out to XLR in test on the ADI? Arta is free and it will prove the point. I also use Virtins Multi Instrument.
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,373
Likes
2,719
Location
Perth Western Australia
#14
Yes, tried ASIO and WASAPI drivers, no differences. I will compare with and without 10k terminators later.
I wouldnt bother, we already know what the issue is.
 

KSTR

Active Member
Joined
Sep 6, 2018
Messages
118
Likes
198
Location
Berlin, Germany
#15
Audiotester seems to be end of life. Got the download link, but couldn't purchase the license.
Ah, sorry, I actually meant HpW-Works, I always confuse the two. I would use REW, it's getting better and better with each release. Its name, Room Eq Wizard, doesn't do justice to the vast functionality it offers today, it is a full measurement suite with better features than some $$$ software
 

KSTR

Active Member
Joined
Sep 6, 2018
Messages
118
Likes
198
Location
Berlin, Germany
#16
Yes, tried ASIO and WASAPI drivers, no differences
In any case you should verify bit-transparency of the playback chain with the test files provided by RME. Without ASIO driver used in foobar (and all processing off, volume at 100, etc) I didn't have bit-transparency and saw increased distortion (can't comment on WASAPI, never used it for anything).
 

QAMatt

Member
Manufacturer
Joined
Mar 1, 2019
Messages
14
Likes
28
#18
Hi Earfonia, this is Matt from QuantAsylum. In general, your test equipment needs at least 5-10 dB of margin to the equipment (device under test, or DUT) you want to measure if you want a definitive answer. That is, if you hope to measure the DUT's THD of -110 dB, then your test equipment needs to be spec'd at -115 or preferably -120 dB. A tester with a -110 dB THD spec measuring a device with a -110 dB of THD could report a result anywhere from -116 to -104 dB of THD, depending on whether or not the harmonic contributions from the tester and DUT were in or out of phase.

Price is a reasonable proxy for that margin. That is, if you want to measure with confidence a properly designed $1000 DAC with claimed -110 THD+N, you will need test equipment that can measure down to nearly -120 dB THD+N, which puts you in the $10,000+ realm of test equipment.

The expensive testers ($3000 and up) use dozens of relays to keep the ADC and DAC operating at their optimum levels (usually 6 dB steps between ranges, with 9 or more ranges). They do this on both the outputs and the inputs. And the the highest performing equipment (>$10K) will go even further and include a notch filter and/or dedicated analog distortion measurement circuits.

In all cases, even with the most expensive tester, you can't measure everything out of the box. For example, if you drop $30K on an a state of the art audio tester, you won't be able to even slightly detect the distortion of a modern $2 opamp like the OPA1612. In those cases, engineers have been forced to find ways to make a series of measurements that tease that performance out. For opamps, there are circuits such as "distortion magnifiers" (see TI OPA1612 spec for schematic) that will force the opamp to distort 100X more than normal.

And the key with all equipment is understanding where your test equipment's sweet spot is and optimize for that. For example, I just swept the current QA401 device on my bench in loopback (pic attached): On this particular unit, the QA401's best case THD (-115 dB) is found at around -18 dBV, and the best case THD+N (about -101 dB) is found around -4 dBV. The best case SNR (about 112 dBA) is found at 5.8 dBV. The best case N+D (20 to 20 KHz) is about -110 dBV up about -10 dBV. If you have an exceptionally clean DAC, the QA401 ADC N+D can get as low as -115 dBV as the DAC is the larger contributor on the QA401. Overall measured dynamic range on the ADC on this unit is 123 dBA.

Back to your measurement: The RME ADI-2 pro is a $2000 ADC/DAC. You could certainly verify pieces of the RME with the QA401. For example, if you put the RME at +13 dBu output range (10.8 dBV), drop it 1 dB (per their spec) to 9.8 dBV, and engage the attenuator on the QA401, the output will hit the QA401 (post 20 dB attenuator) at -10.2 dBV, that would let you measure THD to at least -112, and likely better because the figure THD figure above was a composite between the ADC and DAC. Thus, you could verify with some confidence their spec of -116. Ditto on noise floor. And of course gain and frequency response flatness.

Measuring the RME in a manufacturing setting with a QA401 would be straightforward becuase you'd spot check at performance levels that were degraded somewhat to ensure tester margin (that way you'd not have to put a $30,000 tester in each test bay--you'd only need a $450 tester in each bay). So, you'd measure noise floor with an RME output at -20 dBV. And then you'd measure THD with an RME output -16 dBV. And flatness, and gain, etc. In each case, you are verifying the manufacturing was correct, not that the design was correct. Verifying the design was correct on a DAC like the RME can really only be done in a lab with a $30K tester.

So, what is the value of the QA401? First, price. You won't touch its performance for $450. Second is the absolute calibration. No knobs. No more screwing around with relative levels. If the QA401 told you last month a measurement was -6.7 dBV, you can pull the QA401 off the shelf and re-measure something in 30 seconds and it will be -6.7 dBV. Second is that we bypass Windows audio. No more hidden volume settings, no more base boost turned on by your sound card. No "You've got mail!" chime in the middle of an audio test. No more transcoding. Look on this site alone to see how much time is spent on hidden transcoding issues. Next is programmability. How long would it take you write a test to verify class D efficiency? Verify THD and noise of an amp at both 4 and 8 ohm load? Measure output impedance at 4 ohms? These are all very quick tests to build for the QA401 that don't need any coding. And when they run, it takes just a few seconds each to run. Finally, the input of the QA401 is much more rugged than your average sound card.

If you'd like to try to put some of the above theory to use, I'd be happy to try and help on the forum or via our support email. Just let me know. We could start with verifying the measurement setup and noise floor of the RME, and then see what could be done on THD.

Thanks, Matt
 

Attachments

Joined
Feb 21, 2019
Messages
10
Likes
3
#19
Hi Matt,

Thanks a lot for the detail explanation! Really appreciate it!

When I have time I will re-test with the suggested values.
 

March Audio

Major Contributor
Manufacturer
Patreon Donor
Joined
Mar 1, 2016
Messages
3,373
Likes
2,719
Location
Perth Western Australia
#20
Hi Earfonia, this is Matt from QuantAsylum. In general, your test equipment needs at least 5-10 dB of margin to the equipment (device under test, or DUT) you want to measure if you want a definitive answer. That is, if you hope to measure the DUT's THD of -110 dB, then your test equipment needs to be spec'd at -115 or preferably -120 dB. A tester with a -110 dB THD spec measuring a device with a -110 dB of THD could report a result anywhere from -116 to -104 dB of THD, depending on whether or not the harmonic contributions from the tester and DUT were in or out of phase.

Price is a reasonable proxy for that margin. That is, if you want to measure with confidence a properly designed $1000 DAC with claimed -110 THD+N, you will need test equipment that can measure down to nearly -120 dB THD+N, which puts you in the $10,000+ realm of test equipment.

The expensive testers ($3000 and up) use dozens of relays to keep the ADC and DAC operating at their optimum levels (usually 6 dB steps between ranges, with 9 or more ranges). They do this on both the outputs and the inputs. And the the highest performing equipment (>$10K) will go even further and include a notch filter and/or dedicated analog distortion measurement circuits.

In all cases, even with the most expensive tester, you can't measure everything out of the box. For example, if you drop $30K on an a state of the art audio tester, you won't be able to even slightly detect the distortion of a modern $2 opamp like the OPA1612. In those cases, engineers have been forced to find ways to make a series of measurements that tease that performance out. For opamps, there are circuits such as "distortion magnifiers" (see TI OPA1612 spec for schematic) that will force the opamp to distort 100X more than normal.

And the key with all equipment is understanding where your test equipment's sweet spot is and optimize for that. For example, I just swept the current QA401 device on my bench in loopback (pic attached): On this particular unit, the QA401's best case THD (-115 dB) is found at around -18 dBV, and the best case THD+N (about -101 dB) is found around -4 dBV. The best case SNR (about 112 dBA) is found at 5.8 dBV. The best case N+D (20 to 20 KHz) is about -110 dBV up about -10 dBV. If you have an exceptionally clean DAC, the QA401 ADC N+D can get as low as -115 dBV as the DAC is the larger contributor on the QA401. Overall measured dynamic range on the ADC on this unit is 123 dBA.

Back to your measurement: The RME ADI-2 pro is a $2000 ADC/DAC. You could certainly verify pieces of the RME with the QA401. For example, if you put the RME at +13 dBu output range (10.8 dBV), drop it 1 dB (per their spec) to 9.8 dBV, and engage the attenuator on the QA401, the output will hit the QA401 (post 20 dB attenuator) at -10.2 dBV, that would let you measure THD to at least -112, and likely better because the figure THD figure above was a composite between the ADC and DAC. Thus, you could verify with some confidence their spec of -116. Ditto on noise floor. And of course gain and frequency response flatness.

Measuring the RME in a manufacturing setting with a QA401 would be straightforward becuase you'd spot check at performance levels that were degraded somewhat to ensure tester margin (that way you'd not have to put a $30,000 tester in each test bay--you'd only need a $450 tester in each bay). So, you'd measure noise floor with an RME output at -20 dBV. And then you'd measure THD with an RME output -16 dBV. And flatness, and gain, etc. In each case, you are verifying the manufacturing was correct, not that the design was correct. Verifying the design was correct on a DAC like the RME can really only be done in a lab with a $30K tester.

So, what is the value of the QA401? First, price. You won't touch its performance for $450. Second is the absolute calibration. No knobs. No more screwing around with relative levels. If the QA401 told you last month a measurement was -6.7 dBV, you can pull the QA401 off the shelf and re-measure something in 30 seconds and it will be -6.7 dBV. Second is that we bypass Windows audio. No more hidden volume settings, no more base boost turned on by your sound card. No "You've got mail!" chime in the middle of an audio test. No more transcoding. Look on this site alone to see how much time is spent on hidden transcoding issues. Next is programmability. How long would it take you write a test to verify class D efficiency? Verify THD and noise of an amp at both 4 and 8 ohm load? Measure output impedance at 4 ohms? These are all very quick tests to build for the QA401 that don't need any coding. And when they run, it takes just a few seconds each to run. Finally, the input of the QA401 is much more rugged than your average sound card.

If you'd like to try to put some of the above theory to use, I'd be happy to try and help on the forum or via our support email. Just let me know. We could start with verifying the measurement setup and noise floor of the RME, and then see what could be done on THD.

Thanks, Matt

Hi Matt, thanks for your input. I did own a 401 however moved it on due to the rising noise floor (with increasing signal level) issue. Thing is (apart from your valid points regarding calibration) I can use a "professional" audio interface, achieve similar results but without the rising noise floor. What do you think the issue is here with the 401 and can it be fixed?
 
Last edited:
Top Bottom