RMAA Tests: (Welcome to add others!)

[sm5]

Here's some RMAA tests.

If anyone has other results to share as far as recording interface RMAA tests: you're welcome to post in this thread as more info on recording interfaces would be nice to know! -Or better yet send it in to be tested also View attachment 85392View attachment 85393View attachment 85394View attachment 85395

 

[pozz]

The drawn-in names are nice touch.

You do loopbacks? How do you set levels?

 

[Blumlein 88]

You could try looping form one device to another in a round robin way. Give you some idea if a device has a better ADC or DAC.

Also over at Gearslutz they have a long running thread where people do a loopback with a standard bit of music and report the results originally in Diffmaker later in some custom written software. We could do that and use pkane's Deltawave. It gives results within a db or so of the same answer.

We could use their file or choose one of our own.

It is interesting how measurements and nulling don't always rank devices in the same order.

 

[Blumlein 88]

The drawn-in names are nice touch.

You do loopbacks? How do you set levels?

If I remember rightly, RMAA sets levels at -1 db for most of their signals. So setting the ADC level to get near or just below -1 db is usually how it is done. I think last time I did RMAA in loopback I sent a 1 khz -1 db tone and set the input to get me the same -1db. Then let RMAA do its thing and nothing clipped or caused issues.

 

[pozz]

For the 828ES ... if the analog output trim is 0 and analog input trim is 0: the levels match and the measurements are pretty consistent at about -1. Probably so that you can easily use outboard gear. USUALLY things are pretty simple for the interfaces I've owned ... but if the I/O levels aren't equal I usually go with 100% output and put the input to wherever it hits -1. (Usually the cheaper interfaces or ones where the input always passes through a preamp) I do try the opposite with AD higher and DAC lower /other settings to sort of see what's the best/most consistent. (I don't have my Hilo measurements anymore but that was pretty simple too as there's set AD/DAC trim settings so you can have them match)

I was actually going to do a test of the Anubis/828ES when I had them both to see about differences in AD/DA, but the Anubis drivers/routing bugged me so I just didn't care anymore

As far as the Diffmaker tests on Gearslutz:
I provided the Lynx Hilo results (I've had two!)/Audient id44/Id22/ID14/Motu828ES/Zoom F4/Tascam UH-7000/Mixpre 10T tests! (I've had Mytek and Prism and the original high end Antelope DACs and who knows what over the years but never measured them)

Diffmaker always gave me crazy numbers when I used it and crapped out a lot ... doesn't help it hasn't been updated in years and there's no real documentation as to precisely what's going on. It's like if something measures great on Diffmaker ... cool - more information is more information: but I personally wouldn't put too much trust on the single figure knowing how questionable the software is in the first place. I haven't looked into Deltawave though.

I quit Gearslutz after nearly 15 years not too long ago & deleted my account It's okay for some things like new product announcements and to see if other people are having driver problems or perhaps something like ... don't send charter oak mics in for repair as you'll never get it back Otherwise it's mostly people giving bad advice (or possibly great advice) but either way, you never know as you don't know what they are listening on or their tastes or if they did proper A/B tests most of the time.

I quit before the Neumann V402 preamp thread: but just viewing that thread made me glad I quit earlier I have to say that Hugh Robjohns reviews that I've read for many years at least sometimes have AP measurements unlike many other reviewers/magazines. Same with Cranborne: they have all the AP measurements and info on their site/manual. Neumann (which does have great measuring speakers) ... has NO specs at all for their $2900 preamp that on their site says is "fully transparent" "natural splendor" and "highest degree of linearity and sonic purity" WOW! too bad there's no measurements to back it up.

Overall I just hope that more manufacturers of pro audio gear simply release their AP specs/measurements as obviously most of at least the big preamp/interface manufacturers are using measurement gear to make their stuff.

Gearslutz has really changed, or maybe it's the same and just the scale is different. I been lurking since early 2000s and it's hard to find reasons to keep checking back. Seems like the new wave of manufacturers have really tried to appeal to every camp at the same time (looking at you, HEDD).

 

[bennetng]

Here's some RMAA tests I've done within the last year or so if anyone is curious.

Obviously you can't differentiate between the AD or DA ... but I've found that generally my results seem to be pretty close to other people's measurements of similar interfaces. Generally when I run a bunch of the tests things usually stay pretty much the same even if I change the sample rate from 48 to 96khz or if I measure the same interface a year or two later.

I find the RMAA measurements on Exound & Prosound IXBT & Archimago's Musings sites usually seem to be pretty good resources as far as comparing interfaces. Most manufacturers with great published specs usually seem to have great RMAA results which is a good sign too.
Not nearly as informative as the tests here with AP ... but to get a general idea of how a device might perform it's better than no information.

*Perhaps saving you some money also ... for example I measured my MOTU 828ES and realized it was pretty close to the RME ADI-2 PRO FS so I saved by not 'upgrading' The FS R is a TINY bit better ... but not personally worth it to lose the 2 preamps and 7.1 surround outputs and 8 analog inputs. Perhaps to someone it is worth the extra cost.

*I did measure a Zoom F4 at the same time I measured the Mixpre 10T ... lost the results but they weren't too far off the Mixpre actually if I remember right. Apollo X had super bad windows drivers and I gave up on that

If anyone has other results to share as far as recording interface RMAA tests: you're welcome to post in this thread as more info on recording interfaces would be nice to know! -Or better yet send it in to be tested also View attachment 85392View attachment 85393View attachment 85394View attachment 85395

If you still have access to these devices, or you saved the resulting waveform, you can attach the SAV files here, so we can see zoomable FFT plots. Actually, more detailed than static AP screenshots, and file size is small enough for this forum (2MB limit).

As for the Anubis, the big differences between dynamic range and noise level could be a sign of stacking multiple ADCs with different input gains.

 

[Blumlein 88]

Diffmaker always gave me crazy numbers when I used it and crapped out a lot ... doesn't help it hasn't been updated in years and there's no real documentation as to precisely what's going on. It's like if something measures great on Diffmaker ... cool - more information is more information: but I personally wouldn't put too much trust on the single figure knowing how questionable the software is in the first place. I haven't looked into Deltawave though.

.

Go get Deltawave. Its free and it is terrific.
https://deltaw.org/

@sm5

It will make you forget you ever heard of Diffmaker. Paul added lots of nice extra features, but the core feature of gain and time aligning files for nulling works very, very well. And it doesn't blow up on you. Plenty of graphs of different parameters as well. Like relative frequency response between gear using the music files.

While there you can try Distort which lets you add customized distortion to a file and see if you can hear the difference. Lets you see if everyone's wives tale about 2nd and 3rd harmonics being euphonic is true or not.

 

[Blumlein 88]

Here are some results from my Antelope Audio Zen Tour. I'll attach the RMAA .sav files in a zip file. Something weird on the left channel at elevated sample rates. It isn't in the device with other software. I could switch and make the left any one of several outputs, and it didn't matter some of the measures on the left channel are corrupted with low frequency fluctuating noise at rates above 48 khz. RMAA is a little better than Diffmaker, but I wish we could do better. So really only the right channel is accurate for this device using RMAA. Monitor output level was 0 db or max. Input level was 0 db gain.

@pkane Don't know if you have any interest, but a nice automated measurement program to do what RMAA does would be nice if it was developed like you have Deltawave and Distort. If you don't have any interest fine, but just a shot in the dark if you are looking for another project and it interests you.

Okay, I have a file for doing various measurements in loopbacks. My results are generally pretty close to what RMAA is showing in the right channel. A db or two different at most. I know from other work the DAC is much less good than the ADC in the Zen Tour. In most parameters the ADC is a few db better than this loopback result. And for instance in distortion the ADC has some 2nd and 3rd harmonic at very low levels while the DAC has harmonics up to the 9th. Also at low levels, but higher than in the ADC. Since it is a recording interface I'd rather the better part be in the ADC.

Also even with the DAC being a bottleneck this device nulls better than most. Which is interesting. While not quite at the top, it is just below the very best on the nulling results thread at Gearslutz. Well in front of some very expensive gear. It also nulls much better than the RME ADI Pro unit which has better measurements for DAC and maybe for ADC so why does the RME null less well? I don't know.

 

[bennetng]

Lynx E44 & LynxTWO-A RMAA results with SAV files:
https://www.gearslutz.com/board/gea...974574-lynx-e44-amp-lynxtwo-rmaa-results.html

SoX with some scripting can be handy for automated tests as well:
https://www.audiosciencereview.com/...to-sox-audio-tool-as-a-signal-generator.4242/

SoX's spectrogram can be tuned in a very unforgiving way. For example, let's save the text below as a Windows .CMD or .BAT file and put it in the SoX directory, then drag one or multiple audio files on the .CMD file to get spectrogram png files.

@echo off
pushd %~dp0
for %%1 in (%*) do sox -V %%1 -n spectrogram -z 180 -w dolph -h -o "%~dp1%%~n1.png" -c "%%~nx1"
pause

Download the test files here:
https://src.infinitewave.ca/faq.html

Here is SoX's default resampler quality (high) appeared on the website:

...and how it looks like with high contrast palette:

The background is completely black only when using the very high quality (-v) switch:

The plots above are generated with 32-bit integer reference file and saved as 32-bit integer. The 24-bit reference file itself is much "noisier" even before resampling:

 

[bennetng]

Here's RMAA tests with the save files for those who asked: I only have the M4 and 828ES left so I can't retest any of the others.

The first two are MOTU M4 3-4 out to 3-4 in.
The 3rd is M4 3-4 output into the 828ES input 7/8 inputs. Makes sense it's higher as the AD is the higher end chip on the 828ES.

The last two are 828ES loopbacks: 7-8 out into 7-8 in. (one was MME out to ASIO in ... last was Asio to Asio)

As far as XLR output on the 828ES or 1/2 output that goes through the volume control on the M4 ... they are pretty close: maybe a tad lower: but I didn't want to pull out the XLR to TRS cable to do that test again.

*Also for whoever uploads results: a picture of the numbers would be nice to include too even if you have the save file: for people who just want to see the general results and don't want to download any files.


View attachment 85963

Very nice results. Also it has everything: word clock, ethernet, thunderbolt, ADAT, coaxial, MIDI, dual headphone output, hardware DSP... and at an unbeatable price.

 

[Rja4000]

You could try looping form one device to another in a round robin way. Give you some idea if a device has a better ADC or DAC

You need levels to remain the same for all devices to be able to compare.

 

[Rja4000]

It is interesting how measurements and nulling don't always rank devices in the same order.

Probably some filter in the loop ?

 

[Blumlein 88]

Here's RMAA tests with the save files for those who asked: I only have the M4 and 828ES left so I can't retest any of the others.

The first two are MOTU M4 3-4 out to 3-4 in.
The 3rd is M4 3-4 output into the 828ES input 7/8 inputs. Makes sense it's higher as the AD is the higher end chip on the 828ES.

The last two are 828ES loopbacks: 7-8 out into 7-8 in. (one was MME out to ASIO in ... last was Asio to Asio)

As far as XLR output on the 828ES or 1/2 output that goes through the volume control on the M4 ... they are pretty close: maybe a tad lower: but I didn't want to pull out the XLR to TRS cable to do that test again.

*Also for whoever uploads results: a picture of the numbers would be nice to include too even if you have the save file: for people who just want to see the general results and don't want to download any files.


View attachment 85963

I notice you have the same problem I had. Your left channel THD is much higher than the right. And I am pretty sure you'd find you could reverse channels and yet the left would read high rather than following the connection change.

 

[Blumlein 88]

Probably some filter in the loop ?

Usually that appears to be the case. Often in these recording interfaces it is where and how steeply they cut off the low end response. Not common for these to be DC coupled.

 

[bennetng]

I notice you have the same problem I had. Your left channel THD is much higher than the right. And I am pretty sure you'd find you could reverse channels and yet the left would read high rather than following the connection change.

Same thing from Amir's measurement. Wonder if using input other than USB makes a difference or not. I meant, even when there is ESS hump, different levels of hump in different channels are unique to these MOTUs.

 

[Blumlein 88]

Same thing from Amir's measurement. Wonder if using input other than USB makes a difference or not. I meant, even when there is ESS hump, different levels of hump in different channels are unique to these MOTUs.

No I think the problem isn't in the MOTU. It is in the RMAA program.

I had the left channel issues, and I was using an Antelope Audio interface. It happened whenever sample rates were above 48 khz.

Maybe @sm5 can repeat the RMAA tests at 44 and 48 khz. See if that left channel looks a whole lot better.

 

[bennetng]

No I think the problem isn't in the MOTU. It is in the RMAA program.

I had the left channel issues, and I was using an Antelope Audio interface. It happened whenever sample rates were above 48 khz.

Maybe @sm5 can repeat the RMAA tests at 44 and 48 khz. See if that left channel looks a whole lot better.

But not in my own 96kHz test? Did you check the recorded waveform with external programs?
https://www.audiosciencereview.com/forum/index.php?attachments/x-fi-titanium-hd-zip.29512/

RMAA has issues in 88.2k and 176.4k tests at least in the current version (6.4.5), but other sample rates seem fine, including digital-only tests like this:
https://www.audiosciencereview.com/...ds/the-sound-quality-of-dsd.14773/post-523597

 

[Blumlein 88]

But not in my own 96kHz test? Did you check the recorded waveform with external programs?
https://www.audiosciencereview.com/forum/index.php?attachments/x-fi-titanium-hd-zip.29512/

RMAA has issues in 88.2k and 176.4k tests at least in the current version (6.4.5), but other sample rates seem fine, including digital-only tests like this:
https://www.audiosciencereview.com/...ds/the-sound-quality-of-dsd.14773/post-523597

I didn't check the recorded waveform of the RMAA file during the test. I have sent that file thru and recorded it with other software. There is no distortion or noise in the result from the device which fits with left channel results reported by RMAA. Also I'll reiterate though not shown I did this switching various inputs to be the left. So of 4 possible inputs, all would show high distortion if they were assigned as the left channel for RMAA. All would test properly when assigned as the right channel. I also had two monitor outs, and had the same result with either of the pairs of monitor outputs.

 

[Blumlein 88]

Ok, your right. It isn't left and right. It is a bit confusing if you look into what it is. Apparently it is tones at different levels. The lowest being -30 db which is either is actually THD+N or the noise floor confuses it thinking it is seeing harmonic distortion. I guess my particular device has a low noise floor at 44 and 48 khz and maybe some ultrasonic noise at higher rates.

 

[bennetng]

Oh I see what you were talking about. I usually ignore the last three tests as they don't show up in the html report. I was talking about the 1kHz THD test, not the sweep THD test.