Once the difference between the original and the time and gains aligned loopback copy has been extracted, it is audible and its spectral analysis shows that the part of the energy between 40 Hz and 10 kHz is larger than the part outside this band. Whether it is still audible when mixed with the original is another question, the answer of which is highly subjective. Audio sample and spectrum display for the MOTU 828es here attached.

Is there some published study results proving that non-linear phase difference effects (= phase difference effects not related to a time shift) would be less audible than other ones (frequency response level, non linear distortion) having the same RMS level?

Anyway phase linearity matters to high end audio gear brands:
Phase Response (35 dB Gain, 50 Hz - 20 kHz bandwidth, +27 dBu Out) < 2 degrees deviation

List of the results.

Loopback tests requested by forum members: Apogee Symphony Desktop, Eventide H8000FW and H9000R (for confirming the one at the top of the list of the results), Pacific Microsonics Model One and Model Two, Slate Digital VRS-8, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro. Of course any other one welcome!

I think you posted the spectrum of the original file, not the difference. Since DeltaWave produces all the spectrum and phase difference charts, here's the analysis for the same MOTU, attached.

Motu-delta.png is the spectrum of the difference
Motu-compared.png is the spectrum of the original2 and motu loopback, overlaid
delta-phase.png is the actual phase difference, in degrees
zoom.png is the zoomed-in portion of the spectra where the filter causes larger differences.

I recently came across this forum where lots of prosumer-ish DACs are tested, here is a recent one

https://www.audiosciencereview.com/f...ced-dac.20259/

You have to kind of dig around in there a bit, or I haven't found a way to see an updated list with all of them, but it's interesting to see something like the Soncoz LA-QXD1 or Schiit Modius, which are about $200 hang in there with RME's top offering and outperform (on this scale) benchmarks and graces.

Has anyone dabbled with the prosumerish brands? Obviously no AD/DA with these, but it's interesting to see some other reviews that are in more of an objective space...It would be cool to see how some of these $100/ch dacs stack up in this loopback test.

Wait what, now I am confused. Was the 828es not measuring way worse at the list of results at the beginning of this year, around -40dB or something like that?

Was the measurement wrong?

EDIT: Would it be possible that the other Motu AVB devices are measured wrong as well?

Yes because of a one sample shift between L and R channels.

No.

My bad! I have corrected my mistake. It is the spectrum of the wav sample attached to the same post, the first 34s section of the difference. This new plot, showing most part of the energy between 40 Hz and 2 kHz, makes your statement

still more questionable.

List of the results.

Loopback tests requested by forum members: Apogee Symphony Desktop, Eventide H8000FW and H9000R (for confirming the one at the top of the list of the results), Pacific Microsonics Model One and Model Two, Slate Digital VRS-8, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro. Of course any other one welcome!

You have to be careful how you respond to some of these people, they are hyperliteral and some of them believe in runaway skepticism.

I want to point out the reason I brought up this result is because it is the only one that differed on software alone between AudioDiff and Deltawave. All other results the differences were negligible.

So in terms of understanding relative differences between devices, either software package suffices and there are no differences in measurements that are significant (unweighted).

For these people, they need to be told it's "unlikely" given there is no significant differences between the results.

Hey, you know that I can see your comment as well, do you?
I asked, because the 828es and the other AVB converters have the same parts according to MOTU. I kind of thought it is unlikely they have the exact same behaviour because they measured around -50dB and not -40dB like the 828es, but still.

Can anyone answer this please?
Is the delta between the original test file and result, as big as running the result file through the unit again? If the unit processes in a certain way and all ducks are already lined up in the row the 2nd time, the result might be a lot higher?
If so measuring the differences between the 1st and 2nd run might be more interesting.
What if the sourcefile was recorded via an 8000? You would then just measure how far every card is off from the 8000? Not how well it’s holding up.
The source has already an imprint from a converter.

Time alignment is accomplished by both, Didier's script and DeltaWave to an accuracy of a tiny fraction of a sample. Any initial clock misalignment will be removed and have no effect on the result.

No, it's somewhere around -60 dBFS for 1 µs time alignment error, -120 dBFS for 1 ns, -180 dBFS for 1 ps.

List of the results.

Loopback tests requested by forum members: Apogee Symphony Desktop, Eventide H8000FW and H9000R (for confirming the one at the top of the list of the results), Pacific Microsonics Model One and Model Two, Slate Digital VRS-8, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro. Of course any other one welcome!

The 828 mkii performs so well because it has 8 converters and DC coupled output. These two things alone are huge. But yeah, that one Motu 828 result being lower (the 5\6 line in) was the outlier that I noticed. It differed by 16 points between the two software packages which is why I called it out.

The other debates involving where these should sit on the list is a debate that is mostly about dB(A) vs unweighted.

After going through all these results and doing some analysis, I find it hard to fault using this thread as opposed to the dB(A) weighted lists being created outside here. There are a couple of runaway skeptics here bashing on it, but I don't really see much merit to their hypothosis.

Can they run the Eventide 8000 again without correcting for anything? (expecting 79ish uncorrected)

Hi,

here are some new files :

Motu 828 MK2 Firewire 3-4 to 3-4 : Link

Motu 828 MK2 USB 3-4 to 3-4 : Link

Motu 828 MK2 USB 3-4 to 5-6 : Link

Presonus Firestudio Mobile Main Out to Line In 3-4 : Link
I'm not sure gain was perfect for this one.

I tested the Motu USB twice (could have been the same with the Firewire one) to see if there was a difference between using the same chip for D/A and D/A and using a different chip (they go by pair, so 3-4 is not the same chip than 5-6, even if they are both AK4528).
I can try later with the Main Out as it's different DAC (AK4382) to see if there's any change

Don't know if someone already talked about that, but is it possible that the "better than supposed" results of the three old Motu (828 MK2, 2408 MK3 and 24 I/O) would be due to using a chip doing both D/A and A/D instead of two different chips like in the majority of other devices ?

I have a Khadas Tone2 Pro (costing $200 and ranking a bit lower than the D30Pro but higher than Soncoz LA-QXD1 or Schiit Modius), so I could try it. It even has balanced output, but the main problem would be to sync it with the ADC as the only sync signal it could provide would be the SPDIF Out, and it's not enabled on the current firmware, should be in the next).
My guess is that good result of this DAC is more synonym of low distorsion than transparency if used in a loopback...

Try this link and then select device type, etc.: https://www.audiosciencereview.com/f...pment_Reviews/

MOTU 828 MK2 USB - 3-4 to 5-6.wav
31.480 µs, 0.1036 dB (L), 0.0910 dB (R), -56.7296 dBFS (L), -58.0631 dBFS (R)

MOTU 828 MK2 USB - 3-4 to 3-4.wav
-13.872 µs, 0.1037 dB (L), 0.0911 dB (R), -56.7294 dBFS (L), -58.0631 dBFS (R)

MOTU 828 MK2 FW - 3-4 to 3-4.wav
8.877 µs, 0.1280 dB (L), 0.0969 dB (R), -56.7335 dBFS (L), -58.0707 dBFS (R)

To be added to the next issue of the

list of the results.

Loopback tests requested by forum members: Apogee Symphony Desktop, Eventide H8000FW and H9000R (for confirming the one at the top of the list of the results), Pacific Microsonics Model One and Model Two, Slate Digital VRS-8, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro. Of course any other one welcome!

This thread and the big converter lag thread should be combined into a converter data base.

Perhaps this is a forum Idea. To have a data base of gear info. Yes the reviews are one approach. But having a few people creating a usable data base about things would be the best way to make the info available to the public.

I've bookmarked both because I consider them essential.

Big converter lag? Would you mind to post the link?

The obvious solution here is to host a google spreadsheet so new columns can be added.

Hi Didier,
maybe I missed it but I didn't see the reason why there was a one sample shift.
Does it comes from the unit, or from user having created that shift when cutting the file for example ?

I don't know. Same shifts happened in tests done by Actualsizeaudio involving a Metric Halo ULN8 and in the test of the MOTU 828es done by sm5 (s and m in lower case, automatically replaced here by uppercase, I can't do anything against this stupid automatic advertising link insertion).

List of the results

Loopback tests requested by forum members: Apogee Symphony Desktop, Eventide H8000FW and H9000R (for confirming the one at the top of the list of the results), Pacific Microsonics Model One and Model Two, Slate Digital VRS-8, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro. Of course any other one welcome!

Thanks, two different devices... I would bet on software part then, but would be good to know what they used, and see if it's the same.

Here is a recording using Apogee Symphony Desktop using macOS system drivers (not Apogee's) and the clean preamp ("SD-MP") with input gain set to 0 on the device. I recorded in Logic 10.4.8 on macOS 10.13.6 with the test file track & output gain set to 0.

https://mordent.io/aud/ApogeeSymphonyDesktopDAAD3.wav

-1.732580 ms, 6.1836 dB (L), 6.1488 dB (R), -48.7185 dBFS (L), -50.0059 dBFS (R)

To be added to the next issue of the

list of the results.

Loopback tests requested by forum members: Eventide H8000FW and H9000R (for confirming the ones at the top of the list of the results), Pacific Microsonics Model One and Model Two, Slate Digital VRS-8, SSL 2+, RME M-32 AD Pro --> M-32 DA Pro, RME M-1610 Pro, Audient iD14 MKII, Prism Dream DA-2 --> AD-2, BlackLion Audio Revolution 2×2, RME MADIface Pro. Of course any other one welcome!