The No.1 Website for Pro Audio
Evaluating AD/DA loops by means of Audio Diffmaker
Old 5th April 2020
  #1921
Quote:
Originally Posted by EdwardZ7 View Post
This is E-MU 0404 USB results, not 0404 which is a PCI card.
Correction done. Thanks.

List of the results.

Last edited by didier.brest; 6th April 2020 at 09:29 PM..
Old 7th April 2020
  #1922
Gear Nut
9 years later didier.brest is still constantly updating this post. Give this guy a hand for committing so much time to this! I feel like I say this every couple years here, but thanks man we all appreciate it! You should be getting paid for the amount of advertising you're providing here.
Old 14th April 2020
  #1923
Three of my DACs into my new Mytek Brooklyn ADC. All devices running on their own clocks (unsynchronized mode).

https://bit.ly/34JBURj

Thanks!
Old 14th April 2020
  #1924
Quote:
Originally Posted by Actualsizeaudio View Post
Three of my DACs into my new Mytek Brooklyn ADC. All devices running on their own clocks (unsynchronized mode).
Results from ADM with sample rate correction enabled, operated on the first half of the sample

Metric Halo ULN8 ---> Mytek Brooklyn ADC
-639,6msec, 0,290dB (L), 0,154dB (R). rate adj=-0,0336 ppm.Corr Depth: 41,5 dB (L), 43,5 dB (R) Difference: -61.7 dBFS (L), -63.0 dBFS (R)

Zod Modified RMA AES8 ---> Mytek Brooklyn ADC
-2,284sec, 0,289dB (L), 0,171dB (R). rate adj=-2,1270 ppm.Corr Depth: 43,2 dB (L), 45,3 dB (R) Difference: -63.1 dBFS (L), -63.9 dBFS (R)

Zod Modified RMA Dual 1794 ---> Mytek Brooklyn ADC
-3,33sec, 0,467dB (L), 0,386dB (R). rate adj=-2,3123 ppm.Corr Depth: 43,7 dB (L), 45,8 dB (R) Difference: -63.9 dBFS (L), -65.1 dBFS (R)


Very good indeed ! Please could you record a synchronous loopback file ?

Edit: These tests results have been added to the current list of the test results..

Last edited by didier.brest; 23rd May 2020 at 10:16 PM..
Old 14th April 2020
  #1925
Quote:
Originally Posted by didier.brest View Post
Very good indeed ! Please could you record a synchronous loopback file ?
Sure!

The interesting thing here is that the ZOD/RMA dacs do not have word clock input. They simply derive their clock from the digital signal they receive, in this case AES for the AES8, and Spdif for the 1794. Furthermore, Mytek states in the manual that the Brooklyn ADC will perform best on its own internal clock. That's why I performed an unsynchronized test.

What will be interesting to me is how the system will perform when the ULN8 setup is clocked off of the Mytek. It will also be interesting to see if the Mytek suffers when clocked off of the ULN8 as the manual suggests it might.

Thanks!

Last edited by Actualsizeaudio; 14th April 2020 at 02:55 PM..
Old 14th April 2020
  #1926
Here you go: https://bit.ly/2yVnRfj

Set of tests where the Mytek Brooklyn is the master and the ULN8 is clocked off it. Another set of tests where the Brooklyn is clocked off the ULN8. Note, the Brooklyn is an ADC only, it has no DAC. Also, the ZOD/RMA dacs do not have an internal clock.

Thanks!
Old 14th April 2020
  #1927
If you will indulge me one last test (my last loopback test ever)!

ZOD/RMA 1794 DAC into ZOD/RMA PCM4222 ADC with Mytek Brooklyn as master clock via word clock.

https://bit.ly/2z1***T

Thanks!
Old 14th April 2020
  #1928
Quote:
Originally Posted by Actualsizeaudio View Post
Set of tests where the Mytek Brooklyn is the master and the ULN8 is clocked off it.

Metric Halo ULN8 ---> Mytek Brooklyn as master
ADM: -3,528msec, 0,423dB (L), 0,289dB (R)..Corr Depth: 37,3 dB (L), 54,6 dB (R) Difference: -40.0 dBFS (L), -39.2 dBFS (R)
Matlab: -3.526784 ms, 0.3963 dB (L), 0.2663 dB (R) Difference: -39.6 dBFS (L), -39.6 dBFS (R)

This poor result seems caused by one sample shift between L and R channels. I have compensated this shift by removing the first sample and adding a null sample as last sample of the R channel, so shifting ahead the R channel by one sample. Here's the new measurement result:

ADM: -3,515msec, 0,423dB (L), 0,289dB (R)..Corr Depth: 40,9 dB (L), 42,5 dB (R) Difference: -60.2 dBFS (L), -61.2 dBFS (R)
Matlab: -3.51514 ms, 0.4222 dB (L), 0.2878 dB (R) Difference: -61.2 dBFS (L), -62.6 dBFS (R)

I guess, but has not checked, that there is the same issue on

Metric Halo ULN8 as master ---> Zod Modified RMA Dual 1794 ---> Mytek Brooklyn ADC
ADM: -3,397msec, 0,561dB (L), 0,481dB (R)..Corr Depth: 38,7 dB (L), 72,2 dB (R) Difference: -39.9 dBFS (L), -39.3 dBFS (R)
Matlab: -3.395631 ms, 0.5354 dB (L), 0.4595 dB (R) Difference: -39.6 dBFS (L), -39.6 dBFS (R)

and on

Metric Halo ULN8 as master ---> Mytek Brooklyn
ADM: -2,399usec, 0,000dB (L), 0,000dB (R)..Corr Depth: 37,2 dB (L), 53,4 dB (R) Difference: -39.8 dBFS (L), -39.3 dBFS (R)


I have not yet processed the other files. What do you think about this shift between L and R channels ?

Last edited by didier.brest; 15th April 2020 at 10:26 AM.. Reason: correcting measurement report mistakes
Old 15th April 2020
  #1929
Quote:
Originally Posted by didier.brest View Post
This poor result seems caused by one sample shift between L and R channels. I have compensated this shift by removing the first sample and adding a null sample as last sample of the R channel, so shifting ahead the R channel by one sample.

I have not yet processed the other files. What do you think about this shift between L and R channels ?
Wow, that is messed up.

For the first test I sent you where the Mytek was the master, I recorded into Acon Acoustica off the USB output of the Mytek. For this test, since all units were on the same clock, I recorded directly into Pro Tools. Out of one track in Pro tools, and into another track. That's obviously a common workflow, and should work fine.

Why there would be a 1 sample shift in L/R tracks is beyond me. I suspect that the issue lies inside the Metric Halo Mio console which is the digital mixer the runs on the ULN8. I'll study this more and try and find the issue.

Thanks for the heads up.
Old 15th April 2020
  #1930
Well, we just uncovered something pretty scary with this setup (going back into the Metric Halo via digital signal). This is not something I do in my normal workflow, so I have never uncovered this, but it seems that Didier found a huge bug here. I printed some 2 frame pops of tone to a track and recorded it back into ProTools via AES into a ULN8 and then over MADI into Pro Tools. There is a 1 sample shift in the right track vs the left track as seen in this pic below. This is why performing tests like this is important. It's not just for entertainment sake. Sometimes you uncover something.

I'm going to report this to Metric Halo and get back to this thread when I get some answers.

THANKS!
Attached Thumbnails
Evaluating AD/DA loops by means of Audio Diffmaker-screen-shot-2020-04-14-11.07.23-pm.png  
Old 15th April 2020
  #1931
Measurement after one sample ahead shifting the right channel:

Metric Halo ULN8 as master ---> Mytek Brooklyn ADC
ADM: -3,52msec, 0,420dB (L), 0,286dB (R)..Corr Depth: 40,9 dB (L), 42,5 dB (R) Difference: -60.2 dBFS (L), -61.2 dBFS (R)
Matlab: -3.52113 ms, 0.4215 dB (L), 0.2863 dB (R) Difference: -61.2 dBFS (L), -62.6 dBFS (R)

Metric Halo ULN8 as master ---> Zod Modified Ross Martin Audio Dual 1794 ---> Mytek Brooklyn ADC
ADM: -3,384msec, 0,561dB (L), 0,481dB (R)..Corr Depth: 43,0 dB (L), 44,7 dB (R) Difference: -62.7 dBFS (L), -63.5 dBFS (R)
Matlab: -3.383950 ms, 0.5612 dB (L), 0.4809 dB (R) Difference: -63.4 dBFS (L), -64.7 dBFS (R)

Other reports to be issued later on.
Old 15th April 2020
  #1932
Quote:
Originally Posted by didier.brest View Post

Other reports to be issued later on.
Thanks.

I've filed a bug report with MH and will also be trying to make files later. If I get new ones before you do the tests I'l upload them.

The one I'm most curious about is the PCM4222 test. I'm thinking the very first time I tried this test it had this same flaw and I didn't know it.

Also, preliminary results here tend to speak truth that the Mytek is at least a small amount less accurate when clocked externally, though I doubt one would hear the difference.
Old 15th April 2020
  #1933
Quote:
Originally Posted by didier.brest View Post
Metric Halo ULN8 ---> Mytek Brooklyn as master
(...)
ADM: -3,515msec, 0,423dB (L), 0,289dB (R)..Corr Depth: 40,9 dB (L), 42,5 dB (R) Difference: -60.2 dBFS (L), -61.2 dBFS (R)
Matlab: -3.51514 ms, 0.4222 dB (L), 0.2878 dB (R) Difference: -61.2 dBFS (L), -62.6 dBFS (R)

Quote:
Originally Posted by didier.brest View Post
Metric Halo ULN8 as master ---> Mytek Brooklyn ADC
ADM: -3,52msec, 0,420dB (L), 0,286dB (R)..Corr Depth: 40,9 dB (L), 42,5 dB (R) Difference: -60.2 dBFS (L), -61.2 dBFS (R)
Matlab: -3.52113 ms, 0.4215 dB (L), 0.2863 dB (R) Difference: -61.2 dBFS (L), -62.6 dBFS (R)
Quote:
Originally Posted by Actualsizeaudio View Post
Also, preliminary results here tend to speak truth that the Mytek is at least a small amount less accurate when clocked externally
Small amount indeed. Below measurement resolution.
Old 15th April 2020
  #1934
Quote:
Originally Posted by didier.brest View Post
Zod Modified RMA Dual 1794 ---> Mytek Brooklyn ADC
-3,33sec, 0,467dB (L), 0,386dB (R). rate adj=-2,3123 ppm.Corr Depth: 43,7 dB (L), 45,8 dB (R) Difference: -63.9 dBFS (L), -65.1 dBFS (R)

Quote:
Originally Posted by didier.brest View Post
Metric Halo ULN8 as master ---> Zod Modified Ross Martin Audio Dual 1794 ---> Mytek Brooklyn ADC
ADM: -3,384msec, 0,561dB (L), 0,481dB (R)..Corr Depth: 43,0 dB (L), 44,7 dB (R) Difference: -62.7 dBFS (L), -63.5 dBFS (R)
Matlab: -3.383950 ms, 0.5612 dB (L), 0.4809 dB (R) Difference: -63.4 dBFS (L), -64.7 dBFS (R)
That's funny, I had only looked at these results. Still, that's basically no difference.
Old 15th April 2020
  #1935
Quote:
Originally Posted by didier.brest View Post
Small amount indeed. Below measurement resolution.
B.J. over at Metric Halo has stated (I believe) that the 3D units should behave exactly the same regardless of how they are clocked. It looks like they do, at least on the D/A side of things.
Old 15th April 2020
  #1936
Quote:
Originally Posted by Actualsizeaudio View Post
The one I'm most curious about is the PCM4222 test. I'm thinking the very first time I tried this test it had this same flaw and I didn't know it.
Zod Audio modified Ross Martin Audio 1794 DAC into Zod Audio modifed Ross Martin Audio PCM4222 ADC with Mytek Brooklyn as master clock via word clock
Raw loopback file
ADM: -2,955msec, 3,848dB (L), 3,753dB (R)..Corr Depth: 30,7 dB (L), 35,9 dB (R) Difference: -35.6 dBFS (L), -33.3 dBFS (R)
Matlab: -2.968431 ms, 3.7945 dB (L), 3.7528 dB (R) Difference: -34.7 dBFS (L), -34.7 dBFS (R)
With R vs. L one sample ahead shift
ADM: -2,942msec, 3,856dB (L), 3,751dB (R)..Corr Depth: 31,7 dB (L), 33,8 dB (R) Difference: -34.9 dBFS (L), -34.4 dBFS (R)
Matlab: -2.953458 ms, 3.8097 dB (L), 3.7572 dB (R) Difference: -35.6 dBFS (L), -35.0 dBFS (R)

Last edited by didier.brest; 15th April 2020 at 08:08 PM..
Old 15th April 2020
  #1937
Quote:
Originally Posted by Actualsizeaudio View Post
That's funny, I had only looked at these results. Still, that's basically no difference
The small difference may be caused by one test being done on the full sample with a common clock and the other one being done on the first half of the sample with each converter running on its own clock, this being called the unsynchonized mode:

Unsynchronized mode. The DA and AD converters run each on its own clock. Measurement performed with sample rate drift correction activated in Audio DiffMaker and limited to the first half of the sample. Because the RMS level of this first half is 1 dB lower than the RMS level of the full-size sample, one must increase the difference RMS level figures by 1 dB for comparison with the synchronized mode.


Here's the result with master clock from Brooklyn ADC (always with R vs. L one sample ahead shift):

Zod modified Ross Martin Audio Dual 1794 ---> Mytek Brooklyn ADC as master
ADM: -3,368msec, 0,564dB (L), 0,484dB (R)..Corr Depth: 43,0 dB (L), 44,7 dB (R) Difference: -62.8 dBFS (L), -63.7 dBFS (R)
Matlab: -3.368526 ms, 0.5634 dB (L), 0.4837 dB (R) Difference: -63.4 dBFS (L), -64.7 dBFS (R)

to be compared with this previous result

Metric Halo ULN8 as master ---> Zod Modified Ross Martin Audio Dual 1794 ---> Mytek Brooklyn ADC
ADM: -3,384msec, 0,561dB (L), 0,481dB (R)..Corr Depth: 43,0 dB (L), 44,7 dB (R) Difference: -62.7 dBFS (L), -63.5 dBFS (R)
Matlab: -3.383950 ms, 0.5612 dB (L), 0.4809 dB (R) Difference: -63.4 dBFS (L), -64.7 dBFS (R)

Two remaining AES8 ---> Brooklyn tests to be measured later on.
Old 16th April 2020
  #1938
Quote:
Originally Posted by Actualsizeaudio View Post
I'm thinking the very first time I tried this test it had this same flaw and I didn't know it.
I have measured again your following tests already in the List of the results. but with the same R vs. L shift compensation done before measurement for your present tests. The comparison of the old (without shift) and the new results (with shift) show that there was not the same flaw at this time (2017-2018).

Metric Halo ULN8 DAC to Ross Martin Audio PCM4222ADC, ADC master or DAC master: Old -37.5 dBFS (L), -38.7 dBFS (R) - New -35.7 dBFS (L), -35.7 dBFS (R)

Ross Martin Audio AES8 DAC to Ross Martin Audio PCM4222ADC master: Old -49.7 dBFS (L), -50.9 dBFS (R) - New -39.3 dBFS (L), -39.3 dBFS (R)

Zod Audio modded Ross Martin Audio AES8 into Zod Audio modded RMA PCM4222 slave to Metric Halo ULN-8 3d : Old -49.6 dBFS (L), -51.0 dBFS (R) - New -39.3 dBFS (L), -39.3 dBFS (R)

Zod Audio modded Ross Martin Audio AES8 into Metric Halo ULN-8 3d master: Old -56.8 dBFS (L), -58.2 dBFS (R) - New -39.6 dBFS (L), -39.6 dBFS (R)

Last edited by didier.brest; 17th April 2020 at 08:51 AM..
Old 16th April 2020
  #1939
Quote:
Originally Posted by didier.brest View Post
I have measured again your following tests already in the List of the results. but with the same R vs. L shift compensation done before measurement for your present tests. The comparison of the old (without shift) and the new results (with shift) show that there was not the same flaw at this time (2017-2018).
I can't thank you enough for that. I was teching this whole thing out, and then got busy with a deadline. You have been a huge help. I owe you a few beers.
Old 16th April 2020
  #1940
With R vs. L one sample ahead shift:

Metric Halo ULN8 as master --> Zod Audio modded Ross Martin Audio AES8 DAC ---> Mytek Brooklyn ADC
ADM: -3,883msec, 0,188dB (L), 0,148dB (R)..Corr Depth: 42,5 dB (L), 44,1 dB (R) Difference: -62.2 dBFS (L), -63.2 dBFS (R)
Matlab: -3.82945 ms, 0.1870 dB (L), 0.1488 dB (R) Difference: -62.8 dBFS (L), -64.2 dBFS (R)

Zod Audio modded Ross Martin Audio AES8 DAC ---> Mytek Brooklyn ADC as master
ADM: -3,867msec, 0,173dB (L), 0,142dB (R)..Corr Depth: 42,5 dB (L), 44,1 dB (R) Difference: -62.2 dBFS (L), -63.3 dBFS (R)
Matlab: -3.867748 ms, 0.1733 dB (L), 0.1457 dB (R) Difference: -62.8 dBFS (L), -64.2 dBFS (R)

To be added to the next issue of the list of the results.
Old 17th April 2020
  #1941
Gear Maniac
 

Quote:
Originally Posted by didier.brest View Post
With R vs. L one sample ahead shift:

Metric Halo ULN8 as master --> Zod Audio modded Ross Martin Audio AES8 DAC ---> Mytek Brooklyn ADC
ADM: -3,883msec, 0,188dB (L), 0,148dB (R)..Corr Depth: 42,5 dB (L), 44,1 dB (R) Difference: -62.2 dBFS (L), -63.2 dBFS (R)
Matlab: -3.82945 ms, 0.1870 dB (L), 0.1488 dB (R) Difference: -62.8 dBFS (L), -64.2 dBFS (R)

Zod Audio modded Ross Martin Audio AES8 DAC ---> Mytek Brooklyn ADC as master
ADM: -3,867msec, 0,173dB (L), 0,142dB (R)..Corr Depth: 42,5 dB (L), 44,1 dB (R) Difference: -62.2 dBFS (L), -63.3 dBFS (R)
Matlab: -3.867748 ms, 0.1733 dB (L), 0.1457 dB (R) Difference: -62.8 dBFS (L), -64.2 dBFS (R)

To be added to the next issue of the list of the results.
Isn't the mytek the most transparent so far?
Old 18th April 2020
  #1942
Lives for gear
 
funka's Avatar
 

Nope, it is still Forssell MDA-4 paired with an old Focusrite Blue 245:
Quote:
Focusrite sensitivity 0 dBFS = 20 dBu, serial number = 10
4.0 dB (L), 4.0 dB (R) Corr Depth: 66.9 dB (L), 69.8 dB (R) Difference*: -81.2 dBFS (L) -81.1 dBFS (R)
Old 18th April 2020
  #1943
Gear Maniac
 

Quote:
Originally Posted by funka View Post
Nope, it is still Forssell MDA-4 paired with an old Focusrite Blue 245:
apparently anything paired with an old Focusrite Blue 245
Old 19th April 2020
  #1944
Lives for gear
 
funka's Avatar
 

The DAC must also be great, otherwise, the score is "bad".
It works great with Forssell, very nice with Madison.
With Lynx Aurora 16, it is not that great.
From what has been tested, we can deduce that Focusrite blue 245, PCM 4222 and MSB Studio ADC master are very good ADCs.
The Mytek Brooklyn ADC also seems to be one great ADC.
The Pacific Microsonics HDCD is also another one, not measured here.
Old 19th April 2020
  #1945
Gear Maniac
 

Quote:
Originally Posted by funka View Post
The DAC must also be great, otherwise, the score is "bad".
It works great with Forssell, very nice with Madison.
With Lynx Aurora 16, it is not that great.
From what has been tested, we can deduce that Focusrite blue 245, PCM 4222 and MSB Studio ADC master are very good ADCs.
The Mytek Brooklyn ADC also seems to be one great ADC.
The Pacific Microsonics HDCD is also another one, not measured here.
But how come the Madison pairs so well with the Blue 245 and achieves a phase out beyond let's say a Lynx, which is besides the Mytek ruling current transparency values? Because the Madison by itself is really bad compared to current contenders.

I mean how is that happening? Have you done the files of the focusrite blue with the SPL?

And what was the setup mode? Like how where they paired? What did the D to A / A to D?
Old 19th April 2020
  #1946
Lives for gear
 
funka's Avatar
 

Madison alone loop D/A/D was not great, yes. Their A/D is the bottleneck.
But Madison DAC with blue AD was great, nearly as great as Forssell + blue AD.
That shows how great Madison DACs are.
I did the files, yes. As far as I remember, Blue AD was the clock master for Aurora 16 and Madiface, with Word Clock.
Old 19th April 2020
  #1947
Gear Maniac
 

Quote:
Originally Posted by funka View Post
Madison alone loop D/A/D was not great, yes. Their A/D is the bottleneck.
But Madison DAC with blue AD was great, nearly as great as Forssell + blue AD.
That shows how great Madison DACs are.
I did the files, yes. As far as I remember, Blue AD was the clock master for Aurora 16 and Madiface, with Word Clock.
That doesn't make sense, have you tried pairing the Madison with another great AD and compared the results? I feel like it has something to do with the filters being used at the same time that creates this.

I mean the forsell and the Madison aren't that far apart, just one dB according to the measurements.
Old 19th April 2020
  #1948
Lives for gear
 
funka's Avatar
 

Why this wouldn't make sense? ADC and DAC are not same circuits/chipsets in the SPL Madison.

I did those kind of crossed tests with the Forssell MDAC-4.
The score is only high paired with the Blue 245, not with Studer D19 / Aurora 16 or a Forssell A/D also.

The Aurora dac paired with the Blue 245 also scored average.
Old 20th April 2020
  #1949
Gear Maniac
 

Quote:
Originally Posted by funka View Post
Why this wouldn't make sense? ADC and DAC are not same circuits/chipsets in the SPL Madison.

I did those kind of crossed tests with the Forssell MDAC-4.
The score is only high paired with the Blue 245, not with Studer D19 / Aurora 16 or a Forssell A/D also.

The Aurora dac paired with the Blue 245 also scored average.
Yes but have you tried another good AD with the SPL to compare it with the blue results?
Old 20th April 2020
  #1950
Lives for gear
 
funka's Avatar
 

No, but I could do it with Studer D19 and Aurora 16.
πŸ“ Reply
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
πŸ–¨οΈ Show Printable Version
βœ‰οΈ Email this Page
πŸ” Search thread
♾️ Similar Threads
πŸŽ™οΈ View mentioned gear