• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

E1DA Cosmos ADC

MC_RME

Addicted to Fun and Learning
Technical Expert
Audio Company
Joined
May 15, 2019
Messages
871
Likes
3,613
Wowza - how did I missed that? Downloading now... - not - no link?
 

Marty111

Active Member
Joined
Aug 16, 2021
Messages
110
Likes
8
I received my Cosmos ADC today. I want to use it to record the sound that comes out from the balanced output of my Presonus TubePre (version 1), but I don't know what voltage I should select on the Cosmos ADC.
I plan on using the Cosmos ADC as my recording studio device.
 

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,752
Likes
4,646
Location
Liège, Belgium
For multiple tones you need a measurement function that removes all stimulus tones from the FFT result. The remaning rest is Total Distortion plus Noise, TD+N (not THD+N). Not sure REW supports this at all. AP does.

Virtins Multi-Instrument 3.9 does it too.
I never had a chance to compare to AP results though.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,732
Likes
6,101
Location
Berlin, Germany
For those of you that have a Cosmos ADC (I don't... yet) could somebody run AlexCP's 7kHz+13kHz 1:1 twin-tone with REW (using separate DAC channels mixed together passively)? I found this one the most revealing twin-tone as it tests for loop gain around 20kHz, too bad it's not a standard, yet (see this post and following posts).
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,732
Likes
6,101
Location
Berlin, Germany
Skirting caused by random jitter is typically included in the fundamental when calculating THD+N while distinct spurs count as noise.
We might need another test that looks at the skirt and close-in sidebands vs. zero signal noise: Notch out the fundamental and all harmonics very narrow-band using the inverse envelope of the window function (plus some margin) to truncate the skirt as little as possible. Identical idea to what is used for multi-tone TD+N values.
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,732
Likes
6,101
Location
Berlin, Germany
As for multi-tone in general and more importantly with loop-able ones (frequencies on bin centers), those are tricky wrt established crest factor which massively affects the result.
At the smallest possible crest factor, approaching values well below 2, the sequence degenerates into a repetitive endless "sine sweep" (Newman phase distribution of the cosine generators' start phases, giving the lowest possible crest factor), whereas with the highest crest factor when all cosine oscillators have zero start offset, it tends to become a "single-event" impulse (with pre- and post-ringing from the missing frequencies). At factors in between anything can happen, for example I've set up two overlayed Newman sweeps which basically represents a twin-tone shifting in discrete frequency steps which of course tests IMD differently than a more random distribution.

While the 31/32 components true random multi-tone is a good test signal it can sometimes lead to strange data. I recently tested some phone dongles, the well-known Apple USB-C and another one from JSAUX, the latter being only 16bit. In all tests the JSAUX was worse but with MT31 the Apple had a higher remaining distortion/floor. With MT16 (the NID type from REW) it was much better to see what's going on and the results matched the other tests (1kHz THD spectra at -3dB, -30dB, -90dB):
Multitone 31.png


Multitone 16.png
 
  • Like
Reactions: Tks

HiZ

Member
Joined
Apr 26, 2021
Messages
37
Likes
8
Hi @mansr ,
do you think that this board would work to generate to MCLK , then sending the 4 signal to another board converting I2S (with MCLK this time) to SPDIF/AES ?

View attachment 160699
The point that I'm not sure how to handle it is the constant output of 211kHz :
  • Chipset: CS8421
  • Input Sampling Rate: Up to 192kHz (16 / 32Bit)
  • Constant Output Sampling Rate: 211kHz
  • Power Supply: 5V DC
I suppose that it's due to the 27MHz chip, so would it possible to replace it with a 24.576 one ?

EDIT : just found one with a 24.576 chip :
aliexpress.com/item/33035364226.html

View attachment 160700
or this one :

View attachment 160701


Do you think I can do the function I want with this board?
 

trl

Major Contributor
King of Mods
Joined
Feb 28, 2018
Messages
1,978
Likes
2,542
Location
Iasi, RO

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,670
Likes
10,301
Location
North-East
For those of you that have a Cosmos ADC (I don't... yet) could somebody run AlexCP's 7kHz+13kHz 1:1 twin-tone with REW (using separate DAC channels mixed together passively)? I found this one the most revealing twin-tone as it tests for loop gain around 20kHz, too bad it's not a standard, yet (see this post and following posts).

Here Cosmos ADC (fed by ADI-2 Pro FS as in my other posts) at 7 & 13kHz 1:1:

1634993045951.png
 
Last edited:

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
344
Likes
919
Location
UK
For multiple tones you need a measurement function that removes all stimulus tones from the FFT result. The remaning rest is Total Distortion plus Noise, TD+N (not THD+N). Not sure REW supports this at all. AP does.

REW provides TD+N for multitone and TD+N along with the various IMD figures for two and three tone signals, but only when using REW's signal generator - that way it knows what the test signal is.
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,670
Likes
10,301
Location
North-East
REW provides TD+N for multitone and TD+N along with the various IMD figures for two and three tone signals, but only when using REW's signal generator - that way it knows what the test signal is.

That makes sense, John! It gives me an idea. If I have the generator running with the right multi-tone setting outputting to some null device while feeding RTA data from an non-REW source, it should then measure the distortions correctly, yes? Something like this:
1634992954285.png
 
OP
mansr

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,703
Location
Hampshire


Do you think I can do the function I want with this board?
That board does USB to/from other interfaces. If what you want is an S/PDIF output from the Cosmos, that will not help.

The various ASRC based boards mentioned above will work if a slight degradation of the signal is tolerable.
 

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
344
Likes
919
Location
UK
If I have the generator running with the right multi-tone setting outputting to some null device while feeding RTA data from an non-REW source, it should then measure the distortions correctly, yes?
Yes, as long as the external signal has the same characteristics as the one selected in the REW generator. Easy enough for dual tone, but could be tricky for multitone, need identical bin choices.
 

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,670
Likes
10,301
Location
North-East
Yes, as long as the external signal has the same characteristics as the one selected in the REW generator. Easy enough for dual tone, but could be tricky for multitone, need identical bin choices.

The file I'm using to play back was generated by REW generator, so bins should coincide unless that's affected by the DAC/ADC loopback itself.
 

HiZ

Member
Joined
Apr 26, 2021
Messages
37
Likes
8
That board does USB to/from other interfaces. If what you want is an S/PDIF output from the Cosmos, that will not help.

The various ASRC based boards mentioned above will work if a slight degradation of the signal is tolerable.
It's so hard...
I want to print it as AES/EBU as the cosmos' performance, but I don't know how...
 
Top Bottom