E1DA Cosmos ADC

[Rja4000]

They are there for the 1KHz .

Yes, but not for IMD

 

[pkane]

Yes, but not for IMD

I've not found the REW display of distortions for anything but single tone to be all that useful.

 

[mansr]

There's another solution I thought about as I can sync from SPDIF/AES/ADAT or... Word.
The LRCK pin is supposed to output the Word clock (it's the same), but how to get that out ? just linking this LRCK pin and Ground pin to a BNC connector, with maybe something to amplify and regulate the signal ?

Yes, a simple line driver connected to the LRCK pin would do it. You'd need to check the specs of the receiving equipment for the expected signal level.

What I don't understand is why the above board would absolutely need the MCLK signal.

The WM8805 chip on that board needs this signal in order to operate. As can be seen in the photos, it is wired to the input connector (via a buffer). The MCLK signal must be supplied externally, or the chip won't do anything at all. It's not part of the I2S spec, but it is still needed by this device. The description says as much too, though it's strangely worded (poor translation).

 

[Grooved]

Yes, a simple line driver connected to the LRCK pin would do it. You'd need to check the specs of the receiving equipment for the expected signal level.


The WM8805 chip on that board needs this signal in order to operate. As can be seen in the photos, it is wired to the input connector (via a buffer). The MCLK signal must be supplied externally, or the chip won't do anything at all. It's not part of the I2S spec, but it is still needed by this device. The description says as much too, though it's strangely worded (poor translation).

Thanks @mansr !
That's right that I was not sure of the text, but after checking the board, it matches what you said

So the easiest way to get a sync seems to build a BNC Word clock output, and still use the Cosmos ADC as recorder via USB. It leads to use two devices with the computer softwares, less easy than what @HiZ want to do if I'm not wrong : sending the digital audio signal to a digital input of only one interface used by the computer (getting Cosmos A-D conversion, but routing it with the same interface doing the D-A)
He may confirm if I'm right

At this moment, after matching the levels, I tried to do a loopback test and check it with @pkane DeltaWave, and the result was not so great, but it's certainly because of the missing sync. Will check further later

 

[nagster]

Grade 0..

139.0US $ |E1DA Cosmos ADC Hi Precision Analog To Digital Converter For Audio Testing Bundle 1=Grade C Bundle 2=Grade B Bundle 3 =Grade A|Digital-to-Analog Converter| - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

 

[HiZ]

Grade 0..

139.0US $ |E1DA Cosmos ADC Hi Precision Analog To Digital Converter For Audio Testing Bundle 1=Grade C Bundle 2=Grade B Bundle 3 =Grade A|Digital-to-Analog Converter| - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

That's amazing....

 

[Grooved]

You'll need to connect those signals to an AES/EBU transmitter such as Texas Instruments DIT4192 or Cirrus CS8406. Both of these require a master clock at 128 or 256 times Fs. This can be generated from the word clock using a clock multiplier, e.g. the Cirrus CS2300-03. You'll also need power supply, output transformer, and some passive components. Maybe there's a ready-made board available somewhere.

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 ?

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

or this one :

 

[mansr]

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

Those boards resample the input to a fixed output rate. That's not what you want.

 

[Grooved]

Those boards resample the input to a fixed output rate. That's not what you want.

Right, just realized that some minutes ago, even if it should work if using only 192kHz on the Cosmos ADC, shouldn't it ?

 

[mansr]

Right, just realized that some minutes ago, even if it should work if using only 192kHz on the Cosmos ADC, shouldn't it ?

It would work, but the quality of that ASRC chip probably isn't great.

 

[Grooved]

It would work, but the quality of that ASRC chip probably isn't great.

Certainly, I thought at first there was a way to run it as slave and bypassing the conversion. I only found in the CS8421 data sheet that it can run as slave, and even bypassing the conversion but I didn't find if it still output the MCLK in this case.
Anyway, the CS4821 may be able to do that but this board is not designed like that, it would require to build one from the scratch.

 

[Weeb Labs]

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

Either the Twisted Pear Metronome or my own Constellation board (much cheaper but unavailable to order until next week) would work well. The reference clock output on my board (and possibly the Metronome) is active regardless of IO mode, unless the RCLK switch is in the off position.

 

[mansr]

Either the Twisted Pear Metronome or my own Constellation board (much cheaper but unavailable to order until next week) would work well. The reference clock output on my board (and possibly the Metronome) is active unless the RCLK switch is in the off position.

Those also use an ASRC.

 

[Weeb Labs]

Those also use an ASRC.

They do. The SRC4192 exhibits a SINAD of 140dB, which is perfectly adequate for any purpose.

 

[mansr]

They do. The SRC4192 exhibits a SINAD of 140dB, which is perfectly adequate for any purpose.

There's also the jitter aspect which that figure doesn't take into account.

 

[Weeb Labs]

There's also the jitter aspect which that figure doesn't take into account.

The sum of all uncorrelated jitter is represented by the dynamic range and by extension SINAD/THD+N. Correlated jitter is a somewhat more complex matter but that's going to be largely dependent upon the clock source and not the ASRC.

 

[mansr]

The sum of all uncorrelated jitter is represented by the dynamic range and by extension SINAD/THD+N. Correlated jitter is a somewhat more complex matter but that's going to be largely dependent upon the clock source and not the ASRC.

Skirting caused by random jitter is typically included in the fundamental when calculating THD+N while distinct spurs count as noise.

 

[pkane]

Yes, but not for IMD

Here you go, but I don't think that REW is measuring distortions properly for multiple tones:

 

[MC_RME]

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.

Technical Library

www.ap.com

 

[pkane]

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.

I have an app that does that, also (a bit cheaper than an AP ). Some examples I posted a while back:

https://www.audiosciencereview.com/...ds/multi-tone-audio-testing.12865/post-382712

At some point, I'll use my own software, but REW is just so much more convenient