E1DA Cosmos ADC

[IVX]

pkane, I have no ADI2 and can't confirm that. APx555b IMDvs level plot looks like that

 

[pkane]

pkane, I have no ADI2 and can't confirm that. APx555b IMDvs level plot looks like that

Thanks, Ivan. I'm guessing there's some interaction between RME and the ADC then, since RME by itself (in a loopback) doesn't generate the same hump in an IMD sweep.

 

[IVX]

how looks ADI loopback IMD then?

 

[staticV3]

how looks ADI loopback IMD then?

If I understand pkane's findings correctly:
RME loopback: no hump
APx555->Cosmos ADC: no hump
RME->Cosmos ADC: IMD hump present

 

[pkane]

how looks ADI loopback IMD then?

This is from a while back:

 

[IVX]

pkane, I just downloaded the Multitone to test IMD but can't see an adequate result yet, only -.25db when AP2700 shows -108db. THD+N AP2700 also shows quite different -6db from your result. Also, I think a bad idea to send 0dbfs volume command every time Multitone started. Cosmos ADC always switches on the MONO after that. BTW, AP2700 input level is -8.36dbfs but Multitone shows -6.7db.

 

[pkane]

pkane, I just downloaded the Multitone to test IMD but can't see an adequate result yet, only -.25db when AP2700 shows -108db. THD+N AP2700 also shows quite different -3.4db from your result. Also, I think a bad idea to send 0dbfs volume command every time Multitone started. Cosmos ADC always switches on the MONO after that. BTW, AP2700 input level is -8.36dbfs but Multitone shows -6.7db.View attachment 174640

Just change gain from 0dB to something smaller. You can type in the number if it’s not in the drop down list.

 

[IVX]

THD+N 6db mismatch at 1kHz -116db. BTW, REW seems adjusted only for 1k THD+N and >1k or <1k THD+N are also wrong, at 100Hz AP2700 6-7db better! I gonna test Arta at 100Hz and 6kHz.
PS: no, regarding REW it was my mistake because FFT was 64k and spectral leakage made 100Hz THD+N so bad. At 6kHz REW shows 1db less THD+N vs AP2700, similar to Arta.

 

[IVX]

Yes, Arta is about Ok. Not 100.00% matched but 2ppm vs .18ppm is Ok to me. But as I remember, Arta's IMD is completely mismatched to Ap..

 

[pkane]

THD+N 6db mismatch at 1kHz -116db. BTW, REW seems adjusted only for 1k THD+N and >1k or <1k THD+N are also wrong, at 100Hz AP2700 6-7db better! I gonna test Arta at 100Hz and 6kHz.
PS: no, regarding REW it was my mistake because FFT was 64k and spectral leakage made 100Hz THD+N so bad. At 6kHz REW shows 1db less THD+N vs AP2700, similar to Arta.

Hmm, not sure why you're getting a 6dB difference.

Here's the same signal being processed by REW and by Multitone. The differences appear to be within a fraction of a dB:

Now at 100Hz:

 

[IVX]

I don't know, maybe because I moved down the Windows Volume slider during "warm up"? I thought your "Gain" is some correction coefficient but not Windows Volume slider yet.

 

[IVX]

I have changed the "Gain" to -60db and next -120db, the Windows Volume slider anyway jumping to the 0dbfs every Multitone start.

 

[pkane]

I have changed the "Gain" to -60db and next -120db, the Windows Volume slider anyway jumping to the 0dbfs every Multitone start.

I see what you’re saying. The gain in Multitone is set by changing the amplitude of the generated signal, not by setting the system volume. Let me see if there’s something I can do to keep it from changing system volume to 0dBFS.

 

[Hee]

dac : spartan cue 48khz/s, trs 4v out
adc : cosmos adc 192khz/s, xlr 4.5v in

CUE performance was not the best, but there was no hump.

Oh, sorry.
Please hold for a moment. I found something a little different.

 

[Hee]

Spartan cue is audiointerface.
cue out 48khz/s -> cue in 48khz/s.
cue out 48khz/s -> cosmos in 48khz/s
Even if I try again a few times, there are some hump. All settings are the same.

 

[pkane]

I have changed the "Gain" to -60db and next -120db, the Windows Volume slider anyway jumping to the 0dbfs every Multitone start.

Hi Ivan,

Can you please test to see if this is fixed in Multitone v1.0.13? I found one place where the system volume was being set to 100% and removed it. If you start Multitone, it'll prompt you to update to the latest version or you can download the update from here: https://distortaudio.org/multitone.html#download-multitone

 

[dominikz]

I don't see why you want to go back, the firmware seems stable and improved. It would be interesting though to hear which DC values are measured now 'in the field'. I gave my numbers already.

A bit late to the party, but here's mine:

Around -56dB before FW upgrade and -95dB after.

Is there an AP IMD sweep for Cosmos ADC available? SMPTE 60Hz/7kHz 4:1?

I've done one using RME ADI-2 Pro FS as the source using my Multitone app and it shows a bit of the infamous IMD hump. Would like to confirm if this is the case with a better source:

View attachment 174631

This is with Topping E50 as source (balanced, left channel only, E1DA Cosmos ADC stereo mode and 4,5V sensitivity):

 

[IVX]

Hi Ivan,

Can you please test to see if this is fixed in Multitone v1.0.13? I found one place where the system volume was being set to 100% and removed it. If you start Multitone, it'll prompt you to update to the latest version or you can download the update from here: https://distortaudio.org/multitone.html#download-multitone

 

[pkane]

View attachment 175038

I see. Can you please try measuring using the Multitone-generated sine wave, instead? It's not at exactly the same frequency as the signal generated by the AP. This is in order to put it in the middle of an FFT bin. When calculating TD+N, the notch filter assumes that the frequency is at the same bin that was generated by Multitone, but in your test, it's not.

EDIT: Thinking about it, I'll add another test signal option, "External" to allow for external signal generators. This will help when using AP or REW or other external generators for testing.

 

[manisandher]

Is there an AP IMD sweep for Cosmos ADC available? SMPTE 60Hz/7kHz 4:1?

I've done one using RME ADI-2 Pro FS as the source using my Multitone app and it shows a bit of the infamous IMD hump. Would like to confirm if this is the case with a better source:

View attachment 174631

I thought the infamous IMD hump was only exhibited by ESS DAC chips. Assuming your RME is an older unit like mine, it uses the AK4493 chip.