REW J-test measurement on RME ADI-2

[Blumlein 88]

If I understood correctly those spikes below the yellow line are part of the test signal:

View attachment 246418

But what about those ones with question marks?

If it were a digital only transfer of the J-test those would not be there. Those appear to be an example of close in jitter. Everything done with zero jitter would follow the yellow line.

 

[Blumlein 88]

I think it's like you said above. I feed my DAC on its spdif or usb input with the external signal. And I use one of possible ADI-2 options to pass it directly to the usb output to be able to capture it on my PC.

Maybe I'm dense tonight. Okay, an example. I could use my spare Topping DAC which is fed the Jtest signal from my PC USB. I connect the analog output of my Topping DAC with the RME interface analog input and it goes back to the RME USB connection to be captured by the PC. Is this like what you are doing?

 

[onlyoneme]

Maybe I'm dense tonight. Okay, an example. I could use my spare Topping DAC which is fed the Jtest signal from my PC USB. I connect the analog output of my Topping DAC with the RME interface analog input and it goes back to the RME USB connection to be captured by the PC. Is this like what you are doing?

Not at all. I have my PC connected to the RME DAC over usb. I feed my DAC using PC over usb, or using another device over spdif. I pass incoming signal to the usb output on DAC using one of RME options. I capture that passed signal on my PC.

 

[msmucr]

Then there has to be some mistake in your setup of the DAC/interface or REW.
In your described scenario with internal loopback you'll get just sole clock source in the system, the J test should be perfect.

With regards to observed skirting, try to check whether you don't have enabled option "Treat 32-bit data as 24-bit" at soundcard options pane. ADI-2 Pro via its USB interface supports full 32-bit audio I/O (eg. the last 8 bits in the word aren't zeroed).

For the reference, here's my loopback via ADI-2 USB in/out and enabled internal loopback. Both 44100 and 96000 is perfect, as expected. I've switched to linear X scale in REW, so you can see, the dominant frequency is exactly in the middle of spectrum (the 1/4 of sampling rate).
If you'd use SPDIF or AES for I/O, then FFT should look similar just with higher noisefloor naturally, as it is 24bit.

Michal

 

[onlyoneme]

Then there has to be some mistake in your setup of the DAC/interface or REW.
In your described scenario with internal loopback you'll get just sole clock source in the system, the J test should be perfect.

With regards to observed skirting, try to check whether you don't have enabled option "Treat 32-bit data as 24-bit" at soundcard options pane. ADI-2 Pro via its USB interface supports full 32-bit audio I/O (eg. the last 8 bits in the word aren't zeroed).

For the reference, here's my loopback via ADI-2 USB in/out and enabled internal loopback. Both 44100 and 96000 is perfect, as expected. I've switched to linear X scale in REW, so you can see, the dominant frequency is exactly in the middle of spectrum (the 1/4 of sampling rate).
If you'd use SPDIF or AES for I/O, then FFT should look similar just with higher noisefloor naturally, as it is 24bit.

Michal

You are my hero. I had it enabled. After disabling it looks exactly as it should I guess:

 

[Rja4000]

Here is what I get with Virtins Multi Instrument.
I used SPDIF Out - In to create the loopback here, so the TosLink is part of the Jitter

Log X scale

But I do prefer the Linear X Scale

If you switch test signal to 16 bits, you see immediately the impact on level
(500Hz at -91dBFS vs -139dBFS - a 48dB hit, which corresponds to 8 bits)

 

[msmucr]

You are my hero. I had it enabled. After disabling it looks exactly as it should I guess:

Glad it helped . It looks good to me.

Michal

 

[Rja4000]

In theory, the relative level of harmonics should be like this

Log scale

Linear scale

And in absolute level for the 24 bits Jitter test

And the table of expected levels (dBFS, absolute level, and dBr, relative to 24kHz level)
NB: 24kHz is at -3.01 dBFS

Frequencies are for 96kHz sampling rate.
Divide them by 2 for 48kHz.