REW errors when measuring DUT noise, distortion, etc

[rpschultz13]

I'm using a Cosmos ADC DM to measure the noise, distortion, THD+N, etc of a guitar pedal (DUT), using an Siglent AWG as input to the DUT, using a Mac. Latest beta of REW.

I frequently have to restart REW. It will measure for a few seconds then give me 1 of 2 errors:
the soundcard did not provide any input data from Record Interface (this is DM), please check that it is connected.
OR
unable to access the selected device du to: line with format PCM_SIGNED 192000 hz, 24 bit, stereo, 6 bytes/frame, little-endian not supported. Try a different sample rate.I'm on a Mac. Using a Signlent AWG. my DUT is a guitar pedal. DM shows a green signal.

I have both Audio MIDI setup and REW set at 192k. REW calls the ADC DM "Record Interface".
Cable from AWG to DUT is a BND-TS
Cable from DUT to ADC DM is TRS-XLR with ring (xls pin3) floating, since DUT has a TS jack.
AWG is outputting 1khz and anywhere between 1-4 Vpp (or more if needed).
DUT is set to roughly unity gain.

Any idea on why REW gives these errors and makes me restart it so frequently to work?

 

[John Mulcahy]

I suspect it's due to dropouts between coreaudio and the Java runtime. Is it reliable at 96 kHz?

 

[rpschultz13]

No different at 96k. In the distortion box, I'm assuming I have to manually set frequency when using an external AWG. Although it says Manual Fundamental 999 V? Shouldn't that be Hz?
Feel free to critique my settings, I'm very new to REW.

 

[John Mulcahy]

Seeing the same error messages at 96k suggests a problem with the Mac's connection to the Cosmos ADC.

Manual fundamental is for use with a notch filter to tell REW the level of the pre-notch signal. There is help. The fundamental frequency is found automatically from the largest peak in the input, if the fundamental has been notched out you can use "Get fundamental from generator" and set the generator to the required frequency. You should not use a rectangular window when the replay and record do not share the same clock source and the stimulus is not at an FFT bin frequency.

There are threads on diyaudio that include REW and the Cosmos ADC and may be relevant, e.g. this one and this very long thread.