Macbook, Focusrite Scarlett 2i2 and cal

[KeithNancollis]

Hi,
I bought a Focusrite Scarlett 2i2 (2nd Gen) to use with REW and my Macbook Pro (El Capitan) for measuring my loudspeakers.
When doing a soundcard calibration the frequency response is somewhat curtailed (-3dB at 18kHz). Exactly the same result was obtained when measuring the response of my amplifier using the Scarlett. Although the cal is not required as I am using a UMIK-1, this will surely compromise HF when performing loudspeaker measurements.
Any idea where I am going wrong?

 

[John Mulcahy]

It needn't necessarily have any impact on your UMIK measurements, the roll-off is more likely to be on the input side than the output. The Line/Inst setting for the input you used may have an influence on the loopback measurement.

Edit: make sure Direct Monitor is off.

 

[KeithNancollis]

Conclusions so far :-

Direct Monitor does have to be off; you get an erroneous result if it is on (it was actually off when I did the cal).

Line and Inst on both channels give the same result.

Your thought that the problem be most likely on the input side led me to wonder if the long cable I used for the loopback was forming a low pass filter with the input circuit but changing the cable had no effect. This was confirmed by loading REW on to an old XP machine using the same long cable with an' E-MMU Tracker Pre' sound card (drivers not available for my Mac, hence the Scarlett) and this gave a flat response out to 24kHz.

I presented the problem to Focusrite support who concluded that the software wouldn't provide an accurate test of the interface, and when the interface was tried with more expensive software quite different results were obtained (no more details though).

 

[John Mulcahy]

Was the REW sample rate the same as the rate that was configured for the device in Audio Midi settings? If not the OS would resample, so you end up seeing the response of the resampling rather than the card.

 

[KeithNancollis]

I hadn't thought of that (hadn't actually uncovered Audio Midi Setup on the Mac).
Anyway, the OS automatically sets the sample rate to 192k when the Scarlett is plugged in, hence the discrepancy. Setting to 48k does indeed solve the problem, with a flat response (-0,2dB) to 20kHz.
Confidence restored, I'll try to implement my MiniDSP into the proceedings.
Many thanks for your help.