Wrapping my brain around soundcard/input level calibration.. increasing noise floor?


Thread Starter
Apr 26, 2021
I recently started up with REW and interested in taking some response and noise measurements of some audio equipment. The equipment appears in Windows as a USB soundcard and provides a line-out with a 0 dBFS of 1 Vrms. My goal is to be able to characterize its output with response, distortion, and noise floor.

I am currently using a Steinburg UR22 mkII as the acquisition device. Going through the soundcard calibration process with the loopback method, I've been able to match the -12 dBFS output with a -12 dBFS input using the input gain dials. And in the generator calibration, I've measured it to have a ~2.5Vrms

For measuring the audio equipment, I would be using the device's line-out in place of the Steinburg's line-out and then connect to the Steinburg's line in. The output device in REW would be changed to the audio equipment. After doing this, checking the levels appears to show that a -12 dBFS output produces a -33 dBFS input. (I think) this is due to the 1 Vrms of the device and the 2.5 Vrms of the Steinburg. What I am unclear about is a) if I should increase the input gain dial on the Steinburg to match the -12 dBFS to obtain a good measurement, b) and if applying input gain would also be increasing the true noise and distortion to skew the measurement of the device?

I have been wracking my brain :blink: around this question for the last few days, if applying gain is required and if it changes the acquired measurement. Any thoughts/suggestions would be much appreciated!

John Mulcahy

REW Author
Apr 3, 2017
Yes, apply some more input gain to increase the levels. The improvement in noise floor should be more than any increase in distortion until you get close to the full scale input of the soundcard. You could test that by doing a stepped level measurement with a loopback connection.
Top Bottom