Calibration process and noise smoothing

[trobbins]

I am using an EMU0404 soundcard with 192kHz ASIO and have a 100x scope probe as an input. The probe is compensated for flat response, but has a bandwidth of about 40kHz using a loopback measurement. Normally that is not a concern. I typically use the setup for amplifier gain and phase spectrum measurements and use a few sweep repetitions and a long length to acquire pretty smooth repeatable plots. I sometimes need to use the full 2Hz to 96kHz sweep capability, so I just went through the calibration process to normalise the bandwidth inadequacy of the probe and soundcard loopback setup and can see that the cal sweep is quite short and with only one rep, and the resulting spectrum has substantial jitter (albeit within 0.1dB), and that jitter appears to be then implicit in the cal file, and so any subsequent measurement sweeps show up that jitter and it doesn't look like I can filter the jitter out.

Any thoughts on if I'm missing something in the cal process? I guess I could modify the cal file to remove all the entries within the region where the response is flat so REW interpolates that region as flat, and maybe use Excel or something to smooth the response of the sloping extremities.

 

[John Mulcahy]

I'll change the code to use longer sweeps at higher sample rates. In the meantime you can make a loopback measurement with a longer sweep, add an offset to the data so it goes through zero at 1 kHz, export that measurement as text then load the text file as a soundcard calibration file.

 

[trobbins]

Thanks John - that method works a treat - smooth as silk.