Measuring 10dB SPL difference between ASIO and Java drivers

[Croy9000]

I notice the SPL meter (UMIK-1) when measuring a quite room is about 10dB different when using Java vs ASIO drivers (so not the LFE boost thing). For instance Java will measure 34dB, while ASIO with measure 43dB. I have tried ASIO4ALL as well as all the FlexASIO backends. The Java driver measurement matches my handheld meter.

This also shows up in sweeps. Here you can see the expected +10dB boost to the LFR (blue), but both Left and LFE are getting another incorrect +10dB added.

I have tried rebooting as well as dropping back to v5.20.7. Am I missing something simple?

 

[John Mulcahy]

Could you attach an mdat file?

 

[Croy9000]

Attached. Just to clarify, my signal chain is REW (HDMI) > AVR (Sub1 Out) > MiniDSP HD > Amp > Sub. Left main speaker is disconnected. All 3 of these sweep measurements are measuring the same subwoofer.

 

[John Mulcahy]

Try setting the mic input volume to 54% (it was 80% for the Java measurement). REW is compensating for 10.3 dB of gain that the OS applies at an 80% volume setting, but there's some confusion around why Windows offers a volume control for the UMIK-1 with USB-C connector since the mic doesn't have an internal volume control. On my Win10 build (21H2) the gain is applied and the correction is needed, it appears that on yours it isn't. Using ASIO bypasses the Windows volume control for those newer UMIKs.

 

[Croy9000]

Ok yep that fixed it. Both drivers report the same now. Also the measurements now match my handheld SPL much closer too.

FYI I am using Win 10 (21H1).