Calibration of measurements

[twelti76]

I have been having some difficulties getting consistent calibrated measurements. For example I have a set of two different measurements done on two different days that look exactly the same as far as cal settings, but show up around 20 dB different.
Same "dBFS @ 94 dB SPL" setting, same mixer settings, inputs, outputs and DUT same.

My procedure was to calibrate using the SPL meter (external source) and an acoustical calibrator, all set to 94 dB SPL. Is this a sound way to calibrate, and is there some other way I might have gotten this offset within REW?
See below. Measurements are not exactly the same in every way but...
I should add that the 154 dB data offset and the corresponding -57 dB sensitivity arising from my calibration is very very close to what I calculate knowing the sensitivity of the MEMS mic being measured.

 

[John Mulcahy]

Is this a sound way to calibrate

Yes. It suggests the replay path gain is changing by 20 dB, since any change in the input path gain would be picked up by the cal.

 

[twelti76]

Makes sense. I guess I have to figure out how that could have happened! The current (today) measurement is the 20 dB lower one. I can confirm that the output gain on the mixer is at 0 dB now, which it should have been for the previous older measurement. Thing is, the RME output max is +6 dB so there is no way to have gotten the extra 20 dB as output gain in the older measurements. Seems a mystery. Also, these measurements were made in my own ears so I would definitely have noticed a 20 dB difference.

 

[John Mulcahy]

Pretty strange, but the IR peak is 20 dB higher in the Jun 11 measurement so the incoming data had higher level. The end frequency of the measurements is different, but that doesn't affect the level.

 

[twelti76]

Makes sense. I guess I have to figure out how that could have happened! The current (today) measurement is the 20 dB lower one. I can confirm that the output gain on the mixer is at 0 dB now, which it should have been for the previous older measurement. Thing is, the RME output max is +6 dB so there is no way to have gotten the extra 20 dB as output gain in the older measurements. Seems a mystery. Also, these measurements were made in my own ears so I would definitely have noticed a 20 dB difference.

Also, I listened to the current one and it is fairly loud, almost distorting. A 20 dB louder stimulus would grossly distort and blow my ears out.