Data offset when doing SPL measurement

[Emile Blanc]

Dear REW community,

I have been doing some SPL measurements using REW and a UMIK-1 for two measurements with the exact same setup I have been getting different results due to a different "Data offset" in the Measurement info panel...
Does anybody knows how this offset is calculated and what it means?
My measurements have a 20dB difference and I don't know which one is right. The UMIK has the same calibration file and everything else is the same as you can see in the attached pictures.

Thanks in advance
Emile

 

[John Mulcahy]

That's odd. The offset is calculated from the mic sensitivity cal figure and the volume setting. If you look in the log files (their location is in the About REW dialog) there should be entries showing the mic sensitivity cal figure for each startup, worth checking what those show. The volume setting is showing as zero, that applies 31 dB of attenuation on a UMIK. Better using a volume setting closer to the 0 dB gain point - for macOS that is about 0.318, for Win7 about 0.1.

Also odd is the Win 7 OS though you appear to be on a mac, are you using Parallels or similar to run a Windows version? Better running the macOS version, Java on macOS has better audio support than on Windows.

 

[Emile Blanc]

Dear John,
Thanks for your answer,
In the log files the mic sensitivity is always 0,75dB which is the value from the calibration file of the UMIK...
I set the gain to 0 because I am doing high SPL measurement and I need more headroom. If I set the gain to any value above 0, I have clipping... I already noticed that when the gain is set to 0, the measurement is 15dB lower than if it is set to any other level. But this difference is always the same so it's not a problem.

I am using a windows computer to perform the measurements and I just transferred them to a Mac for convenience but the difference was already present under windows.

 

[Emile Blanc]

I just found my mistake...
In one of the measurements, I added the calibration curve of the microphone after the measurement so it didn't recalculate the data offset value. It just applies the curve.

 

[John Mulcahy]

OK, good to have an explanation. It is a little tricky to recalculate the offset if the cal file is changed since it depends on the OS type and volume setting. OS type is saved with the measurement, but the volume setting is part of a text string describing the input rather than being saved separately. It is something I can look at though.

 

[Emile Blanc]

Yeah it might be complicated. I will just take a special care in loading the calibration curve.
One thing I don't understand is the 15dB offset when setting the UMIK gain to zero compared to any other setting and also the headroom difference. I have clipping when the gain is set to 0.1 and more than 30dB headroom when it is set to 0...

 

[John Mulcahy]

The Win7 gain mapping for the UMIK is not great, it has about 4 dB of gain at a volume setting of 0.1 and 31 dB of attenuation at 0. Better with a more recent version of Windows.