Merge function without compensation?

[bcodemz]

I'm trying to use the merge function in REW to splice some nearfield and farfield measurements together. I'm getting some strange results at times with completely messed up frequency response for one part of the graph. I'm pretty sure it is due to the phase alignment that REW tries to do.

Is it possible to do a merge that is a pure SPL merge?

 

[John Mulcahy]

Can you attach some example measurements showing that?

 

[bcodemz]

Sure, here it is. See the 60 and 70 degrees merge just fine, but the 80 degree measurement doesn't.

Background, one of them is an outdoor ground plane measurement of a driver. The other one is a quasi anechoic measurement. However, the outdoor measurement has filters and a HPF applied, which caused phase shifts. The quasi anechoic measurement did not have any filters applied to the driver.

 

[bcodemz]

Looks like the attachment didn't work, probably due to size? I removed the 60 degree measurements to bring it under 10MB

 

[John Mulcahy]

The far field measurements have a very big reflection at 15 ms. In the 80 degree case it is larger than the direct sound, so t=0 has ended up at the reflection meaning the initial part has a big negative time delay. If the delay is removed (by Estimate IR Delay, for example) they merge without problems. It would probably be best to either use a timing reference or remove delays before merging.