Phase response and loopback issues

[fcana]

Dear All,
maybe i'm doing a giant config error or misunderstanding something obvious; sorry in case, don't ban me please !
I'm using REW for measuring a FIR filter implementation in the same machine on which I'm running also REW... using soundflower for interchanging signals between the two applications (REW and Max).
Now, if I don't use a loopback signal between the apps everything seems working fine. Bypassing the filter results in a flat frequency response and a constant 0deg phase; when the filter is ON, SPL and phase measurements are as expected in the filter design.
Different strange thing is obtained when I enable loopback signal. Nothing change when the filter is bypassed... but everything changes in phase measurement when the filter is on . Phase response (with loopback signal passed flat through Max) is now doing a full 360° turn every ca. 2kHz @ 96kHz !
The question: is loopback feature influencing phase response analysis in a right way?
I'm attaching three measures for explaining better the results... Please, ask if something above was unclear or wrongly explained!
Thank you very much,

I'm on REW V5.20 beta 22

 

[John Mulcahy]

There is a time delay in the measurement (which is noted in the measurement notes) which can also be seen in the Group Delay plot. About 50 samples.

 

[fcana]

thank you very much John.
so, if i understand well, this ripetitive turning is due just to the elaboration delay and it is inestimable without a loopback connection right?
[EDIT] yes, obviously. if i put a time delay offset in the measurements everything go back "right" as before. thank you very much for your time!
regards

 

[fcana]

PS. now i found the exciting "Estimate/Shift IR" feature in SPL & phase estimation controls... REW is definitively great!
in the estimation it is missing a little something (in the examples above it calculates 0.508 ms instead of the real 0.516 ms, find attached a pic)... but if i put the right value 0.516 on the timing offset of the measurement window before starting measuring everything goes as expected!

 

[fcana]

As a [feature proposal] would be great to have the possibility to add a "manual" delay in the "SPL and phase" measure control as the amplitude offset!