This post will shed some more light on what led to all the delays regarding the crossover tuning of the Directiva.
I had already mentioned (
here and
here) the small uncertainty regarding the phase alignment of woofer and tweeter to each other. It could be that the reference points, when measuring the drivers, could have had minimally different distances from the baffle.
Now that Rick has updated his measurement equipment, we can be pretty sure that his measurements should be fairly accurate.
When comparing the measured frequency responses of tweeter and woofer by Amir's NFS and Rick's gated measurements, there are sometimes significant deviations.
Comparison of Rick's gated measurement versus Amir's NFS measurement of the
SeasDXT tweeter in the Directiva cabinet without XO, after scaling the sound pressure level:
That looks pretty good on the whole - considering that different tweeter (of the same series) and measurement methods were used. There are some deviations, among others, in the effective range of a possible crossover, 1-3kHz, and around 14kHz.
The latter could be due to the "decreasing fine resolution" of the NFS measurement ("fitting error") at frequencies >10kHz. In any case, my own measurements of the SeasDXT in different speakers show a comparable dip:
Therefore, we assume that Rick's measurements are more accurate in this area.
Comparison of Rick's gated measurement versus Amir's NFS measurement of the
Purifi woofer in the Directiva cabinet without XO, after scaling the sound pressure level:
The deviations in the important frequency range 1-4kHz took us*** days and days until we were absolutely sure that the deviations were not due to the measurement equipment.
Either the two Purifi woofers show significant frequency response differences in the range 1-4kHz, or there was a disturbance in Amir's measurement in this range.
It is not about small things, but about values that can change the character of the speaker. Here are the differences of the two measurements, when normalized to Rick's measurement:
Currently Rick tunes his speaker in the 1-4kHz range according to his measurements and ignores the NFS for this range.
Reliable measurements are everything on a project that is spread across continents. And only with reliable measurements can VCAD play to its strengths. Otherwise, there are always irritations when auditory impressions do not match the crossover simulations, etc.
*** to be honest, Rick had the work and I thought I knew everything better