Enjoying this thread, thx.
A friend on DIY forums has done a lot of cardioid modeling but hasn't committed to a build yet...nice to see a build in play.
Hey, with regards to measurements...i'm not an expert, but i sure do make a lot of them building DIY speakers.
I use REW, ARTA, and SMAART (the most)....taking lots of multi-way polars on a spinorama outdoors, (and indoors in winter time).
REW and ARTA should give the same results, when excess delay, TOF (time of flight), is removed the same. If not, something is amiss.
Since you guys are working with John M, i'm sure all will sync up.
Afaict, all the FFT/DFT measurement programs find impulse peak the same way, the peak of the ETC energy-time curve (log view).
But that provides a confusing catch with the usual impulse response, the linear view (that shows polarity.)
Catch is that linear view gives one vote to each frequency towards finding impulse peak . Which overweight's high-freq response.
This is easily seen by the heights of impulse responses. 10k-20kHz is one octave and gets 10,000 votes. 100-200Hz is one octave and gets 100 votes.
In the case of the Purifi 6.5, looking at its raw response curve down to say -10 dB, i'd judgement call that bandwidth to be 60Hz to about 6 kHz.
Each single frequency in that bandwidth gets a strong vote towards where in time the linear impulse peak falls.
Which means roughly >3000 Hz equals influence of below 3000Hz.
So when the driver is rotated, and the relatively beaming upper end response of its response rolls off, the impulse peak looses its previous time bearing.
And as folks have noted, room reflections want to join the voting then.
The two workarounds i use for measuring drivers used below HF/VHF:
are measure outdoors, so reflections don't get a vote. (belaboring the obvious)
and put a temporary linear phase xover in place somewhere above the final intended xover frequency range.
The temporary xover works indoors too...it forces the measurement software to look for peak ETC energy in the range the driver will be used at....without influence from bandwidth that won't be used.
I use it outdoors as well, as it simply quicken and narrows the time alignment process.
I've come to realize time and phase alignment of a speaker,
really mean time and phase alignments of the measurements I took of the speaker.
The closer i can get the raw measurement to reflect the actual passband that will eventually be in use, the less the fixed delay manipulation that goes into the xover.
Key i think, is getting a raw measurement of each driver section that has overall phase trace as flat as possible
for the intended passband.
Hope all this is welcome...don't mean to be teaching/preaching/or anything remotely close.
Just want to maybe share some things that i've found helpful making measurements.
And nothing guaranteed. ymmv.