How do we get the breakup modes of a particular speaker, is there any particular software, testing method for it.
I would even go as low as 1/5 or less depending on the severity of the breakup. The H5 is oftentimes horrid due to the breakup (see W22EX as an example) and that cannot be tamed by notching the peak.Hopefully individual distortion levels rather than THD / a lumped total.
1/3 isn't a rule as such, although it can certainly serve as a useful ROT / guideline -depends how steep your XO slope is, where & the quality of the motor design.
I would even go as low as 1/5 or less depending on the severity of the breakup. The H5 is oftentimes horrid due to the breakup (see W22EX as an example) and that cannot be tamed by notching the peak.
If the low-pass is around 700 Hz or higher, you will start to see amplification of HD5. No clue whether the driver has significant HD6/HD7/..., but if it did, you might want to cross even lower. Such a shame; the W22EX could still be a world-class driver if all they did was stamp in a few stiffening ribs a la SB65WBAC25 to push up the breakup a bit more so that you could wholeheartedly use the driver for what its size would typically prescribe. Or add a bit of damping and live with a slight loss in sensitivity... a very advantageous compromise IMO.Ouch, 18dB rise @ 1KHz from the primary bell-mode. That'd melt your ears.
I haven't worked with the W22 but from the HFC measures it starts to get in trouble at about 2KHz (you can see the glitch in the impedance indicating a resonance issue / energy storage), which from my perspective would realistically rule out an XO higher than, say, 1.4KHz with a minimum of LR4 slopes for that reason alone. Still, assuming for sake of example that wasn't there & ignoring potential issues with the polars for a moment, you could probably get away with, say, 1.8KHz or (with fingers crossed) about 2KHz if you stamp on the driver hard enough -LR6 or LR8 acoustic, probably with notched / Chebyshev type II filters to avoid excessive component numbers / losses. That should shunt it sufficiently low to avoid excess HD amplification.