Originally Posted by noah katz
"create two drivers, one with 30% less BL. "
The 30% is only at the endpoints of the excursion range being considered.
But even if it were 30% less in that whole range, and SPL is proportional to BL (not sure if that's right), it would be only 1.5 dB compression.
It isn't necessarily proportional. I don't have my laptop with me (that has LSPCad) or I'd simulate something quick.
Here is the Kepler, my 2-way monitor speaker that uses a midwoofer with XBL^2 and about 10mm of x-max.
This is a deviation from linearity measurement done in the chamber at the National Research Council of Canada (NRC). The first graph is at 96dB 1M, the second at 101dB @ 1M. Compared to subs, this is a tiny voice coil @ 1.5" in diameter and the former is about the same in length.
I don't know why I'm showing this other than to say that the compression in real-life at normal listening levels, is not that severe. The 96dB @ 1M is louder than you would see in a home environment under normal use. It is a good metric for transients and you can see it shows less than a dB of compression in the midwoofer. The 101dB @ 1M is extremely loud, much louder than even the hard-core is going to listen at even to show off the system. Even then... I'd say about 1.5dB of compression in a small midwoofer isn't bad.
How much of this is power vs. BL compression? I don't know... the test doesn't deviate between the two.