If I remove the 1.5µF cap (doesn't seem necessary) Put 22 ohms across the piezo and add a 15 ohm in series with the parallel leg of the 8 ohm L-pad and push the x-o frequency up to 4500 calculated on 16 ohms (2.2µF), The R1 resistor would be set for the half way position on the L-pad, giving partial cut and gain. (I know that it would not give the same dB cut as on the face-plate of the L-pad, but I wasn't planning to use it anyway.)
Do you reckon it would work ok?

My calcs seem to indicate that the resistance seen by the x-o capacitor would vary between 15.2 and 16.9 ohms. This is probably quite a bit more stable than a normal driver.

Also, as asked further down, have you ever used the Eminence Gamma 12 in a PA box, I had great trouble getting a sensible box size, ended going fully sealed and fully stuff about 95 litres, hope it works. If it doesn't, I'll swap it out with a beta 12 and port that section of the box.

Thanks for the assistance. Let you know how I end up.

=== 10kHz Filter ===

Series Capacitance = 1.0uF
Parallel Resistance = 16 ohms

freq resp
=============
20kHz -1 dB
10kHz -3 dB
5kHz -7 dB
2.5kHz -13dB

That will give you some subtle "sparkle" and is probably what you're looking for.

Alternately, I'll describe the exact filter you've asked about, which is a first order 20kHz network. If this is what you want, you'll need a capacitor that's exactly half as big as the one listed above. So you'll want a 16 ohm resistor across the tweeter, and a 0.47uF capacitor in series with the tweeter/resistor connection.

This will give the following response curve:

=== 20kHz Filter ===

Series Capacitance = 0.47uF
Parallel Resistance = 16 ohms

freq resp
=============
20kHz -3 dB
10kHz -7 dB
5kHz -13dB
2.5kHz -19dB

A KSN 1038 using this second filter might add some "air," but it will be crossing over pretty high.