Here's the custom filter executed with polypropylene caps paired to within 0.5%. "The FET-input op amps are low-noise OPA1562 smd types which are best for actives filters. Distortion is around 0.001 – 0.002%, about the same as my generator's own distortion. The frequency response is exactly as it should be, -6dB/40Hz at 24dB/octave. This filter should have no self sound, just the effects of steep attenuation."

Here is the filter curve for the sub…

… here that for the satellites.

Here is the 2nd-order crossover with Jantzen Amber copper caps and silver bypass of the 3-way sound|kaos speakers themselves …

… followed by their pair-matched impedance plot (port tuned to 35Hz)…

… and their amplitude response at 1.8m, in line with the only low- not high-passed AlNiCo widebander (disregard plot below 100Hz and above 10kHz due to sidefiring woofers and upfiring Raal ribbon).

How would the combo of sound|kaos + Zu bed in with my new active filter? Which of the three options would win out and why? As described here for Acelec's Model One displaced by the Vox, Pál Nagy's fixed active filter was wildly superior to the continuously variable job of Zu's Hypex plate amp. On that two-way speaker, bass and midrange share the same cone. The most surprising gains which I'd not given proper prior consideration occurred in their dynamic range far above the filter transition.

That's because liberated from true low bass, the ScanSpeak mid/woofer's voice coils no longer heated up the same. They no longer tried in vain to handle 1st-octave content beyond their reach. Without said stress, the inverse law of related voice-coil Ω rise no longer put down its foot on that dynamic brake. By the same token, the influence of the ports on the speaker's low-frequency impedance chart had been cut. The amplifier driving them no longer saw the related phase angles and impedance fluctuations. That too equaled less stress thus better performance.

The most practical upshot was that the icOn 4Pro SE had transformed a 2-way into a "half 3-way". Executing a superior 40Hz high-pass in the passive domain would mean very large very reactive filter parts. It's why it's virtually never done. Shifting the job into the active domain thus ahead of the amplifier nixed the usual limitations. It really was a win all around.

As a 3-way of admittedly unusual design, for the Vox it now were dedicated low-passed woofers which fell into the active filter's cut. Thus benefits to greater midrange dynamics applied less. What still did was perfectly seamless continuity of full bandwidth down to a solid 25Hz and a happier small main amp.