Do they just sound better? What is their actual sonic/performance merit?
I read some discussion of caps in hifi setups about more pleasing audio reproduction, but haven’t found a clear explanation about their use in resonant filters vs. other types of caps.
Tolerance! 2.5% or better is hard to beat. Also, they have excellent temperature stability. The tonality thing is more religion.
And they will keep their capacitance over time better than most capacitors, and the ESR is more stable at varying temperatures.
That’s why I installed mine 
Martin
when you start upgrading your bypass caps to polystyrene and noticing that they induce a warmth to the audio quality of your Shruthi your pants have become too fancy.
There they are, the virgin fairies pubic hairs at full moon to polish your Caps on PCBs from alien copper carved from meteorites - long time no see…
hey, coat hangers made from alien copper have been proven to be more sonically accurate than Monster Cables.
But only for infrasonic coat vibration…
In all seriousness…
-Low drift over time
-low, consistent temperature coefficient of 120ppm (better than other plastic dielectrics-very good for designing stable VCOs)
-lowest distortion due to capacitance being affected by voltage (really only important to ultra finicky audiophiles-one test I read cited distortion at -160dB versus -120db for C0G ceramic)
-very low leakage and “memory” effects (excellent for sample and hold circuits)
Disadvantages
-can’t handle high temperatures very well, so not at all easy to surfacemount, and need to be handsoldered with care
-bulky for the capacitance
More here
Awesome! Thanks for the info. I have some temp drift in my SMR4 mkii that I suspected could be ameliorated with polystyrene caps. Since PPS is manufactured exclusively by Toray (chem giant) here in Japan according to that link, it may be easier to get those and try 'em out too. I’ll check in Osaka when I’m there this weekend.
If you mean drift in the filter cutoff/tuning/volts-per-octave response, then that is most likely due to the 3300ppm temperature drift of the expo converter transistor-if you are already using C0G ceramics for the filter caps then they have effective zero temperature drift.
What you could try-the more of these the better but I’ve put the (IMHO) most effective ideas first:
-replace R24 by a 2K +3300ppm tempco from Magic Smoke or equivalent (if the tempco you use is 3000ppm or 3500ppm don’t worry, you’ll still be reducing the problem by a factor of 10…)
-match Q2 and Q3, using either the Møög or Fritz methods
-put Q2 and Q3 in thermal contact, at least get them touching, but ideally use some heatsink compound or epoxy glue
-put your new tempco R24 in thermal contact with Q2 and Q3 as well by keeping the leads long and bending them in order to position the resistor body to touch the thermal goop
Coverup’s problem is the temperature coefficient of the OTA itself, which causes a drop of gain preventing the filter to go into self-oscillation.
Changing the caps will not solve this, or at least not in the expected way. It might bring an improvement by “matching” the poles together, reducing the gain required to reach self-oscillation, but the very same thing could be obtained by lowering the resistor that sets the OTA biasing current.
You caught me! 
Thank you for the correction. And gwaidan, that’s good info, cheers.