Short answer
Rod Elliott of Elliott Sound Products (ESP) treats passive-crossover capacitors as a high-stress job: they see real AC voltage and current, unlike tiny coupling caps in a line-level circuit. In his Capacitor Characteristics article he writes that bipolar (non-polarized) electrolytic caps sold for loudspeaker crossovers are, in his view, suited only for systems where fidelity is not a major concern. He calls polypropylene the optimum dielectric for that duty, with polyester usually all right on lower-powered systems. Electrolytic caps, polarized or not, change value over time and are simply not suitable for high-fidelity systems in that framing.
Why electrolytics drift
ESP’s point on electrolytics in filter and crossover service is stability: capacitance, ESR, and related parameters move with temperature, frequency, and age. In Design of Passive Crossovers he repeats the warning for impedance-correction and resonance-compensation networks: bipolar electrolytics can be used where audiophile grade is not required, but they are not stable over time: capacitance falls with age and the network drifts. He prefers polyester, polypropylene, or oil-filled paper when the network must hold its values. Bipolar parts stay attractive because they are physically small for a given microfarad rating. ESP’s tradeoff language is size and cost versus long-term accuracy and measurable distortion when significant voltage appears across the part.
Polyester vs polypropylene
For passive crossovers, ESP’s stated optimum is polypropylene. Polyester (PET / Mylar) is the practical step down he allows for lower-powered systems. Both are film dielectrics. He also notes temperature coefficients differ (polyester typically positive ppm/°C, polypropylene negative in his table), so keep crossover caps away from hot resistors when you can. ESP’s point is stability for the electrical job, not that polypropylene sounds better.
High-current positions: film-and-foil
Where capacitor current is high, ESP flags metallized-film construction: the metallized layer is very thin and may not carry full current as well as a film-and-foil part. He recommends searching manufacturer data for current handling, and says a film-and-foil type is more suited to high-current duty than metallized film. That matters more on high-power passive networks than on a small desktop two-way.
Voltage rating and tolerance (as ESP states them)
ESP does not give one voltage rating for every crossover. His guidance is to rate parts for the voltage and current they will see, prefer dielectrics that stay stable with time and temperature under sustained high power, and avoid building a network on wide-tolerance electrolytics that then need constant tweaking. Tolerance and aging show up as crossover-frequency error, which is why he pushes film types for anything meant to stay accurate.
Where the cap sits in your network
Part type is only half the job:
- Topology and order: first vs second order crossover and active or passive crossover.
- Level matching with resistors: how to pad a tweeter (L-pad).
- Physical layout, lead dress, and heat: crossover layout mistakes.
Parts sourcing notes live on the parts and kits hub.
Quick pick table
| Situation | ESP-leaning pick |
|---|---|
| Passive crossover, fidelity matters | Polypropylene film (optimum per ESP) |
| Lower-powered passive system | Polyester film usually all right (ESP) |
| High current through the cap | Prefer film-and-foil over metallized film (ESP) |
| Small / cheap / “fidelity not a major concern” | Bipolar electrolytic (ESP’s limited-use framing only) |
| Network that must hold values for years | Avoid bipolar electrolytics; use film (ESP) |
Before you order parts
Which kind of inductor? See crossover inductors: air core vs iron core.