[quote author="dustbro"]I've been trying to do some research on the sonic differences between mylar and poly caps. Some articles I've read said that mylar is OK to use in Power supplies, but not in an audio path... some sites claim that there is no sonic difference between the two. I havent set up an A/B test yet... wondering what you guys think? is there a difference?[/quote]
There are maybe a half dozen important performance parameters that differentiate capacitor types from each other. Which metrics are important often depends on how they are being used in a circuit. For example a DC blocking cap with almost no working terminal voltage change in use will be quite a bit different than a small film cap in an EQ circuit, or worst case IMO for capacitors is in passive speaker crossovers where they are working pretty hard, passing high current with large terminal voltage changes with signal frequency.
For active EQ circuits I would read up on voltage coefficient (actual value changes with terminal voltage, a direct source of distortion), and perhaps to a lesser extent DF (dissipation factor). In heavier lifting application equivalent series inductance and resistance can cause errors. It all comes down to how does Cap XYZ deviate from a perfect or theoretical capacitor in your circuit.
Some applications, like sample and holds, are very sensitive to DA (dielectric absorption) but this is not as much of an issue in light duty blocking applications.
Years ago I liked polystyrene as a really nice dielectric that was also quite inexpensive. Unfortunately polystyrene was not very robust for commercial manufacturing and has fallen out of favor.
While I am not a huge fan of listening tests, it might be interesting if you have multiple identical audio paths to load alternate channels with different test components and null two channels from each other to reveal differences "in circuit" and with real music test signals. Of course be careful as a value difference can diminish depth of null in some circuits. Listen to the null, and if nasty, one or both of your test devices is nasty. I'm not quite sure how to identify which one is which.
JR