I guess that the big cap I had on the input side of the booster wasn't nothing; it took forever (in computer time) for the voltage to rise. I found some inverse e to the t over tau formula that gave me about .1uF for 98% of my voltage in a 15 ms rise time... if I was using that formula correctly! Anyway, I did replace the big cap with a .1uF and it seems to be working okay.

Do I need any input cap for the booster circuit? I'm running the PWM at 72KHz so the inductor on-time is only a few microseconds. If the panel response time is on the order of a millisecond or so, it would seem like I wouldn't need an input cap at all... Does that sound right? (Perhaps I've been reading too much; I've been reading about component "stress" in SMPS'es and am concerned about "stressing" something...)