Thanks, guys.
Update:
* The "nothing after the diode" means the scope registers nada here. I haven't actually attached the Arduino yet.
* The diode is the right way around, and looks good according to my multimeter.
What I hear is that the design is basically sound, and that we need some bug-fixing/tweaking. That's progress, too.
What I'll try, according to your suggestions is:
1) Add a 100uF across Vin/GND, that's the largest I have (actually I tried this, but it dampened the output signal from the LM386, so I removed it). But it goes back in.
2) Change the input cap to 1uF
3) Replace the diode, just in case
Pito: Could I bother you with a further expansion of t=k*R*C? I assume that R is resistance in Ohms, the C capacitance in F, and t the time in s? But k?
Thanks in advance.