I'd be happy to share the discovery. I expected I'd be done with the prototype by now, but other things have been keeping me busy.
What I found works much better is a simple op-amp or comparator, with the piezo connected from ground to one input, and the other input biased at a few milliamps. I added a 22 megohm resistor across the piezo just to provide a DC path for the tiny input current of the op-amp.
Any op-amp or comparator with a maximum input offset of a few millivolts or less an an input bias current of 1 nA or less should work fine. There are literally hundreds to choose from.
With this circuit and the big piezo from a Radio Shack sounder (the element, when removed from the housing, just fits inside a Mag D tailcap), I can get prolonged outputs of over a minute. With smaller piezos, it's not hard to get outputs of several seconds. This circuit is far more sensitive than the '555 circuit, but it will not latch like the '555 can if you are careful (or careless?).
The problem with these things is that they are very sensitive to impacts. The slightest bump, especially with something hard, can cause a short output pulse. In my designs it's easy to de-bounce the signal inside the microprocessor, but with something like a TaskLed driver, you can't do that. You could reduce the sensitivity by increasing the bias voltage on the amp, or perhaps by adding a lowpass filter on the output.
Don
