As my primary axe is my Martin D-25 (D-18 with koa sides and
back) I've explored the topic a bit - Doc and I passed a few e-mail
on it a year ago or so. The long and the short of it is the piezo
pickup wants to see about 10M of load - really high-z device, and
the signal can swing as high as 10 v P-to-P. Using a grid leak bias
input stage you can get a 10M, or so, input impedance (really
resistance) but design is critical to limit noise pickup and signal
cutoff. I've experimented with a few preamps, including one that
used old battery portable radio tubes (1S4, 3V4, etc.,) with some
success.
 
But, what I wound up doing is rewiring the second input jack of the
vibrato/reverb channel on my 1967 Vibrolux Reverb with a 3 meg
series resistor and a 1 meg grid return resistor (often erroneously
referred to as a "grid leak"). With this setup I get "reasonable"
signal level - I tried a 8 m/2m setup but got too much signal
attenuation. Subjectively, the SPL is about 1/2 of the level I get
with my 60 Strat plugged into the first input jack but it suffices for
most purposes (I play with 4-5 hand drummers sometimes). What I
like about the mod is that input jack 1 is "normal" while using input
jack 2 switches in a different load for the Martin with minimal
alteration of the amp.
 
Still, "when I get around to it" I am going to modify my old Bogen
"all purpose" amp (60w Bogen head with 8417s built on top of a
sealed plywood cabinet with 12 Yamaha speaker and tweeters for
general "stand alone" PA use) and add a grid leak input stage.
 
Rob