From: "Peter Barbella" on Fri, Mar 11 2005 7:40 pm

Yes, there is a trigger. Here is the schematic.
http://mysite.verizon.net/nonobar/circuit.jpg

There's actually a capacitor coupling the trigger pulse into the
diodes (I
forgot to show it).

The two bias resistors going to -15 are critical as are the
commutating
capacitors (not shown) across the 3.3k resistors.

It wants to work, but in a very unstable manner. I think I simply may
be
taxing the rise times of the 3904s.

The legacy transistor 2N3904 (and 2N3906) aren't "taxed"
any rise times. They are old workhorse transistors
which will work on up to 30 MHz (f_t is 150 MHz or
better)...provided the biasing AND the rather large
base-emitter capacitance is compensated. A good text
(also a "legacy" one) is Millman and Taub's "Pulse,
Digital, and Switching Circuits" which has much good
basic information on discrete device non-linear-
operation circuits.

The -15 VDC supply isn't necessary nor is it "sensitive"
if the trigger-switchover voltages are selected with
reasonable care. Ten percent tolerance resistors can
do the job with a single-rail supply of +5 VDC.

The "commutating" capacitors (common term is just
"speed-up") across the collector-to-base resistors
MAY be critical since those compensate for the base
to emitter capacitance and thus set the high frequency
response in that part of the circuit.

Since +5 VDC supply is there, one or two Schmitt
trigger inverters or gates from the oscillator output
(74HC family, for example) to a 74HC74 dual D F-F
will work fine. A quarter for each device and there
will be gates and a flip-flop left over to use
somewhere else. No sweating out component values
for discrete device circuits, response of the 74HC
devices known and dependable.

For that matter, the crystal oscillator itself can
use the remaining inverters of a Schmitt trigger hex
inverter. An ordinary non-Schmitt inverter (74HC04)
can do the oscillator and squaring-up task as well.
No sweat on that.