Mark - Those last specs seem closer to what you need. Depends mostly on how wide the pulses are. With 220 MHz unity gain bandwidth, amplifier risetime would be about... ... 1/(220x3.5)usec or about 1.4nsec. Pulses would need to be several times as long in duration to be recognized, so a 5 ns long pulse would do okay. Some single component amplifiers also do well in this frequency range. For example, the lowly 2N2222A has more than 200 MHz bandwidth. Used in common emitter (voltage amplifier) configuration, the standard amplifier mode, the low input impedance make a pretty good match to a 50 ohm PMT output. The key to making bipolars (or FETs) work fastest is to have them biased on at all times.
If a PMT is terminated in 50 ohms, then for 1mv signal, you will need a current of 1/50 mA or 20 uA. 20uA over 5nsec makes a charge Q of 0.1 pC. For an electron gain of 10^8 This calculates to about 10^-3 photo-electron sensitivity. A more practical output signal of 10 to 100 mV would still require only about one event per second at the PMT input. Noise currents and other stray photocurrents would be substantially larger than this, unless you cool the PMT.
Because of the high stage(dynode) resistance (R ~ 10.0 Megs/Stage) in the PMT, the number of pulses per second is limited to about 1/RC where the R is the stage resistance and C is stage stray capacitance. A 1 pc stage capacitance would give a time constant of 10^-5 sec for example, allowing a pulse rate <100Kp/sec.
Incidentally, most PMT's require several hrs for the glass fluorescence to die away, if exposed to any sort of fluorescent or UV lighting.
Dave Cooper
Created on Monday, June 18, 2001 3:19 AM EDT by David Cooper