Jim,
What gate frequency are you using in your example?
Depending on the frequency you are using, you could have a synchronization error. This refers to the measurement variation due to the gate and source edge not being synchronized. In the example you mentioned, you should be using a 20MHz timebase, where each source cycle lasts 50ns. If the gate frequency you are using is of 50KHz, the number of 50ns periods that occur during one cycle of a 50KHz signal is 400 +/-1, which yields a frequency measurement of between 49.88 and 50.13KHz. On the other hand, if the gate frequency you are using is of 5MHz, then the number of 50ns periods that occur during one cycle of a 50MHz signal is 4+/-1, which yields a frequency measurement between 4 and 6.67MH
z.
Again, the reason why you might be getting very erratic readings could be because of the frequency that you are outputting from the optical tachometer into the gate of your board.
Please let me know if this helped. Thanks!
LA