Hi James,
I have not checked your code but here are some suggestions which you may have already tried.

The DC motor will produce quite high noise spikes, so try de-coupling the DC motor power from the micro and measurement circuits.
Display the signal from your magnetic pulse generator on a cathode ray oscilloscope and look for narrow spikes. If there are spikes filter the signal to get rid of them.

Try reducing the source impedence of the signal from the pulse generator to make it less susceptible to electrical noise.

A simple check to identify what pulses the micro sees is to toggle an I/O pin of the 2051 every time it sees a pulse from your pulse generator. You can now display both these signals on a CRO and there should be a 1:1 relationship.

Make sure that the interrupt is edge triggered and not level triggered.

I suspect you have noise on your signal and the micro is actually counting these and jumping out of the loop prematurely.

Good luck,
Tony