I've been testing out a CRT monitor we have here, and sending out a digital pulse (using a digital IO card) that is synchronised to the vertical retrace of the monitor (using XNA software). Alongside that I am measuring the output from a light sensor attached to the screen which is about 250 micro-seconds behind the pulse but otherwise exactly in synch with it, i.e. always exactly the same delay after the pulse.
When I attach an oscilloscope to the digital pulse signal, I notice that the signal is not entirely stable, the durations at which the level is high and low vary by as much as a millisecond. This observation has also been verified in the software using high accuracy timers.
For my application, timing is extremely important, therefore I'm asking whether anybody knows the answer to this question: IS IT CORRECT THAT CRT MONITORS CAN HAVE FREQUENCY INSTABILITY?