How accurate is the Hour Counter in the Statistics Log?

Does anyone know how accurate an hour counter log will be when the events that cause the hour counter log to activate are sometimes in the range of 0 to 2 seconds? I am running a program where I need to monitor when a motor goes over a certain max RPM but quite often it is only going over the max RPM for less than 2 seconds. In most cases I only care about times when the motor is above the max RPM for longer periods of time which I know the hour counter will add up accurately, but I received a clone where my event counter showed that the motor had gone over the max RPM more than 3000 but the hour counter said that the total time above the max rpm was only 0.40 hours which works out to less than 0.5 seconds per instance.