Fedia is right

In a similar project I counted the frequency using both methods (counting pulses in 1 sec and f =1/T) The 1/T method is more accurate for lower speeds. I ran both methods concurrently using the 2 timers, and then decided which one would be more accurate based on the pulse count, and if the frequency was more than about 200 Hz, I went with the pulse count rather than 1/T. The pulse count is accurate enuff to make this decision. Using this I could accurately count from 0 - 9999 HZ without a sweat.

200 HZ was an value I used due to design constraints, but for finding which method is best for which range, look up "Electronic Instrumentation" by Helfric Cooper. The formula used is extremely simple but you should understand the theory behind it.

But once u decide to use f=1/T method, be warned that you will face 16 and 32 bit divisions. Erik is also right that as the range goes on increasing the complexity increases exponentially. You will have to use several different methods in a wider range.

kundi