Interval Timer Problem

--- Quote Start ---

I'm using it as a cycle counter.

So I need to change the timeout period to 1sec instead of 1 msec if I want to measure a time round about 0.0008 sec.

--- Quote End ---

You definitely need a timeout period larger than the measured time.

As I said above, if your timer is dedicated to the measuring task and doesn't serve other functions or peripherals, you can operate it freely.

So, I suggest this:

- set timeout period larger than the maximum expected interval to be measured

- timer initialized in stopped status

- reset and start the timer when you start your measure,

- stop the timer at the end of measure

- read the duration

--- Quote Start ---

I also want to measure it exactly as possible.

--- Quote End ---

This greatly depends on your application, namely how you perform the measurement:

- poll a PIO or any other condition and start/stop/read the timer: poor precision, dependent from your polling rate, unless you can afford using the 100% of cpu time for the polling loop and don't execute any other task

- use a irq triggered by start/stop conditions: common solution for greater system effinciency and good precision in the us range

- use a hardware counter solution which automatically identifies start/end conditions and places the measurement in a register which you can read without effort: maximum precision at the expense of a slight increment in hw complexity (fpga resource usage)