0
\$\begingroup\$

Imagine a pulse train around 80Hz. Using two DAQ at the same time I sample this pulse with 5000Hz and 10000Hz. I get slightly different pulse widths in each case. I use an algorithm which detects rising edges and find each pulse widths.

Do you think sampling freq. have effect on detecting pulse widths in this case?

\$\endgroup\$
  • \$\begingroup\$ How many samples occupy the width of the pulse for both sampling regimes? \$\endgroup\$ – Andy aka Apr 19 '16 at 10:53
1
\$\begingroup\$

You are aliasing, because you are not respecting the Nyquist criterion.

An 80Hz pulse train has components going up to very high frequencies. You will alias these differently depending on your sampling frequency.

If you low pass filter your pulse train, such that all components above fs/2 are removed, then you will get consistent results. However, note that once you have low pass filtered, you will need to use an ADC to capture the signal.

\$\endgroup\$
0
\$\begingroup\$

If your sampling is accurate and you are using a timer that is operated at the sampling frequency and the timer is properly gated from the rising edge to rising edge detect then there will always be at lease one sampler period interval of variation on the measured pulse width.

On the other hand if you are using some software techniques to detect the edges and start/stop the sampling interval counter then there are a number of things that can affect the overall accuracy and repeatability of the measurements. Since you have not done anything to describe in detail the algorithm that you used it is totally impossible to make an expert evaluation of your question.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.