Which Method Is Best?

This depends on the frequency to be measured, the rate at which you want to monitor the frequency and the accuracy you desire. Take for example, measuring a 50 kHz signal. Assuming that the measurement times for the sample clocked (with averaging) and two counter frequency measurements are configured the same, the following table summarizes the results.

Table 24. 50 kHz Frequency Measurement Methods
Variable Sample Clocked One Counter Two Counters
High Frequency Large Range
f x 50,000 50,000 50,000 50,000
f k 80 M 80 M 1,000 80 M
Measurement time (ms) 1 .02 1 1
N — — — —
Max. frequency error (Hz) .638 31.27 1,000 .625
Max. error % .00128 .0625 2 .00125

From this, you can see that while the measurement time for one counter is shorter, the accuracy is best in the sample clocked and two counter large range measurements. For another example, the following table shows the results for 5 MHz.

Table 25. 5 MHz Frequency Measurement Methods
Variable Sample Clocked One Counter Two Counters
High Frequency Large Range
f x 5 M 5 M 5 M 5 M
f k 80 M 80 M 1,000 80 M
Measurement time (ms) 1 .0002 1 1
N — — — 5,000
Max. frequency error (Hz) 62.51 333 k 1,000 62.50
Max. error % .00125 6.67 .02 .00125
Again, the measurement time for the one counter measurement is lowest but the accuracy is lower. Note that the accuracy and measurement time of the sample clocked and two counter large range are almost the same. The advantage of the sample clocked method is that even when the frequency to measure changes, the measurement time does not and error percentage varies little. For example, if you configured a large range two-counter measurement to use a divide down of 50 for a 50 k signal, then you would get the measurement time and accuracy listed in the 50 kHz Frequency Measurement Methods table. But if your signal ramped up to 5 M, then with a divide down of 50, your measurement time is 0.01 ms, but your error is now 0.125%. The error with a sample clocked frequency measurement is not as dependent on the measured frequency so at 50 k and 5 M with a measurement time of 1 ms the error percentage is still close to 0.00125%. One of the disadvantages of a sample clocked frequency measurement is that the frequency to be measured must be at least twice the sample clock rate to ensure that a full period of the frequency to be measured occurs between sample clocks.

  • Low frequency measurements with one counter is a good method for many applications. However, the accuracy of the measurement decreases as the frequency increases.

  • High frequency measurements with two counters is accurate for high frequency signals. However, the accuracy decreases as the frequency of the signal to measure decreases. At very low frequencies, this method may be too inaccurate for your application. Another disadvantage of this method is that it requires two counters (if you cannot provide an external signal of known width). An advantage of high frequency measurements with two counters is that the measurement completes in a known amount of time.

  • Measuring a large range of frequencies with two counters measures high and low frequency signals accurately. However, it requires two counters, and it has a variable sample time and variable error % dependent on the input signal.

The following table summarizes some of the differences in methods of measuring frequency.

Table 26. 5 MHz Frequency Measurement Methods
Method Comparison Sample Clocked (Averaged) One Counter Two Counters
High Frequency Large Range
Number of counters used 1 1 1 or 2 2
Number of measurements returned 1 1 1 1
Measures high frequency signals accurately Good Poor Good Good
Measures low frequency signals accurately Good Good Good Poor