Verifying frequency characteristics of high performance devices requires accurate frequency measurement. This document contains an overview of various frequency measurement methods and instruments and is intended to help the users of Si Time MEMS oscillators take accurate frequency measurements.
Common Frequency Measurement Issues
Measurements Made With Different Frequency Counters Do Not Match
A discrepancy between measurements results made with different frequency counters may be caused by one or more of the following reasons.
- Two instruments use different frequency references. Base models of frequency counters are usually equipped with inexpensive TCXO-based frequency references with 1 to 5 ppm frequency stability and a few ppm/year aging rate. An error introduced by the frequency reference will add an error to the measurement results. The picture below illustrates how measurement results of a frequency counter may differ when using an internal TCXO reference compared to an external high-precision rubidium reference. Refer to section 3 for frequency reference selection guidelines.
Frequency measurement results of a 25 MHz oscillator using an oscilloscope frequency counter synchronized to an internal TCXO reference and to an external rubidium reference. Gate time 1s. The difference between the measurement result s is 4.6 ppm
Gate times or instrument specifications are different. Different measurement results may appear when using the same frequency reference if instruments are using different gate times. Moreover if gate times and references are the same, but instrument resolutions are different, the results may not match at low gate times. Refer to section 4 for details.
Frequency displayed by the frequency counter is much higher than expected
Poor signal integrity may inadvertently increase or even double the frequency measured by the frequency counter. This is often encountered in probing schemes where the instrument inp ut is configured to a high impedance mode (for example, 1 MΩ). Section 6 discusses signal integrity impact on frequency measurement and provides probing recommendations.
Frequency counter measurements with different gate times do not match
The frequency measurement error of a frequency counter is inversely proportional to the gate time. As illustrated in the picture below, the shorter the gate time, the larger the error.
Frequency measurement error (ppb) versus gate time. Two time interval measurement accuracies represent different frequency counters. The plot assumes perfect time base
Oscilloscope frequency measurement shows large spread
Oscilloscopes take frequency measurements for every period of the input signal. Depending on the scope settings and capabilities measurement results may be averaged over multiple captures or over all signal periods within a single capture. Frequency measurement conducted on a single period is highly affected by signal period jitter and internal noise in the oscilloscope, causing the results to change by thousands of ppm. Collecting thousands of samples and taking the average significantly reduces the error, but this method still doesn’t provide ppm-level accuracy that can be easily achieved by using a frequency counter. The picture below shows an example of frequency measurement using a high-end oscilloscope.