Timebase Calibration Overview
An explanation of how the timebase is tested and the assumption of a uniform instrument clock.
The timebase can be tested by inputting a wave of known frequency and counting the number of zero-crossovers that the instrument sees. Other techniques are possible but the zero-crossover occurs at the point of maximum gradient of the wave so the accuracy of analysis in this region is better.
If an instrument has a timebase that varies over a short period but is correct over longer periods this can be difficult to detect. Fortunately in hardware terms this scenario is unlikely.
Timebase Error Explanation
An explanation of why the timebase is accurate to +/- the interval between two sample points.
For this explanation an example sine wave is used that has a length of eight times the gap between samples. The graphs and discussion below consider cases where it is either exactly this or infinitesimally more or infinitesimally less.
Demonstration of how a wave can coincide with the sample points to maximise the measured duration.
If a zero-crossover occurs just before sample 0 and just after sample 8 then this the length is infinitesimally more than 8 times the gap between samples. In this case the final zero-crossover will not be seen until sample 9. This means the instrument overestimates the length by one sample length and obtains the value of 9 tims the gap between samples.
Demonstration of how a wave can coincide with the sample points to minimise the measured duration.
If a zero-crossover occurs almost a whole sample gap before sample 0 and just before sample 7 then this the length is infinitesimally less than 8 times the gap between samples. In this case the instrument underestimates the length by one sample length and obtains the value of 7 times the gap between samples.
Using these two extreme examples it can be seen that both the maximum overestimate and underestimate are the same and equal to one sample gap. So, the error in this method can be treated as +/- the gap between samples. A counter-argument to this is that interpolation can improve the accuracy but the Tcalibration approach is to be cautious and also make the analysis simpler to understand and audit by not using interpolation.