Sampling Parameters
All of the tests enable setting the sampling parameters. If you're not familiar with the parameters here are our suggestions.
Here is the standard sampling block->
Sample Rate : The number of samples per second. High sampling rates will run faster and contain data at higher frequency (up to rate/2). High sample rates also have less accuracy at the very low frequency range (< 50Hz or so). The frequency distance between data points is Sample Rate / FFT Size.
FFT Size : the size of each sweep. Since the total sweep time is FFT Size / Sample Rate increasing sweep size increases the sweep time. The larger the FFT size the better the low frequency accuracy. Increasing FFT Size also increases frequency resolution and reduces the visible noisy floor (not the actual noise floor). Conversely increasing the FFT Size increases the possibility of a slight signal glitch occurring during the sweep - ruining the FFT.
Averages : averaging the sweep involves averaging in the frequency domain. The total test time is then Averages * sweep time. An average of one displays a single sweep. Increasing the number of averages generally reduces the visible noise spikes in the spectrum, revealing distortion and hum spikes. It also improves accuracy of all readings just by averaging out random perturbations. We recommend using at least 2 averages in general use and 4-6 is probably better. For sweeps like THD vs Frequency or THD vs Amplitude 1 average is usually adequate since distortion readings are not improved with more samples except for a slight reduction in variance.
Windowing : this is an advanced topic. In general the FFT result is very good if the oscillation time is an integer divisor of the total sweep time - which almost never actually happens. Windowing is a way of partially ignoring the packet end effects. It produces a good fft result even without that condition. There are good resources describing the effects of various Windowing methods.
A large percentage of the time using a Sampling Rate of 96000 samples per second, FFT Size of 64K, and Windowing of Hann will work with good performance.
We recommend a faster sampling rate for chirps because the chirp signal gets noisier at the top frequency range. Extending the top range by having a higher sample rate enables more accuracy in the 40-50KHz region which can be useful. We also recommend more data points (so maybe an FFT Size of 128K or 256K) if the chirp extends over lots of octaves to ensure enough of the sweep time covers each frequency. The best thing to do with chirps is limit their frequency range to only the frequencies you actually care about instead of being lazy and using 20-20KHz for everything.
We recommend using Hann windowing (good amplitude accuracy and very little smearing of the frequency response) except in the rare case when you have an external signal generator with variance at the start time. Then it may make sense to use Kaiser windows, which will have slightly lower frequency accuracy but more resilience in the face of start-time variation.