Has there been any comment from the Pico team anywhere about the frequency graphing drop outs as mentioned here: topic16831-10.html
that I have missed.
If you activate low pass filtering in the channel options, I left it at 1kHz, and apply a maths channel for freq(A), the trace becomes stable across all buffers.
When using frequency based maths it is often useful to filter out some of the noise as this can be interpreted as crossing points for the frequency calculations, causing misleading values. Remember the software has the same base code as the test and measurement version, which is designed for very high frequencies compared to those found in the automotive world.
The issue with those traces are the signal levels are very low, the traces are set at x3 to make them more visible in the display, and there are glitches of noise present. This make it difficult to calculate the mid level crossing points, with the necessary boundary settings, to be able to determine the frequency. Where you see jumps in the frequency trace, zoom in on the live channel data and you will often see a spike.
I had suggested AC coupling this signal to centre it around the zero line and allow for a fuller scale signal trace, but even this has a certain amount of ripple which is throwing out the calculations. If you zoom in on the frequency jumps you can usually see something different happening in the live trace.