Thanks IanL for the reply and your useful notes on Head Echoes. Interestingly another way of looking at the Doppler shift is using the Wikipedia definition for bistatic Doppler which gives DeltaF = (1/lambda) . d(RG + RM)/dt where DeltaF is the frequency shift, lambda is wavelength, RG and RM are the ranges of moving body from GRAVES and the monitor station respectively. Thus d(RG + RM) / dt is rate of change of the sum (RG + RM). Although I haven't done the algebra I am sure they are different formualtions of the same physics.
Wikipedia notes also notes under bistatic Doppler "... that objects moving along the line connecting the transmitter and receiver will always have 0 Hz Doppler shift ..." By this I infer that DeltaF will be zero when crossing over the Great Circle between GRAVES and Monitor station. This could at least give a line over which the meteoroid has passed
So, choosing a nice round figure for the actual velocity of the meteoroid of say 24 km/s the detected signal would last about a quarter of a second.
I too, very many versions away, started my Conditional Action scripts on those of Paul Hyde. For Head echoes I have modified my most recent incarnation to capture a screen shot whenever the rate of change of the frequency (DeltaF/dT) is above a given amount. Measuring this frequency slope has been identified in recent IMO journal articles: "Forward Scattering : an interesting formula to calculate the velocity of a meteoroid that generates a head echo"; Pierre Ernotte; WGN, the Journal of the IMO 46:6 (2018) pp 198 and the subsequent paper " Visualizing sporadic meteor radiants and their dynamics by radio forward scattering"; Wolfgang Kaufmann; WGN, the Journal of the IMO 46:6 (2018) pp 201.
I needed to modify the size of the FFT to increase the Spectrogram scroll rate and improve time resolution (at the expense of frequency resolution) I am still playing with this so won't give any details at the moment, but here is a typical screenshot.
I have been able to extract the frequency and time data from pixel data of images using software and have sensible figures for the frequency slope albeit with wide uncertainties. The software gives me 20 ms per pixel on the time axis and just over 2 Hz per pixel on the frrequency axis. My spectrogram indicates 2000 Hz corresponding to zero Doppler and the GPSDO on the RSPDuo ensures confidence in the absolute frequency.
In a roughly 24 hour period I captured some 1000 screen shots of which 50 were of the kind of quality that would allow slope extraction, and so it can be seen that I need to improve matters quite a bit to get the right ones 😞 The software will probably allow me to automatically extract the details but not sure yet. I also am working on getting a log of the frequency and time down the slope.