my prototype now can be used with DHO804 for bode plot (barely) so i tried to redo test by TimFox here...
https://www.eevblog.com/forum/testgear/rigol-dho914s-bode-plotR = 100 ohm, C = 33nF, so cutoff freq = 48kHz... (1 / (2.pi.R.C) )
i can confirm the same bugs and wobbly effect in old FW1.14.
1) the bode plot procedure is finding optimum vertical scale for each frequency, but its inefficient imho. its just brute force set to minimum vertical scale and increase the scale until signal fits the monitor for each point. this results one click for each channel for every point to be plotted (2 clicks per point), i think this can wear out front end relay quickly. the more efficient way is only do brute force search during the beginning, and increase or decrease vertical scale from there on the next point, if signal too small, decrease scale, if signal too big, increase scale, cant you think of that rigol? how many brains you have over there?
2) i tried to set acquisition to average to reduce noise and hopefully wobbly effect of the plot on low level signal, but unfortunately the dso changes to normal acquisition during bode plotting, sigh. where is another brain?
3) suggestion: another trick to get better bode plot. is to convert input signal to FFT, find largest magnitude bin, that is our fundemantel, get it phase as well. then convert the output signal (CH2) to FFT, and find the same bin element, find its magnitude and phase, substract both values and you got a point on bode plot. i think this will be much cleaner.
so i guess if bode plot is too important, controlling DHO804 from PC to adjust time and vertical scale appropriately, command the AFG to give incremental frequency at desired voltage level, download the data to PC,and do our own procedure/algorithm for bode plot is much preferable... only the good thing about internal AFG/AWG is we only need one cable connected to PC rather than two in case of external AFG unit... fwiw...