The classic way to do this for any application that does *NOT* require an X:Y plot was to keep the monitor intact and turn it on its side so the frame scan could be used for the frequency scan. One could tap into the timebases for the ramp waveforms, but it was usually easier to simply build Miller integrator ramp generators, and use a pair of astables for line and frame sync pulse generation, that also reset the integrators. The frame ramp provides the control voltage for the L.O. VCO to scan the spectrum, and the rectified and low-pass filtered detector output goes to a comparator, where its compared with the line ramp, and the comparator output drives a resistor network to generate the actual video signal. If you want a line graph rather than a bar graph, capacitively couple it! Its worth adding a couple of cursors - more comparators with a pot on one input, driving a different RGB pin to the spectrum signal (or a different gray scale level via the resistor network), and for the frequency cursor, you need a couple of flipflops as a 2 bit shift register after the comparator so it's clocked through one scan line at a time by the Hsync pulse, and XOR the Q outputs together to drive the RGB pin. Then all you need is an accurate readout of the pot wiper voltages to be able to take measurements off the screen.