Just being rhetorical, I wonder why someone wants multiple tiny PVMs.
kitty666cats's idea to try the phase knob is a good idea as I think the problem is rooted in phase. Hue or tint adjust is another form of the same thing. Maybe dialing down chroma helps a little. I assume PVM has a phase pot inside you can adjust.
I'm pulling most of what I say out of my @ss and doing a writing exercise so take my advice in light of that. Component is nowhere near as convoluted as NTSC/PAL but is the least understood in our scene. The Y is the brightness (luma), the "blue" is U aka Pb-Y and the "red" is V aka Pr-Y. The Y contains green, blue and red information scaled by how bright we perceive them, meaning all color info will be recovered with matrix math. In video processing, the term YUV is more useful.
You should test Composite or S-Video as well. Way television processes them is by converting into YUV on the way to RGB for the electron guns. Much amping along the way. Can narrow down the problematic circuitry if they show ghosting or not.
LukeEvansSimon brought to my attention that some televisions, I believe more so older ones, actually don't convert YUV to RGB and send the YUV to the electron guns instead. In such a case, I believe the OSD is in YUV form for being the display's "native" format.
Is the color magenta versus purple? Can see in these two YUV color space representations that magenta and green are equal in having max chroma value but opposite in phase by a 180° shift. Purple still overlaps with shades of green.
The basic problem is a sharp transition from max luma (white) to zero luma (black) should show grey if too slow but never color. Can find from matrix math equations or the 3D plot that shades of black/grey/white contain no U or V, only Y. Pb and Pr would be 0 volts input. That there is any magenta/purple-green is a mistake in the video processing that mismatches the Y to the U/Pb, and Y to V/Pr if seeing magenta due to corruption of the proper phases.
Possible causes, not mutually exclusive:
1) PVM's chip that filters the U if not also V has end of life capacitor(s) that shift the filtering past the phase margin. Any analog low pass or high pass filter shifts the phase. 1st order low pass shifts from 0° to -90° and 1st order high pass 90° to 0° in the passband. Sharper transitions need higher frequencies and therefore get greater shifts. Probably looking at 5th or 6th order filters here where a typical phase margin could be 50° versus 90° in 1st order.
If parasitic capacitance or phase distortion earlier in the chain is high enough, every signal that is off at least 50° hits 180° shift and gets inverted. Output is unstable and oscillates between U and -U and you could see a swap between magenta and green. 0V is still 0V but now Y is paired against the wrong U, if not also V.
Issue could equivalently be a capacitor in the filter to cancel out parasitic capacitance is end of life. In any case, you have to replace the video processor IC that handles this. Probably easier to swap the whole board.
2) Coupling capacitor on the U chain is end of life that reduces the RC time constant enough to miss trigger timings. Equivalently, oscillation could occur from too much capacitance on the output, such as from a loose output resistor that is supposed to prevent instability. This you can replace. Check solder joints, leaked fluid creating a short, something along those lines that is fixable. V chain too if seeing magenta.
