Splatter is formed when you overmodulate a waveform so that you are
actually driving the modulator off. No?

This is my theory and I'm sticking to it for now.

When you overmodulate the carrier and drive it off then you are
creating quasi-squarewaves. Squarewaves of course (theoretically)
contain an infinite spectrum of harmonics. This is why we put a click
filter on a cw transmitter. It slows down the steep edges and removes
some of the higher order harmonics that produce clicks and out of band
components.

So now we are summing a squarewave that is rich in odd harmonics and a
sine carrier. Stick this into a non-linear (toob or transistor) device
and you got the harmonics of the squarewave and the carrier frequency
mixing up nicely. You could do an inverse fourier transform here to
prove it to yourself, but have I ever lied to you before?

If you were to modulate a 1MHz carrier with say a 1kHz squarewave then
you would have components at n=odd multiples of 1kHz away from 1Mhz.
9.999Mhz, 1.001Mhz, 9.997Mhz, 1.003Mhz.... and so on. Each successive
harmonic would be smaller, but they would go on for some time.

Throw some music into the mix and then the splatter is not fixed to a
set of nice odd multiples of any one modulating frequency, but a whole
band of frequencies. If these splatter components end up on a
particular frequency up or down the band where you are then you don't
really have any means to get rid of them. You can narrow up your
filter some maybe, but that isn't really getting rid of anything.
Maybe making the situation more tolerable. When I was a little kid we
had a bootleg cb radio guy in my town that was constantly splattering
the whole band. Probably overdriving a linear.

So point here is I am doubtfull that a filter would help the situation,
but maybe some kind of cancelling technique might work. This MFJ1025/6
sounds like an interesting piece of gear.

I think I'll drag my Racal RA17 back into the shack for some boatanchor
fun this weekend.