So far the only theory that accommodates the majority of issues is a "shorted-turn" caused by the ADF antenna - I think the majority of those interested in this thread are on that page currently. This lines up with the directional aspect of the problem. How about noise....?
The AGC circuit is probably the culprit. In most receivers the AGC voltage is taken from the detector. The dc voltage is derived to be proportional to the average level of the carrier and adjusts the gain of the radio-frequency (rf) and intermediate-frequency (i-f) amplifiers. The AGC tends to keep the input signal to the audio frequency amplifier constant despite variations in rf signal strength. Remember that all circuits have an inherent "self-noise", so, given we are losing signal due to the shorted-turn effect, the net result is to run the receiver at very high gain, hence the noise.
This theory fits as far as the new antenna on the aircraft underside NOT being so affected (out of the shorted-turn ADF antenna influence).
So what doesn't fit?
The only part I'm perhaps struggling with is why turning off squelch does not seem to produce the same "noise" on an open channel; i.e. tune off station, defeat the squelch and I would have expected a similar noise experience (since the AGC should ramp the receiver to max gain). However I have not looked at the specs or circuit of the radio in question and there are various tricks of circuit design to prevent ''no-carrier-at-all" situations blasting noise... so that may be it.
The last question - why the RPM relationship? Vibration... above X RPM we have some vibration causing the ADF antenna to behave as the theorized shorted-turn. This would probably explain the intermittent effect also, since the exact vibrational mechanism and cause of the shorted-turn behavior are currently unknown.
Thoughts?