Mine also sometimes shows targets moving fast, laterally. It’s because it cannot determine the azimuth accurately. There may also be airframe shielding of the antenna(e).
within 2cm of each other in length translates to a miniscule difference in attenuation
why do you think it’s the attenuation that matters? I’d expect it’s the phase angle that matters; 2cm physical difference translates to 3cm free space difference, together with a wavelength of ~30cm, this translates into a 35° angle error
I thought it worked in the amplitude only.
Obviously you are the expert on RF 
I thought it worked in the amplitude only.
I don’t actually know, I just make an educated guess 
But you basically proved in your post that it cannot work using amplitude differences.
I would expect it to work like LORAN in reverse, i.e. measure the time of arrival difference (via correlation of the carrier wave, so the antennae should be less than lambda/2 ~15cm apart) at two closely spaced antennae, to get a hyperbolic location equation. Do that (at least) twice and you should get an intersection point
Having been exposed to a PhD (humanities) for 11.5 years, I should know by now to never argue with a PhD (real engineering)
But amplitude must come into it, otherwise the phase lag cannot be measured (instantaneously). Otherwise, why do they insist on same length cables?
Reportedly, the Garmin box doesn’t need same length cables (don’t have the IM immediately to hand).
Otherwise, why do they insist on same length cables?
Because a difference in cable length translates into a difference in propagation time through the cables, which directly adds to the arrival time differences at the antenna. So the receiver sees the sum of the arrival time difference with the (electrical) cable length difference.
But amplitude must come into it, otherwise the phase lag cannot be measured (instantaneously)
Well yes RF is an AC waveform, it cycles through positive and negative amplitudes rougly 1 billion times per second. So yes amplitude is part of the measurement. But only that positive/negative thing is important, the absolute level not so much.
This is why I do the measuring with the network analyzer as described in the other post. Using time domain measurement one can measure how long it takes for a signal to travel over each cable, each cable should have the same electrical length for the best possible performance. This is why using difference batches of cables is to be avoided, especially if only mechnical length is measured.
Because a difference in cable length translates into a difference in propagation time through the cables, which directly adds to the arrival time differences at the antenna. So the receiver sees the sum of the arrival time difference with the (electrical) cable length difference.
OK… but then the 5cm spec is way too loose, given that the two antennae are only 5cm apart to start with
Well yes RF is an AC waveform, it cycles through positive and negative amplitudes rougly 1 billion times per second
That’s what I like about life. Every day is a learning experience!
It seem to be intermittend on one of the antenna’s. Does it actually swap positions, or does it position all trafic on one side only?
Intermittend contacts can destroy the amplifier inside the unit, so anyone with these kind of issues should have investigated this in an early stage.
OK… but then the 5cm spec is way too loose, given that the two antennae are only 5cm apart to start with
No it is not, it is a doppler technique with 1/4 lengt antennes spaced 1/8 wavelength.
The propagation of RF signals trough air is different than trough transmission lines. @tomjnx is right on this topic. It is important to understand the difference between electrical length and mechanical length.
The amplitude (or attenuation) is important to detect aircraft which do reply to other than the TAS active interrogations. In that case amplitude can be a measurement for distance. When an reply on it’s own interrogation is received, it does know the time that this reply took, and thus the distance.
