Just a comment on distortion on weak signals.
This was a point of tech discussion I have seen and experienced in my own sets.

Basically if a "demodulating" diode in a radio is not fed with a "healthy" signal it will run down in its non linear part of the rectification curve.
The distortion products are therefore higher and the recovered audio has a nasty 'edge' to it.
I ran into trouble with IF signals at the detector diode under 10 volts (say).
If I reconfigured the AGC levels/Gain/ and raised the diode voltage to 15 or 20 volts, the recovered audio became cleaner.
I think there is a dissertation in the Radio Designers Handbook complete with pages of maths, non of which I understand!
The broad subject to research is "types of Detectors and the inherent distortion" and so forth.

This is probably the reason why some sets sound "nice" and some "not so nice".
Having an audio system with a lot of treble cut also helps as the teeth gritting products are high in the spectrum.
The worst set I made for this trouble was a TRF wide band radio with a 3-speaker network Hi Fi speaker box and 6V6 push pull with a extended frequency output transformer winding. Just a tooth gritting lot of reception until you turned the treble control to -30 db cut!
Upping the detector voltage nearly fixed it.

Its not something that is immediately picked up by testing with a clean sine modulated RF source and doing THD tests on a steady signal.
That sort of single tone signal sounds clean, but when you ran program signal as complex as music, not so much human speech, the problem is apparent.

So much to learn.

Fred.

Some where I did read an article where, I believe, IBM created a variant of 6BE6 to use in computers, as there were issues with it for what they intended to do with it.

Based on observations over many years and the voltmeter on the UPS which is always on, when it is on; Any fluctuation in supply is potentially going to introduce a variable and it does often vary during meal times, as the loading changes. I would be interested in the manufacturers detailed data in the 6BE6 as it might be similar to 6A8 which is just one pentagrid that is affected by AGC and fluctuating voltage. In the main, to stabilise a master oscillator it is fed regulated power.

The IF tubes should not drift in frequency as they are just amplifiers and should only increase, or decrease in amplitude in accordance with any voltage changes & bias changes due to the action of the AGC (If applied).

2WN on 1431 is supposed to be a "local" station to me as well, but it isn't very strong, and is completely overwhelmed by the hash from the digital electricity meter. 2FC (576) from Sydney is a much better proposition.

Distortion has so many causes. I would tend to look more at the speaker, the components around the output valve and AGC circuit, before getting into the complicated stuff.

One thing to take notice of, is that the N78 requires a 7000 ohm speaker transformer, while the 6AQ5 is happy with 5000 ohms. Make sure the valve in each radio matches the transformer. The valves have the same pin connections, so could by accident be swapped around. The N78 will overheat with the 5000 ohm transformer. A 6AQ5 will be fine with 7000 ohms though.

Measure the G2 and B+ voltages. The 6AR7GT and 6BE6 share a common G2 connection, the maximum allowable voltage being 100 volts. However the for the 6BE6, the1/3 theory should also be considered. So at 210 volts B+, the maximum G2 should be 70 volts or so. Doubling the dropping resistor (as suggested) from 10k to 22k should do the trick. Make sure of course that the associated decoupling capacitor is ok.