That's the thing I've noticed generally with Atlas and some Fox Valley locomotives is when you remove the shell for whatever reason, the mechanism is always noisier than prior to disassembly. I don't exactly know why. For that reason, I dread opening up Atlas locomotives. And it's funny how the noise becomes apparent after the shell is replaced. You can run the mechanism alone for hours and not hear a thing. The shell goes on an there's the noise. Grrrr....
I'm assuming these are not equipped with DCC and if you check this forum there might be some traffic about that situation. Some decoders create a slight buzzing noise, some don't.
Regardless of DCC or not, here's what I've done to both.
I've suspected it has something to do with the shell being slightly distorted when removed from the mechanism. Maybe if I left the locomotive alone for a few days after assembly that might cure it but I'm not that patient. I need to solve it immediately it seems.
There are many "noise abatement" solutions tried including packing sections of foam to fill the empty voids in the locomotive to reduce noise and vibration. Nothing really noticeable there. Tried electrical tape on the frame and had some success there. Swapped trucks end for end and sometimes the noise is reduced.
Done the Beardon fix on most of the problem locomotives and that helps, but doesn't work on the First Gen Kato F-Units (where the motor is completely enclosed by the frame): too much slop in the drive train as a result.
The other thing that was tried was loosen the two screws that secure the mechanism to make it a little wider than from the factory to prevent shell vibration and that seemed to work.
Your solution is an approach I've not thought about and I will try it out the next time I dissect an Atlas.
Thanks for the tip!