If ferrites haven't been tried, on *all* wires going to and from the controller, then I suggest that these may indeed help. It's the wires that act as antennas. With the antennas "disconnected" by the ferrites then you only have the circuit board and components as possible
noise emitters. The dimensions of these are small enough that the emissions may be significantly reduced.
Putting a metal shield (mu-metal or otherwise) around the device will accomplish very little unless you filter and bypass the wires to the shield. I used to do this stuff for a living when I was designing
electronics products. Mu-metal by itself can shield close-in magnetic fields, but it doesn't sound like this is your problem.
I do agree that this sounds like very bad interference, which may indicate a defective controller, or perhaps a poor controller design.
The best approach for ferrites is to run the positive / negative pairs through the same
core. With #2 wires you may have difficulty finding ferrites large enough. Depending on the ferrite material and the
current level, it's probable that running a
single wire through a
core will be OK.