Yes what Aaron says is true a LED connected as a headlight will work perfectly in the forward movement but in reverse it will flicker because of the back EMF generated by the motor, it is irritating to say the least, but most locomotives ready to run have a light circuit in it that filters it out to stop this problem sounds like this is a case of the Chinese cheapening it up a bit more and not including the circuit for the lights.
It will not matter what you clean if this is the problem, it is just a thing that happens although it shouldn't though.
Just while we are on this subject would a higher value resistor on the LED lead cure this problem. I have one or two locomotives that do this most are home made installations though.
Not too sure about the BEMF theory. Has anyone proved this? I am a little rusty on my DC motor theory but for BEMF to be generated the motor needs to be running. BEMF is generated by the armature windings cutting through the field permanent magnet lines of magnetic flux. This induced current opposes the current actually driving the motor and is almost a type of inbuilt regulation. The current flowing through the motor does not change direction, it is simply opposed slightly. I can not see how BEMF would affect the LED. The current does not reverse, BEMF decreases as the motor speed decreases. This opposing current is also fairly constant in relation to motor RPM. I would doubt that any flickering of sorts would be created. If you spun the motor by hand it may do something but I would love to see that also.
Obviously BEMF decoders use this opposing current to regulate the motor in which their driving, however I thought this occurred in the off period of the PWM cycle. The decoder handles all of this and the lighting function would be isolated from any BEMF feedback. I would imagine anyway.
As this R class is purely a DC locomotive I just can't see how the BEMF theory would work. I would be more open to the idea if it had a decoder as there may be a fault with it. Maybe one of the steering diodes is faulty or there is a slightly bridged solder joint. Who knows?
If it were my R Class I would send it back for repair or replacement. Eureka is aware of it and knows what the model should be doing.
Not sure? Would love to hear the theory spelt out. I know Aaron will be all over it ha ha.