[Kevin Sevcik]

Your servo is likely seeing the commutation noise from that last motor. The signal to the servos is a standard PWM signal, so electrical noise can appear as pulse width commands to the servo. I'd recommend keeping your servo cable runs as short as possible and away from the motor power cables. You can reduce the noise from the motor by putting a capacitor across the motor leads, as close to the actual motor as possible. Per R66, this can be a 1 microFarad or smaller capacitor. You can also put a small resistor across the PWM signal leads to the servo to act as a shunt load. This would go between the white and black wires, again preferably close to the servo itself. I'm not sure on the value for this resistor. I think I'd start at 1k and slowly work down till it stopped dancing or stopped moving at all.

This resistor is good at filtering the noise from power wires, which is usually inductively coupled to your signal wire. This means it generates small currents in your signal wire. The signal wire is connected to a high resistance input, so small currents make for big voltage swings. By adding a shunt to the circuit, you're reducing the resistance, which reduces the size of the voltage swings. Your only problem is that eventually, the sidecar wont be able to drive enough current through a small resistance, and your servo will pretty much stop moving at all.