Shaft voltage is directly attributable to two different things: dissymmetry in the magnetic path, and generation of a static charge.
Internal to the machine, the construction of the magnetic circuit (laminations and frame) and the layout of the conductors (winding) contribute to a potential dissymmetry in the flux path. After all, most coils are hand-made - which leads to marginal, but noticeable differences.
External to the machine is the source waveform. If it is distorted (for example, contains harmonics in VFD) then the resulting magnetic flux generation will also be distorted.
The problem with several grounding systems for motors is that it is the waveform distortion causing the shaft voltage - and it is occurring at a very high (e.g. switching) frequency. This means it attenuates quickly: just putting on a brush is not always sufficient. A large cross-section / short run ground path to the facility ground plane is often required as well.
In my work this problem started more than 10 years ago and adequate protection is now used. There are 2 types of current that flow:
The Type 1 is caused by the fast edges of the PWM waveform and circulates around the motor. This Type 1 current can be prevented by insulating one bearing usually the NDE bearing. These bearings are readily available.
The Type 2 is also caused by the fast edges of the PWM waveform and circulates back to the inverter by any route. This Type 2 can be prevented by either:
- An insulated shaft coupling plus 2 insulated bearings
- A good shaft earthing system, like 360 degree AEGIS brush at the DE end, plus insulated NDE bearing.
I have recorded these voltages and currents using 2 small carbon brushes on a 500 kW motor with 2 insulated bearings. Can also use filters at the inverter to have lower dv/dt applied to the motor. I have also experienced bearings failing by incorrect installation without an adequate end float margin.