VFD Noise on Incoming Line

My 2 cents ..

Of course check the grounding of the cables as suggested.

If you check a signal from the pressure sensor (one of them) with a scope (a spectrum analyzer would be better) and determine what your noise frequency is, that is a start.

The noise from the front end of the drives, through the diode bridges, is line frequency * 3, 3 phases. If your noise is at about 180 Hz, the suggested EMI filter on the front end is likely a good solution.

The 350 HP drives likely default to 500 Hz switching frequency. But you should check. Whatever the switching frequency is, if that is where your noise is, changing to 5 KHz or some other available number should make a big difference. The motors will heat up a bit more, and so will the drives. But at the loads you are talking about, the drives won't care. If the sensors are happier with that, you`re done.

The DC Reactors are on the DC bus and need to stay.

If the noise is lower, like 60 Hz, someone could have replaced an instrument power supply with a crappy one and it could be letting 60 Hz noise through. Replacing the instrument power supply is quick and pretty easy. Make sure that the instrument grounds or shields are connected, one side only, to a good bus as suggested already,

If your noise is not either of those, 180 Hz or 500 Hz (or whatever the switching frequency of the drives are), but some other frequency or perhaps a combination of frequencies, it sort of depends on how much money you would like to throw at the problem.

A DV/DT filter (replacing the load reactor) will reduce the spikes and should reduce the 'volume' of the noise. Sized for full load amps fo all 4 motors combined. But it will still be there. And the filter will heat up the VFD cabinet.

A sine wave filter, sized for the full load amps of each motor, installed after the overload for each motor, should get rid of more of the noise. Reducing the `volume`at least, but it should also get rid of some of the higher frequencies entirely. But it is an expensive solution and will heat up the cabinet. Will it be enough? I don't know.

I have no input to give on how much to spend when filtering the VFDs versus filtering the sensors. I don`t have experience with filtering sensors

A few comments on the above postings: First, the input noise frequency from the diode pulses will be six times line frequency, not three times. On 60hz, that would be 360hz. Remember, there are three phases and two diode pulses per phase, one + and one -.

Second, while it is good practice to only ground one end on instrumentation shielding, that is not true of power grounding. Power grounding as in shielded motor leads, conduit, and enclosures should be grounded on both ends effectively forming a ground bond connection between the motor(s) and the VFD enclosure. That is what a ground bond is: a connection between TWO grounded devices.

Third, ordinary conduit connections do not form reliable electrical connections. They need to be strapped at each joint to be sure the circuit passes thru the joint.

Fourth, reactors are simply inductors and filter higher frequencies without regard to the current passing thru them. As long as they are not saturated with too much current, their reactance is equal to 2 x pi x hz x inductance. Note that current does not appear in the formula.

Finally, since this is a semi-mobile application, I would start by being sure the motor frame and all interposing conduit and enclosures are copper-wire bonded back to the VFD enclosure ground. This can be done with bare wire run on the outside of the conduit with strapped connection to each piece or component in the conduit/enclosure system. And, yes, ground both ends, one at the VFD and one at the motor(s). #8 gauge should be large enough.

If that doesn't do it, we will need to talk about input and motor lead reactors. Before coming back to us, it would be good to identify with some precision what reactors are already in place.

Hope this helps.

DickDV said:

First, the input noise frequency from the diode pulses will be six times line frequency, not three times. On 60hz, that would be 360hz. Remember, there are three phases and two diode pulses per phase, one + and one -.

Click to expand...

Hmm. I count 1 + pulse and 1 - pulse as 1 Hz. Whatever. 360 pulses per second, 180 + and 180 - ... we are agreed.

Second, while it is good practice to only ground one end on instrumentation shielding, that is not true of power grounding.

Click to expand...

Yes. Agreed. Sort of. Bonding is for both ends, as far as I know. So the bonded ground cable keeps both the VFD and the motor at ground potential - so there is no shock hazard during faults, no induced voltage at the motor frame.

But VFD cabling has a cable ground .. not sure what it is actually called so I'll call it the bonded ground? .. plus there is one drain per phase .. What I've seen (admittedly in brochures) is those three go to ground on one end only. If they are connected on both ends they don't really drain the noise, do they?

Fourth, reactors are simply inductors and filter higher frequencies without regard to the current passing thru them. As long as they are not saturated with too much current, their reactance is equal to 2 x pi x hz x inductance. Note that current does not appear in the formula.

Click to expand...

Agreed. I think I was discussing dv/dt filters (which have an LC circuit) and sine wave filters. But all of them have rated currents. And although you can size filters many ways, one way is for the full load amps of the motor. For a sine wave filter , running at FLA makes the waveforms look most like a sine wave on a scope. There are only little high frequency pulses visible on the main 60 Hz (or whatever wave the VFD is putting out) that are quite low amplitude, at switching frequency and .. odd harmonics .. I think.

thingstodo said:

Yes. Agreed. Sort of. Bonding is for both ends, as far as I know. So the bonded ground cable keeps both the VFD and the motor at ground potential - so there is no shock hazard during faults, no induced voltage at the motor frame.

But VFD cabling has a cable ground .. not sure what it is actually called so I'll call it the bonded ground? .. plus there is one drain per phase .. What I've seen (admittedly in brochures) is those three go to ground on one end only. If they are connected on both ends they don't really drain the noise, do they?

Click to expand...

Sorry but DickDV is correct on VFD cable all the symetric grounds as well as the cable shield should be grounded on each end other wise you run the risk of having common mode current flowing in your ground grid.

PBuchanan said:

Sorry but DickDV is correct on VFD cable all the symetric grounds as well as the cable shield should be grounded on each end other wise you run the risk of having common mode current flowing in your ground grid.

Click to expand...

OK. I guess I learned something today. Do you have any links that explain the common mode current .. dumbed down a bit?

https://www.hottconsultants.com/pdf_files/APEC-2002.pdf

http://www.mtecorp.com/pages_lang/w...e-Problems-and-Effective-Filter-Solutions.pdf

http://literature.rockwellautomation.com/idc/groups/literature/documents/wp/drives-wp019_-en-p.pdf

Looking at the construction of the cable, the phase wire shielding and the overall cable shielding is not made of foil but rather of heavier stranded mesh or, in some brands, is made of a spiral wound solid copper sheet.

Clearly, this much copper is provided to handle the loop currents that can flow when a ground bond is installed. The purpose of the ground bond in VFD systems is to force the motor to the same ground reference as the VFD. Grounds can float all over DC and AC if not bonded and the Short Circuit Fault and Ground Fault detector in the VFD cannot properly detect current leakage to ground. The result is nuisance Short Circuit and Ground Faults which nobody wants. Of course, the shielding also minimizes the amount of high frequency electrical noise that is radiated throughout the adjacent spaces.