VFD noise reduction

[megloff11x]

I have a 480V motor system that is not surprisingly picking up high frequency noise bursts at the VFD PWM frequency. I really knocked it down on the rigid frame with some moderately expensive multiply layered shielded conduited cable designed just for that purpose. However, I have some moving parts that are still good antennae because their moving coupling is not well bonded to the return at all frequencies, particularly the higher ones.

My own understanding of knocking down noise is that you need to provide a low impedance path back to the noise source's return point (usually grounded) for all frequencies. A fat wire is fine for low frequency but is an inductor and high impedance for high frequency. High frequency needs planes or conduit or braided shielding or all of the above and just the right geometry.

Does anyone have any solutions they'd like to share for mitigating noise pickup on moving metal parts?

And would anyone like to comment further on noise reduction strategy from a scientific basis?

Matt

 

[kc8dxx]

1. Find the source of the noise. We use a portable oscilloscope with a high frequency scope probe. We hookup the scope probe to a 3" pickup coil made by taking hookup wire and wrapping it around a tube 4-6 times. Consistent with being safe, move that pickup coil around the equipment, and note the location and frequency of noise that appears. Very simple and handy technique.

2. Find the noise frequency that is bothering you. No need to deal with frequencies that are not hurting you.

3. MAKE SURE that the equipment is installed per it's manufacturer's instructions, especially if any equipment has EMC installation notes. For motor systems, it is not unusual to have the OEM require filters on the field/armature lines. Those filters make a huge difference. Those filters can be large and heavy. Shielded power conductors are a necessity. For a 350HP motor, the power conductors can run $25/FT.

4. Shield all lines, bond all surfaces, ground all frames. Braided wire is better for conducting high frequency noise than low frequency noise, due to skin effect. At high frequencies, unbonded metal can become an antenna! Bond metal using braided wire, protected as necessary, to a single point on the equipment.

5. RF filters & capacitors may be necessary on signal & power lines. Filters must be solidly bonded to the panel frame. Noisy lines must be kept physically separate from lines that need to be quiet. If you run noisy power wire to a filter input, then run the filter output next to the input wire, you've just defeated the purpose of the filter.

EMC is a big issue in many industrial sites. Google is your friend. Many companies have further noise reduction techniques available online.

 

[megloff11x]

Thanks. I've done all of the above. The switching frequency is 8kHz and I get a burst every 8kHz on the scope. The elements of the burst are at higher frequency.

The rigid parts are fine. It's the moving parts and not having my bonding job wear out over time that are of concern. I've used braid in the past to strap moving parts but it's looking like I'll need to use a LOT more braids on this one.

The drive output is filtered a bit internally but they don't allow for external filtering the sharp PWM, or I'd do it.

I'm just checking if I missed anything or if there are new insights or worse, if I'm doing something dumb. I found the special conduit on google and it surprised me how well it worked.

Thanks again.

Matt

 

[kc8dxx]

If the HF noise is causing a problem, I would suggest contacting the manufacturer and asking them for suggestions. If they sell in Euro land, ask them for their European installation instructions, which typically have lots of EMC mitigation instructions included. Whose drive is it?

 

[John Valdes]

Have you tried a line/load reactor? I use one everytime I install a VFD. Also, if the motor is more than 100' away from VFD, I install a load reactor at the VFD. Really line and load reactors are the same thing. They change the impedance. Check www.baldor.com and search reactors. Theirs are suited for both line and load. Installing these reactors has solved noise and nuisance trips for me. Actually Baldor will sell the reactor at 1/2 price if you buy the VFD from them. That is also the first thing they ask you when you call for advise. "Do you have a line reactor" Some VFD's come with one built in. ABB is one.
As far as moving parts bonding: They make a device that consists of a grounded mount and carbon brush. You can attach it to a grounded surface, then you position the carbon brush on the rotating shaft. I will see if I can locate a vendor for this product. Hey....Ever tried insulated bearings? Both inner and outer race are non-conductive.....Bartlet Bearing Co.

 

[peteo]

Matt,

I'm not sure that I understand the issue here. If, as I suspect, your motor and so forth, are rattling away at an audible frequency above 8 khz.

Every physical system has a natural frequency which it responds to, just like a tuning fork. The equation is fn = sqrt (k/m), where m is the mass and k is the stiffness coefficient, found by trial and error. Tuning forks respond to audible frequency, highrises to very slow (1 Hz) movements.

You cannot change the mass without adding weight. You will alter the stiffness each time a belt tension, extra bolt, etc., is altered in the physical system. VFDs do allow you to alter the carrier frequency, but what it sounds like is that you've got a 'big brass drum' equivalent which takes very little stimulation to rattle. More or less stiffness will change the frequency response curve shape but never get rid of the fundamental frequency issue.

Perhaps there is some combination of bearings, mounting plate, motor frame, which you can isolate and then deal with. A piece of hose used as a stethoscope, is how mechanics track down noise locations in moving machinery. I'm not sure this is even responding to what your question was.

 

[petersonra]

Line reactors are there mostly for increasing the source impedance and probably won't do much for noise issues. Load reactors may help.

Best answer I can give you is that we build panels with hundreds, maybe thousands of VFDs every year, and almost never have noise issues.

Adding line and load reactors to every VFD increases the cost and the panel footprint, and does nothing to solve the problem.

One thing that does tend to help is to use the relatively new VFD cables. Even wiring the motors with SO cord makes a big difference.

 

[megloff11x]

peteo

It's electrical noise, not mechanical. Every 125u seconds (8kHz) I get a burst of noise Voltage spikes. Coincidentally, the VFD switching frequency is 8kHz.

I changed to a newer conduited cable especially designed for 480V VFD driven motors and the noise went away on the frame. I still get a bit of electrical noise on a moving part that is "bonded" to the rest by moving metal parts, and it's annoying a sensor. I'm just trying to nip that last little bit.

A continuously moving "strap" that conducts it away has proven elusive.

I will second petersonra on newer VFD cabling. These new generations of VFD optimized cables are the cat's meow. The braided shielded cables provided by the vendors in the past were mediocre at best. The multilayered somewhat flexible conduit at several bucks a foot (for 6AWG 4-cond - not bad given copper prices) just knocked it to almost nothing where I have solid bonding on the fixed parts.

Matt

 

[peteo]

Just imagining a couple quick and dirty tricks with encoders, would it be possible to

1) increase current through the sensor.

2) put a small cap or RC somewhere in the sensor circuit, it'll behave like a low pass filter.

 

[plate]

peteo

It's electrical noise, not mechanical. Every 125u seconds (8kHz) I get a burst of noise Voltage spikes.

I changed to ...and the noise went away on the frame. I still get a bit of electrical noise on a moving part ...
Matt

Would it be correct to say you are getting mechanical noise (vibration) that you attribute to the electrical switching frequency of the drive (that you see on the scope)?

 

[boater bill]

I have heard very objectionable VFD noise from fan and pump motors just before they fail. Then replace them with inverter rated motors for the higher insulation value and the noise is greatly reduced, almost negligible.
If the switching frequency is set at 8Khz, have you tried going higher or lower to see if the noise follows the switching frequency?
The only problem with going to a higher frequency is the capacitive coupling of the motor leads if they are long, >100 feet or so. I might try 15 Khz just to see what happens. Since 15khz is almost out of our hearing range, it will just annoy the dogs!:grin:

 

[Ranch]

Thanks. I've done all of the above. The switching frequency is 8kHz and I get a burst every 8kHz on the scope. The elements of the burst are at higher frequency.

Has your scope the ability to do a spectrum?

 

[RayS]

You may also want to look into shaft grounding systems, which may or may not help the noise, but may help bearing life by routing the noise current around the bearing

http://ecmweb.com/products/motor_grounding_device_120505/index.html