VFD Noise

VFDs (Variable Frequency Drives), or inverters, manipulate kilo Watts of energy, often times in short powerful bursts, and in the process produces a lot of electromagnetic noise. I have seen that noise couple into USB keyboard cables and activate the special function keys telling the PC to shut down.  I have seen that noise couple into limit and home switches making it impossible to run the CNC rig without constant false limit switch events (adding noise filtering to the limit switch input pins proved this to be the problem). Does this mean that you should not use a VFD? No. This means that you should try and address the issue correctly.

One way to address the issue is to add noise filtering to inputs, or try to make the rest of your circuitry immune to the noise. While there is nothing wrong with trying to make the rest of your circuitry immune to electromagnetic noise, the better option is to FIRST try and eliminate the noise at its source.


Conducted Emissions (Noise)

A good first step would be to place a line filter on the AC input to your VFD or even better a line reactor. This will keep the electromagnetic noise and distortions your VFD generates from going back up into the power cables in your building and then affecting other electronics (like your SmoothStepper).


Line Filter

An EMI line filter will use resistors, inductors (about 100 uH) and capacitors to filter out noise that would flow back up stream into the 60 Hz or 50 Hz sinusoidal power supplying the VFD. When searching for a EMI filter, a decent starting point would be to look at the SCHAFFNER FN2020 or FN2060 series EMI Filters like this one or this one. There are other brands too.

Please note that you should NEVER place any line filters on the 3 phase output of your VFD going to your spindle unless the manufacturer specifically says to do it.


Line Reactor

A line reactor is an inductor in the shape of a coil, and they typically come in a three phase form factor. A line reactor will have up to 100 times more inductance than the filter covered in the previous section. The much larger inductance in the reactor will allow it suppress the harmonic frequencies and transients, reduce power surges and also perform power factor correction.

A restive load (think a toaster or an incandescent light bulb) pulls it's current in phase with the voltage, which means that it has a power factor of 1. A VFD is not a resistive load, which means that the current will not be pulled in phase with the voltage, and a VFD may have a power factor as low as 0.5. When the VFD pulls a large amount of power, it will distort the incoming power's waveform for the rest of your house or building which can cause problems for other electronic devices (and if it is large enough and often enough, it will upset the power company). At a power factor of 0.5 you will need to pull twice the current that the VFD would need to pull with a power factor of 1, and this doubling of the current might even be enough to trip breakers. By adding the line reactor to the incoming power side of the VFD you will achieve some level of power factor correction and bring it up to 0.85 or 0.9, which will also help to reduce the current consumption to only 10 to 20 percent over the expected.

Since most users outside of an industrial setting will only have single phase power, they would only use one of the three coils on the reactor. There are lots of line reactors available second hand on eBay and may be significantly cheaper than new. You would want to find one that is between 5 mH and 10 mH and the higher the current rating the better, but at least what your VFD says is it's maximum, otherwise you will risk burning out the reactor or need to avoid running your spindle at full power. Typically you would want the reactor to be 3% to 5% of the load impedance, but this will be harder for most end users to calculate.

It is possible to connect the 3 line reactors together in series, in order to increase the total inductance, but you must pay attention to the phasing or they will cancel each other out; 2.1 mH +2.1 mH + 2.1 mH = 6.3 mH. Here is a 6 Amp 6.2 mH line reactor as an example.

Please note that you will NOT use a line reactor on the 3 phase output of your VFD going to your spindle unless the manufacturer specifically says to do it do to because of an excessive cable length to the spindle or because of spindle over heating.



Make sure your VFD is wired to ground, as stated in the manual! Follow the grounding instructions specified in the manual or good grounding practices if the manual is less than helpful. In some situations, you may have an excessively noisy VFD and it may be beneficial to have a dedicated earth grounding rod for that VFD so that it's noise will not flow back through the building's wiring to the main grounding rod.


Radiated Emissions (Noise)

Even after you placed a line filter, line reactor and grounded your VFD, it will still be emitting or radiating noise through the air.

  • Wire management. There is a lot of power flowing through the cables into your VFD and out of it to your Spindle. That power flowing will also have noise flowing with it, and both of those will couple into wires that run near you power cables and also just radiate noise in general.
    • The best way to prevent this coupling is to keep all of your other wires away from the VFD, Spindle and their associated power wires.
    • If you need to have a wire go by, keep it as far away as possible and cross at a 90 degree angle.
    • If the wires need to run along side the VFD power wires (i.e. through the cable tray to get up to the Z axis) then consider 24V signals or shielded cable (the next two sections).
  • Use a 24V DC Break Out Board (BOB) for your logic circuitry instead of 5V DC for probes, home and limit switches, and to control relays and other outputs. Using 24 V BOBs will give you almost 5 times the noise immunity compared to a 5V BOB.
  • Differential wiring. Some devices (like encoders) will allow you to use them with differential wiring (instead of A, B, I, Power and GND in the cable, there is A+, A-, B+, B-, I+, I-, Power and GND). Differential is much harder to corrupt with noise than a standard single ended signal.
  • Shielded cables for your logic signals. If your logic wiring is in a very noisy environment, using a shielded cable will cause the radiated noise to couple into the exterior shielding of the cable instead of your logic wires that are protected inside the cable. The shielding needs to be grounded at one end of the cable or the other, never both. When grounding the shield, DO NOT ground it to the ground of your BOB or your SmoothStepper. A chassis or earth ground would be the better place to ground the shield.
  • SmoothStepper noise filtering. The input pins of the SmoothStepper can have noise filtering added to them, which will help to reduce the effects of system noise. It is okay to add noise filtering to see if that fixes the problem, and thereby confirm that you have a noise problem. It is always best to actually reduce your noise sources, or improve your wiring to make it immune to the noise, instead of relying on the SmoothStepper's noise filtering. (When you use noise filtering, it winds up delaying the signals by at least the noise filtering time, or longer if there is lots of noise. This will directly impact the precision of your probing and homing routines. It is okay to add 1000 us of noise filtering to the limit switch and EStop input signals.)


Concluding Thoughts

How would I personally try to make my system as noise immune as possible while not killing my budget?

  • I try to keep my wires as far away from the VFD as possible. Plasma units and VFDs are the two biggest noise offenders I have to deal with.
  • I am switching from 5V BOBs to 24V BOBs. There is not that much of a price difference anymore and the noise immunity is just simply worth it for me. Plus with the preproduction Warp9 BOB it just makes my wiring so much simpler. A 24 V BOB and wiring may very well allow you to avoid using shielded cables all together.
  • While a line filter will help to reduce the noise going back into your building's power lines, it will not fix the waveform disruptions caused by the VFD, and for that you need a line reactor or an isolated transformer. For a small VFD and Spindle in a garage, a line filter may be all you need, especially since your utility will not be hitting you with a power factor correction fee. If you have a larger VFD and Spindle that draws more power, then adding in a line reactor certainly makes sense. If you really need to eliminate noise, then an isolation transformer would be the next step up.
  • If the components I am using allow me to run in differential mode, I would use it especially with encoders.
  • I don't typically use shield cables, except they make sense in the run up to the Z where the spindle or plasma cable is co-located.  IF I had really noisy motors, I would probably use shielded cables there too. However, with a 24V BOB, I don't think I will bother with shielded cables to begin with.
  • Finally I use a USB phone charger (wall wart power module) since it produces an isolated supply for my SmoothStepper's 5V power. Going forwards with the 24 V BOBs (which also power the SmoothStepper), I am using power supplies that use wall power and a transformer (the transformer isolates the 24 V power produced from everything else). It is NOT recommended to use a regulator that takes the low voltage tap off the motor driver power supply, since there is no isolation from the noise produced by the motor drivers, and that noise can flow directly into your SmoothStepper and BOB.













Top of Page