Adventures in VFD, circuits, EMF/EMI: thoughts?

I recently decided to give the VFD / Chinese spindle gig a whirl. I went with a 110v VFD with built-in step-up transformer to the three phase 240v output the 1.5kW spindle requires. The jury is still out on the overall performance and weight of the monster (I do have the upgraded one piece X-axis and I 3D printed the mount in PETG - by hand it seemingly has similar or less play than my Dewalt). I’ve had good results from some initial conservative tests.

Here’s the fun part. My workshop is a small plastic utility/garden shed (much nicer than it sounds!) in our back yard next to a much older wooden structure that serves as a sort of garden shed and former well pump related structure. It has electricity from the house, so when I added my makeshift workshop last summer, I temporarily just extended some heavy extension cords between the two in conduit while getting situated. This worked well for some months - most of which were spent doing other things due to my schedule.

I tackled the VFD in earnest during my extended holiday break. As expected, just spinning up the motor threw a breaker on this old circuit. To get things tested, I pulled a #10 AWG extension from another circuit within the house that I could easily access (cord through the cat door routine) to power the VFD. This worked great and I got it interfaced to the X-Controller, did test carves over a few days, etc.

One of my last goals for the holiday break was pulling a replacement branch circuit to the pump house and properly extending it into the workshop with outlets, etc. We’d recently had service upgraded in the house, so popping in a 20 amp breaker and running a new circuit parallel to the old one into the pump house, dropping off an outlet and extending the run to an outlet or two in the workshop was pretty straightforward. I used all #8 and #12 wire, etc.

Last night I plugged everything into the new circuit, fired up the spindle and instantly lost connectivity to the X-controller. Knowing some people have EMF/EMI issues, I had been thrilled in the week prior that I was never seeing such a thing. No missed steps, no disconnects, etc. It all went swimmingly. Now with it all on one modern, properly grounded circuit I’m having a raft of issues. Upon searching, I find that separate circuits are often the fix… so, ironically, I started more ideally than I’ve ended up in this regard.

Here are some of the things I tried and discovered:

  • it’s coming from the VFD even with the spindle detached… I get the same results just sending the M3 start even with the cable off the spindle.

  • I get the same results even if I start the VFD run manually on the front panel WITH it detached from the X-Controller (meaning the 0-10v motor start/speed control disconnected)… so it’s not that signaling connection. In this situation, the VFD is just as detached as a Dewalt would be.

  • I found an old thin USB cable with a ferrite core on one end and it improved the situation, but went from absolute failure every time to intermittent failure and hosed up communications in Chilipeppr talking to the controller. Heading the right way, though!

  • I found a really short translucent USB cable with mesh weave inside… perhaps a foot long vs. the usual 3-6 feet. This actually works seemingly 100% as long as I don’t keep the x-controller and VFD on the same circuit - I have to put the X-Controller on the old circuit and keep the rest on the new one. It still fails if they are together on the new circuit.

It’s pretty clearly a momentary spike of sorts that is hosing the USB since I can improve on it with different USB cables. I’ve ordered some USB cables with ferrite cores on both ends to try later this week… but it still makes me nervous about forthcoming missed steps I wasn’t seeing to date if I move to one circuit.

I can live with keeping these split across two circuits, though it irritates me that I can’t seem to combine them yet. Arguably, two is better anyway, of course. Things I do plan to try:

  • My VFD is very close to my X-Controller and I can probably move it pretty easily to test proximity.

  • I will try rerouting the VFD power cable more even without moving the entire unit somehow.

  • I’ve not tried doing this with my unshielded end stop switches disconnected from the X-Controller… perhaps they are a path for this since everything else is shielded that goes in or out of the controller and/or the VFD?

  • I’ve seen discussion of attaching shields to ground which I doubt is the case on all of my X-controller connections. I know I have that in the VFD since last week I found the ground pin in the spindle was unconnected and I went to great pains to avoid the great pains of being fried by grounding the spindle body, etc.

Right now, I’m running with a short USB between the Raspberry Pi (JSON serial for chilipeppr) and the X-Controller with the controller on a separate circuit from the VFD. I can live with refining this… but I’m a systematic problem solver and can’t rest until I’ve tested all the permutations and know for sure what I have to do to solve the issue once and for all. :slight_smile:

Love any feedback from those who better understand wiring / EMF/EMI situation I describe!

1 Like

You might find this article interesting;

1 Like

Fascinating. I’m inclined to put my oscilloscope in various spots in my workshop power to see what kind of chaos the VFD is putting on the circuits.

Does this suggest the issue is EMF affecting the USB, or a disruption to the stability of the AC wigging out the X-Controller? I’m assuming EMF since the short USB cable and the one with the ferrite core behaved better (shorter and shielding being less of an antenna, so to speak)…


  • Aaron

Personally, and I’m not a bit winnie, I poke ferrite cores on anything I can see. I’ve even opened my incoming service panel and added them next to the circuit breakers. When in doubt, why not?

Done! Just purchased a near identical set (albeit from Amazon). Going to systematically see what I get as I add them. Any value going onto wires right in the breaker box as well? That was mentioned above.


putting them on the plug going to the outlet should have the same effect as being on the breaker box.
things to try.
separate grounds for xcarve and vfd
grounding frame of xcarve
put vfd in a grounded metal enclosure
put distance between vfd and other systems i.e. xcarve and pc

You might try purchasing/ installing an appropriate reactor (inductor) to the input of the VFD. I work with VFD’s and Power Conditioning Units for Fuel Cell Power Plants on a daily basis. It depends on the VFD, but most will modulate at a frequency of 3 - 4 kHz which if the recitifier stage and DC Link is not of “great” design/ construction this can inject quite a bit of harmonics back onto the AC circuit that is feeding it. The reactor will also help prevent damage to the VFD due to transients on the AC circuit. In the event you have a well constructed VFD it might actually have a reactor on the AC input.

This high switching frequency can also produce high levels of EMI aspreviously mentioned so giving it some space from the X-Controller would be a fantastic idea.

When/ If I ever make the switch this is something that I definitely will do.

EDIT: Haaa…just read the white paper above…




This jumped out at me: insulated cords might not belong in conduit due to thermal issues. Heat might affect voltages, etc…

Valid point. That said, draw was never high on them (never felt a warm cord under use) and I’ve left that method behind now with the new wiring. When I temporarily had the VFD on a #10 cord at the outset, it was never in conduit and also never got warm under load (I checked frequently). I’ve never seen the VFD report more than 1.3 amps under load which shouldn’t be pushing limits on a cord that size. It’s now on #12 wire that transitions to even larger #8 wire for the new run directly to the breaker box.

Does that address what you were thinking?

That should cover it. :slight_smile:

#12 is not per code. #10 and # 8 are. How long the cable is being run is an important factor. Conduit is not an issue and should be run if possible. Thermal issues = 0.