VFD and EMI

[123CNC]

Looking for help in finding the 'best/quickest' fix. Added a vfd to a knee mill and when the spindle powers on getting a fairly steady interference signal into the z-axis encoder signal. The encoder wiring is shielded back to the replacement PC controller in a cabinet separate to the vfd. The vfd motor output lines do parallel the encoder wiring a good distance, but this was factory and worked OK on normal 3-phase line power and original controller, is now a problem with vfd and PC retrofit.

Should I try to get shielded motor cabling, and any good source (3hp spindle)? Or reroute new shielded encoder signal cable away from motor leads? What is the most common/typical problem with VFD and signal noise? I'm counting on some 'been there, done that' experience to fix the 'real' problem, cheapest, fastest, and effectively. Especially since sheilded power cables seem a little harder to find.

[Al_The_Man]

What kind of cabling is the motor wired up with?
Belden make a special VFD cable, but it is very expensive.
Many are wired up with single conductor stranded TEW or close, the three 3 phase conductors should be tightly twisted for the full length and the ground conductor run with them but not twisted. Liquid seal metallic flexible conduit is a good choice to run the conductors in, this acts a shield when also grounded properly.
How does the Z encoder noise manifest itself?
Al.

[123CNC]

The motor wiring, from what is visible, doesn't have any markings. It is four separate conductors (not cable and notmanually twisted either), stranded wire, and I would guess MTW, run through liquid-tite PVC conduit from cabinet to motor, with a fairly long open run in the cabinet.

With the spindle off multi-axis moves and encoder feedback all checkout OK. When on, the Z axis is clearly having a problem in that it will move to a commanded static position yet proceed to drift up and hit the max-Z limit all the while maintaining a (fairly) static feedback position. The interference seems to be fairly constant as the drift/controller's response is fairly constant, not erratic or ramped.

I thought about the metal flex conduit, but don't know where the biggest source or contributor is, vfd and in cabinet, or wiring/signals to devices.

[Al_The_Man]

I prefer Metallic flex conduit, twisted conductors is also important.
Also I would ensure all grounds are taken back to a common ground point together with the Service ground, also it may pay to look at grounding the encoder supply to this plate.
If the encoder is supplied from the PC power supply, the grounded common point for it can wend its way back to the PS ground through the MB screws etc, instead if a sound point external, for example, the ground bus, together with all shields.
Al.

[123CNC]

Al,

Thanks for the suggestions, I will give it a go when I can allot enough time and pickup some flexible metallic conduit. The encoders are on a separate power supply, but I need to verify its grounding path. I may try to run the encoders, or at least the z-axis, from the PC PS on the remote chance it provides better isolation.

You continue to live up to your moniker, you are 'The_Man.' Although smoking has clearly stunted your growth some.

Thanks again and I will try to remember to post my findings later.

[TOTALLYRC]

Originally Posted by 123CNC Looking for help in finding the 'best/quickest' fix. Added a vfd to a knee mill and when the spindle powers on getting a fairly steady interference signal into the z-axis encoder signal. The encoder wiring is shielded back to the replacement PC controller in a cabinet separate to the vfd. The vfd motor output lines do parallel the encoder wiring a good distance, but this was factory and worked OK on normal 3-phase line power and original controller, is now a problem with vfd and PC retrofit. Should I try to get shielded motor cabling, and any good source (3hp spindle)? Or reroute new shielded encoder signal cable away from motor leads? What is the most common/typical problem with VFD and signal noise? I'm counting on some 'been there, done that' experience to fix the 'real' problem, cheapest, fastest, and effectively. Especially since sheilded power cables seem a little harder to find.

The reason it worked with real three phase power and not with the vfd is that the vfd output generates noise in way that can be picked by sensitive devices.
I believe that my hitachi manual specifically states that control wiring must cross at 90 degrees and not run parallel for any distance.
My btc-1 which came factory with a vfd, had almost all the wiring twisted.
When I did the retrofit, I made sure that the wire twisting was replicated.

I used to have similar problems with my rc cars and just twisting the servo wires going from the servo to the reciever would solve most problems.

My understanding of the problem is that wires act as antenae and when you twist them you make the antenae lenght so short that any signal it picks up is well beyond the range where it makes any difference. I am sure there are other reasons why it works but work it does.Twisting the wires from the VFD to the motor will also be helpful.

Consider running the vfd wires in grounded metalic conduit if physically possible.. This way you have contained the noise and it won't affect anything that you add in the future.
Shielding the encoder wire just protects that wire.
Of course somtimes you have to shield both wires.

Mike

[Al_The_Man]

Originally Posted by 123CNC The encoders are on a separate power supply, but I need to verify its grounding path.

I normally take all my power supply commons to a single ground plate.

Originally Posted by TOTALLYRC My understanding of the problem is that wires act as antenae and when you twist them you make the antenae lenght so short that any signal it picks up is well beyond the range where it makes any difference. I am sure there are other reasons why it works but work it does.Twisting the wires from the VFD to the motor will also be helpful.

Twisting conductors have a mutual cancellation effect of adjacent conductors.
Both VFD supply and motor wires should be twisted, but the VFD to motor are the most important.
Many VFD manuf supply and recommend a 3phase choke on the VFD output for suppression also.
Al.

[123CNC]

Al,

Thanks for yet more info. My main goal was to attack from the most troubled area, i.e. most effective single fix, and thus

"...but the VFD to motor are the most important."

Is exactly what I was hoping to hear. If it doesn't completely fix it, I will move to the next best fix, line to vfd, then maybe chokes, then ....

I know you are a bigger fan of the higher quality industrial grade VFD's, but aside from the added bells and whistles are there designs much better in the way of EMF emissions?

[Al_The_Man]

Originally Posted by 123CNC Al, I know you are a bigger fan of the higher quality industrial grade VFD's, but aside from the added bells and whistles are there designs much better in the way of EMF emissions?

I am not sure whether one make has any more emission problems than another.
Just that the procedures I have followed, I have never yet experienced a real problem with interference , and this is with up to 3 10hp VFD's in one enclosure.
It's something I have never cut corners on.
Al.