Shielding PWM'd power cables

Thread Starter

cmartinez

Joined Jan 17, 2007
8,220
One of my machines has an electrical setup roughly as shown.

1596199205513.png

There are a couple of 90VDC 1/10 HP motors being PWM'd at approximately 20% duty cycle at 7KHz. They're probably consuming less than 1.5 amps each. The machine has been working fine for a while, but recently the MCU controller has been getting noisy signals from the inductive sensors, and hence it has been misbehaving.

Unfortunately, I could not find an off-the-shelf cable and connector harness for the sensors that is also shielded. And so the lines running from the sensors to the distribution box are ordinary 24ga three-conductor wire for each sensor. And even worse is that those wires run in parallel with the cables powering the motors through the same plastic conduit chain. Once the sensor's wires reach the distribution box, they are then connected to a single multi-conductor shielded wire all the way to the MCU, where its shield is connected to ground. The other side of said wire (the side in the distribution box) has been left floating so as to avoid possible ground loops.

Question, would it be better if the wires feeding the motors were also shielded so as to contain whatever EMI is being produced? If so, should their shield be grounded at both sides, or at one side only?
Would changing the sensor's unshielded cables for shielded ones solve the problem?

@MaxHeadRoom
 

MaxHeadRoom

Joined Jul 18, 2013
28,619
To take a leaf out of the VFD/CNC type of installation, there is motor cable manuf. just for this purpose, if running your own cable, it is imperative that the motor conductors are twisted together the whole length, the motor earth ground wire ran along side.
There should be no other L.V. conductors ran in the same cable, pipe or tubing.
Equi-potential earth bonding of the equipment should also be carried out.
Max.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,220
it is imperative that the motor conductors are twisted together the whole length
Gotcha, thanks ... Thing is, that said motor conductors run along 10m of plastic conduit chain that is constantly being flexed. That would be a problem for twisted 18ga cable, wouldn't it?

I'm sure that twisted cable would work best, but wouldn't using shielded cable for the motor conductors improve things a bit?
 

MaxHeadRoom

Joined Jul 18, 2013
28,619
What I have used in the past for VFD motor conductors if some flexing is needed, is flexible metalic conduit.
It is automatically grounded at each end.
You can get it down to 3/8", if needed.
It is a bit laborious, but I have twisted the motor conductors together using THHN/MTW control wire, with the GND conductor ran along side.
If flexing friction is a concern, you could use wiring gel as a lubricant, although messy, the cleaner way is Talc.
Max..
 

nsaspook

Joined Aug 27, 2009
13,086
I suspect most of your EMI is magnetic field sourced (motor drivers PWM switching currents). Electrostatic (electric field) shielding on single-ended signals helps somewhat with that but twisted cables on the motor drive cables (and signal cables) will result in a net reduction of total EMI from magnetic field cancellation if the currents are balanced in the twisted cables.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,220
I suspect most of your EMI is magnetic field sourced (motor drivers PWM switching currents). Electrostatic (electric field) shielding on single-ended signals helps somewhat with that but twisted cables on the motor drive cables (and signal cables) will result in a net reduction of total EMI from magnetic field cancellation if the currents are balanced in the twisted cables.
Many thanks for sharing your thoughts, NSA. The currents in the cables should be balanced, since they come from an H-Bridge. I'm going to try first with a shielded cable. But then go nuclear with a twisted pair of my own craft if I have to.

Question, how tight would the twists have to be in the cable. That is, how many twists per unit length would you recommend?
 

MaxHeadRoom

Joined Jul 18, 2013
28,619
This is a constant bane in the DIY CNC (CNCzone) etc, spurious tripping of L.S. inputs etc, I have recommended good earth grounding as well as equi-potential bonding of the various earth points to prevent it, as well as ensuring all motor and power cables are twisted and shielded.
This has traditionally solved their problems.
With equipotential bonding, the old practice of grounding one end of the shield has been dropped in favour of ground both ends.
I have Siemens paper on it if you need it.
Max.
 

nsaspook

Joined Aug 27, 2009
13,086
If you use motor cable shielding it should be ferrous (magnetic) for it to be most effective if the EMI source is primarily low impedance switching currents. EMI is electromagnetic so you need to deal with the characteristics of the source emitter via electric or magnetic coupling. Good bonding and grounds reduce current loops, shielding (electric and magnetic) reduces emission across space but the goal is to reduce the EMI source as much as possible.

https://www.machinedesign.com/mecha...le/21833014/eliminating-emi-in-motion-systems

To correct this problem requires a threefold attack. First, reduce the emissions. Second, shield the receiver to break the coupling. And third, force a current path to ground via the intended connections.

https://www.analog.com/media/en/training-seminars/tutorials/MT-095.pdf

Figure 2 shows some of the general ways noise can enter a circuit from external sources.

Impedance mismatches and discontinuities
Common-mode impedance mismatches → Differential Signals
Capacitively Coupled (Electric Field Interference)
dV/dt → Mutual Capacitance → Noise Current
(Example: 1V/ns produces 1mA/pF)
Inductively Coupled (Magnetic Field)
di/dt → Mutual Inductance → Noise Voltage
(Example: 1mA/ns produces 1mV/nH)
Figure 2: How EMI finds Paths into Equipment
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,220
I've just noticed that my cable supplier sells shielded and twisted cable. And I've just tested its shielding, and unfortunately it's non-magnetic.
 

Thread Starter

cmartinez

Joined Jan 17, 2007
8,220
Here's what I think I'm beginning to understand, please correct me if I'm wrong:

  • Intermittent (or alternating) power through a wire will cause a varying magnetic field radiating perpendicularly from its axis. The higher the power, the stronger the magnetic field.
  • Since said magnetic field varies with time, it will induce a current in all nearby cables around it. Assuming that those cables form a circuit of some sort and are not open ended.
  • It is those induced currents that are mainly responsible for interfering with low signal cables, especially if those cables run parallel with the power cable in question. The longer they run in parallel, the stronger the interference.
  • Since it is a magnetic field we're talking about, it takes a tube of ferromagnetic material to "enclose" said magnetic field by bending its lines inward as much as possible and keeping them from extending beyond its boundaries.
  • Also since said ferromagnetic material is also subject to the varying magnetic field, eddy currents will be created on its surface which will help dissipate the energy being absorbed.

I had always thought about electrical interference as being of of a sort of radio emissions only. And about the cables receiving said interference acting as antennas that will in turn transmit said interference into the circuit. And I am sure that is still the case, it's just that I hadn't visualized before the magnetic part of the problem.
 
Last edited:

nsaspook

Joined Aug 27, 2009
13,086
What's important is the source and transmission impedance resistive and reactive characteristics when you look at types of shielding. High power, high impedance EMI sources will be dominated by electric field energy emissions, high power, low impedance EMI sources will be more magnetic in nature in the reactive near-field. There is only one EM entity (Electromagnetic Field Strength Tensor) but it exists (and is modeled) in a higher dimensional space that will be expressed in 'normal' 3D space frames of reference in a range from a pure electric field to a pure magnetic field and all combinations of the two between.


 

MaxHeadRoom

Joined Jul 18, 2013
28,619
There is also the cancellation effect of EMI when current conductors used in a common circuit are twisted throughout the length, resulting in cancellation of EMI.
Siemens & Schneider publication on Grounding and equi-potential bonding. Particularly CH6 on shielding
Max..
 

Attachments

Last edited:
Top