50Hz 3 Phase Circuit - Twisting to reduce EMI?

[Dinoduck94]

TL;DR Summary

Help finding the right formulas for calculating reduced EMI when twisting cables.

I have done the maths to work out the Induced Voltage on a communication cable, from a bunch of 3 phase circuits in a panel - with the minimum distance between them I calculated it to be about 1.5nV; but for a larger system where there could be larger loads on the cables I was thinking about what action you can take to reduce the Magnetic Field Density around the cables - without adding additional components.

I know that twisted pairs are used for to do this (to a degree) but I can only find formulas for higher frequencies - is anybody able to point me in the right direction for formulas that calculate how many twists per meter I would need on 3 phase 50Hz circuit (all Tri-rated singles) to negate or reduce its EMI; and what the reduction would be, depending on the number of twists used?

Thanks in advance.

 

[berkeman]

Power cables are not usually twisted that I know of. Data cables (like Ethernet, USB, etc.) are always twisted. That is where you get the immunity, not from trying to twist the power cables. Data networks should be pretty immune to power frequency noise anyway. If there is high frequency noise on the power cables (from switching power supplies, etc.), that may be more problematic.

Are you experiencing some problems?

 

[Dinoduck94]

Power cables are not usually twisted that I know of. Data cables (like Ethernet, USB, etc.) are always twisted. That is where you get the immunity, not from trying to twist the power cables. Data networks should be pretty immune to power frequency noise anyway. If there is high frequency noise on the power cables (from switching power supplies, etc.), that may be more problematic.

Are you experiencing some problems?

I'm not experiencing any problems, just as stated above - I'm thinking about a larger system to the one I have, if the Induced Voltage in a data cable is too much then twisting the AC cables together should reduce that - and I want to know the formulas for calculating that.

The theory on data cables is that twisting reduces the radiated emissions caused by differential mode signals, my thinking is that the wave forms of a 3 phase AC circuit would mimic differential mode signals at certain areas of the waveform and such the radiated emissions could be reduced from twisting.

 

[berkeman]

The theory on data cables is that twisting reduces the radiated emissions caused by differential mode signals

No, that is not why data cables are twisted. First, radiated EMI comes from common-mode RF currents flowing on the cables (especially if the length of the cable is a multiple of a quarter wavelength at the RF noise frequency).

Data cables are twisted to minimize the B-field pickup (the induced differential noise voltage picked up by one twist is canceled by the next twist's pickup voltage), and to ensure that any E-field noise pickup is only common-mode in nature (which is rejected by the communication transformers typically used).

The main technique for reducing coupling from power cables to nearby twisted pair data cables is just separation. Often, power cables are run in grounded metal conduit (by code in the US for industrial installations), and the data cables generally are not run inside the same conduit. And power frequency noise coupling is rarely an issue in my experience -- the higher frequency noise that I mentioned from power supplies and VFD motors can be an issue if the data cables are run right next to the power cables, and the comm PHYs don't have a reasonable CMRR.

 

[Dinoduck94]

No, that is not why data cables are twisted. First, radiated EMI comes from common-mode RF currents flowing on the cables (especially if the length of the cable is a multiple of a quarter wavelength at the RF noise frequency).

Data cables are twisted to minimize the B-field pickup (the induced differential noise voltage picked up by one twist is canceled by the next twist's pickup voltage), and to ensure that any E-field noise pickup is only common-mode in nature (which is rejected by the communication transformers typically used).

The main technique for reducing coupling from power cables to nearby twisted pair data cables is just separation. Often, power cables are run in grounded metal conduit (by code in the US for industrial installations), and the data cables generally are not run inside the same conduit. And power frequency noise coupling is rarely an issue in my experience -- the higher frequency noise that I mentioned from power supplies and VFD motors can be an issue if the data cables are run right next to the power cables, and the comm PHYs don't have a reasonable CMRR.

Thank you, at least I've learned something today!

Although, There was an experimetn carried out by California State University where they showed twisting 3phase 50Hz cables drastically reduced the Radiated Emissions - there is just no formulas derived from the data. (http://www.phys.ufl.edu/ireu/IREU2013/pdf_reports/Daniel_Vander-Hyde_2_experimental.pdf)

 

[hutchphd]

The paper shows "drastic" (10 fold) reduction only for standard two wire single phase power. This is not new science..
One of the joys of three phase is its inherent cancellation of dipole emissions. I am no expert here. Does anyone know a good basic reference for this??

 

[anorlunda]

It sounds like some confusion between RF EMI and power frequency magnetic coupling. We do not normally use the word emissions for magnetic coupling.

@berkeman is right, I don't recall twisted power cables. But they can be bundled in a cable with a shield level around the outside. The shield can contribute both tensile strength and electromagnetic shielding.

 

[berkeman]

Although, There was an experimetn carried out by California State University where they showed twisting 3phase 50Hz cables drastically reduced the Radiated Emissions - there is just no formulas derived from the data. (http://www.phys.ufl.edu/ireu/IREU2013/pdf_reports/Daniel_Vander-Hyde_2_experimental.pdf)

That is for a very special situation, where power frequency B-fields were interfering with the Virgo Gravitational Wave Detector. That is certainly unwanted interference, but nothing like what most of us deal with in the MHz-GHz range for traditional Radiated EMI compliance.

This is the frequency range they were testing over (power frequency range of 10s to a couple 100 Hz):