Twisting wires

[mikerascally]

Hi,
I am a mechanical engineer...thus the question.

I know that when you twist wires together you are protecting from EMI and RFI. My question is...if you do not twist can these two interferences or any other for that matter actually create/induct a current in the wires?

Thank!
Mike

 

[IRstuff]

The twisting does not prevent induced currents. The object is to make the induced current the same in both wires. Typically, the two wires would represent a signal and its return. When the noise is equally induced in both the signal and its return, a properly designed receiver will ignore the common-mode noise and correctly extract the desired signal. The degree of twist drives the degree of common-mode coupling, BUT, it also increases the wiring capacitance, so there are limits to how much twist is allowable. A typical value might be 7 twists per inch.

TTFN

 

[richs]

Hiya-

Yep. You sure can. Also, by having the wires not twisted
AND not together, you can have the conductors be influenced
by different changes in magnetic flux. Differences in
changes in magnetic flux will cause a potential difference
between the conductors (Faraday's law). If both conductors
are in close physical proximity, then the changes will
affect both conductors and raise or lower the potential to
an external datum point.

You might have seen some "differential" amplifiers in
circuits that tend to reduce the raising or lowering of
this "common potential" difference. The term used is
"common mode rejection".

While possibly not "quite" as good as twisted pair, having
the conductor pair in close parallel phyisical proximity
will do just about the same.

A "better" method of shielding this is to enclose the
wire or wires inside some shielding material. This can
be anything from foil to braided shield, to hard conductive
pipe (hardline) depending upon the engineering contstraints
and requirements. This shield is then connected to a
ground point or points if it is part of the conducting
pair.

These issues are usually for signal or power supplying
wire pairs close to signalling wires. Or the reduction of
RFI in wires.

Besides, they look neater.............(sorry).

Hope this helps. There is quite a bit on "electrical
shielding" on the internet these days.

For a "light" overview, and I am NOT affiliated with this
site, you could point your browser to:

http://www.engineersedge.com/instrumentation/instrumentation_table_content.htm

Which I found myself while googleing to answer your
question.

Hope this helps.

cheers,

Rich S.

 

[jimkirk]

Another tidbit...If you have several parallel runs of twisted cable, twist them at different pitches to help reduce crosstalk. If you take apart a CAT5 UTP cable, you'll find that the twists per inch is different for each pair. Twisting them at the same pitch can set up some interesting resonant peaks of crosstalk frequency.

As IRstuff said, the object is to induce the same signal into each wire. Twisting helps force the wires to the same overall geometric center, so each sees the same interference, up to around the wavelength of the twist.

There are even higher order solutions involving braiding 4 or more wires in such a way that works even better than simple twisting. And one reason you use coax cable is that the inner and outer conductors do, in fact, share a common geometric axis, hence the name, co-axial.

 

[itsmoked]

I'm having some difficulty with the above assessments about twisting. It seems that I am reading that it doesn't reduce coupling it "just evens it out" !!! This doesn't fly with my understanding. (even vaguely!).

What twisting does is "cancel" the inductive coupling that is occuring.

How? Just draw a stylized picture of a twisted pair. Then draw a B-field arrow into adjacent loops. You will see that the induced currents are canceled by alternate twisted loops. This is the entire point of twisted pairs and reduces inductively coupled noise.

Capacitive coupling is dealt with by the outer shield which does nothing to combat inductive coupling unless it is steel or iron, (not a normal cable material) (Inside a steel conduit would work well however!)

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[jimkirk]

Hi itsmoked,

You are correct that the current is canceled by alternating loops, as long as the wavelength of the B-field is very long compared to the twist pitch. But for a non-uniform B-field the canceling is not perfect. This is the reason for varying the pitches of adjacent cables.

If I have a B-field that changes directions with each twist of the pair, what happens? The situation is helped by the fact that twisting forces the loops to be relatively small, but other than that there will be little cancellation.

For a co-ax cable yes, the shield deals with capacitance coupling (acts like a shielded cable), but the geometry (and magnetic properties) deals with inductive coupling. An off-center inner conductor is still protected from capacitive coupling, but keeping the center conductor co-axial with the shield will reduce the effects of inductive interference. (All assuming adequate common mode rejection of the receiver.)

Another approach I've seen for differential signals on PCBs is a quad of traces, using two layers, with no ground plane near, something like

(+) (-)
x
(-) (+)

where "x" denotes the geometric center of each pair. Since each pair has the same geometric center, each pair sees essentially the same interference, so there will be common mode interference, but little differntial mode. Again, this assumes the interference varies little over the geometry of the traces, and adequate CMRR.

Also, this may or may not be better than running a single pair in close proximity to a ground plane, depending on the nature of the interference.

 

[VEBill]

ItSmoked: "What twisting does is 'cancel' the inductive coupling that is occuring."

Yes for differential. No for common mode.

 

[richs]

Hiya-

"Yes for differential. No for common mode."

Against a common datum point (i.e. ground) yep. That's
why we have diffential inputs!

 

[VEBill]

It's worth keeping the common mode coupling in mind in case you run your twisted pair beside, for example, an arc welding cable and you run into the practical limitations of differential inputs.

 

[mikerascally]

Wow! This whole electrical area is way out of my radar.

So in Mechanical Engineering layman terms....

If I twist unshielded wires I will stop the creation of unwanted current from an external source be it EMI, RFI, etc.. to pass through the wires?

In theory, If I have a battery/switch on one end of 100 meter twisted wires and a light bulb on the other end. If I bombard it with EMI,RFI, etc... it will not light...unless I flip the battery switch?

And if it was untwisted it would?

 

[VEBill]

Putting it simply, twisting the wires helps to make sure that both wires see the same EM environment. There is no difference between them. Thus there is very little EM-generated voltage difference between them. Your light bulb should operate normally.

On the other hand, if the bulb was a florescent tube mounted to a grounded metal beam, then it might flash with nearby HF transmissions due to common mode coupling, even with twisted wires. Common mode coupling isn't helped with twisting.

In other words, there's always exceptions. Looking for an all-encompassing simple answer is not the best approach.

 

[mikerascally]

Ok, so if I have something else than a bulb at the end...for example a solenoid that requires a low voltage to activate...would EMI/RFI influence be able to activate it if the wires are not twisted?

 

[MikeHalloran]

A solenoid with a low voltage requirement would probably have a high current requirement, and would not be particularly susceptible to interference. It would probably be a strong source of interference.

Perhaps a better example of a susceptible circuit is an audio input cable, say between preamplifier and amplifier. Twisted pair is usually not sufficient protection for such a circuit; you will commonly find twisted pair inside a foil shield or a braided shield or both.

Another practical example is USA domestic telephone premises wiring. If your house was erected when there was just one phone company, it probably has 'quad' wire, which is twisted only by accident. The extra pair of wires was intended only to carry current to light Princess phones. If you try to hook up a second line using the unused wires, you will be able to hear conversations from the other line, and incoming calls on the other line may cause your modem to hang up. Newer premises wiring is done with four twisted pairs of smaller wire, so you can have four lines and little crosstalk at audio frequencies.





Mike Halloran
Pembroke Pines, FL, USA

 

[pauljohn]

A point which may or may not be valid to this thread.

If you do not twist wires that supply the light on a ship's compass or that even pass nearby you will definately affect the accuracy of the compass. I found that out the hard way!

Paul
(from somewhere in the atlantic)

 

[itsmoked]

If you only knew where..

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com