6

I read that the twisted-cable wiring is used for eliminating the common mode noise. As far as I know, in differential signalling the two wires carry their mirrored images so any common mode noise like EMI will cancel at the differential amplifier and the signal will remain since a-(-a) = 2a.

But does that mean using twisted pair wiring makes sense only for differential signalling? If not why? And I read Alexander Graham Bell invented this type of wiring to for telephone wires. But were telephone wires differential signals that time? I'm confused.

JRE
  • 67,678
  • 8
  • 104
  • 179
user1245
  • 3,955
  • 4
  • 37
  • 87
  • 2
    Twisted pair wiring makes sense for many types of signals. If it is not a differential signal, one of the wires can be GND, and the other can be the signal. This will improve signal integrity for signals that must travel over longer wires. – user57037 Mar 04 '17 at 18:08
  • I still don't understand what is the reason using twisted pair in single-ended signalling. What is wrong with this type of wiring for long distances they are also very close each other: http://www.av2buy.co.uk/image/cache/data/2%20Core%20Burial%20Gel-500x500.jpg There is no twisting and I dont think there will be capacitive coupling since they are enclosed in same shell. I guess it is so hard to explain – user1245 Mar 04 '17 at 18:21
  • Two wires close together also have great benefits. Ribbon cable with alternating GND wires works very well, even if the wires are not twisted. I think the main benefit is from placing the wires close together, but I am not 100% sure. To understand this better, you really need to understand transmission lines a little bit. It is hard to explain without going into transmission lines. – user57037 Mar 04 '17 at 18:30

3 Answers3

8

OK so the key is in your question, COMMON MODE NOISE...

Consider the following image..

enter image description here

Without twisted pair the line closest to the noise source gets more noise than the other line. So the noise is no longer common mode.

By twisting the wires you negate, or significantly reduce this effect.

Telephone systems are, or were, all transformer based, which is effectively differential.

So what of single ended..

Well arguably, there is no such thing. You are always passing the difference between something... but I won't open that can of worms...

enter image description here

In a simple system, you are not only passing the signal but also passing a reference point... usually we call it ground.. Again. if there is more noise in the signal line than the ground line, that gets picked up by your receiver.

Of course, that all goes out the window if there is some other ground paths or the grounds are shared with other signals.

Ultimately, it depends on the application. If it's single ended... where there are only two wires connecting between each device, and the receiver is battery operated or powered from the communication line, twisted pair helps a lot. If there are multiple connections or both the transmitter and receiver signal grounds are grounded locally to the mains... not so much.

In fact, in some instances over-use of twisted pairs can make things worse...

For example, running single ended signals through twisted pair ribbon cable is generally a bad idea.

enter image description here

The fact that you are bringing your signal wire closer to the neighboring signal wires increase the cross coupling between them. It is better to use a standard flat ribbon cable with ground wires between each signal acting as a kind of shielding.

Trevor_G
  • 46,364
  • 8
  • 68
  • 151
  • I saw some guys twisting the wires when they use single-ended signals. Are they wasting their time? How does twisting have a good effect when the signalling is single ended? First part amazing explanation but I still dont get how in a single ended system twisting has a positive effect. Ground is always ground what ever the noise imposed on it. I would be glad if you can expound on this. – user1245 Mar 04 '17 at 17:46
  • :) I probably added that answer while you typed that @user134429 – Trevor_G Mar 04 '17 at 17:47
  • I actually read it before but I dont get the conclusion here. Better twisted pair even it is single-ended? – user1245 Mar 04 '17 at 17:49
  • It depends on the application. If it's single ended like a phone... where there are only two wires connecting between each device, and the receiver is battery operated or powered from the line, twisted pair works great. If there are multiple connections or the receiver is grounded to the main line... not so much. – Trevor_G Mar 04 '17 at 17:51
  • thanks but why it helps a lot? i cannot picture it – user1245 Mar 04 '17 at 18:02
  • To put it simply. If you have a 5V signal, and if 1V of noise gets added to the signal AND 1V of noise gets added the ground reference line.. your receiver still sees 5V between the signal line and the reference. – Trevor_G Mar 04 '17 at 18:07
  • Ok let me put it like this with two cases where both cases carry single-ended signals: "Case 1: No twisted pair. 5V signal and GND are both exposed to 1V noise. 5V becomes 6V, GND becomes GND so the signal we see becomes 6V." "Case 2: Twisted pair is used 5V signal and GND are both exposed to 1V noise. 5V becomes 6V, GND becomes GND so the signal we see becomes 6V." .... What is the difference between these two cases? Are you saying GND will become 1V. But GND is normally earth-grounded – user1245 Mar 04 '17 at 18:17
  • That is ONLY true if both ends are grounded LOCALLY. If one end is a little grey box with a battery in it.. it's "GND" is the signal GND. Basically.. the whole box will float up 1V if you like. – Trevor_G Mar 04 '17 at 18:21
  • But that has nothing to do with being twisted or not twisted. – user1245 Mar 04 '17 at 18:23
  • yes it does. if they are not twisted.. the signal may get 1.25V and the "GND" may only get 0.75V of noise if they are not twisted. = 0.5V signal error. – Trevor_G Mar 04 '17 at 18:24
  • The whole point of twisting is to try and make sure both wires pickup the same common mode noise. – Trevor_G Mar 04 '17 at 18:25
  • BTW.. In truth the receiver doesn't need to be isolated from local ground, but the incoming "GND" reference should be, and the receiver circuit biased to that point. – Trevor_G Mar 04 '17 at 18:26
  • I think I got what you mean. Sytem is single-ended but on signal part floating like a battery. And the reciever's GND is local ground or earthed. In this situation it makes sense what you say. And if they were sharing same local ground twisting wouldnt have affect but this time there would be some ground loops. I hope I understood now. – user1245 Mar 04 '17 at 18:30
  • :) Good man.. you got it. As I say, the receiver does not have to tie the signal ground to local ground. You can design your op-amp or whatever to be biased to the incoming ground instead. That is treat it like a signal. But it gets complicated with multiple signals and grounds. – Trevor_G Mar 04 '17 at 18:39
  • 1
    As an aside, the voice signal in a POTS system isn't single-ended, but floating. DC is provided via a transformer or twin-coil relay, and possibly a baretter. This circuit was referred to as a transmission bridge. – peterG Mar 04 '17 at 18:53
1

Even for single-ended signals, the twist means that magnetic coupling from adjacent wires will cancel, as every couple of inches it will reverse polarity.

Where a number of pairs will be in close proximity, they will often be arranged to have a different twist rate (number of inches per twist) to minimize coupling. (Wikipedia - twisted pairs)

As you say, capacitative coupling will not cancel if the signal is not differential but that is less of a factor at low frequencies with low impedances as used with phone lines.

Dave Tweed
  • 168,369
  • 17
  • 228
  • 393
Kevin White
  • 32,097
  • 1
  • 47
  • 74
  • 2
    This is potentially mistaken - twisting only works for cancellation if the *return* current (or other mirror image) is paired with each signal. That may or may not be true in a single -ended system - it depends on the details of connection at each end. – Chris Stratton Mar 04 '17 at 17:31
  • I copy my comment here: I saw some guys twisting the wires when they use single-ended signals. Are they wasting their time? How does twisting have a good effect when the signalling is single ended? I still dont get how in a single ended system twisting has a positive effect. Ground is always ground what ever the noise imposed on it. I would be glad if you can expound on this – user1245 Mar 04 '17 at 17:50
1

Here's the key: it has nothing to do with the voltages on each line, and everything to do with the current path of the signal.

A differential driver pushes as much current out one lead as it takes in on the other. That means the current is balanced.

Why is this important? Since the net current is zero, the differential receiver does not need to sink or source any current --- so it does not need a low impedance path between any wire of the transmitted signal and it's local supply rails.

This means that the two wires form a loop that is (mostly) isolated from the rest of the system and any EMI impinging on that loop will only induce current in that loop, and not in the loop formed by balanced pair and any ground connecting the system.

When you use single-ended lines, you can't say what path the return signal will take: will it go through the ground wire of the signal? Through the ground wire of another signal? Through the common ground between the two systems? It's a complicated mess, and EMI will induce current both in the signal wire and in every other connected ground between the two systems such that the induced ground currents are equal and opposite to the induced signal current. But since the loop formed by the signal wire and any ground wire that isn't twisted around the signal wire isn't twisted, there will be little cancelation.

Now, if you have two systems that don't otherwise share a common ground, and there is only one signal wire and one ground wire running along a twisted pair between them, then all the current doing down the signal wire must run back along the ground, and this is, in effect, a differential pair, even if you didn't use a differential driver to create it. In this case, twisting the pair will absolutely help. But the moment you ground the two systems to each other, you form a ground loop and the benefit of the twisted pair vanishes.

user310340
  • 11
  • 1
  • "When you use single-ended lines, you can't say what path the return signal will take". Sure you can. In the case of AC signals, above say 20 KHz. the return path will be that path minimizes the loop inductance of the current path. In the case of low frequencies or DC, the return path will be the path of least resistance. – SteveSh Mar 29 '22 at 23:51