In differential mode transmission, two wires carry equal and opposite main signals and the receiver detects the difference between the two. Noise sources introduce signals into the both wires equally (not opposite), so the noises thus produces a common-mode signal are canceled at the receiver when the difference signal is taken.

Yes, but the idea of a balanced line (two lines in parallel) is the same... The question needs to explain how and why "noise sources introduce signals into the both wires equally ".

You have said this arrangement needs a differential amplifier. But if there was such a sophisticated amplifier in 1881?

Regards, Cyril

Noises are generated in signal lines by magnetic fields from the environment. So the noise current in data lines is the result of that magnetic field. In the straight cable, all noise current is flowing in the same direction, just like in an ordinary transformer coil. When the cable is twisted, in some parts of the signal lines the direction of the noise current is the opposite from the current in other parts of the cable. Because of this, the resulting noise current is many factors lower than with an ordinary straight cable.

That is, each twist of the lines generates a loop, where the current generated from external fields coming from the source and returning from the load will "rotate" in an opposed sense to that in the next loop. Thus, the contribution of each loop cancels the other (more or less). Thus, the influence of external magnetic fields is greatly reduced.

Dear Rodrigo,

You made me deeply think about your explanation... Sorry, but I can't agree with you... IMO there are two different problems with this arrangement and there is no connection between them...

The first is to make the two noise voltages (caused by the same noise source and simultaneously induced in the two conductors) subtract. This problem is solved in both the arrangements (the straight and twisted cable) by the differential signaling (not by twisting).

The second is to equalize the two induced noise voltages. This problem is solved by twisting the two conductors and the question is how the twisting solves this problem.

Regards, Cyril

Vasile, sorry but I can't understand why "twisted wires acts as low parasitic capacitance shielded cable, while balanced not-twisted line does not". A twisted pair cable is not obligatory shielded (or maybe, you want just to say that it is equivalent to but actually is not a shielded cable?) I would like also to add that IMO the twisted pair has a significant internal capacitance between the two wires (I remember how hobbyists from the early 20's century made small "capacitors" by twisting two insulated wires).

Yes, I remember the ubiquitous Russian TEMP and RUBIN... and, of course, the sophisticated German STADION-:) But now I realize what were these two parallel wires coming from the aerial-:)

Regards, Cyril

if I remember correctly, when differential amplifiers weren't available the twisted pair was simply connected to the primary of a transformer, which had the "primary" purpose of impedance matching ..and a secondary purpose of blocking common mode low frequency (or DC) signals.

As for the problem of electromagnetic susceptibility the tight twisting of the pair can be seen as many small solenoids connected in antiparallel, where the differential noise voltage by one loop is effectively minimized by an (almost) equal voltage of opposite sign induced in next loop. Imagine the simple case of a single twist in a pair and this will become apparent. This holds true, of course, as long as the "twist" spacing is much smaller than the minimum noise wavelength you're concerned with.

Of course radiated and capacitive couplings can still induce common mode voltages, but these can be taken care of easily by your differential amplifier CMRR and a wise usage of ferrite rings..

As though there is some contradiction here - we want to reduce the electromagnetic noise induced in the line but... we have wound the wires to increase the induced noise?!?

So, I continue thinking that twisting is море equalizing than compensating...

Johann, thank you for the profound and comprehensive answer. You have considered the question in all its aspects and finally you have shown very precisely the main idea: the two twisted wires form a homogeneous structure in which each wire is exposed equally on average to the noise radiation.

This simple but genius 19th century Bell's idea makes the impossible possible - as though it places the two wires at the same place. If we generalize this powerful idea, we can see (and apply) it in our routine in situations where we have to expose two long or large objects to an equal radiation. To solve the problem, we divide the objects into many equal parts and intermix (shuffle) them. For example, imagine a new building with two owners that have to share out (divide in half) all the apartments, and they want to be in equal conditions. A possible solution is to divide and arrange the apartments in a chess-board order...

Regards, Cyril

The idea of using a twisted pair, other than more mechanical stability, relates mostly to partial cancellation of magnetic field around the wire since they are in a loop and their current directions are opposite. Therefore, the reduction in magnetic flux reduces the self-inductance of the transmission line. An observant reader is invited to have a look at "Bifilar winding" to make small capacitor(with less inductor!).

Masoud and Vasile,

I continue thinking that the two techniques solve different problems - the balanced pair solves the problem of compensating while the twisted pair - of homogenizing. Actually, the twisted pair solves both the problems; it is a "twisted balanced pair"-:)

Imagine you are in the place of Bell and you have to solve the problem of induced noises. So, you first "balance" the line (place the return wire in parallel to the direct one) and second, you "homegenize" them (by twisting)...

Regards, Cyril

Yes, Cyrill, I also think that Johan brilliantly answered. I also liked your chessboard example of trying to explain it. I had to explain this to my students in networking once by drawing something similar to the attached Figure. In Figure 1, for the case of parallel pair, the interference causes the red wire (the closer to interference source) more pick-up voltage (induced) per unit length (1 mV as example) while less induced (0,5mV) in the blue wire. Total difference at the destination is 3mV. While in the twisted pair case (Figure 2), total difference is 0V at destination because parts (twists) of red and blue wires alternatively subjected to same level of interference and thereby total difference at destination is 0V. I drew this figure for this question hoping to use it in lecturing as well. This is especially useful when teaching networking to non-electrical engineering students who don't recognize terms of impedance, differential mode noise terms, ...etc . By the way, interference in twisted pairs comes mainly from signaling on other pairs running together within the same cable which can have many of them. Thanks. @AlDmour.

I like Aldmour's Figure1 showing the purpose of twisting the wires and how it works. One picture has one thousands words. The idea comes from the problem technicians run into in late 1880's when phone lines and power lines were becoming more noisy due increasing use of electricity. So they did a wire transposition (twisting cables) on every pole or some other distance to equally expose them to electromagnetic field coming from outside.. Then they transferred that method to electronics and equipment.

The second figure given by Ismat Aldmour is called "Transposition of Conductors" in Power System Transmission Lines and is used to eliminate Mutual Induction between the two conductors.

P.S.

My colleagues, I am already here (I was absent a few days)...

Ismat, thank you for the expressive picture and the persuasive and powerful explanations; you moved me! After that, there is no doubt about this genius but so simple idea - the twisting simply equalizes the noises induced in the two wires of the differential pair. This is an excellent implementation of the passive compensation idea (it deserves a separate question).

Adam and Pisupati, thank you for the valuable remarks; I completely second them.

Best regards, Cyril

Cyril,

Am involved in a CCTV ( closed circuit television ) system

and have just switched from coax to CAT cable.  

CAT-5 has a looser winding than  the CAT-6 cable.  

Had been looking for some good discussion on the reasons

for using this twisted-pair cable.  

Your other thread provided good information. 

Glen

You can't use twisted pair cable instead coax cable. But you can use special coax to twisted pair converter and twisted pair to coax converter. It can be buy for some dollars. This device enables any equipment with coax interface to operate over twisted pairs. For example, Super-Simple Coax to Twisted-Pair Adapter proposed by Linear Technology can be see there: 

http://www.linear.com/solutions/1309

http://cds.linear.com/docs/en/lt-journal/LT6552_0204_Mag.pdf

http://cds.linear.com/docs/en/design-note/dn65f.pdf

Similar circuit suggested by Texas Instruments:

http://www.ti.com/lit/an/snoa422c/snoa422c.pdf

Vasile,

Thanks for the engineering advice.

Am using Coax-to-CAT BalUn Toroid Converters. 

Linear runs are 300 ft.  

Using 720 analog cameras. 

Image is really OK in this application.  

Longer runs will be 700 ft,

and one amplifier is indicated for each CAT cable,

to drive a signal stronger.  

I still expect hi-resolution loss of image, but will see. 

Digital is not an option.  

All installations I have seen here

have been CAT-6, with both types of cameras.  

Eligiusz,  

Right on.  

Thanks for the engineering advice.

That is what my research indicated.

Here, Coax-to-CAT BalUn Toroid Converters are 

off-the-shelf device for $4 (US). 

Two required for each twisted pair run. 

Am using 3 cameras and one power per CAT cable. 

The phase-shift comparison was very obvious

when looking at the Oscope pictures. 

.  

I haven't seen crosstalk cancellation mentioned. It's not just outside interference, which for the most part creates a common mode current that any balanced interface will cancel. It's also, or rather primarily, crosstalk. And the right hand rule applies. Current on one conductor will create a magnetic field around that conductor. That magnetic field induces a current in the other conductor(s) of the cable.

Using your right hand, stick out the thumb and curve the fingers. The thumb points in the direction of current flow. The bent fingers are the magnetic lines of force, surrounding the conductor. As the current varies, so does the magnetic field, and it therefore creates a crosstalk current in the adjacent wires.

A tight twist will continuously cancel these induced currents, as the magnetic field created by the current in a conductor will cancel itself out, as the wire twists.

Use your right hand. See how these magnetic lines of force will continue to oppose themselves, as your thumb goes up and down, to point in the direction of the current going through the twisted conductor.

Now, obviously, if you want complete cancellation, you'll never get anywhere with that cable. But this is the idea. And it explains why voice grade twisted pair, cat-3, and cat-5, are progressively more tightly twisted.

By the way, I want to say that the explanations one often sees are incorrect, in my opinion. Any common mode induced current, from outside EM fields, will change very little indeed, whether the two pairs are twisted or not.

First, the outside EM field is not cancelled by the twists. How can it be? Whatever current is induced along the length of the cable is induced in the SAME direction, on each wire of that cable. The twists have no way of making the induced current opposite, in each wire of the twisted pair.

Assume a single external source of interfering EM.

It's true that by twisting the conductors, some short sections will experience less or even no induced current, while other short sections will experience more induced current, depending how the twisted tiny section of wire is oriented with respect to that EM source. But how exactly would the induced current be opposite, from one wire to the other? It cannot be. The external field impinges on both wires exactly the same way, on average, over the length of the cable. Therefore, the induced current will be the same, on each wire.

The other explanation might be slightly more valid. Which is, the amount of induced current in each wire of the pair is more identical, if you twist the wire, because the distance from each wire to that external noisy EM source will average out to be more exactly the same.

Well, you know, maybe. More likely, that effect will be negligible. But at least, this second explanation makes it clear that the only improvement, for eliminating the effect of outside noisy EM fields, is that the common mode current is "more exactly common mode."  In a differential interface, it would be more perfectly rejected.

So, in short, I disagree with the "popular answers," but the one popular answer is slightly more accurate than the other!

Albert,

Thanks for your deep reasoning; it has revived the discussion about the legendary pair.

I have noticed a significant difference in our approaches - until you think in terms of currents I do it in terms of voltages. My notion about this arrangement is as a line of two (twisted) wires that is shorted by a low-resistive generator (perfect voltage source) in the beginning, and loaded by a relatively high-resistive load at the end.

So this arrangement consists of two elements in series with opposite voltages across them... and as it obeys KVL, the resulting voltage across the load is zero. There is a voltage, not a current summation here, since the elements are connected in series... and the voltage summer is a loop. If this was a current arrangement, we should sum the opposite currents by connecting the elements in parallel... and the current summer should be a junction.

Now about the role of the differential amplifier in this arrangement. Actually the summation (subtraction) is implemented by the loop according to KVL; so there is no absolute need of a differential amplifier. For example, if the load is floating, it can be with a simple single-ended input.

Imagine such a simple experiment - twist two wires, short them at the one end, and leave them unconnected at the other end. Then approach the pair by an AC-energized solenoid.

What would be the voltage at the pair output? I think - zero. Do any currents flow here? I think - no.

Then, are the two wires generators? I think no since no currents flow through them (open circuit).

I continue thinking that the homogenization is the only role of twisting...

Twisted pair cable reduces the iseries nductance per unit length of the wire.It has reduced inter wire capacitance as compared to 2-wire sysem but Electro-Magnetic Interference will not be reduced. That requires some kind of shield.

The basic benefit of twisted pairs is a more stable impedance, This of course varies with the length, gauge, insulation thickness, and signal characteristics being transmitted.

Contrzry to Mr. Pawlowski's assertion, you can actually use twisted pair instead of co-ax depending on the application. Bear in mind that T-1 telecom signals

(1.54 Mbps) are routinely transmitted along old copper telephone wires (shielded twisted pair) in DSL applications - albeit not very well.

I was hired to work at a telecom start-up in the early 90's and a task was to analyze timing/jitter/BER between two PCB assemblies. The source was a 155 Mbps STS3 telecom signal, with discrete differential clock and data lines. They were using two short twisted pair connections and wondering why there was excessive jitter and BER. They also had poor quality control in that the length of the cables could vary significantly. It's a wonder the system worked at all, but it did as long as the two cables were extremely close in relative length and the absolute length didn't create a transmission line mismatch - as you can imagine they weren't terminated very well either. In this case, a single-ended semi-rigid co-ax interconnection was a significantly better solution. 

Cyril,

This question about Twisted-Pair concepts and applications 

has good application in my current work.  

Thank you for posting this thread about a seemingly simple subject

and giving it more reality and complexity . 

Glen

Cyril,

Sorry, I don't understand your voltage model. Yes, a perfect voltage source has no internal resistance, as you say. So, connect a perfect voltage source to two wires, twisted or not twisted makes no difference, and only if the load at the opposite end is infinite will no surrent flow.

KVL says that the voltages have to sum to zero around the loop. The voltage source has x volts across it, and the other end of the two wires HAVE to have x volts across them too. So, only if the load is infinite will no current flow. If the load is non-zero, and finite, current will flow.

I = V/R, where V is the voltage that must exist across the resistive load, for the voltages around the loop to add to zero. Otherwise, what voltage will oppose the voltage of the generator, around that loop?

(But I think we have strayed off course, perhaps, from the discussion of why the wires need to be twisted. I do find some "explanations" online that are not very believable!))

Hi Albert,

Thank you again for the so interesting way of reasoning - straightforward, honest, creative.... and most importantly - benevolent. This occurs so rarely in discussions... so I really appreciate it...

I suggest to look at the impressive Ismat's picture - it explains the phenomenon in terms of voltages. The fact that we connected a voltage source in the input or simply shorted it does not change the result at the output where the two induced voltages do not appear (because they have destroyed each other).

So, only the useful input voltage will appear at the output and, if the load has some resistance RL, a current IL = VIN/RL will flow through it. If the load has infinite resistance (open loop) no current will flow...

(Maybe the problem is that I can not well explain my notion in English.)

Hi again Cyril,

No, the problem was with the very late (or shall I say early) hour, not with your explanation. I got what you were saying shortly after, while falling asleep.

Yes indeed, I had seen Ismat's picture already, and it was the argument that I thought was closer to the truth, but did not really make the point. Reason being, if the two wires are close together compared with the distance to the noise  or interference source, let's say only separated by the thickness of the insulation, then the difference in distance to the noise source would most likely be insignificant. This has to be the norm, not the exception, I have to believe. Twisting does indeed help create a more precise common mode voltage (and current) between the two conductors, but the most likely insignificance of the improvement cannot be ignored. Parallel conductors, very close together, would produce much the same effect. While it is true that the noise-induced voltage between the two wires adds to zero, this is only a statement of the fact that a common mode voltage is induced!

What truly eliminates the noise problem, when dealing with common mode currents or voltages? A differential input in the circuit which follows. Not the twists in the wire, aside from that possible slight improvement in commonality of mode!

Assume you feed that signal, with common mode voltage, to an unbalanced input amplifier. One wire goes to the ground or common terminal. The other wire goes to the high input impedance gain stages, and each gain stage most likely reverses its polarity, just to make matters worse! Which means that the common mode voltage in the input lead becomes a signal that becomes amplified, a significant noise problem downstream of the amp.

We have to expect that any current argument has to agree with any voltage argument. I'm saying that the twists in the conductors provide primarily a cancellation of the magnetic fields around each conductor, thanks to the right hand rule, which reduces crosstalk, and that the cancellation of common mode voltage (and current) are a property of differential input circuits, not so much the twists of the wires.

Which explains, by the way, why RS-422/485 maximum lead lengths are so much greater than RS-232. The twisted pair is not the reason - they are common to both. But RS-232 is unbalanced, where the other two are balanced circuits.

So can we make the final conclusions below about the twisted pair arrangement?

• Two operations - equalization and compensation, are performed on the common-mode signals in the wires
• They are implemented respectively, by a twisting and a differential input
• The twisting equalizes the common-mode signals; the differential input compensates (rejects) them

Agreed, but not to ignore crosstalk reduction. For example, cat-5e unshielded twisted pair cable has four twisted pair, for transmit and receive signals in twisted pair Ethernet. Crosstalk among those twisted pair must be reduced as much as possible, because it creates noise in the different individual pairs.

The most concise statement I would make is, twisting the conductors is a way to make each conductor cancel out the magnetic field it generates around itself, so as to prevent each conductor from inducing a signal into the other conductors of that cable. Inducing these unwanted signals into adjacent conductors is to induce noise, which becomes worse with cable length. That's the primary purpose of the twisting.

Incidentally, these days I should have to redesign my home local network (for Internet and digital TV). I have already bought 50 m 4Xtwisted pair... and I should work with the crimp tool...

For that matter, Cyril, in our household these days, the majority of TV we watch is from the Internet. It arrives either over twisted pair Ethernet or WiFi!

I've had a PC dedicated to one TV set already, since 2010. And the other TV has a limited-use streaming content box attached to it (it's called Roku, which only provides a very limited view of the Internet, only for a handful of by-subscription TV sites - the unlimited PC is a lot more fun).

I don't know the situation in Bulgaria, but in the US, the TV networks all have an online presence these days. For 10 years, actually, starting in 2006. You can watch their content on demand, online, bypassing both the airwaves or cable systems. Not everything is available this way (yet), but the handwriting is on the wall, so to speak.

Albert,

For a long time I had not checked how many online TV are in Bulgaria... and now I came in Google to see - they are about 50. But this is not the problem (how mane they are or whether they are analog, digital, online...). The problem is that practically I myself do not watch TV... just because I do not like this "violent" way of presenting ... or because the lack of meaningful content... or maybe just because of the age:)

Incidentally, yesterday I managed to drill a few holes in the walls through which the cables will pass. As usual in such cases, the very drilling lasted minutes while moving furniture and next cleaning - hours:) Tomorrow I will do the wiring...

Cyril,

I share the sentiment "the lack of meaningful content",  

ella se asi! 

Aparna,

referring to your comments, 

I use individual crimp connectors ( ACORNS  )

for patching CAT pairs to Coax-TwistedPair baluns

or patching cables together to extend the length.  

From my reading, and from my experience

repairing Radio Frequency cables , 

it seems that the percentage of impedance mismatch

for a 1/2 inch patch ( folded out from the twisted pair end )  

is very, very small compared

to the overall length of the transmission cable

when the Wave-Length is 74 ft and 32 ft long ; 

ie, the effective in-efficiency of the patched connection is neglible.  

In CCTV our vidio signals are less than 6 MHz, or 126 ft wavelength.  

In Radio, I work in the 7MHz and 14MHz range, which is 148ft and 64ft .  

I do am not involved in GigaHertz RF work, 

and I can only imagine that as the Frequency rises 

and the Wave-Length shortens 

that the relative size of a patch grows important

as an overall percentage of losses incurred by various factors.

I do not think that GHz factors are involved

in the relatively low frequency CAT applications.  

All, 

I edited my last post for clarity. 

Glen

The answers above are great. For more information checkout

1) http://usmilcom.com/TwistedPair.htm

or look at the paper (which is a lot more technical)

2) "A twisted-bundle layout structure for minimizing inductive coupling noise"

https://engineering.purdue.edu/~chengkok/papers/2000/p406-zhong.pdf 

The basics of signal transmission explains this in detail,

In wired signal transmittion, the transmitted signal is basically electrical currents and for the current flowing it creates magnetic fields around it and for every action there is a reaction the magnetic field tends to lower the original current (it creates eddy current)

 to eliminate the opposing current (the magnetic field) twist the wire so the opposing current due to induced magnetic field cancel one another, hence better signal transmission or low SNR

Alternatively you can have straight laminated cables but it will reduce cable flexibility and increase cable diameter

This is an important detail - magnetic fields cancel each other, not currents...

Another detail is that they will cancel even if not twisted... but not so successfully...

the data signals are converted to electrical signals to be sent over the twisted pair. Please confirm if this electrical signal sent over twisted pair is ac or dc?

Mainly AC... but it can be DC as well... Then only the induced external disturbances will be compensated.

 If data signals are converted to AC electrical signal to be sent over twisted pair, at what AC voltage does the signal transmission occur over twsited pair. Where i am getting confused is how a 48V DC signal for power over ethernet over twisted is sent over the same conductor carrying data signal in ac waveform?

Thank you. currently IEEE is developing standard to transmit 100W DC over twisted pair cable, where the POE signal will be sent over data carrying conductors. POE also allows DC power to be transmitted over conductors carrying data signal(electrical). Not sure how both data signal(electrical) and POE DC signal co-exsist in the same conductor of a twisted pair?

Probably DC signals are common-mode signals while AC is a differential signal. Also, inductive filters may be used to separate them (but these are only quesses)...

Srinivasan, all you need, to transmit both DC power and pulsed baseband digital over a twisted pair Ethernet cable, is a way at the receiving end to filter out the DC from the pulsed baseband Ethernet signal. So, a low-pass filter is used to retrieve the power content, and a high-pass filter is used to retrieve the digital data content.

However I don't see how this applies to the twists in the cable. They are orthogonal discussions. And I agree with Vasile, that DC power distribution would not necessitate twisted pair. Twisted pair is used for this, simply because that's what we need for the data content, to prevent crosstalk. A DC power component won't create a varying magnetic field around the conductor, so it won't induce currents in adjacent conductors.

Data and power use the same pair for 1000BASE-T and beyond, where all 8 conductors of the Cat-5e cable are required for data transfer. In slower Ethernets, where only four conductors of the Cat-5e are used for data, the spare four conductors are used for power. So that's an easy design!

Thank you so much,  Vasile, Albert and Cyril.

thanks, useful information...

But more practically, twisted pair made it easier to run cable, especially through pipe that was laid underground and through buildings. With one twisted pair you could essentially run one stiffer cable rather than two floppy wires. The twist created strength. There are even apocryphal stories that claim that rats were trained by AT&T to run through long pipe runs with the end of the twisted pair cable in their mouth to be able to make long cable runs.

Interesting...

For AC or DC, it is the same reasons as very well explained before, to eliminate noise from induction. Also, twisted pair cables are more stronger. In few cases, Hydro-Québec (in province of Québec, Canada), they are using the ground wire (located at the top of the 735Kvolts power lines) to get power from it, and they converted down with transformer to 240volts AC. On a scope, it is almost sinusoidal.

Following..

Indians also used a mixed color thread in their worship and tied it to in hands where the red color is in dominant in compare of yellow and white.

Thanks. Good answer.

One can only add that the 19th century is a century of amazing discoveries.

Some great insights here. Thanks to all the contributors!

Twisted pairs are used to both minimize the pickup from external electromagnetic fields as well as to minimize the radiation of signals from the wires to external devices. This occurs because the coupling between the wires and the external world are reversed with each half turn of the twisting. Thus, if the signals are approximately constant along the length of the twisted pair, the signals are largely cancelled out.

Thanks for the clarification. You are absolutely correct. My answer assumed that interference was nominally same along wire and, as you point out, at high frequency, phase change along length of wire could be very significant- especially if at odd multiples of 180 degrees per length of twist.

Twisted cables are more suitable for cancelling magnetic effects created by high current. Parallel cables for low current, easy to remove line noise or interference. Filter of abnormal signal easy at destination. Twisting of data cable may not save from interference as current flow in each wire depends on data but can be filtered out from minimum limit for high data. Twisting of number of conducting wires is to improve strength of wire.

By FFT you may get result.

The main idea for twisting was to cancel the generated electromagnetic interference around the wires.

Two wires are twisted to get proper adjustment so that no space could left around the wire to generate disturbance due to magnetic field.

The twisted pair consists of separate links. When inducing electromagnetic interference in these links, they are directed counterclockwise. Therefore, they are mutually compensated.