0
\$\begingroup\$

I think in some Communication Protocols it is sometimes recommended to twist the signal wires. Why and in which Protocols or Applications?

\$\endgroup\$
  • \$\begingroup\$ Where did you read that for I2C? \$\endgroup\$ – Majenko Sep 2 '16 at 13:25
  • \$\begingroup\$ @Majenko: nxp.com/documents/application_note/AN10658.pdf talks a lot about sending I2C over a few twisted pairs, even though I2C is really intended peripherals wired to processors on the same board. :-) \$\endgroup\$ – davidcary Aug 15 '19 at 3:55
2
\$\begingroup\$

Why in some Communication Protocols the Wires are twisted?

It's called twisted pair and it's a method of reducing EMI from the cable and making it more resilient to EMI that it receives.

Consider the emission scenario - if the wires are side-by-side (in a plane) then there will be a measurable far-field magnetic disturbance (due to wire signal currents) all along the plane of the wires until you reached the exact centre between the two wires. At the centre, the fields will be equal and opposite.

By twisting the wires you are repeatedly swapping the conductor positions and thus the far field magnetic field is significantl reduced. Consider these three scenarios: -

enter image description here

It's the same for the electric fields too providing you are driving differentially. They even do it on old overhead telephone wires to prevent crosstalk to other subscribers: -

enter image description here

For example in I2C: SDA and SCL?

If you mean why are SDA and SCL twisted together then this answer doesn't cover that and I can't explain why there would be an advantage.

\$\endgroup\$
  • 1
    \$\begingroup\$ I2C is not really designed for off-board communication, so why there would be someone recommending twisting wires that shouldn't even really exist is anyone's guess... \$\endgroup\$ – Majenko Sep 2 '16 at 13:32
  • 1
    \$\begingroup\$ In fact twisting I2C SDA/SCL is probably the worst thing you could do in terms of signal integrity - they are already weak signals and any capacitive coupling from twisting will wreak havoc. \$\endgroup\$ – Tom Carpenter Sep 2 '16 at 14:17
  • \$\begingroup\$ @TomCarpenter twisting doesn't really increase mutual capacitive coupling over a close parallel run. What it does is gives a differential pair a similar experience of external interference, which can then be removed as common mode noise/interference. \$\endgroup\$ – Chris Stratton Sep 2 '16 at 14:26
  • \$\begingroup\$ @ChrisStratton, I2C has not differential component, so common mode noise is destructive regardless of whether you twist the SDA with the SCL or not. But passing I2C through a twisted pair cable vs. two seperate wires that aren't closely coupled will be a higher mutual capacitance. I'm thinking from a practical point of view where a twisted pair cable is specifically used instead of keeping the two signals separated. \$\endgroup\$ – Tom Carpenter Sep 2 '16 at 15:01
  • \$\begingroup\$ @TomCarpenter - the key point is it is not the twists themselves, it's the closely parallel run, so it is really only a difference if you would otherwise not have a closely parallel run. \$\endgroup\$ – Chris Stratton Sep 2 '16 at 15:07
2
\$\begingroup\$

I'm not sure why you would do it for those two nets in I2C, however twisting the two wires together is an important technique for differential signals, such as USB or RS-485.

Keeping the two wires twisted together ensure that the two wires are exposed to the same noise, which then gets cancelled out when the signal is processed (if you want more information on differential signals, there are some good pages online, Wikipedia gives a good introduction here)

Another reason to twist wires together is to give the signal line a reference to work to, by twisting a ground line in with the signal line (which is effectively what is happen with differential signalling, only the reference is the other signal line rather than ground). A reference is very important when dealing with high frequency signals ("high frequency" depends on the size of the signal loop you're looking at). A good reference path will improve signal integrity and reduce EMI.

Again, you could twist the wires together to reduce in the induction seen in the path, which would allow for cleaner and faster signalling.

Yet another reason is convenience: having two wires twisted together means they'll move as one when you're moving the unit around, which can make your life easier.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.