Electromagnetic interference

[Vev]

I saw this on the EAA site and thought it was a very nice experiment to show one method of reducing electronic interference.

 

http://www.eaavideo.org/video.aspx?v=595153256001

 

Cheers

 

Vev

 

 

[Alpi]

Thats a great bit of practical information right there.

 

Much appreciated.

 

 

[Guest davidh10]

The wires don't actually have to be twisted to have the cancellation effect, just be run parallel and close together in respect to whatever may be interfered with by the field.

 

Some may remember the old aerial telephone lines on poles. many wires on multiple arms, run in pairs about 10cm apart. These too exhibit the same cancellation effect, but in the case of having multiple pairs on a cross-arm, and the wires about 10cm apart, there was a net effect between the closest wires of adjacent pairs. The solution, to put a single twist at each pole, so that the opposite wires of each pair are alternately adjacent.

 

You may ask, why were the wires about 10cm apart instead if closer. The answer is in transmission line theory. The impedance of the lines was 600 ohms (not DC resistance, but impedance seen at the operating frequency). In order to get maximum power transfer, the transmission line must be the same impedance as the source and the load, otherwise you get standing waves, power loss and radiation from the wires.

 

In multi-pair cables, not only are the wires twisted as pairs, but the pairs are twisted into bundles in such a way that the same pairs are not running adjacent continuously. The bundles are then twisted in a like manner.

 

This whole field cancellation thing only occurs if the current in both wires is equal, opposite and in the case of alternating current, in phase. So if you are using this to try and cancel the RFI emanating from wires supplying a high power LED lamp, for instance, you need to run both the supply and earth return wires all the way from the battery to the light. You cannot use a chassis return and expect the same effect.

 

You also need to avoid creating "earth loops". A situation where a circuit is earthed in multiple places and the current can thus take multiple paths. Anyone working with stereo systems and interconnection of amplifiers and audio sources will know that earth loops can introduce significant "hum" into the system. This applies to the shield on shielded cables as well, not just to the active wires. Shields should only be connected to the aircraft chassis at one point. This may mean that the distant end must be left unconnected and simply insulated.

 

 

[Vev]

The wires don't actually have to be twisted to have the cancellation effect, just be run parallel and close together in respect to whatever may be interfered with by the field.Some may remember the old aerial telephone lines on poles. many wires on multiple arms, run in pairs about 10cm apart. These too exhibit the same cancellation effect, but in the case of having multiple pairs on a cross-arm, and the wires about 10cm apart, there was a net effect between the closest wires of adjacent pairs. The solution, to put a single twist at each pole, so that the opposite wires of each pair are alternately adjacent.

 

You may ask, why were the wires about 10cm apart instead if closer. The answer is in transmission line theory. The impedance of the lines was 600 ohms (not DC resistance, but impedance seen at the operating frequency). In order to get maximum power transfer, the transmission line must be the same impedance as the source and the load, otherwise you get standing waves, power loss and radiation from the wires.

 

In multi-pair cables, not only are the wires twisted as pairs, but the pairs are twisted into bundles in such a way that the same pairs are not running adjacent continuously. The bundles are then twisted in a like manner.

 

This whole field cancellation thing only occurs if the current in both wires is equal, opposite and in the case of alternating current, in phase. So if you are using this to try and cancel the RFI emanating from wires supplying a high power LED lamp, for instance, you need to run both the supply and earth return wires all the way from the battery to the light. You cannot use a chassis return and expect the same effect.

 

You also need to avoid creating "earth loops". A situation where a circuit is earthed in multiple places and the current can thus take multiple paths. Anyone working with stereo systems and interconnection of amplifiers and audio sources will know that earth loops can introduce significant "hum" into the system. This applies to the shield on shielded cables as well, not just to the active wires. Shields should only be connected to the aircraft chassis at one point. This may mean that the distant end must be left unconnected and simply insulated.

Hi D10,

 

I'm often surprised by the knowledge on these forums by the members ... thanks mate!

 

Cheers

 

Vev

 

 

[Guest davidh10]

No worries Vev. That's the point of, and the value of these forums. We all have different backgrounds and experience. Collectively, thousands of years of experience; benefits us all.

 

I have also read lots of interesting things here including some very informative fuel posts by yourself.

 

I'm committed to life long learning.

 

 

[Guest Andys@coffs]

David's point about the cable not needing to be twisted is valid, however Im concerned that the inference people draw from that is that twisted is no better than straight and real world conditions show that twisted is generally (thats not an absolute, with EMI rarely is anytrhing absolute!!)much better than straight where the EMI source is close to the wire such that the difference in distance from the source to each of the parallel lines is different. David did cover that in his first paragraph but I suspect that for most people it will have slipped through to the keeper. if the distance between the 2 lines as a percentage of the distance to the source of EMI is high then twisting wont probably help. if, as in ethernet where a cable run sits on top of a fluro fitting, it is significant then you want as much twisting as you can get.

 

Ethernet cabling is a really good example where Cat 5 cable (lower speed) is tightly twisted pairs, and Cat 6 (higher speed) is even more tightly twisted in an attempt to further reduce EMI impacts.

 

Andy

 

 

[Dieselten]

Twisted-pairs for noise-cancellation work best on balanced lines with inputs to either coupling or impedance-matching transformers (noise cancels in the transformer input winding due to phase reversal from one end to the other) or inputs into differential amplifiers (op-amps), where noise-signals of equal aplitude and phase are cancelled by summing the inverting and non-inverting inputs. Twisted-pairs don't work very well on unbalanced lines (such as a single active wire with an earthed shield surrounding it).

 

Ferrites on power input cables, good bonding and grounding of equipment chassis, correctly-terminated cables (using high-quality connectors) and careful choice of cable-runs (and antenna-positioning) can go along way to strangling the interference. Strobe-inverters are especially troublesome and may require multiple treatments to reduce the inverter-noise or whine to an acceptable level - if it can't be eliminated entirely.

 

Sometimes you have to be highly creative and ingenious to reduce or eliminate RFI/EMI.