Faraday cages and cables

I think a foil/screened cable could be considered a faraday cage provided the screen is full and of low resistivity.


A coax cable may be considered as a screened cable again provided that the braid/foil is full enough at the frequencies of interest.  This will rely on good contact between the sides of the foil and /or braid wires within the screen at the frequencies of interest. 


IIRC a coax cable is designed to keep the signal in the cable at the frequencies it is designed for and has specific impedance.  A screened cable is designed to keep external signals out and these may be at different frequencies and field strengths and will often have  non symmetrical field geometries (unlike as coax which is general cylindrical) and so require denser or thicker screens so they may not always be interchangable.

A coax (co axial) cable is a type of screened cable with a single conductor in the centre and one or more conductors wrapped around it in some way. Triax, with two separate screening layers, is quite common. Screened pairs, triples, and quads are also common. Screened twisted pairs work very well with balanced signals. Single conductor coax is better when you have an unbalanced signal.


Various types of screen are used, often in combination, depending on the shielding requirements, spiral braid (just wrapped around), normal braid, metalic foils,metalised plastic foils. The number and size of wires in the braid as well as the angle of the wires may be calculated to give a specific performance.


A Faraday cage gives a path for the electric field to travel along the surface of the cage rather than through it (very basic desciption) so you could describe the screens on cables as Faraday cages.


Best regards


Roger

agree, the braid on a coax is indeed a form of electrostatic shield, so the volts between inner and outer are not (to first order) affected by the electric fields outside the jacket.

Actually, to define how good (or bad) the shielding is, we often talk about the transfer impedance of a coax (not to be confused with the characteristic impedance or surge impedance.)


The transfer impedance tells you how much the voltage difference between inner and outer is upset if a current that is not part of the wanted signal is flowing on the outer.


Imagine, if you had a very simple wanted signal and  put a 9V battery between inner and outer at the source end, and measured with a meter at the load end, you would also see 9V between inner an outer at the other end, the wanted "signal" is getting there un-affected..

IF however you put say 1 amp down the braid from some other source, then because that causes the braid voltage drop to slope, but not the inner, it is as if a small part of the current outside has got in series with the wanted signal when you measure it - so a transfer impedance of one ohm per meter, would mean a change of 1V (could add or subtract- depends on the direction) the wanted signal, per amp or induced braid current.


For completeness, the characteristic impedance, typically 50 ohms, 75 ohms or other is the ratio of inner  core current to voltage between inner and outer for a length long compared to the measurement time. (and is also the impedance you use as a load to make it look infinitely long, and suppress multiple transit effects)

PS - so , unsurprisingly a thin wispy higher  resistance braid, makes a lot poorer shield than an outer conductor made from a solid copper tube, and a well braided jacket or two is somewhere midway.

Shielding effectiveness, and transfer impedance are inversely related.

Thanks guys - that's been very helpful!


I'd been reading something that was attempting to categorize different types of cables - basically into co-ax, 'unbalanced' (e.g. plain multicore) and 'balanced' (twisted pair) - but things then seemed to get into a bit difficulty when it came to screened twisted pair cables (S/FTP etc). I was wondering if things might be easier to describe if screening and twisting could be described just as techniques for reducing interference that could then be applied in various combinations within a single cable, rather that something that defined the type of the cable overall. A TV Co-ax, a 2-core microphone cable and a 25-core screened cable for the old RS 232 system, being fundamentally the same - just differing in the number of cores. Then in things like S/FTP cables both techniques are applied together. Similarly in the likes of SCART cables where some cores may be protected by screening, but not others.


  - Andy.

Feeling distinctly rusty on the theory of all this but I seem to remember that a in a twisted pair screened cable the signal is travelling down the conductors and is shielded from outside interference by the screen. With more than one twisted pair in a screened cable the lay or length of a twist is different for each pair to minimise interference.

In a co-ax cable the signal is a radio wave travelling down the insulation and effectively bouncing off the centre conductor and the screen. The more fresh air in the insulation the better but there has to e enough solid material to keep the shape circular.

The important thing about twisted pair signals is that the same signal is sent down both wires, but inverted in one wire. At the receiving end, the differerence between the two signals is taken to extract the data. Interference tends to induce the same spike on both wires (common mode interference), and taking the difference effectively subtracts out the interference.

The important thing about twisted pair signals is that the same signal is sent down both wires, but inverted in one wire.

That's got me thinking. I think there are examples where the data line is paired with a 'ground' wire rather than a -ve data (some old versions of IDE and SCSI come to mind,  but I'm just working from memory) - in that case the receiver end can still reject noise in just the same way (by subtracting the signal on the 2nd wire of the pair from the first - it ends up subtracting (0V+noise) from (data+noise) to get just data. Presumably the difference though is the amount of noise the cable creates - with a +ve data line and a -ve data line at a reasonable distance away the two will cancel each other out reasonably well - so reducing the noise seen by other conductors in the system - but with a signal & 0V pair there's no cancelling, so other conductors nearby see increased noise.


  - Andy.

Quite so - the idea of ground being some magical '0V' is not really true, certainly between points that are far apart in terms of the speeds of the signals in question - there simply is no time for electrons to rush back up the wire to let the origin know the conditions have changed at the load end. 

So instead you take a local reference with you, along the same route as the wanted signal,. and measure relative to that.


In many ways the ground is not a level voltage at all, and making the analogy of height and voltage, it behaves more like the floor seems to when you are very drunk, and have to lie down and hang on to it as the room bounces up and down.

Genarally higher frequencies and short pulses aggravate this.


Actually if you have a fast scope, and the ability to generate single fast pulses, you can make a pulse inverter, by cutting some  coax in the middle and wiring the incoming centre to the outgoing ground and vice-versa, and then mounting  it with a couple of coax sockets in a metal box.. Any pulse short compared to the transit time, proportional to the length of the cable (say 2nsec per foot there and back.) does not have time to "know" it is shorted out by the input  and output coax jackets being connected again, and the pulse that sets off as positive on the inner, emerges as negative on the inner. Longer pulses d however see the short circuit, and are truncated.