In aircraft systems should the shields be grounded at one end or both?
Common sense suggests grounding audio shields at both ends is a bad idea (it creates an obvious loop), but it makes sense to ground signal wires for RS232/ARINC and similar at both ends.
What about high power wires like pitot heaters? A twisted pair should eliminate most magnetic effects, so does a shield add anything?
Why does it make sense to ground signal wires at both ends?
Signal is signal whether audio or digital. On audio you hear the interference. For signal wires the same interference is there, you just can’t hear it. I say ground these at one end also. A shield grounded at both ends isn’t a shield, it’s a ground.
I didn’t ground my pitot but did ground lighting (especially LED strobe)
A cable shield can have different purposes:
Grounding at both ends is not recommended if
and thus inducing a heavy current to flow through the shield , which creates a voltage across the two ends of it which then may add itself to the signal.
With differential signals (balanced audio or RS422/485/ARINC/etc data) the signal is on a twisted pair and the shield is just acting as a shield, so grounding both ends is better than grounding one end. The Vcm range is maintained via the interfaces at the ends not being isolated i.e. by the circuitry inside the driver chips.
But there is another factor: all analog and digital interfaces have a finite common mode voltage (Vcm) range and going beyond that can affect comms or even blow them up (e.g. -7V to +12V for RS422/485) so often grounding at both ends (and accepting some current flow in the cable shield, in return for which the Vcm is reduced to some small figure) is the lesser evil in digital comms. That’s unless at least one of the two interfaces is isolated/floating which is very rare in avionics (except ethernet which is normally transformer-isolated).
I do a lot of this in my day job (in the context of analog signals and RS232/422/485 comms) and one has to look at the individual case. With an aircraft I would avoid running heavy currents through the airframe anyway, so e.g. the starter current flows there and back via its own two cables, and similarly with stuff like the pitot and wingtip lights. From what I have seen, this is not done e.g. the starter motor grounds to the crankcase and the crankcase is connected with a thick wire to the firewall, so the starter current does flow back to the battery via the airframe – this is OK because that happens only when starting. But the alternator ground goes exactly where??. The main reason for that is to avoid large-area loops generating magnetic fields which can mess up the compass system. If you return these currents via the airframe, you have no control where they are flowing, and the creation of a loop is almost inevitable.
I have seen a fair bit of trouble in avionics installations with interference on audio. IMHO this is mostly caused by installers using the airframe as the signal return, and at the same time using the airframe as the return for heavy and pulsating currents e.g. wingtip strobes. Audio in GA systems is not balanced (in the “professional audio” sense of using a 600 ohm transformer at each interface) so you have just the one signal wire to play with, and the shield is the signal return, so the shield connection(s) need to be done carefully.