Active screen (on a PCB or shielded cable)

[marcdraco]

I'm currently learning how to design PCBs - properly, i.e. ones that belong in the 2000s...

So I've been absorbing as much as I can find and I saw something (might have been on Dave's channel) about what I think I heard as an "active screen".

Admittedly I was utterly carked at the time (lack of sleep) but I think the narrator was discussing surrounding what looked like a very sensitive, high-impedance differential input stage with a drive shield. I assume the same thing applies to any input but rather just than putting copious ground pour around the area (even some experts seem to disagree on this) it appeared they were putting some sort of negative feedback signal on the shield.

Forgive me if this was a fever dream but I've discovered so many new things over the last few months (particularly about how energy moves through dielectrics) I've torn up the manual and started again.

Thanks,

A. N. Noob

[Cerebus]

It's quite common to put a guard ring around a sensitive high impedance node to protect it from current leakage across the surface of a pcb. Ideally this guard ring is at the same potential as the node to be protected and is fed from a low impedance. If it's at the same potential as the sensitive node there can can be no leakage current flow because there's no potential difference. A common place to take the guard signal from the output of the stage being thus protected, often this is the same as in the - input to an op amp, which is probably where you got the 'negative feedback' impression from.

The same technique is used to drive the intermediate shield in triaxial cables as a way of both reducing leakage currents across the cables dielectric and effectively cancelling the capacitance of the cable.

If you have an inverting op amp setup you connect the guard ring to ground, to match the potential of the virtual ground at the - input.

Examples for both common op amp topologies:



Guard rings should be bare copper, without solder mask, so that they make electrical contact with any surface contamination that may be on the board.

[T3sl4co1l]

So, a driven shield, guard trace or bootstrap?

It's actually more of a positive feedback, but with limited gain (ideally exactly 1, from a voltage follower), so it doesn't (usually) cause instability*.  There are a few other connections like that, where the output feeds back to the positive input in part, but it doesn't cause instability: Howland current pump, some active filters (the feedback effects peaking or ringing for a sharp response), etc.

*But that does suggest when it can go bad: when extra phase shift and gain shows up from an external network e.g. input filters, wiring L/C, unterminated cables, etc.

In the case of driven shields, you'll likely use it very little.  I haven't had need of it myself, and I've been doing this for a long time.  It's the kind of thing that, when you need it, you need it, and it might simply not be obvious if you haven't seen it before -- so it pays to have articles out there discussing it.  But keep in mind, the amount it's discussed, is far out of proportion with how often it's actually needed!

The same goes for slotted ground planes, you need it very rarely if ever; and when you do, it's great.  The insidious part is, it's very easy to misapply, and not even be aware of your misdeed.  It's a lot harder to test things when you're rev-ing a dozen things at once on a PCB, and you don't know the results until the next prototypes arrive.  EMC is a notoriously difficult subject.

So, for that part: solid ground planes are a surprisingly effective and practical "default" case.  They can still be misapplied, but it's much harder to do, and doesn't go as badly when it does.  And there are far fewer cases where you don't want ground plane (whether as slotted planes, or completely absent).

Tim

[marcdraco]

Thank you both!

Not just for the clear explanations but the name of the dratted thing! This has been driving me stupid for days now with frantic googling but the problem with search engines is that they only give you the right answer if you ask the right question... Artificial Ignorance to go with my own.

[Terry Bites]

Driven screens can lead to oscillation. They do work wonders in reducing cable capacitance when you go off board. where high source impedance combined ith Ccable can causue unaccpetable frequency response. Any long driven sheild need to be driven with just under unity gain via small reistor, say 100R. You still want an outer gnd screen and you'll need triax cable for this.