19

I have heard about Guard Rings many times, and I know they are supposed to avoid currents in places where there shouldn't be no currents, but I never found a good text to read more about it.

Can someone describe them properly or please recommend some material to further reading?

Daniel Grillo
  • 7,659
  • 18
  • 51
  • 69
mFeinstein
  • 4,293
  • 11
  • 45
  • 84

4 Answers4

20

From my answer to an earlier question:

A guard ring is traditionally used to protect high impedance nodes in a circuit from surface leakage currents. The guard ring is a ring of copper driven by a low-impedance source to the same voltage as the high impedance node. This would typically be the input pin of an op-amp.

Here's an example of a classic guard ring layout for a metal can op-amp from National Semi's AN-241:

Guard ring layout

The way this works is, say there's a low-impedance node nearby, like V- in the picture. Current can't flow from V- to the susceptible input pins, because it will reach the guard ring first, and be consumed by the source that's driving the guard ring. At the same time, the guard ring won't drive any leakage current of it's own onto the susceptible node, because it's kept at a very similar potential.

Community
  • 1
The Photon
  • 126,425
  • 3
  • 159
  • 304
  • In this case can I assume the guard ring will be tied to GND? – mFeinstein Apr 17 '14 at 19:40
  • Yes, additionally in this case you're assuming the inputs are both near ground potential. If that isn't the case in your application, you'll need to find another low impedance line which is near the expected input potential. – helloworld922 Apr 17 '14 at 19:58
  • 3
    @mFeinstein, not necessarily. The guard ring should be tied to a low-impedance source at the same potential as the inputs. See my earlier answer for an example where one of the inputs is considered to have low enough impedance to drive the ring protecting the other input, for example. – The Photon Apr 17 '14 at 20:05
  • @mFeinstein, thanks for the accept, but you might get more answers if you wait a day or so before giving it out. There are a number of other users here who might have something interesting to add. – The Photon Apr 17 '14 at 20:07
  • 1
    @ThePhoton I like the way you think about it, I got complains about many users when I say I just want to wait for other ansers. – mFeinstein Apr 17 '14 at 20:16
  • 1
    Photons answer describes the effect of any low impedance track. The guard works by tracking the signal and thus cancelling the guard capacitance to signal loading. Noise Current is suppressed only by virtue of capacitance ratios of signal and noise sources. –  Apr 18 '14 at 20:16
  • @Joe, good point, adding a sentence to clarify this point. – The Photon Apr 18 '14 at 20:19
5

Guard rings are analogous to the guard conductor for triax cables. When using triax cable the outer braid is connected to ground and acts as a conventional electrostatic shield. The inner braid is the guard and will be driven by the measurement equipment to approximately the potential of the input signal on the inner conductor. That minimizes leakage currents involving the input signal since the surrounding material is very close to the same potential. It also minimizes the effects of coupling noise via the cable capacitance, including some microphonics, by maintaining both sides of the capacitance that ties to the input signal at the same potential.

An excellent reference for small signal effects is "Low Level Measurements Handbook" published by Keithley Instruments. It's available from their web site or you might be able to talk a friendly rep out of a paper copy for frequent reference.

wilk
  • 176
  • 1
  • 3
  • Interesting, I have only once heard about triax cables, and I am pretty sure the outer conductor was driven by mais earth, the middle one by the DUT ground and the inner conductor by the DUT signal to be measured...thanks for the book link! – mFeinstein Apr 27 '14 at 06:34
  • Maybe you will want to update the link with the 7th edition link... – mFeinstein Apr 27 '14 at 06:36
  • 1
    7th edition: http://www.tek.com/dl/1KW-30010-0%2520KeithleyLowLevelHandbook_7Ed.pdf – sbell Oct 16 '14 at 15:10
  • 1
    Old links are broken - This one works - https://download.tek.com/document/LowLevelHandbook_7Ed.pdf – Aditya P Jan 27 '19 at 06:26
4

Rather than shunting common mode noise to ground where coupling capacitance can draw current on AC signals, Guard Rings use the output signal to shunt E fields nearby by eliminating the current flow with no voltage difference to the input, thereby reducing the effective capacitance of stray noise and also reduce loading effects of voltage drop from a Capacitive load.

So think of guard as a method to shield and reduce load capacitance effects which induce noise voltage from grounds. Ground noise coupling to signal is significantly reduced with guarding.

  • 1
    Could you elaborate more on that? I can hardly see how a stripe of a shield can protect a pin, if the pin is not enclosed by the shield – mFeinstein Apr 17 '14 at 19:39
  • The guard must drive the shield –  Apr 18 '14 at 01:45
  • but these systems never have a shield – mFeinstein Apr 18 '14 at 03:43
  • When you have very high impedance inputs, stray noise is easily coupled. The key to protection is to have low impedance tracks beside, above, below to shunt stray noise, if external, then shielded balanced differential. Ground is good, but Active Guard is better for high for impedance. Often this is not good enough for RF so CM chokes are used instead. –  Apr 18 '14 at 16:15
  • I see....why does high impedance inputs are so vunerable to stray noise? Stray noise have just some uV or mV, this sould be like no current flow at all (i am imagining a huge resistor connected to 1uV source)...what am I missing? – mFeinstein Apr 18 '14 at 20:01
  • E fields are coupled by stray capacitance to High Z signals not huge resistors. h fields are coupled by high current and mutual inductance. Guard voltage shunts the stray field with a low impedance (good) and same ac as signal (better) up to bandwidth of amplifier –  Apr 18 '14 at 20:11
  • Oh got it, so the high Z "makes" a capacitor, since there will be 2 sides of conductive material and the middle is too high z that allows to build a charge. You said "Hih Z signals"...but is it right to say it this way? I mean signals dont have impedace, just the connections, or is it the way it's said? – mFeinstein Apr 19 '14 at 22:20
  • 1
    All signals have impedance at source, path and destination. All insulators are dielectrics. All dielectrics have capacitance (relative to air) –  Apr 20 '14 at 14:19
  • Yes but, then the term "signal impedance" is relative to the point being observed isnt it? The source can be low Z and the destination high Z, so it depends, right? – mFeinstein Apr 20 '14 at 19:42
0

Guarding is also used on ion chambers for radiation measurement to prevent surface current leakage at the connector. There is an innermost signal connection, middle guard and outermost connection. The inner connection is to a wire inside the chamber and the outer is to the body of the chamber. When a few hundred volts of bias is applied between those 2 conductors then surface and cable leakage can easily swap the nanoamp signals of interest. The solution is to drive the guard conductor to the same potential as the inner conductor. Leakage current only flows from outer to guard and the signal on the inner is protected. It's pretty easy to get less than a picoamp of leakage with a hundred volt bias (it is, however, pretty tricky to measure that leakage accurately).

The Keithley handbook is a great reference on the subject and a lot of other things that can cause errors when measuring small signals.

wilk
  • 176
  • 1
  • 3
  • Did you mean "leakage can easily *swamp* the nanoamp signal"? Additionally, please provide more information than merely "The Keithley handbook" to provide the content the question is asking about. – user2943160 Jun 30 '16 at 18:49
  • 1
    Why have you posted a second answer saying pretty much the same thing? There's an edit link under the original post. – Transistor Jun 30 '16 at 19:09