guard trace blog?

[ftransform]

So, I read about guard traces and there seems to be a mixed opinion, some people say they work for reducing noise, others say its only the added distance that makes an effect.

Can dave jones construct some experiments to illuminate this debate?

do guard traces have any use in circuits that are not electrometers?

[weilawei]

I'm reviving the dead because, in EEVblog #667, Dave mentions guard traces in an annotation, without further explanation. As someone who's not an EE, but a hobbyist, I'd like to see a video on them. Fundamentals Friday is my favorite segment (but I still love all the others).

[EEVblog]

Yep, it would require a suitable test mockup to show the effect.
Maybe not trivial, but I haven't given any thought what would be required there.

[nitro2k01]

Guard rings exist to prevent accumulating a voltage offset from a small (nA) stray current going over the surface of the board, into a high impedance input. This is why these traces are exposed, to be able to absorb these surface currents. Linear tech describes this here:

http://www.linear.com/solutions/1853

Here's an image from that article with a negative feedback opamp reading a high impedance sensor, showing how the stray current is absorbed by the guard trace, instead of being injected to the positive terminal. Guard traces are relevant whenever there is a high impedance source of some kind. I've seen it in an IR thermometer, for example.

[weilawei]

Wow, The Dave Jones himself! (Okay, I'm done being awestruck.) I love the show, wouldn't miss an episode for the world. Keep it up!

nitro2k01, that's an awesome little image that readily conveys the purpose of guard traces. I'm not sure I can measure down at nA levels with my very limited test gear, but I'm thinking it might be a fun experiment anyway. Thanks for that!

[nitro2k01]

I think you could actually use the layout in that image pretty much right away as a demonstration, together with a suitable opamp like the LTC2057, as suggested in the article. Replace the high impedance sensor with, say a 10 Mohm resistor to ground to simulate a high impedance sensor, and configure the opamp to produce a high gain. Put some low impedance traces like a supply rail around the guard ring. Now make a copy of that layout and remove the guard trace. Make boards of both the designs, and observe what the opamp does on the output side for both the boards. Maybe there are other variations that could be tried as well for demonstration purposes, like covering the guard traces with solder mask and seeing how effective they are then.