Hello,

I searched on guard rings but i didn't find an answer...

I am using guard traces to protect a very high impedance line from influence from conductance across the PCB surface. Everything I read talks about guard rings. However, is it important for any reason that they be a ring? How about simply a trace that ends without looping back?

This is a very low frequency signal, but very high impedance. I am buffering it with a voltage follower using an op amp with very low input bias current. I am guarding it with the output of the follower, which will approach the same potential as the input signal. It is much better to my layout, however, to run the guard trace between the input pin and the power supply pin to the op amp, but *not* to loop it all the way around. I figure if it gets in between the power and input pin, that's doing the function of protecting the input signal from any conductance on the PCB or surface contaminants.

Any disagreement on that?

 

These guys talk about that a little bit:

http://www.eetimes.com/design/analo...d-band---or-not-to-guard-That-is-the-question

Easiest way is to just lift the sensitive pin off of the PCB and keep the connection to it floating. Another way is to use a teflon standoff for the pin.

 

These guys talk about that a little bit:

http://www.eetimes.com/design/analo...d-band---or-not-to-guard-That-is-the-question

Easiest way is to just lift the sensitive pin off of the PCB and keep the connection to it floating. Another way is to use a teflon standoff for the pin.

Thanks for the link. I have done a guard ring, in the end. I've also read Bob Pease and others, as well as some data sheets, that describe the technique.

Mainly, I was wondering whether it's important for some reason to use a ring... I think that simple lines would work if they protect the area you need.

I attach a picture of my layout... this is a quad op amp with low fA input bias current. You can see there is an input to the non-inverting of gate 'd' on the chip, and it's protected by a ring that is formed from the output of that op amp, which is used as a follower to buffer the high impedance input. I also use a low-pass filter which you can see from the 2M resistor and a film capacitor. I use a very low leakage capacitor so that it will not bias the voltage... poly dielectric.

You can also see that I increased the spacing of the ground plane on the bottom to about 80 mil... to increase the distance from ground to any pin... although I keep all the high impedance stuff on the top plane with no ground pour.

Does anyone know the resistance through the thickness of a typical FR4 PCB?

Any more insights or intuitions on extremely high impedance work is much appreciated.

 

National Semiconductor have written much about this. And it is very often very good explained. I suggest you do a google search on these words -> national semiconductor guard ring
The information id often embedded in datasheets. But even if you do not use that component. The information on guard rings will be uneversial.