Q
Can you give advice for implementing a good guard layout on the PCB for theinstrumentation amplifier AD623?
A
The good news is that we don’t need too much detail from you in order to helpyou, so it won’t be necessary to get any design files from you.
Any time there is a high-impedance source coming into an amplifier, there is an
increased chance for leakage (through the PCB) to become a problem in these
high-impedance nodes. As you have noted, guarding is a popular technique to
solve this problem. The key to a successful board design is to maintain any of
these sensitive nodes (at any given point) away from nodes that could be at
different potentials.
AD623, like many of the industry standard inamps, uses the standard op-amp
pin-out, which places one of the input pins right next to the negative supply
voltage (pins 3 and 4)(see attachment)
Most likely, this is where the problem resides. Also, unlike opamps, the input
pins could be at very different potentials, which could mean that leakage would
become a problem between them. Placing a guard ring (or two) around the input
pins would help significantly. How do you make a guard ring? In the case of an
inamp, there may be a few options available, depending on the size of the input
signal.
But before I go into that, let me tell you a possible simpler solution. There
are inamps with a different pin-out that would minimize the issue. I’ll use
AD8220 as an example(see attachment).
If you notice on the pin diagram, the input pins are located in opposite
places, away from the supplies and away from each other. As long as this is
laid out carefully, leakage issues will not be a concern due to the chip
pin-out. The RG pins next to each input are usually 0.6V (nominally) away from
their respective input pin and these are relatively low-impedance, which means
they won’t interfere with each other. The RG pins then are “bootstrapped” to
their respective input pins, therefore they don’t present a large leakage path.
Obviously, the supply pins are far enough away that their risk is also reduced.
So, in other words, if the customer is concerned with leakage, it is possible
that moving to a different pin-out is a simple solution to your problem.
Attachments:
