Advertisement

Article

To guard (band)… or not to guard? That is the question

As designs continue to push the capabilities of instrumentation amplifiers, issues that have been taken for granted must now be considered. This month we take a look at a technique to applied a very basic item, the printed circuit board and how to protect from leakage currents. I have enjoyed phone calls from several engineers that have doubts concerning guard rings. This question is seen in the application of op amps, instrument amps and log amps. Actually any application where the signal is a low current from a high impedance source is a candidate for this issue.

The procedure for establishing a guard ring is two simple steps. First, identify the node that is sensitive to stray currents. This will be a point in the signal path that is associated with very high impedance. It is generally one of the inputs of an op amp, instrumentation amp or even a log amp.

Next, identify a low impedance point that is at the same potential as the sensitive node. This is usually the other input of the op amp.

In the design of the printed circuit board draw a ring in copper around the sensitive node. Connect this ring to the low impedance drive point as shown in the figure examples below.




Low input bias current op amps might require precautions to achieve best performance. Leakage current on the surface of circuit board can exceed the input bias current of the amplifier.

For example, a circuit board resistance of 1012 Ω from a 15V power supply pin to an input pin produces a current of 15pA”more than one hundred times the input bias current of some op amp. To minimize surface leakage, a guard trace should completely surround the input terminals and other circuitry connected to the inputs of the op amp. Through-hole devices should have a guard trace on both sides of the circuit board. The guard ring should be driven by a circuit node equal in potential to the op amp inputs. A blanket statement concerning the substrate termination of any IC family cannot be made. It is best to consult the data sheet to determine the proper termination for each device.

Follow-up on the last column

A couple of readers wrote to me with their experiences using op amps as comparators, which was the subject of my previous column. The first experience spoke of the interaction between devices on the same chip. This interaction could be caused by a couple of situations. While the obvious cause would be the high speed edge of the comparator switching action radiating into the second side, another possible answer could lie in the basic design of the chip.

The typical op amp has a sub-circuit that generates the various bias voltages for each stage. With multiple op amp chips some designs have one bias circuit for the entire chip while other designs have a bias circuit for each op amp. For those multi channel devices with a single bias circuit, if one channel is saturated the voltages may be distorted for the other channels. Driving the comparator function from one rail to the other would cause a transient that could appear in the output of the linear channel. A device with separate bias circuits for each op amp would not show this interaction.

Another reader told of a situation where the input resistance and bias current specifications were for the case where the amplifier was operated in the linear range. The current flow into one input pin depended on the voltage at the other input. This sounds like the op amp was one of those that had a pair of diodes connected across the input cathode-anode/anode-cathode to protect the input transistors from over voltage. This is a common circuit with bipolar input op amps and would cause problems when the device is used as a comparator.

Thanks for readingComments Please.
klein_bill@ti.com
And be sure to copy, you-know, sohr@cmp.com

0 comments on “To guard (band)… or not to guard? That is the question

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.