Guard Rings and Leakage Control in Sensitive PCB Design
When a circuit measures picoamps, the resistance of the board itself becomes part of the circuit. Surface contamination, humidity and flux residues create leakage paths that can be larger than the signal being measured, and the errors they produce look like drift, offset or noise rather than like a defect. Guarding is the standard technique for intercepting that leakage.
Why Leakage Currents Matter
A clean, dry laminate has a very high surface resistivity, but it is not infinite. At low humidity and with a clean surface, the leakage between two adjacent tracks may be negligible; with a fingerprint, a flux residue or condensation, it can rise by orders of magnitude. The path also changes with time, as contamination spreads and the board absorbs moisture, so a design that looks adequate on the bench can drift once it is in the field.
The effect is only visible when the impedance of the circuit is high. A sensor with a high output impedance, an electrometer input, a photodiode amplifier or a pH probe all work at currents where a leakage path of a few hundred picoamps is significant. In a low impedance circuit the same path is irrelevant.
Sources of Surface Leakage
Contamination is the largest contributor. Flux residues from assembly, ionic contamination from handling, solder mask residues and moisture absorbed into the laminate all provide a conduction path. The path is not a simple resistance either; it varies with humidity and with the voltage across it.
Thermally generated currents inside the device are a separate matter and cannot be fixed by layout. The important distinction is that guarding addresses the board level contribution, while device selection and temperature control address the internal one. How much moisture a laminate can take up depends on the resin and the construction, and the relevant data are compared in this guide to laminate material properties. Confusing the two leads to a lot of effort in the wrong place.
<img src="https://www.gopcba.com/wp-content/uploads/2026/06/视觉交互PCBA.jpg" alt="Guard ring traced around an op amp input on a high impedance PCB layout” />
How a Guard Ring Works
A guard ring is a conductor that surrounds the sensitive node and is held at a potential equal to that node. Because there is no voltage difference between the guard and the signal, no current flows between them, and any leakage from the surrounding environment is collected by the guard instead of by the signal.
The guard must be driven, not simply grounded. A grounded ring around a node sitting at one volt would create a potential difference of one volt across the surface, which is exactly the leakage path the guard was meant to eliminate. The guard has to follow the signal.
Guarding at an Amplifier Input
The classic application is an operational amplifier in a transimpedance configuration, where the inverting input is a virtual ground. The guard ring surrounds the input pin and the summing node and is connected to that same node, or to a buffered copy of it, so the leakage path is shorted out rather than passing through the feedback network.
The ring should also surround the feedback resistor and any other component connected to the summing node, because leakage anywhere on that net contributes error. Components on the node should be placed inside the ring, with no tracks crossing it other than the ones that must.
Driven Guards and Bias
Where the sensitive node is not at a convenient potential, a buffer amplifier can drive the guard at the same voltage, which is the usual solution for a high impedance sensor input. The buffer’s own offset and drift then matter, because any difference between the guard and the node becomes a leakage driving voltage.
In some instruments the guard is deliberately biased to steer leakage away from the measurement. This is common in cable and probe designs, where a driven guard conductor surrounds the signal and is maintained at the signal potential along the whole length.
Layout Geometry and Spacing
The guard ring should be a continuous conductor with no gaps, drawn on both sides of the board where the signal appears on both, and connected through vias at intervals. A gap in the ring provides a path for leakage, and a ring on one layer only leaves the other layer unprotected.
The ring needs its own clearance from the node it protects, because the two must not short, and that clearance must be small enough that leakage across it is negligible compared with the leakage the ring is intercepting. The ring also needs to be wide enough to be fabricated reliably, which sets a practical minimum.
Cleaning, Coating and Contamination
Guarding reduces the effect of contamination but does not replace cleanliness. A board with visible flux residue under a guard ring will still be unreliable, because the contamination can bridge across the ring or create a path to a neighbouring net. Assembly and cleaning have to be controlled together with the layout.
Where the environment is humid, a conformal coating over the guarded area helps by keeping moisture off the surface. The coating has to be applied after cleaning and must wet the surface properly, because a coating that has lifted provides a path for moisture and can make the situation worse than leaving the surface bare. The assessment of cleanliness before coating is described in this guide to judging a finished board.
Guarding Digital and Mixed Signal Nodes
Guarding is not limited to analogue measurements. A high impedance digital node, such as a reset line or an oscillator input, can be protected from leakage and from coupled noise by the same technique, and the effect on reliability is similar.
In mixed signal designs, the guard is usually tied to the analogue ground reference rather than to the signal, so the separation of the analogue and digital return paths is part of the shield design. The interplay between guarding and overall noise performance is covered in this guide to mixed signal design.
Verification and Testing
Guarding is verified by measuring the leakage that remains. A simple approach is to measure the current into the node with the guard connected and disconnected, at different humidity levels, and compare the results. A guard that halves the leakage is working; one that makes no difference may not be connected as intended.
Where a design is expected to meet a leakage specification, the test method and the conditions should be defined, including temperature, humidity and the settling time allowed. Without those, two measurements of the same board can differ enough to make the comparison meaningless.
FAQ
Can a guard ring be connected to ground? Only if the sensitive node is also at ground potential. A guard must be at the same voltage as the node it protects, which is why transimpedance inputs can use a grounded guard and floating sensors cannot.
Does a guard ring eliminate the need for cleaning? No. The guard reduces the effect of leakage but contamination can still bridge across the ring or create a path to an adjacent net. Cleaning and, where needed, conformal coating remain part of the solution.
Why is my leakage still high after adding a guard? Common causes are a gap in the ring, a ring connected to the wrong potential, contamination bridging the clearance, or leakage inside the device rather than on the board. Measuring with the guard disconnected and connected separates the board contribution from the rest.
