Using Current Reversal to Reject Offset in Low Resistance Measurements

A four-wire resistance measurement can still contain an offset voltage unrelated to test current. Reversing a known current allows the resistive contribution to change sign while a sufficiently stable thermoelectric offset remains approximately unchanged. The difference between the two voltage readings can therefore isolate resistance more effectively than a single polarity reading.

Write the measurement equation

For equal current magnitudes, write Vplus = I times R + E and Vminus = negative I times R + E. Then R = (Vplus minus Vminus)/(2I). The Keithley handbook describes current reversal and offset-compensated approaches. Their benefit depends on the unwanted voltage remaining sufficiently stable during the pair of measurements.

In an illustrative test, a 10-milliohm connection carrying 100 mA produces 1 mV. A stable 10-microvolt offset causes a one-percent error in a single-polarity calculation, but cancels in the ideal reversed pair. If the source delivers unequal positive and negative currents, use their measured signed values in the denominator rather than assuming symmetry.

Choose the reversal interval deliberately

Each reversal needs electrical settling before sampling. Waiting too little admits inductive transients, source settling, or instrument filtering. Waiting too long allows thermal offsets to change between readings. Test several intervals and look for a region in which the result is stable. Include instrument aperture and switching delays in the recorded timing.

• Confirm that the board path can tolerate both current directions.
• Isolate parallel active circuits that would make the response nonlinear.
• Use separate force and sense contacts at the defined resistance boundary.
• Record actual current and voltage for both polarities.
• Repeat at a lower current to expose self-heating or contact nonlinearity.

Know what reversal cannot remove

Reversal does not correct a misplaced sense contact, changing resistance caused by heating, or a semiconductor junction in the path. It also does not establish that a connector behaves identically at its service current. A clean low-current resistance result and a high-current thermal validation answer different questions.

Define whether the result includes solder joints, connector interfaces, or copper between pads. Apply Kelvin connection principles to that boundary and compare the thermal behavior with the actual current pulse profile. Store the paired raw readings so a later reviewer can see offset drift instead of receiving only a calculated resistance.