Often there is a need to measure very small electrical resistances, in the range of milliohms. In high-current applications they become fairly important. Commonly available multimeters however have rather restricted accuracy at the low ranges. Fortunately it is fairly easy.
A simple workaround is using a millivolt range and measure voltage across the tested sample. The voltage is produced by forcing a constant current, in this case convenient 100 milliamps, by the means of a current source. Each millivolt of the voltage on the multimeter then equals 10 mΩ.
A cheap and easy one is the venerable LM317.
The device was designed as a compact attachment for a multimeter. Pretty much all ones have the inputs as banana plugs spaced 20 mm apart, with a COM and V terminals adjanced. The adapter was therefore decided to be a small circuitboard with integrated banana plugs.
A third-party design
was used
as a starting point. Instead of an internal battery an external 12V power supply was used, by the means
of a standard 2.1mm barrel connector.
Due to absence of a prescribed 50 Ω trimpot and a suitable 18 Ω resistor, a replacement equivalent resistor was designed from available parts. A considerably high accuracy was achieved with three parallel 33, 33, and 100 ohm resistors, with a parallel 100-ohm trimpot with a series-connected 68-ohm resistor. For additional stability, 1% metallized resistors were used. The current was adjustable from 96 to 102 milliamps across the entire range of the trimpot, making it easy to precisely adjust the value without having to source a multiturn trimpot. (The 68-ohm resistor may have to be lowered, depending on the tolerances of your LM317 and other resistors. If the trimpot runs out of scale, lower the resistor.)
A small green LED was added to indicate the presence of the power.
The trimpot was used to adjust the current through the test leads to precisely 100.0 milliamps.
With 12 volts power and longer periods of activity with next-to-zero ohm loads, the LM317 dissipates a considerable amount of heat. Out of concern that such heating could influence the parts and distort the measurements, a copper heatsink was fabricated from a piece of copper roof flashing sheet.
A label
describing the adapter's parameters was attached to the adapter's back side,
which was purposefully left flat. It was then sprayed with a transparent plastic "paint", sold as conformal coating,
to achieve higher durability by an effect similar to lamination.
To eliminate the effect of the resistance of the measuring leads, a four-wire attachment was employed.
Each clip is connected not by one but by two wires, with one for the high-current path and one for the
voltage sensing. (The series resistance of the wires would play a significant role as it'd add to the
measured resistance of the load and skew the measurement. However, for voltage sensing the series resistance
of the wires is insignificant in comparison with the 10-megaohm input impedance of the multimeter. Two wires
are therefore used.) This technique is known as four-terminal sensing
.
A length of a double-strand cable was used as a test lead. The wires were soldered together to the alligator clips. It worked fairly well, however it turned out that the clips add some 5 milliohms to the measurement, which was sufficiently annoying to warrant a correction.
To correct this problem, additional sensing contacts were added to the alligator clips. The contacts were made from a copper metal sheet, acquired as a grounding strip. (Roof copper flashing was too thick.) The metal was cut and bent to a long U shape, which matches well the shape of the clip. The sensing wire was disconnected from the power wire and the clip and attached to the sensing contact. With this assembly, direct connection of the alligator clips shows 0 millivolts, the contact resistance being below the resolution of the multimeter.
The sensing contacts were secured to the alligator clips with a drop of glue. The top and bottom parts of the clips were isolated with kapton tape, to provide a degree of safety against accidental short circuits.
 Bare adapter board |  Bare clip |  Attached clip |  Board with leads |
 Board, detail |  Board detail, back side |  Board detail with heat sink |  Board detail with heat sink, back side |
 Clips |  Clip v2, side view |  Clip v2, detail view |  Clip v2, detail front view |
 Measurement of a 35-milliohm ammeter shunt |  Finished adapter, back side |  Finished adapter, front side |