The Kelvin method of resistance measurement is a
technique whereby very low
values of resistance may be accurately measured. The method
is to pass a constant current through the test object and remotely
measure the potential difference across the object with a voltmeter. It
is not
new, nor is it rocket science. The use of the Kelvin method is required
for resistances below one ohm on account of the inability of common
bench top and portable multimeter to work meaningfully below about 5
ohms. This is due to the unknown or unknowable resistance of
the test fixture and probes.
To eliminate the uncertainty of the probes, a four wire test set is required and constant current
source and millivoltmeter will be required for meaningful measurements
of DC resistance below 1 ohm. I have found that a practical test set
can be constructed using a 100 milliampere constant current source,
these probes and DC millivolt meter to give meaningful readings at the 100
micro-ohm scale.
The
motivation for this project was provided by the acquisition of a
Wavetek 300 PLL UHF signal generator that is beset by many
problems. The first problem was that tantalum capacitors across its +/-
18V DC supply rail had shorted. These are inside seal metal enclosures
and the prospect of having to open all of them was not to be
entertained. I needed the ability to find multiple short
circuits
on a power bus, and as it turned out being able to measure resistance
differences of the order of a tenth of a milliohm.
This
objective launched an interesting journey of discovery because we have
been conditioned to think of resistances below one ohm to be short
circuits. This journey showed that not all short circuits were created
equally !
In the first attempt of a Kelvin probe, you can
observe the two banana jacks attached to a brass bar and a brass probe
made from 1/8 rod. This turned out to be an unsatisfactory probe
because the resistance of the brass rod was actually greater than the
resistances I was attempting to measure. The rods resistance was about
1.5 milliohms. The other major problem was that even though I had
sharpened the probe tip to a needly point, the variation of probe tip
to substrate resistance showed over one order of magnitude variation
depending on how I held the prob. Measurements were simply
not
repeatable. They provided some encouragement , however, as things like
my crowbar could now be measured or rather the non zero resistance of
my number one crowbar was not, in fact , zero !
The next probe
The
next probe was designed to overcome the discovered shortcomings. The
probe tip is a piece of 3mm tungsten TIG welding electrode. Tungsten is
an extremely hard metal, sensibly inert and resistant to oxidation and
corrosion at room temperature. I also discovered that you can braze
tungsten, even though it is utterly impossible to solder directly to
tungsten. I brazed a tungsten tip to my probe and terminated
the
voltmeter wires directly on the tip. The hard metal can be
ground
into a very sharp point that stays sharp during use. The contact
potential of tungsten to other base metal also seems
small.
The measurements made with the tungsten probe are also sensibly
repeatable down to the 100 micro-ohm scale.
My constant current
source is nothing more than a five volt regulator and a 47ohm or 10
resistance. When measuring below one ohm, this arrangement
makes
a sensibly accurate constant current source. I have found
that
100milliamperes is sufficient current for good measurements. I had
allowed for 500
milliamperes as well. This should have provided more accurate measurements
of extremely low resistance however it is
less repeatable. I expect that this is due to the current
heating
at the very fine needle points and now thermal emf distorts
the
voltage reading.
I can very thoroughly recommend this
design of Kelvin probe for those seeking accurate multimeter style
measurement of less than one ohm and those that are chasing an elusive
short circuit. It is very handy for verifying transformer taps.
|