Minimize Measurement Errors in Component
Test Fixtures
Robert Brown - November 01, 2000
Buying a more accurate instrument to measure resistance and impedance won’t always improve the
accuracy of your measurements. Your problems could be caused by measurement errors that
originate in your test fixture. When designing a test fixture, you must minimize stray capacitance,
extraneous resistance, and unwanted EMI. Cables, relays, and contacts all contribute to
measurement errors. Fortunately, you can minimize those errors by applying simple techniques.
You can guard your DUT to minimize stray resistances and common-mode signals if your LCR meter,
milliohmmeter, or megohmmeter has a guard terminal or connection. A floating measurement
instrument is described as “guarded” when it has an additional conductive guard between its low
terminal and ground. Guards take the form of PCB traces, ground planes, and rings. When properly
connected to an instrument’s guard terminal, the guard shunts common mode currents away from
the measuring circuit. Guards are popular in high-resistance measurements made with a
megohmmeter. A metal guard plane can effectively negate alternate resistance paths in parallel with
the resistance you’re measuring.
Figure 1. A guard ring eliminates parallel resistance paths that can cause errors in materials
resistance measurements.
Guard rings also improve resistivity measurements of materials and are common in measurements
made by sensor manufacturers. When measuring resistivity, you measure resistance from top to
bottom through the sample material. Place one electrode on top of the material and place the other
electrode on the bottom. With two electrodes, a meter measures the resistance as a parallel
combination of the resistance through the sample and the surface resistance across the top surface,
down the side, and back across the bottom surface. The parallel combination produces a low
resistivity measurement. You can effectively eliminate the surface resistance path by placing a guard
ring (Fig. 1 ) around one of the electrodes.
You can also shield your measuring circuit against noise from unwanted EMI by constructing a metal
enclosure around the measuring instrument and component under test. I recommend using a metal
fixture with a hood to completely enclose the DUT. Connect the enclosure to earth ground.
Pay Attention to Cables
Shielded cables can also reduce noise from unwanted EMI. Remember that braided shields in coaxial
cables don’t provide 100% coverage against EMI. The higher the percentage of braid coverage, the
more effective the shield. Connect cable shields to ground, but connect each cable shield to ground
at one point only. Otherwise, you could introduce ground loops, which would add noise to your
system.
Cable length also affects measurements. If your meter manufacturer requires a specific length of
cable, then use cables of that length only. If your meter manufacturer doesn’t specify cable length,
then keep cables as short as possible.
Keep your cables organized to save significant time in troubleshooting. Separate and, if necessary,
shield cables that carry different signal types. Running all cables from the fixture to the instrument
in one piece of conduit can introduce signal errors. Large analog signals can easily interfere with
digital control lines, and digital lines can interfere with low-level analog signals.
When cables move, triboelectric noise develops that can produce erratic measurements on high
impedance DUTs. You can minimize triboelectric noise by using cable ties to keep the cables
stationary.
Use the Right Components
Proper use of guards, shields, and cables can reduce measurement errors, but you can improve your
measurements even further. Relays are still a mainstay in switching analog signals, both low-level
and high-level voltages and currents. When choosing a relay, you must check the manufacturers’
specs for contact resistance, insulation resistance (especially over a humidity range), and thermal
EMF.
Contact resistance in relays, pogo pins, and other contacts will affect your measurements, especially
when you measure low-impedance components. Specify components with the lowest possible contact
resistance compared to the impedance measurements you’re taking. Mercury wetted reed relays,
which have low contact resistance, work well in low-impedance, low-signal-level applications.
Unfortunately, as relays, pins, and contacts age from use, your fixture’s contact resistance will
change. Use four-terminal Kelvin connections between a DUT and a meter. Regularly replace your
fixture’s components to minimize contact resistance.
A relay’s insulation resistance affects high-impedance measurements. Most relays have an insulation
resistance of 1 GV or slightly higher, but insulation-resistance measurements using a megohmmeter
often exceed 100 GV. That makes relays extremely difficult to use in such applications because the
meter’s excitation signal will leak through the relay, causing a low measurement value. Fortunately,
you can compensate for the relay’s low insulation resistance with your meter as long as the
insulation resistance doesn’t change over time.
Most LCR meters have an “open compensation” or offset adjustment that compensates for insulation
resistance. To perform the adjustment, first measure open circuit resistance and capacitance. Then,
use the measured resistance and stray capacitance to correct future measurements. Remember,
though, that the meter bases future measurements upon a previous “compensation” measurement. If
the resistance or capacitance in the fixture changes, then an error will occur even in the
compensated measurement. If the relays’ insulation resistance changes because of a change in
humidity, your compensation will no longer work. You can also use the offset adjustment to
compensate for contact resistance.
If your fixture will operate on a production line where humidity levels could exceed 85% RH, you
should select relays rated for high humidity. Otherwise, you’ll get measurements that change by the
hour. High humidity levels also will reduce the overall life expectancy of the relay. A combination of
an electrical arcing and humid conditions can produce nitric acid in environmentally sealed relays.
The acid can corrode the contacts, resulting in a relay that fails too early in its life.
Where dissimilar metals meet and a temperature gradient exists, you’ll get unwanted thermal EMF
voltages in your system. The voltage level depends upon the metals and temperature of their
junctions. When measuring low-level voltages with a milliohmmeter or an LCR meter, you must
compensate for these thermal EMFs with your meter. You also can minimize the errors by using lowthermal-EMF relays.
Because a relay’s characteristics change with age, temperature, and amount of use, you’ll have to
replace them regularly to maintain consistent measurements. Therefore, select relays that you can
get easily, and always keep spares in stock. T&MW
FOR FURTHER READING
A Guide to LCR Measurements (P/N 035010), QuadTech, Maynard, MA, June 2000.
www.quadtech.com/resources/appnoteindex.html.
Bennett, Emeric, “Applications Dictate Capacitance Measurements,”Test & Measurement World,
September 2000. p. 25.
LCR Measurement Primer, QuadTech. www.quadtech.com/primer.
Strassberg, Dan, “Using Relays to Switch Analog Signals Is Neither Silly Nor Trivial,” EDN, July 20,
1992. p. 118.
Van Der Burgt, Marty, Solving Signal Problems. Belden, Richmond, IN.
www.belden.com/products/tpbroad.htm.
Robert M. Brownis an applications engineer with QuadTech. He has more than 13 years of
experience in instrumentation and holds a B.S.E.E. from University of Maine at Orono. E-mail:
rbrown@quadtech.com.