Professional Documents
Culture Documents
STRAIN GAUGE
CIRCUITRY:
WHEATSTONE BRIDGE
Foreword (a word from our sponsor…) expensive, or affordable but unsuitable. To date, we have delivered
some very fine systems, the latest with microcontroller technology,
Instruments for Engineering Measurement was established in more than 16-bit resolution and computer interface. We stay away
1992, at first as a “systems integrator” and an importer of high from carrier frequency amplifiers, because of cable capacitance
quality loadcells from – the previously unknown – Kaliber problems, and DC amplifiers have come a long way in long term
Instrument and Measurement Technology Company in Budapest, stability. Good amplifier design is of paramount importance, and it
Hungary. Because there are a great number of loadcell is possible to produce systems at a reasonable price which are a
manufacturers in the world, each offering better and better loadcells happy marriage of old-fashioned analogue technology, and modern
at lower and lower prices, our Customers were limited to those who digital processors, rather than try to eliminate analogue problems
needed something special, or at least a loadcell which could endure with software !
severe dynamic loading and adverse environments without being In November 1997, we purchased the entire wire grid strain
damaged. gauge manufacturing process from Kaliber, who decided to give up
These Customers were few and far between, and our attention the wire grid strain gauges, and to start making foil grid strain
turned to another of Kaliber’s products: the wire grid strain gauge. gauges for their own loadcell and transducer production. Since we
This type of strain gauge was manufactured a long time ago by started with the wire grid strain gauge manufacture, we have made
Philips, TML, HBM, and possibly others. The available literature (T. a few improvements to the form of the strain gauge. The copper
Potma: Strain Gauges) refers to wire grid strain gauges on a paper terminals are larger and do not extend to the edge of the gauge.
backing, and the great difficulties in obtaining accurate readings Also, for half and full bridge measurement, we offer matched strain
with these. Problems were experienced with production, especially gauge pairs, whose elements are chosen from the same batch,
with attaching terminals to the grid, without contaminating the paper even from the same sheet that they were made. The huge benefit
backing with solder fluxes. Relatively high cost of production, a high of this was observed at an outdoor site near Stilfontein, where the
reject rate, coupled with the lack of modern-day adhesives, and a ambient temperature varies greatly. With a temperature variation at
special “feel” needed to apply these strain gauges propagated the the metal surface of +30 to +13 °C, (just before sunset to just
idea that strain gauge application was a “black art” known only to a before sunrise) we observed a measured value which varied by 1
select few. The select few, in turn, capitalised on this idea, and kept count, with a measuring range of 0,2 mv/v representing 327 counts
the experience to themselves. on the digital display. This, with no temperature compensation at
With the development of printed circuit techniques, strain gauges all, and with a constant mass applied, therefore the measured
rapidly became a mass produced item, with all kinds of grid shapes value contained the span temperature error as well as the zero
and sizes imaginable, which could be made with the photoresist point temperature error. To save installation time, we pre-wire the
and etching method. Ultra-small strain gauges were feasible, strain gauges as half bridges with external terminals, and we
especially for applications in loadcells and other transducers, where actually did an experiment to check our installation speed. Sanding
the measuring element is necessarily small, requiring miniature and degreasing of the surfaces, sticking 16 gauges in 8 places,
strain gauges with grid lengths of 3 mm and less to measure the wiring up, insulation testing, sealing, running the cable and
strain on the small load bearing element. In stress analysis too, installing the amplifier, connecting it to 220v power took 1 hour and
many tiny strain gauges could be made on one small backing to 20 minutes for 2 people.
measure strain gradients on small parts. We have found that the wire grid strain gauge offers some
In the early 80’s, HBM stopped the manufacture of wire grid strain benefits: the cross-section of the grid is circular, therefore the
gauges, and TML continued with them until the early 1990’s, then gauge factor is very predictable, being within 1% of the theoretical
stopped also. value of 2,05. The temperature coefficient of the gauge factor
The wire grid strain gauges of Kaliber were of a much more happens to inversely match the temperature coefficient of the
sound design than the paper-backed type. Phenolic resin was used elasticity of steel, within a few parts per thousand. The temperature
as a backing, and constantan wire of 20 - 30 microns diameter coefficient of the zero point is zero, which means that we can not
used for the grid. The manufacturing process allows the grid to be produce “temperature compensated” strain gauges. Rather, we
completely embedded in the backing, which is not more than 50 - suggest to use a matched pair in a half bridge, and then the zero
100 microns thick, and heavy duty copper solder terminals to be point temperature error cancels completely. The application
integrated into the strain gauges. Of course, this allows easy process that we use, together with their large and robust
handling, and attaching of measuring cable directly to the strain construction makes it possible to apply these strain gauges even in
gauge. Our first experiences with these strain gauges was that they dusty, windy conditions, and get measuring results as good as
were easy to apply and extremely stable, especially at strain levels when applied in a clean laboratory. We have discovered that readily
below about 500 µm/m. Their relatively long grid (10 and 20 mm) available application materials produce excellent results. Our
allows low error measurement, with excellent long term stability. covering material, first applied in 1997, is re-enterable, yet protects
Soon we discovered that customers who were in the market for the strain gauges for years in mint condition, despite our
strain measurement were not interested in learning the application competitors’ comments that this type of application is strictly short-
process, rather they were happy to give us an order to do the term…
complete job. We accepted the challenge, and did not invoice until In recent months, our strain gauges have been exported to
the job worked to the customer’s satisfaction. In this way, we did Europe, Australia, and Israel, each time with accuracy which was
our first strain gauge job, instrumenting of a 100t garage type above the client’s expectation. Although our experience shows that
press, in Welkom. This was rapidly followed by the strain gauging this is a precise and predictable technology (and not a black art),
of the big 750 t press at the same place, then a 100t Amsler press. there are some applications where we find temperature gradients,
The presses were calibrated by SABS, and were found to be well and these can cause headaches with regard to temperature error.
within the accuracy requirements. In the case of the 750 t press, Besides applying temperature compensation as the loadcell
the overall linearity was better than 0,2%, much better than any manufacturers do, simple prevention such as thermal insulation is
other measuring system used on that press. Kaliber themselves did an effective solution. As wonderful as the strain gauge technology
not believe that their product was that good ! But, the long term may be, some applications demand extra attention, and the lesson
temperature stability was not so good, and the zero point had to be to be learned is that we never stop learning. Each project is a
adjusted at the beginning of each test. thrilling, challenging and learning experience.
Soon, we started manufacturing our own amplifiers, because
what was available in the market was either good and frighteningly Louis Eder
Resistivity of copper:................................................................................................ 15
FORMULAE ............................................................................................................................................. 33
1. QUARTER BRIDGE ..................................................................................................................... 33
2. HALF BRIDGE.............................................................................................................................. 33
3. FULL BRIDGE.............................................................................................................................. 34
UEXC. R2
UOUT = -------------
R1 + R2
UEXC
UOUT = -------
2
= 0,5 UEXC
A change of resistance, induced by strain, of 0,24 Ω will cause an output voltage of:
UEXC. 120
UOUT = -----------------
120,24 + 120
120. UEXC
UOUT = -----------
240,24
It is quite difficult to measure a change between 0,5.UEXC and 0,4995.UEXC with any
degree of accuracy.
quarter bridge.
Sir Charles Wheatstone, b. Feb. 6, 1802, d. Oct. 19, 1875, was an English
physicist and inventor whose work was instrumental in the development of the
telegraph in Great Britain. Wheatstone served (1823-34) an apprenticeship as
a musical-instrument maker. His work in acoustics won him (1834) a
professorship of experimental physics at King's College, London, where his
pioneering experiments in electricity included measuring the speed of
electricity, devising an improved dynamo, and inventing two new devices to
measure and regulate electrical resistance and current: the RHEOSTAT and
the Wheatstone bridge. He worked on magnoelectricity and submarine
telegraphy, and he suggested the STEREOSCOPE. In 1837 he designed,
with William F. Cooke, an electric telegraph system that became standard in
Britain in 1840.
UOUT = UA - UB
UEXC . R1 UEXC . R3
= ----------- - -----------
R1 + R2 R3 + R4
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
120,24 120
= ----------- - -------
240,24 240
= 0,000 499 5 V / V
or 0,4995 mV / V
Let us examine the following case where the strain is compressive, and the resulting
∆R negative:
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
119,76 120
= ---------- - ------
239,76 240
= -0,000 500 6 V / V
or -0,5006 mV / V
Here we see evidence of the inherent non-linearity of the Wheatstone bridge when
used for quarter bridge strain measurement. ERROR% refers to the difference
between tensile and compressive strains of equal magnitude.
half bridge.
∆R2 = - 0,075 Ω
UOUT R1 R3
------ = --------- - ---------
UEXC R1 + R2 R3 + R4
120,25 120
= -------- - -----
239,925 240
= 0,5011982 - 0,5
= 1,1982 mV / V
∆R2 = + 0,075 Ω
UOUT R1 R3
------ = --------- - ---------
UEXC R1 + R2 R3 + R4
119,75 120
= -------- - -----
240,075 240
= 0,4988024 - 0,5
= - 1,1976 mV / V
Once again, the ERROR reflected in the graph shows the difference between the
signals of two strains of the same magnitude and opposite polarity.
Note the convention of expressing the output of the Wheatstone bridge as a voltage
ratio. Because the bridge itself does not generate any current, its output will always be
a fraction of the excitation voltage. Increasing the excitation excessively is not possible
because of self-heating of the strain gauges, whose current should be limited to less
than 25 mA generally. For 120 Ω gauges, this means a maximum of 6v excitation, and
an output at 1000 µε of 3 mV.
full bridge
whose properties we will now examine.
R1 = R2 = R3 = R4 = 120 Ω
∆R1=∆R2=∆R3=∆R4 = 0,24 Ω:
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
120,24 119,76
= ----------- - ----------
Compiled by Louis Eder / Instruments for Engineering Measurement
12
240 240
= 0,501 - 0,499
= 0,002 V / V
or 2,0000 mV / V
We can see that reversing polarity of the strain will simply swop around the two terms,
but the output will be of the same value (and opposite polarity).
NO LINEARITY ERROR.
In each case, two pairs of strain gauges can be used, each pair being subjected to
equal and opposite strains.
Let us now examine the case where the cantilever beam is strained in a direction that
we do not want to measure:
∆R1=∆R2=∆R3=∆R4 = 0,24 Ω:
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
120,24 120,24
= --------- - ----------
240,48 240,48
= 0,5 - 0,5
Now that we have considered quarter, half and full bridge circuits, let us examine a
few practical applications where usage of bridge circuits helps to solve problems...
On this structure, we have no means of using the full, or even the half bridge, only one
active strain gauge is used. The other three-quarters of the bridge is in the measuring
instrument.
We have added extra resistance to the active quarter bridge, in the form of connecting
wire. The 0,22 mm2 wire is commonly available, and its resistance is as follows:
Resistivity of copper:
0,017 . 10
RL1 = -------------
0,22
The return wire has the same resistance, so we have a total extra resistance of
0,7727 . 2
= 1,545 Ω
Compiled by Louis Eder / Instruments for Engineering Measurement
15
This results in the following equivalent circuit:
R2 = R3 = R4 = 120 Ω
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
121,785
= --------- - 0,5
240,24
= 6,9305 mV / V
R2 = R3 = R4 = 120 Ω
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
121,545
= --------- - 0,5
240
= 6,4375 mV / V
Of this signal, 0,493 mv/v is due to the strain of 1000 µε, and the rest, being 6,4375
mv/v, is due to lead resistance. This is a large value, but strain amplifiers will have
sufficient adjustment to balance this value to zero.
Another, more tricky problem is the fact that copper, in spite of being one of the best
conductors of electric current, has quite a large temperature coefficient of resistance
α r(Cu) = 0,004 Ω / Ω / °C
With our wire resistance of 1,545 Ω, and an expected ambient temperature range of
(say) 0°C to 25 °C (typical Highveld winter's day in a factory), the wire resistance
values will be:
= 1, 6995 Ω
Because our amplifier cannot tell the difference between strain induced resistance
change and temperature induced resistance change, it will measure a strain caused
by a change in resistance of:
∆R
---- = k. ε
R
∆R
ε = ----
R.k
0,145
= --------------
121,545 . 2
= 596,4 µε
for the ambient temperature change alone ! (Assuming, of course, that the
strain gauge itself is fully compensated to the metal of the member being measured,
and gives no output change for the member's own temperature expansion.)
Using the Wheatstone bridge's properties, part of the problem can be compensated. It
is simply a matter of where we connect the measurement point:
Note also, that one of the lead resistances, RL2, now belongs to R2, not to R1 any
more .
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
120,7727 120
= -------------------------- - ------------
120,7727 + 120,7727 120+120
= 0 mv/v
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
120,85 120
= -------------------------- - ------------
120,85 + 120,85 120+120
= 0 mv/v
We have got rid of the temperature problem, and our amplifier no longer has to
balance out a large offset value, but we should also look at the error in the measured
strain value:
τAMB = 0°C , ε = 1000µε, RL1=RL2=0,7727 Ω, ∆R1 = 0,24 Ω
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
121,0127 120
= ------------- - -----
241,7854 240
= 0,4963 mv/v
UOUT R1 R3
------ = ----------- - -----------
UEXC R1 + R2 R3 + R4
121,09 120
= ------------- - -----
241,94 240
= 0,4959 mv/v
The temperature change produces a span error of 0,08% which can probably be
ignored.
Compiled by Louis Eder / Instruments for Engineering Measurement
20
A very useful strain gauge arrangement for normal strain measurement is shown here.
It encompasses complete temperature compensation (in the accuracy range that
can be considered acceptable), and produces minimal linearity error and the
biggest signal that can be had under the circumstances.
It uses two strain gauges applied in the main strain direction, and two at right angles,
making use of the Poisson effect.
Taking a value of R=120, k=2, ν=0,3 and values of ε from -1000 to + 1000 µm/m, we
get the following:
Having examined various bridge circuits, it would have been noted that in each case,
stress was placed on the fact that no current may be drawn from the output
connections, otherwise the bridge circuit will not function correctly. This imposes
certain requirements on the input circuitry on the measuring amplifier, regarding input
resistance. At the same time, we should also consider that (especially in the early
days of strain measurement), the only instruments that were available for measuring
the output of a bridge were moving-coil galvanometers, which had a very low input
resistance.
{gal-vuhn-ah'-muh-tur}
A galvanometer is an instrument that measures the amount of electrical current by converting electrical energy into the
physical displacement of a coil, which in turn moves a pointer or light beam. The device was named for Luigi GALVANI, an
18th-century Italian physiologist and physicist.
In a galvanometer, a coil of fine wire is suspended between the poles of a permanent magnet, so that when the coil is
magnetized as current passes through it, the like poles of magnet and coil repel each other and cause an attached pointer to
deflect across a calibrated scale. When a light beam is used instead of a pointer, a mirror is mounted on the side of the
moving coil and a fixed beam of light is directed at the mirror. As the coil turns, the reflected image of the light moves along a
translucent, calibrated panel. The coil may be mounted on a spindle, whose ends turn on rubies or very hard steel.
The direct-current ammeter is a type of calibrated galvanometer that measures larger currents a calibrated galvanometer
may also be used as a direct-current voltmeter, which measures direct voltage using Ohm's law. Galvanometers are
currently being replaced by modern digital instruments.
Leslie W. Lee
Bibliography: Cooper, William D., Electronic Instrumentation and Measurement Techniques (1978).
The actual "bridge" instrument was quite refined. Dial calibration was directly in µm/m,
and there was also provision for gauge factor adjustment. One manufacturer of such a
Bridge was the Swiss HUGGENBERGER, and there is at least one such instrument
still in (semi)active use in South Africa.
In fact, the highest precision strain meters (until recently) still employed the
"compensator" or null balance technique, as it is the most precise, offering inherent
linearity compensation and independence of measuring amplifier input current.
Let us examine the influence of the current drawn from the Wheatstone bridge on
accuracy.
UG
I = --------
RB + Ri
The voltage appearing at the input of the amplifier, by Ohm's Law, will be:
URi = I. Ri
Ri
= UG . --------------
RB + Ri
Ri
= 100 . ( 1 - -------------- ) %
RB + Ri
As users of strain gauge amplifiers, we expect that this instrument will provide us with
an acceptably accurate measured value. The graph shows the span error introduced
by amplifier resistance which is less than infinite. An especially irritating factor is
having an altered apparent strain reading when using strain gauges with a different
resistance.
In many applications, it may be better not to rely on the self-compensation, but to use
half- or full bridge circuits wherever possible. In that case, the self-temperature
compensating properties of the strain gauges become of no consequence. If the full
bridge is not feasible, a half-bridge circuit offers a good compromise, and if even this
is not practical, we need to look at the possibility of temperature compensating
"dummy" strain gauges.
The advantage of this connection is that minimal lead resistance (span) error is
introduced. It is useful for single measurement points.
This connection method is widely used for other resistance-type transducers, such as
resistance thermometer devices (RTD’s), for example, the very precise platinum wire
thermometer (Pt 100).
Although this connection method introduces double the span error of the previous
circuit, it has the advantage that one dummy (R2 circuit) can be used for a number of
measurement points, each of which uses only one active strain gauge (R1).
BEAM A BEAM B
For this example, we need to measure the average (or total) strain on the two
members shown.
We can, of course use two strain amplifiers to individually measure the strains, then
mix their outputs. However, if the individual strains are not important, we can use a
more elegant and more economical method, that is to connect the bridges together in
one of the following two circuits, either parallel, or series, as follows:
BEAM A BEAM B
This connection method will reduce the bridge resistance to half of that of a single
bridge. It may mean that the supply current capability of the amplifier will be exceeded.
Some strain amplifiers (especially the cheaper ones) can not accommodate a bridge
resistance below 350 Ω without special modifications.
This is the circuit used for connecting multiple loadcells to a single amplifier, such as
in platform scales and weighbridges.
SERIES CONNECTION:
This connection is not used for loadcells, because each loadcell is a complete bridge,
and the bridge has to be broken to allow connection in series.
1. QUARTER BRIDGE
UOUT k.ε
------ = -----
UEXC 4000
Output (mv/v) is equal to Gauge Factor times strain (µm/m) divided by four thousand
(the thousand is to convert the micro in the strain to milli in the volt).
NUMERICAL EXAMPLE:
With a gauge factor of 2, and a strain of 500 µm/m, output wll be:
2 . 500
-------------- = 0,25 mv/v
4 000
(See Page 6)
2. HALF BRIDGE
Output (mv/v) is equal to Gauge Factor times the difference of the strains (µm/m)
divided by four thousand.
NUMERICAL EXAMPLE:
For measuring normal strain, with a gauge factor of 2, and strain1 of 500 µm/m
(tension), strain2 of -150 µm/m (compression), the output wll be:
2 . (500 – (-150))
-------------------------- = 0,325 mv/v
4 000
(See Page 8)
Output (mv/v) is equal to Gauge Factor times the difference of the strains (µm/m)
divided by four thousand.
NUMERICAL EXAMPLE 1:
For measuring normal strain, with a gauge factor of 2, and strain1 of 500 µm/m
(tension), strain2 of -150 µm/m (compression), the output wll be:
NUMERICAL EXAMPLE 2:
July 1996
Modified March 2000
Louis Eder, Instruments for Engineering Measurement, Germiston.