You are on page 1of 6

Sensors and Actuarors A 55 ( 1996) 173-178

Experimental analysis of a new strain-gage signal conditioner based on a


constant-current method
P. Cappa *, Z. Del Prete, F. Marinozzi
University of Rome La SapietWJ, Department of Mechanics and Aermzautics. ,;u Euclo.uiona /8, 00184 Rome. Jtaly
Received 29 November 1994; revised 15 Decembcr 1995; acccpted 27 February 1996
Abstract
A strain-gage signal-conditioning scheme, which appears to be innovative, has been proposed. lt is bnsed on a potentiometric circuit with
twin constant-current supplies. From the experimental tests carried out on our prototype, the foliowing have been highlighted: a satisfying
sensitivity ( ;; l ,_m m -), a o.;alibration curve that might be considered linear ( non-linearity errorless than l #'m m -) ,low valuesofzero-shift
( < l ,.m m-
1
) throughout 60 hour test, and, finally, a global temperature coefficient that can be considered negligible (0.1 ,.m m - oc-).
Keywords: Constant..:urrent method ; Signa) condilioning ; Stnn gages
l. lntroduc:tion
Among the different strain-gage measurement techniques
proposed, in 1983 the possibility of getting the electrical
resistance variation directly had been analyzcd [ l ] ; such a
configuration was later called the DRM ( direct resistance
method). Compared with the Wheatstone bridge, the DRM
needs a greater number of connection lead wires ( four for
each strain gage) in order to reduce the effects induced by
the lead wires themselves and does not allow the compensa-
tion of the effects induced by temperature variations. How-
ever, it greatly simplifies the measurement equipment as it
can easily be made of an ohmmeter. Such a configuration was
later experimentally examined [2,3] and the advantages of
the DRM hve been confirmed. The effects induced by a
relay-switch contro! unit, not devoted to strain-gage meas-
urements [4,5], bave been analyzed and the DRM's
capability to compensate the voltage drops across the meas-
uring leads has emerged. l t is useful to elucidate that to ohtain
the same results with a traditional Wheatstone bridge needs
highly sophisticated intermediate circuits [6-9]. Then, the
possibility of adopting the DRM even for dynamic measure-
ments utilizing high-speed digitai multimeters [IO] was
examined. Finally, tbc possibility of improving the metrolog-
ical performance by acting on the electrical current direction
to reduce the thermoclectrical phenomena that occur at the
switch contro! unit contacts has been evaluated [ Il].
* Correspondingaulhor. Tel: ( + 396) 44585273; Fax: ( + 396) 4881759;
Email: cappa@inge-ns.ing.unirornal.it.
0924-4247/96/$15.00 lt' 1996 Elsevier Science 1:-'A. Ali rights reserved
PII S0924-424 7 ( 96)01257-5
The tests which bave been carried out so far by utilizing
off-the-shelf instrumentation bave shown the potentialitiesof
the DRM, and the development and realization of a strain-
gage channel foundd on such a princip!e has appeared of
interest. The aim of this paper is, in fact. to compare the
results obtained with a prototype of the DRM dedicated to
strain-gage measurements with those acquired from a com-
mercially available Wheatstone-bridge integrated circuit.
supplied by a constant voltage and characterized by high
metrological qualities. In the following, the prototype that
has been realized will be referred to as DRM-SG, whereas
the Wheatstone-bridge integrated circuit will be referred to
as WB-IC. The signal conditioner proposed he!e and metro-
logically examined is analogous to others conditioners for
electrical resistance gages [ 12,13] and is also capable of
performing a direct A/D conversiol! and a digitai correction
of the signal by means of an iroplemented autocalibration
procedure.
2. Desc:ription of the new straingage signal c:onditioner
The main diffi.:uli} for the realization of a DRM-SG
scheme consists in the need fora strain-gage constanl-current
supply virtually working with no ftutter, that is, with a max-
imum ripple equal to 0.0002% of the nominai value, so as to
appreciate strain values of ;;;;; l J.i.l1l m - considering a gage
factor F= 2. Moreover, in order bave a wide range of indus-
triai uses, it is desirable that such a current supply should be
cheap and easy to find.
174 P. Cappa et al./ Sensors .nnd ActU<'Inrs A 55 (1996) 173-178
SG
2
Fig. l. Completeelectrical scheme ofthe DRM-SG circuit: IC
1
= voltage ultrahigh precision reference; IC
2
= ultralow offset voltage dual op amp; IC, = precision
instrumenlalion amplifier; SG
1
, SG
2
= strain gages witlo nominai resistance of 120 n.
With the aim of realizing the specifications mentioned
above for the strain-gage current supply, the scheme of Fig. l
was proposed. In this configuration each supply J/2IC
2
can
generate a constant current of a value controlled only by the
voltage reference IC
1
and by the resistance R equivalent to
the resistance network R
1
-R
4
, and not affected by the resis-
tance value of the strain-gage SO and by its connection wires.
As an advantage, the proposed scheme needs a constant volt-
age reference, in order to generate a constant current, and
puts out a voltage signal that it is proportional to the strain-
gage SO resistance variation tl.R. Only two Iead wires are
necessary for strain-gage connection to the signal conditioner
and do not significantly affect strain measurements. It is easy
to see that the simple scheme proposed is able to compensate
temperature effects induced both on the gage and on the
connecting wires. Furthermore, there is a simple way to con-
nect in series more than one strain gage, as they are applied
on transducer elastic elements ( pressure transducer,load celi,
etc.).
In order to achieve the previously indicated features, i t has
been decided to fit the well-known double constant-current
potentiometric circuit [ 14], so the 'differential' scheme
shown in Fig. l has been proposed. The indication l ~ t C
2
shows the choice of a dual operational amplifier ( op amp)
for the current supply, made of two op amps on a single IC.
The signals V
1
and V
2
coming from the two arms are the
inputs for an instrumentation amplifier IC
3
and the resistances
R
1
-R
4
are utilized to null out the unstrained configuration and
to regulate the current intensity supplied to the strain gages
S0
1
and S0
2
It seems useful to observe that the scheme in
Fig. l intrinsically has the properties of a Wheatstone bridge,
which are: ( l) possible increase of sensitivity by utilizing
two or four strain gages, appropriately cemented on the meas-
ured point, and ( 2) compensation of temperature effects
As regards the DRM-SO sensitivity, assuming only one
active strain gage is used and first imposing the unstrained
configuration, it is
(l)
where R"Q: i ~ the strain-gage resistance and 1
1
and 1
2
are the
current values flowing in the two DRM-SO arms.
When a resistance variation tl.R
50
, of the strain gage S0
1
occurs, we gel
(2)
Considering VoaM.so=Rs
0
,1
1
and from Eq. (2) we obtain
__g_=tl.Rso,=FE
VoaM-so Rso,
where F is the gage factor and E the strain.
(3)
Comparing the DRM-SO potentiometric circuit with acon-
stant-voltage-supplied WB-IC, both with only one active arm
and, considering that the strain gages are supplied with the
same current intensity either for the DRM-SO or the WB-IC,
it obviously results that
(4)
where Vw
8

1
c is the bridge voltage supply for the WB-IC and
VoaM-so as previously noted, is equal to when the DRM-SO
potentiometric circuit output has been nulled. Thus, as is well
known fora linearized quartrr-bridge configuration:
(5)
and therefore comparing the DR1vt-SG sensitivity, SoaM-So
with that of the WB-IC, Sw
8

1
c. i t is:
P. Cappa et al./ Sensors and Act110tors A 55 ( /996) 173-178
175
:l
7 '...i\
6 \
es '\ r.
'l\
g " \
Lll 3 ', l'
J \
"'\
\
,-
1
,.l
WB rl
l
l
l
-l
l
l
,_,
l
l
-2000 -1000 o 1000
2000 3000
lnput strain [Junlm)
Fig. 2. Non-linearity error for DRM-SG and WB-IC as a functioo of lhe imposed input.
AE/VoaM-so.,.,.,{AE/Vwa-IC)='H' IC
SoaM-SG ARso/Rsot \ARso/Rso. 6oUWII-
(6)
that is, the proposed DRM-SG configuration is intrinsically
linear and, compared to a constant-voltage-supplied WB-IC,
has twice the sensitivity.
3. Test procedure and results
With the aim of evaluating the metrological performance
ofthe DRM-SG circuit, it has been decided to carry out three
different kinds oftests aimed at determining the non-linearity
of tbe calibration curve, the zero-shift and the temperature
coefficient.
As regards the commercially available WB-IC, ti'. has been
decided to utilize a wide-bandwidth strain-gage signal-con-
ditioner integrated circuit. Such an IC supplies the bridge
with a constant voltage and is equipped with an instrumen-
tation amplifier.
Considering medium-sized strain gages with a nominai
resistance of 120 n, applied on a material with agood thermal
conductivity, the value of the electrical current that can ftow
through the transducers without causing noticeable self-heat-
ing phenomena [ 15] has been chosen equalto IO mA. Con-
sequently, on the quoted electrical current, the value of the
voltage supply for the WB-IC is equalto 2.4 V. Il must b,}
underlined that the voltage value obtained in sucb a way turns
out to be lower than the IO V value for wbicb the IC manl-
facturer assures the best metrological performance; the 2.4 V
value is also less than the minimum value, equalto 4 V, for
which the manufacturer suggests the appropriate circuit set-
tings to adopt. Such considerations bave made necessary the
experimental determination of the WB-IC behavior for the
specific voltage supply that bas been chosen. Finally, a board
on wbicb both the DRM-SG and the WB-IC bave been assem-
bled has been realized in such a way as to reasonably consider
a uniform temperature distribution around the ICs.
The sensitivity of the DRM-SG circuit and the WB-IC has
been experimentally assessed by choosing the configuration
witb one active arm and imposng known strain values in the
range 3000 pm m - by means of a Micro Measurements
1550 A Strain Calibrator. The data collected bave confirmed
the double sensitivity of the DRM-SG prototype compared
lo that of the WB-IC (SDRM.so=2.41 mV (pm m-
1
) -,
Swa-IC = I.IOmV (pm m -) -). With the aim ofhighl ;ht-
ing the non-linearity error for the two circuits under
comparison, the data computed from the best straigbt lines
through zero are shown in Fig. 2; t is possible to obse-ve that
in the examined measurement field equalto 3000 pm m -,
the DRM-SG shows an error always between l pm m-
1
,
whereas for the WB-IC the error s included in the range - I
to +9pmm-.
In the zero-shift tests, as well as for the temperature-coef-
ficient evaluation, a constant input has been imposed by util-
izng two 120 n precision resistances from Micro Measure-
ments, specific for strain-gage equipment (temperature coef-
ficient equalto :t: l ppm oc-. stability of25 ppm per year);
furthermore, test-area temperature variations havl" beenmon-
itored with a PtlOO thermometer. The precision resistances
bave been connected on the two active arms for the DRM-
SG and in a half-bridge configuration for the WB-IC. lt was
decided to warm-up the DVMs utilized to measure the two
v out for at least 3 h, whereas, because of the reduced number
of electronic components used, the significant part of the
warm-up transient for the DRM-SG and the WB-IC is limited
to the first 5 min. In the two zero-shift tests, theenvironmental
temperature has been within the l "C interval. Each ofthe
zero-shift tests lasted 60 h and the output of both the DRM-
SG and the WB-IC was acquired every 15 min. The results
of the first zero-shift test are shown in Fig. 3, which shows
that the DRM-SG zero-shift turns out always lower than l
176
-4
40
35
30
e
l25
~ 20
i::
<
5
o
10
e -1o
].
c -20
"i!
u
l-30
~ -40
.so
~ o
20
P. Cappa ~ .. l Sensorsand Actuators A 55 (1996) 173-178
10
20
20 30
Timc[h]
40
Fig. 3. Zero-shift as a function of lime.
40 60 60
Timc[min]
100
Fig. 4. Apparenl sttain as a funclion of lime: hcating test.
40 60 60 100 120 140
Timc[min]
50
180
Fig. S. Apparent sttain as a function oftime: cooling test.
120
180
30
25
60
55
so
45 u
t....
25
30
-5
p.m m-
1
l t is useful to observe that, from the comparative
examination ofthe graphs in Fig. 3, the trend ofthe zero-shift
versus time might be ascribed to the effects induced on the
DRM-SG by the test-area temperature variations. More
specifically, the temperature affects the instrumentation
emplifier IC
3
, see -ig. l, as it is the only element ofthe DRM-
SG circuit for which a sym;:>letrical compensation element is
not present. Such a result, although included within b few p.m
m -, turns out to be more relevant for the WB-IC. The second
zero-shift test has confirmed the results obtained in the first
P. Coppa el al./ SeRSors and Aclllalors A 5S ( 1996) 173-178
177
one; in particular, the effects induced by the environmental
temperature on the WB-IC can be highlighted. Moreover, it
has shown that the DRM-SG has a zero-shift comparable to
thatoftheWB-IC p.mm- comparedto


but it appears to be Iess sensitive to the temperature effects.
To evaluate the temperature coefficient of the DRM-SG,
four tests bave been carried out: two heating tests in the
interval 25-50 C and two cocling ones from 2S down to O
C; for the heating tests a thermal gradient o i ;;; l "C min- l
has been applied, whereas for the cooling or.es the applied
gradient has been 0.4 C min-
1
Tne data bave been
acquired every 15 s, either for the heating or cooling tests.
The results of one heating test are shown in Fig. 4, from which
a mean temperature coefficient can be calculated; for the
DRM-SG it is equa! to p.m m- oc-, wherr.lS forthe
WB-IC a value of p.m m- oc- has been found. The
results relative to one cooling test are given in Fig. 5 and,
even for these tests a significant temperature coefficientequal
to l p.m m-
1
C-
1
has been found for the WB-IC, whereas
for the DRM-SG a value of l p.m m-
1
oc - is confirmed.
From the comparative examination ofthe cooling and heat-
ing tests, a delay ofthe ICs outpu! ,less marked for the cooling
tests, can be observed and seems to be attributable to the
differenl thermal gradient imposed in the cooling tests, which
was 40% of the imposed gradient in the heating tests.
4. Conclusions
From an examination of th, tests carried out in order to
evaluate the metrological performance of a simple and in.:x-
pensive innovative configuration for strain-gage signa! con-
ditioning, the following conclusions can be drawn:
l. An acceptable resolution value ( l p.m m-
1
), a calibra-
tion c"rve tbat can be considered linear (non-linearity
error < l p.m m-
1
in a measurement range of 3000 p.m
m-).
2. The 60 h zero-sbift tests for the DRM-SG bave sbown a
small deviation from the zero ( l p.m m- ! ) ; further-
more, such a value does not seem to be mainly caused by
the variations of test-area temperature, whicb has been
kept constant for the experimental set-up within l C.
3. The temperature-coefficient evaluation tests of the pro-
posed circuit bave therefore higblighted wbat bas previ-
ously been observed, or rather tbe small influence of the
effects induced by temperature on the realized DRM-SG
the temperature coefficient bas been evaluated
to be <1).1 p.m m- oc-.
References
(l] E.J. Nelson, C.D. Sikorra and J.L. Howard, Measuring strain gages
direclly wilhoul signal condilioning. Exp. Techniques, 7 ( 1983) 26-
28.
(2] L. W. Zscbary, K.G. M<:Connellmd N. T. Younis. Accoua1i11g l'or lead
wiJe resistance chmlges md loss of zero iB kmc-term suain
measumnen!S, Sac. E:rp. Mechanks Spring Omf. Exp. Meclwlics.
Albuquerque. NM. USA. U .filM, 1990. pp. 201-204.
(3] P. Cappa. K.G. McConnellmd L.W. 7ac:bary, Zero-sllifl values of
automati<: ar,d il:expensive stnWI gage iiiSinlmenwion syseems, Exp.
Mechllnics, 31 (1991) 88-92.
( 4] P. Cappa. A comparative examilllllioo of IIUtOmati<: sequential direct
systems for strain-gage dala o:eadin&s based OD a low-cost switch-
cootrol tmil. Exp. Techniques. 13 ( 1989) 13-15.
[S] P. Cappa. Umits of stability of IDCIOCIIIIIpll-col!lrolled sca1111iag
systems for strain gage dala readings. E::p. Mec'-ic.. 30 (1990)
300-303.
(6] M. Kreuzer. Comparillg lhe effecl of llllld md Vll
voltage- md current-fed suain-pge circuits. Reporrs Applied Meas ..
l ( 1985) 12-16.
(7] M. Kreuzer, How to avoid enms caused by beat dfects iB strain gage
measumneniS wheo usiag scanniag UIIIS. Proc. VllltA lnt. Omf.
Experimenlal Strm Analysis. Amsterdam. 1M Nerherlmttls. 12-16
May. 1986. pp. 455-464.
(8] WJ. Versel. Compen.<l&lion of leadwire effects willl resislive
straingauges in multi-cbannel suaingauge nstrumenwion, Proc.
VII/rh lm. Conf ExfHrimental Stress Analysis, At:IS1erdam. 1M
12-16 May. /986, pp. 455-464.
[9) B.G. Mahrenbolz. An improved strain-gage signal COIIditioner for
d)'lllllllil: stress measurement. !SA Proc. 37th /nt. IIIStrllmerltslion
Symp .. San Viego, CA. USA. S-9 May 199l, pp. IIIS-1129.
(IO] P. Cappa md Z. Del Prete. An experimenlal anal)'SIS of accuracy md
precision of a bigli speed strain gage system based on lhe direct
resistance rnelhod. Exp. Mechllnics, 32 ( 1992) 78-82.
(11) P. Cappa.Z. Del Prete. K.G. McConncllmdL.W. Zachary,Zero-sllift
evaluation of autumalic strain gage systems based OD direct md
reverse-current method. Exp. Mec'-ics, 33 ( 1993) 293-299.
( 12] LJ.M. Joosten, A hybrid signal COIIditioner l'or measuriag small
resistance varialioos. in applil:alion ofmeasuring crack-length iB meta!
plale unch:r faligue - Uve testing, Colloq. /m. Ekctroniq11e MeSKre
(lnt. Conf Ekctronics Measuremenrs). Paris, France, 26-30 May.
1975. p. 634.
( 13] P. Sente, M. Deladriere md H. Buyse. A 'smart' digilal signal
condiliooer for suain gage bridge 5e1ISOIS, IMEKO TC-4 TeC::.
Comminee on Measurement of Eleclrical Quantiries, lnt. Symp.
lnrelligent lnstrumenrarion for Remote and On-site Measurements,
Brussels, Belgium. 12-13 May. 1993, pp. 188-191.
( 14] J.W. r,111ly md W.F. Riley, Experimental Stress Analysis, McGraw
Hill, New York/Kogakusha. Tokyo. 1978. pp. 243-248.
( 15] Oprniz.arion strain gage e.tcitarion kvels, TN-502. Measurement
Gr,)Up lnc., Raleigh. NC, USA, 1979.
Biographies
P. Cappa was bom in Rome, Italy, in 1956. He receiveda
degree in mecbanical engineering with bonours from the Uni-
versity of Rome La Sapienza in 1980. In the same year he
joined the Mechanical and Aeronautica! Department of the
Faculty of Engineering, University of Rome La Sapienza as
assistant professar. Since 1990 he has been an associate pro-
fessor of sensors and transducers al the University of Rome
La Sapienza. His current interests are innovative trans-
ducers, automatic data-acquisition systems, and biomedical
instrumentation.
178 P. Cappa et al. l Sensors and Actuators A 55 ( 1996) 173-178
Z. Del Prete, bom in 1963, graduated in mechanical engi-
neering within an instrumentation course in 1990 at the Uni-
versity of Rome La Sapienza. He joined the Mechanical and
Aeronautica! Department as an assi: lnt professor in 1992.
His mai n interests in the mechanical field include strain meas-
urement, fatigue test methods, automatic data-acquisition
systems, and biomedica! instrumentation.
F. Marinozzi was bom in Rome, Italy, in 1964. He
graduated in electronics engineering in 1991 at the
University of Rome La Sapienza. He is a doctoral student
in mechanical measurements for engineering. His cur-
rent main research activity involves high-precision
stress-strain measurements with traditional and optical
techniques.

You might also like