Bachelor in Electrical and Computer Engineering

Department of Electrical and Computer Engineering

INSTRUMENTATION AND MEASUREMENT (IM)
L.EEC030 — 3rd curricular year — 2024/2025

Design and Implementation of a Precision Scale

Objectives

• Study, design, implementat and validate the measurements of a scale device based on
strain gauges, capable of measuring up to 200 g with a resolution of 1 g.

Materials

• Breadboard, resistors, capacitors, wires, zener diodes (6V8 or 8V2), opamps (e.g. TL081/2
or LM385) and instrumentation amplifier (INA126)1 .

Introduction

This work presents the design of a precision scale based on a four-active-arm Wheatstone
bridge, with each arm consisting of a strain gauge. The system uses a single-point load cell
bar made of aluminum, featuring two holes in the middle to define weight-sensitive areas (see
Fig. 1). Four strain gauges are pre-attached to these regions—two on the top and two on the
bottom of the holes—all covered with epoxy for protection. Due to tension and compression,
the resistance of each strain gauge increases or decreases, respectively. These strain gauges are
electrically connected in a Wheatstone bridge configuration.

Figure 1. Single point load cell with support for scale. The electrical connections to the Wheatstone
bridge are available by means of wire-to-board terminals.

Each group will be provided with a device similar to the one shown in Fig. 1. This structure
must be returned at the end of the class and cannot be taken home.

1
Resistors, capacitors, and wires are available in the component storage drawers of the lab, while any addi-
tional components should be requested from the lab technician.

Cândido Duarte 1 of 9
 IM ⋄ 2024/2025

Project Proposal

The goal of this project is to design, implement, calibrate, test, and demonstrate a precision
scale system capable of measuring weights up to 200 g. The standard approach enables the
design of a system with 1 g resolution, using DC excitation. The main challenge will be noise
and offset fluctuations. However, by using short wires, proper filtering, and careful breadboard
assembly, you can achieve the desired resolution.

+12 V

voltage VREF bridge

reference excitation

−12 V V+ V−
OUT+
wheatstone differential V1 voltage V2 (active) V3 01.00
bridge amplification amplifier filtering
OUT− DMM

Figure 2. Block diagram.

There are several possible circuit topologies, but all should be based, to some extent, on the
following approach—see the general block diagram in Fig. 2:

• Bridge excitation: Provide a stable electrical excitation (current or voltage) to the

Wheatstone bridge. Some examples are provided later, but you are encouraged to pro-
pose alternatives, as long as the device’s operational limits are not exceeded. The recom-
mended global power supply voltage for this design is ±12 V (bipolar), while the bridge
power supply should remain below 9 V (i.e. V+ − V− < 9 V).

• Amplification: Typically, an instrumentation amplifier is applied to the bridge output

to reject common-mode signals and amplify the differential voltage. In this project, the
INA126 will be used. Initially, you can test it with a gain of approximately 1000 V/V
and adjust it as needed for your final circuit. A second stage may be required to further
amplify the voltage to ensure it falls within the appropriate dynamic range.

• Filtering: Low-pass filtering may be necessary for this design, with options ranging
from simple passive RC filters to first- or second-order active filters, depending on the
level of noise.

• Data conversion: This will be performed using the bench digital multimeter (DMM),
where 1 g corresponds to 00.01V and 100 g corresponds to 01.00V. Design your mea-
surement chain so that the output voltage range is suitable for these values2 .
2
Different ranges/conversion factors are also possible as long as well justified.

Cândido Duarte 2 of 9
 IM ⋄ 2024/2025

Lab Work

The strain gauge implemented in the scale device is shown in Fig. 3, with the connection
terminals identified. The value of R0 is 1000 Ω ± 1% and the resistance variation is about
0.2 mΩ/g.
V+

R0 ·(1−r) R0 ·(1+r)

OUT+ OUT−

R0 ·(1+r) R0 ·(1−r)

V−

Figure 3. Strain gauge Wheatstone bridge with terminals labels indicated (R0 = 1000 Ω ± 1 % and
R0 · r = w · 200 mΩ, with w in grams).

Although the labels used are in conformity with the manufacturer, these may cause some
ambiguity. Do not confuse V + and V − with the amplifier power-supply pins. They share the
same nomenclature but in this work the bridge is never electrically connected directly to the
power supply.

Some design advises:

• In any implementation, do not allow currents above 4.5 mA in each bridge branch, i.e.
the voltage between V+ and V− should be kept under 9.0 V;
• Furthermore, use ±12 V as power-supply to the remaining circuits;
• The wires should be kept with the minimum length as possible. Even in the connection
to the bridge, do not use very long wires, 12 cm should suffice (or longer but twisting
each pair);
• For each block implemented, test it individually. The gain in each block may be adjusted
later, when all the blocks are implemented. At the beginning phase just ensure that each
block works reasonably. Once all the chain is defined (each block adjusted), isolate each
circuit again and characterize it for the report.

Bridge Excitation
A simple approach for a voltage-mode circuit is shown in Figure 4. It protects the bridge
against over-voltage at the same time it provides a stable supply and prevents over-current
through the bridge (the resultant current is less than 4.5 mA in each arm).

Here it is important to keep the zener in regulation, e.g. determining R1 and R2 so that
it is provided about 5 mA to the zener (additionally, R1 = R2 improves the common-mode
rejection—why?). If the design defines Iz > 5 mA, just check from the datasheet if the power
rating of the zener is not exceeded.

Cândido Duarte 3 of 9
 IM ⋄ 2024/2025

+12 V

R1
560 Ω
<9 mA
V+ = +4.1 V

6 mA
R0 ·(1+r) R0 ·(1−r)
Dz
OUT+
8V2
R0 ·(1−r) R0 ·(1+r)

V− = −4.1 V
R2 OUT−
560 Ω

−12 V

Figure 4. Simple voltage-mode bridge excitation.

Check with the technician the availability of 8V2 zener diodes. Nonetheless, also a 6V8
zener can be used, although providing inferior performance (why?). In such a case, new values
for R1 and R2 need to be calculated.

The bridge shown in Figure 4 is available for download in Moodle, for simulation with
LTspice, under the name scale v1.asc. In any LTspice file provided, you may need to install
missing models for some devices, like the zener and amplifiers—check Moodle for the installation
procedures.

Fig. 5 shows another approach in which if R3 = R4 the performance is optimized (why?).

You can use the values depicted in figure, but an adequate justification on how these are
obtained must be provided in the report.

This circuit is also available for download in Moodle, for simulation purposes with LTspice,
under the name scale v2.asc. Try to understand its operation and determine the voltage
values at each node, as well as the current at each branch and verify it with the simulator before
assembling. Also, for guidance, before assembling take note of the pin numbers in a schematic
sketch. For instance, be aware of the pinout differences of the TL082 comparing to the TL081—
it is common to overlook the datasheet (which is the same for both) and accidentally swapping
pins can damage the amplifier permanently.

Amplification

Amplification is obtained by the instrumentation amplifier together with a subsequent am-

plifier stage. For this following stage, simple inverting or non-inverting operation amplifier
configurations can be used for such a purpose—choose one, not both.... You will need an addi-
tional opamp, which can be requested to the lab technician—you can use the LM358 or TL082.

Cândido Duarte 4 of 9
 IM ⋄ 2024/2025

+12 V
OUT+
+12 V
R1 R0 ·(1−r) R0 ·(1+r)
1 kΩ −
1
R2 2
TL082
+ V+ = +4.4 V
12 kΩ R0 ·(1+r) R0 ·(1−r)
Dz R3 −12 V
6V8 22 kΩ OUT−
R4 V− = −4.4 V
22 kΩ
−
1
2
TL082
+

Figure 5. Alernative voltage-mode bridge excitation.

First of all, simulate the circuit in LTspice. Furthermore, always use capacitors in the power
supply rails, like in the previous works. Once assembled, you can check the performance impact
(noise amplitude) with the oscilloscope in ac coupling, both checking the power supply nodes
and the output of the amplifiers, with and without capacitors.

The noninverting configuration is shown in Figure 6 (a). It allows also a precedent low-pass
filter, which can be designed e.g. with R = 47 kΩ and C = 330 nF. Do not add any capacitor
in parallel with P1 (why?).

The inverting configuration is depicted in Figure 6 (b). It also allows an additional filter
option. It has the disadvantage of changing the polarity of the signal, but it also is more
robust in terms of common-mode performance comparing with the noninverting amplifier. If
the polarity is an issue you can interchange the instrumentation amplifier inputs, or use the next
stage in an inverting configuration, or at the output obtain the signal with inverse polarity—
there are many options available. Also the cut-off frequency changes if the optional capacitor
is used.

At this point, since you do not have the complete measurement chain yet implemented, just
admit a reasonable gain sufficient to see the voltage output due to some weights. You can later
adapt the value to the most adequate option—as in RG , which can also be changed accordingly
to the overall amplification requirements.

Offset Compensation

The same opamp used for the improving the gain can be used to compensate the overall
offset in the amplifier chain, e.g. using an opamp subtractor configuration. Another way
is to use a TL081 that has two pins that can be used with a potentiometer to change the
offset (check its datasheet). However, if you have the gain of INA126 higher than 5,000, it is
practically mandatory to cancel the offset in the REF pin of INA126 (why?). To perform offset

Cândido Duarte 5 of 9
 IM ⋄ 2024/2025

P1
R1 10 kΩ

+12 V +12 V

OUT+
−
1
Roptional LM358 V2
V1 2
RG INA128 +
ref

Coptional −12 V
OUT−

−12 V

(a)

+12 V +12 V

OUT+
+
1
R1 LM358 V2
V1 2
RG INA128 −
ref
−12 V
OUT−

−12 V P1
10 kΩ

Coptional
(b)

Figure 6. (a) Noninverting and (b) inverting amplifiers.

compensation at the REF pin you need a dedicated opamp, because compensation requires a
low-impedance driving.

Offset compensation can be a daunting task in the present project, but put some faith on it:
fine adjustments are the key to success! Figure 7 (a) shows an implementation applied to the
first bridge excitation proposed earlier. It makes use of the reference pin of the instrumentation
amplifier, hence driven by a low impedance from the added opamp output.

A reference voltage is obtained from the zener terminals, with added resistors ROS1 and
ROS2 to improve the fine trimming of the offset voltage. Care should be taken for the current in
this added resistor series—limit to 250 µA or calculate R1 again so that the zener can operate
in regulation mode. Furthermore, if you need a higher compensation voltage (not likely, but...),
convert the buffer configuration to a noninverting circuit with gain by adding the required
resistors.

Figure 7 (b) depicts an implementation suitable to the circuit in Figure 5. The essential
difference is basically the connection to −12 V to provide the possibility of having negative trim-
ming voltages, but if a negative compensation voltage is not required, ROS2 can be connected
to ground.

Cândido Duarte 6 of 9
 IM ⋄ 2024/2025

+12 V

R1

V+
ROS1
R0 ·(1+r) R0 ·(1−r)
+12 V
P1 OUT+
Dz
10 kΩ
R0 ·(1−r) R0 ·(1+r)
ROS2 RG INA128 V1
ref
V−

R2 OUT−
−12 V
+12 V

−12 V −
1
2
LM358
+
+

C
4.7 µF −12 V

(a)

+12 V OUT+
+12 V
+12 V
R0 ·(1−r) R0 ·(1+r)
R1
−
1
R2 2
TL082 RG INA128 V1
+ V+ ref
R0 ·(1+r) R0 ·(1−r)
Dz R3 −12 V

OUT− −12 V
R4 V−

−
1
ROS1 2
TL082 +12 V
+
−
1
2
LM358
P1
+
10 kΩ
+

C
4.7 µF −12 V

ROS2

−12 V
(b)

Figure 7. Offset compensation (a) applied to the circuit in Figure 4, and (b) for the circuit in Figure 5.

Cândido Duarte 7 of 9
 IM ⋄ 2024/2025

The secret of fine tuning relies in the values chosen for ROS1 and ROS2 . Let us suppose that
you use a 10 kΩ potentiometer and you have the resistive series driven by a voltage VR (e.g.
obtained at the zener cathode). In essence we have the general concept shown in Figure 8 (a).

Now, let us assume you have already tested the circuit and by trial-and-error you found
out that the voltage required cancel the overall offset is VOS
∗
around 2.5 V. So, we can admit
P P
VA = 2.6 V and VB = 2.4 V.

Moreover, since we only need a positive voltage, we connect ROS2 to ground—so, we jump
to Figure 8 (b) which brings benefits.

Then, for a 10 kΩ potentiometer, since its voltage drop is 200 mV (i.e. VAP − VBP ), then the
resultant current is solely 20 µA—a little bit low, in the order of noisy currents (a 1 kΩ would
be preferable in this case, but anyway...). We get ROS2 = 2.4/20 = 120 kΩ. So, knowing VR
you can find ROS1 .

Furthermore, 200 mV might be a too large value, you might consider reducing it to 25 mV or
50 mV. The main obstacle will be the tolerance of resistances, in these configurations sometimes
ROS1 and ROS2 are obtained through the series of two resistors, each.

VR
VR VR

ROS1
ROS1
VAP
coarse ROS1
P1 ∗
VOS VAP P1
10 kΩ 10 kΩ 220 Ω
P1
+

C ∗
VOS
VBP 10 kΩ P2 ROS2
2.2 µF
+

C ∗
VOS
ROS2 VBP 10 kΩ
4.7 µF 470 kΩ
fine
+

C
ROS2 1.0 µF

−12 V
(a) (b) (c)

Figure 8. Fine tuning the offset voltage (a) bipolar and (b) positive only, and (c) coarse/fine tuning.

If you need a more flexible adjustment, with coarse and fine trimming, then the circuit of
Figure 8 (c) is the way to go—you will need to ask the lab technician one more potentiometer.
Just take into account that in this case, you need to lower the resistance connected from the
zener to the power supply, to accommodate the current increase.

Also, do not take the cursors to their extremes, as this will imply the zener to go out of
regulation. A LTspice simulation should precede the implementation in any case above.

Cândido Duarte 8 of 9
 IM ⋄ 2024/2025

Data Acquisition

To display the measured data, we will make use of the work-bench dc voltmeter (from the
DMM). As mentionet earlier, the calibration is supposed to get 0.01V for a weight of 1 g and
1.00V for a weight of 100 g—choose the manual range of the DMM for this purpose. The offset
needs to be quite accurate to get this working properly, so if you use a range of, let us say, mV
to fine tune the offset, then when you change to the manual range proposed, the offset requires
adjustment again. It will be interesting for you to think why this occurs. Anyway, calibrate
only the gain (for a known weight) after offset compensation done.

Testing and Validation

For testing, in each lab there will be a commercial precision scale to be used for reference,
as well as spheres with weights of 1 g and 0.35 g (with some error associated, but that can be
checked in the reference scale). It is not expected that you get the resolution of 0.35 g, you
need to use more than one of these weights. Nonetheless, it is preferable that you get 3 or 4
weights, using a pen, a key chain or whatsoever, to get 1 g, about 100 g, and near 199 g.

Use several known weights (relying on the reference scale) and perform measurements from
the minimum to the maximum and back, to test hysteresis. Take note and plot the results
to validate the weighting range. Obtain several measurements to find an adequate calibration
factor.

Notes. Avoid testing the weight by putting too much force with the fingers on the scale
plate. Also, minimize the weighting time, only apply weights for short periods of time, leaving
the scale plate free from weight most of the time—because of the load cell creep phenomenon.

Cândido Duarte 9 of 9