www.nature.

com/scientificreports

OPEN Temperature field analysis and

compensation improvement of
load cell
Shudong Zhuang1, Wen Yang1, Yuxiang Zhou1, Ying Zou1, Chang Liu2, Le Zhang1,
Miao Tong1 & Jinlong Ma1
During the operation of load cell, heat is generated by the strain gauge and the electronics on
the PCB board, which leads to temperature gradients within the sensor itself. These temperature
gradients are unstable at different ambient temperatures. Compensation inaccuracies can also
occur when compensating for sensor measurements at different temperatures This paper proposes
a method to change the position of temperature compensation resistors to address errors caused by
the temperature field effect of the strain gauge sensor itself. Without affecting the sensor’s strain
measurement, the correctness of the proposed method is demonstrated through steady-state thermal
simulation results in ANSYS and experimental results, effectively addressing errors caused by unstable
temperature gradients during the operation of strain gauge sensors.

Keywords Load cell, Temperature gradient, Temperature compensation, Compensation resistance, Thermal
simulation

The load cell, also known as a force sensor, has the advantages such as high precision, high reliability, high
sensitivity, high linearity, small size, and mature manufacturing technology1, and is widely used in robotics2,3,
medicine4–7, agronomy8,9, vehicle science and other fields10.
However, during the production and use of load cells, many performance compensations are required, including
hysteresis compensation11, creep compensation12, zero compensation13, and sensitivity improvement14,15. When
the external temperature changes, the thermal expansion and contraction effects and residual stress will affect
the performance of the load cell16. The performance of the load cell varies at different temperatures, which leads
to the complexity of sensor compensation. According to OIML R60 regulations, as long as the performance
compensation at -10℃, 20℃, and 40℃ is within the acceptable range, the performance compensation of the load
cell is considered to be up to standard. Luo L et al.17 showed that within a certain temperature range, the output
voltage, output linearity, and sensitivity of the load cell decrease with increasing temperature, and temperature
compensation is performed on the sensor according to the principle of double Wheatstone bridge compensation.
Dadasikandar K et al.18 studied the use of polysilicon as a piezoresistive material for measuring pressure in high-
temperature environments, conducted relevant simulations from the perspective of materials, and improved the
sensitivity and other performance of the sensor. Yi J H19 studied the factors affecting the thermal effects of load
cells, including different strain gauge resistances, unequal lengths of Wheatstone bridges, and different TCRs
(Temperature Coefficient of Resistance) of thermal compensation resistor nickel sheets. The results showed that
it is impractical to include TCR differences in compensation, and proposed inserting a thermistor into one of the
four bridge arms to compensate for zero drift caused by temperature. Yi J H et al.20 also proposed connecting a
thermistor in series with the bridge circuit to compensate for the zero point of the Wheatstone bridge. The results
showed that as long as the TCR of the temperature compensation resistor is greater than the TCR of the zero
balance resistor, the two temperature output ratio differences obtained by the iterative method converge to zero.
Du D L et al.21 analyzed the nonlinear relationship between the thermal resistance changes of different materials
and expansion mismatch, and experimentally verified that the temperature characteristic curve is also nonlinear.
Hui Chao Shi et al.22 proposed a temperature compensation method based on artificial neural networks (ANN)
to reduce the additional temperature drift of the resonant frequency caused by electrothermal excitation and
environmental temperature changes. Wang S et al.23 proposed a temperature compensation method based on
backpropagation neural networks (BP-NN) and introduced genetic algorithms to improve BP neural networks,
effectively improving temperature compensation. Based on the above literature, this paper studies the changes
in the temperature gradient caused by the self-heating of the load cell, changes the position of the compensation

1College of Mechanical and Electrical Engineering, Hohai University, 213022 Changzhou, China. 2Department of
Physics, California San Diego University, San Diego, CA 92127, USA. email: wen.yang@hhu.edu.cn

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 1

www.nature.com/scientificreports/

resistor without affecting the strain of the elastomer, and provides a new idea for the temperature compensation
of the load cell. Compared with the above compensation methods, this paper starts from the core of temperature
compensation: the position of the compensation resistor, changes the position of the compensation resistor, and
thus obtains more accurate compensation.
Section 2 of this paper introduces the heating of the load cell and the reasons for the formation of the
temperature gradient, and verifies the feasibility of changing the position of the nickel sheet through finite
element analysis. Section 3 uses the steady-state thermal module of ANSYS/Workbench to conduct thermal
analysis of the load cell and obtains the position with the smallest temperature gradient difference. Section 4
conducts experiments on the sensor based on the finite element results of Sect. 3, and compares and analyzes the
experimental data with the finite element analysis data. Section 5 summarizes and prospects the solutions to the
thermal field effects of the load cell.

The circuit and heating reason of load cell

SG1, SG2, SG3 and SG4 in Fig. 1 correspond to four strain gauges, two compression sheets and two tension
sheets respectively; The Ni is the nickel sheet used to compensate the TCR, and the P resistance is to prevent the
nickel sheet from compensating over the parallel resistance.
In the actual operation of the load cell, the four strain gauge wire grid and strain gauge will heat up, and
there is a certain distance between the nickel plate and the strain gauge, which will cause the local temperature
of the strain gauge to be higher than that of the nickel plate, and there is a temperature difference between the
two. the PCB board also generates heat, and some of the heat is transferred to the strain gauge and nickel sheet
through the elastomer, which makes the temperature field inside the sensor very different. Under a certain
temperature condition, the temperature gradient is constant, and we can correct the temperature gradient by
compensation algorithm. However, when the temperature changes, the temperature gradient will change, and
the compensation amount will also change to ensure the accuracy of compensation.
As shown in Fig. 2, the weighing sensor needs to be fixed with iron and bolts when it is tested before leaving
the factory. The main components of the weighing sensor are: elastomer, strain gauge, PCB board (with electronic
heating element), weld plate and weld cup for sealing.
As shown in Fig. 3, the sensor nickel plate is now attached at 1, and the distance from this position to the
strain gauge is basically equal to that of 2, 3 and 4. It can be considered that the heating of the strain gauge has
the same influence on the temperature of 1, 2, 3 and 4. However, the heat of the PCB board will also be conducted
through the elastomer to the strain gauge and 1, 2, 3, 4. The voltage connected to the sensor is 5 V, and the
current flows through the current acquisition module of the PCB board to the strain gauge. After calculation, the
rated heating power of the integrated electronic components on the PCB board is 0.3 W, and the rated heating
power of the strain gauge is 0.07 W. The distance from 1, 2, 3, 4 to the strain gauge is equal, so the heat generated
by the strain gauge heating has the same effect on 1, 2, 3, 4, if only the strain gauge heating and the PCB board
does not heat, then the temperature gradient generated by the nickel sheet attached to any position is basically

Figure 1. Load cell circuit.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 2

www.nature.com/scientificreports/

Figure 2. Load cell assembly diagram and main parts explosion diagram.

Figure 3. Schematic diagram of heat conduction between PCB board and strain gauge.

constant, but the PCB board will generate heat in actual work. Moreover, the distance between the PCB board
and the strain gauge 2 and 4 is equal, and the position of the nickel sheet is stuck at 1, and the distance between
the PCB board and the PCB board is less than the distance between the strain gauge and the PCB board, which
leads to the heat conduction of the PCB board to the strain gauge direction, the temperature impact on the 1
place will be greater than the impact on the strain gauge, which leads to the change of temperature gradient.
The elastomer material of the sensor is stainless steel, its thermal conductivity is 20 W/(m.K), the specific heat
capacity is 460 J/(kg.℃), and the density is 7850 kg/m3.
The component on the sensor PCB with the strain gauge as an internal heat source follows the heat conduction
equation:
∂T q
= α ∇ 2T + (1)
∂t ρc

Where: T is temperature, t is time, α is the thermal diffusion coefficient, ∇2 is the Laplacian operator (which
describes the change in temperature in space), q is the heat generated by the heat source, ρ is the density of the
elastomer, and c is the specific heat capacity of the elastomer.
Convection is also considered, following Yelifu’s law and Newton’s law of cooling:

a = −λ ∇ T (2)
a = h(T − T∞ )(3)

Where: a is the heat flux, λ is the thermal conductivity, ∇T is the temperature gradient, h is the convective heat
transfer coefficient, T is the surface temperature of the object, and T∞ is the temperature at infinity (usually the
ambient temperature).

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 3

www.nature.com/scientificreports/

Materials Density/(kg·m−3) Young modulus/GPa Poisson’s ratio

All sensor elements 7850 200 0.269

Table 1. Material parameters

Figure 4. Flow chart of thermal coupling calculation.

Thermal coupling analysis of load cell under different working environment

The strain change of the sensor at different patch locations and different temperatures was
analyzed
The sensor used in the experiment is a steel sensor, except for the PCB board and strain gauge, all the other
components are stainless steel, and the position of the PCB board is in the non-strain zone and away from the
strain zone, according to the principle of Saint-Venant, we only consider the components in the strain zone, and
in order to facilitate the calculation, the PCB board is also set as stainless steel during the finite element analysis.
Since the thickness of the strain gauge is 20 μm, we directly treat the strain gauge as a surface. Therefore, all
components are set to stainless steel, and its parameters are shown in Table 1.
Due to the different strain sizes at the four locations, the strain in the strain zone of the sensor may be affected
under different conditions. Before changing the position of the nickel patch, the thermal coupling analysis of the
sensor at full load is carried out to verify that changing the position of the nickel patch has little impact on the
measurement performance of the sensor. The flow chart is shown in Fig. 4.
The sensor elastomer model and nickel sheet model established by CREO were imported into ANSYS/
Workbench for grid division. The thickness of the nickel sheet is about 0.1 mm. The nickel sheet is pasted on
the substrate of the polymer material in the form of resistance wire, and then pasted on the elastomer of the
sensor through a layer of glue. The main body of the nickel sheet is the polymer material of the substrate, and the
resistance wire is only a small part, and the elastic modulus of the polymer material of the nickel sheet is 3GPa to
8GPa. In order to consider the maximum tolerance, the elastic modulus of the entire nickel sheet is set to 8GPa.
After the nickel sheet is built, it is pasted at positions 1,2,3,4. Each area in the sensor forms a common node and
the grid scale does not exceed 1:3. The grid division is shown in Fig. 5.
Binding contact is set between each cutting part. The measuring range of the sensor is 220Kg, so 2200 N force
and fixed support are applied to it, and the axial direction is defined as the X axis, and the direction along the
inner wall of the sensor is defined as the Z axis. X, Y and Z axes are shown in Fig. 6.
First, perform a steady-state thermal analysis, setting the initial temperature to 22 °C and the convection to
horizontal cyclic convection. Calculate the heat generation through the power dissipation of the strain gauge and
PCB. The thickness of the strain gauge can be neglected, and the surface heat generation of the strain gauge is
set to 2.1875 × 10^-4 W/mm², while the volumetric heat generation of the PCB is set to 1 × 10^-3 W/mm³. The
sensor’s elastic body material is stainless steel, with a thermal conductivity of 20 W/(m·K), specific heat capacity
of 460 J/(kg·°C), and density of 7850 kg/m³. Perform steady-state thermal analysis on four sensor models with
different nickel sheet attachment positions at ambient temperatures of 40 °C, 20 °C, and − 10 °C. After obtaining
the results, couple them with the static module. The stress-strain cloud diagram is shown in Fig. 7.
In Fig. 7, the circular area in the middle is the strain zone of the sensor. The measurement performance of the
sensor is affected by the magnitude of its strain zone strain variable. According to the strain characteristics of the
sensor, a path with an Angle of 45°from the level is taken in its strain zone, and the strain on the path is analyzed
after the thermodynamic coupling results are calculated. The strain diagram of the path is shown in Fig. 8.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 4

www.nature.com/scientificreports/

Figure 5. Sensor meshing diagram.

Figure 6. Schematic diagram of sensor constraints.

Figure 7. Sensor full load strain cloud.

Figure 1 is the starting point of the path, 2 is the end point of the path. We extracted and compared the
strain on the path of the sensor at different positions of the nickel sheet at three different temperatures, and the
comparison results are shown in Fig. 9.
It can be seen from Fig. 9 that the strain in the sensor strain zone is almost unchanged at each position under
three different temperatures. In order to better compare the data, the strain on each path is averaged, as shown
in Fig. 10.
It can be clearly seen that when the nickel sheet is attached at different positions and the sensor is fully
loaded at different temperatures, the average strain in the strain zone is almost equal, which will not affect the
measurement performance of the sensor. For this reason, experimental verification is carried out in 4.1.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 5

www.nature.com/scientificreports/

Figure 8. Strain zone path diagram.

Figure 9. Path strain plots at different temperatures for each location.

The difference between the temperature and strain gauge temperature at different nickel
patch locations was analyzed
It can be seen from the thermodynamic coupling results that the measurement performance of the sensor will
not be affected when the nickel sheet is attached to each position. Therefore, the temperature field of the sensor
under different temperatures was analyzed, and the four nickel patch areas were mesh encrypted on the basis of
Fig. 5 to calculate the temperature cloud image of the sensor under 40℃, 20℃ and − 10℃, as shown in Figs. 11,
12 and 13.
The average temperature at the strain gauge and four nickel patch locations was derived, and the data were
processed to obtain Table 2.
After obtaining the above data, the temperature difference between the four pre-selected nickel plates at
different temperatures and the strain gauge is calculated. The calculated results are shown in Table 3.
In the table, △1, △2, △3, △4 are the difference between the average temperature of the strain gauge and
the average temperature at 1, 2, 3, and 4 respectively. At different temperatures, the difference between 2 and 4
and the average temperature of the strain gauge is more stable than that at 1 and 3. The average value of each

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 6

www.nature.com/scientificreports/

Figure 10. Path-averaged strain plots at different temperatures for each location.

Figure 11. 40 °C sensor operating temperature cloud.

Figure 12. 20 °C sensor operating temperature cloud.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 7

www.nature.com/scientificreports/

Figure 13. − 10 °C sensor operating temperature cloud.

Setting temperature(℃) Strain gauge(℃) Location1(℃) Location2(℃) Location3(℃) Location4(℃)

40 41.21984 41.15311 40.98741 40.8463 40.98501
20 21.46874 21.41168 21.23566 21.08545 21.23316
-10 -6.48964 -6.51480 -6.72467 -6.90466 -6.72744

Table 2. Thermal simulation temperatures at different locations at different ambient temperatures during
sensor operation.

Setting temperature(℃) △1(℃) △2(℃) △3(℃) △4(℃)

40 -0.06674 -0.23243 -0.37354 -0.23484
20 -0.05707 -0.23308 -0.38330 -0.23558
-10 -0.02516 -0.23503 -0.41502 -0.23780

Table 3. Difference between each position and the temperature at the strain gauge at different ambient
temperatures during sensor operation.

location root-mean-square error×10−3

location1 17.76656
location2 1.10349
location3 17.70715
location4 1.25689

Table 4. Root mean square error of temperature at each position.

group of data is calculated. The average value is the predicted value of this group of data, and the root-mean-
square error of each group of data is calculated.

n 2
i=1 (Pi −O i ) (4)
RSM E =
n

Where: Pi is the actual temperature value, Oi is the predicted value, that is, the average value of this group of
data, the root mean square error obtained is shown in Table 4.
The results of thermal analysis and later data processing show that the root-mean-square error of temperature
difference between position 1 and position 3 is much larger than that between position 2 and position 4. This
shows that when the nickel sheet is stuck at position 2 and position 4, the temperature difference between
position 2 and position 4 and the strain gauge is stable when the ambient temperature changes, and the error
generated by temperature compensation for the sensor will be very small, far less than position 1 and position
3, but the root mean square error at position 2 is slightly less than the root mean square error at position 4.
Therefore, we choose position 2 as the best position for the experiment.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 8

www.nature.com/scientificreports/

Experimental verification
Verify the influence of the nickel sheet paste on the sensor range
In Sect. 3.1, thermodynamic coupling strain analysis was performed on the four pre-selected patch positions of
the weighing sensor. The results showed that no matter the nickel plate was attached at any position, the strain
variable of the sensor was not significantly affected, which also indicated that the measurement performance of
the sensor was not affected. In order to verify the results in Sect. 3, the position of the nickel sheet has no effect
on the measurement performance of the sensor, so the nickel sheet is stuck at position 2 for the experiment, and
then compared with the initial nickel sheet position 1. The initial position of the nickel sheet is shown in Fig. 14.
Based on this idea, the following experiments are designed.
Experimental tools: 220 kg range weight sensor, static gravity machine, fixture for fixing the weight sensor,
nickel sheet, and sampling equipment.
experimental procedure:

1. The nickel plate position is attached to the position 1 (original position), and then the weight sensor is fixed
with a fixture, the force machine preloading, preheating for 30 min before loading to make its internal tem-
perature field stable, the load force is the full load force of the sensor 2200 N, the purpose is to eliminate the
installation stress of the weight sensor.
2. Apply a small force to the weighing sensor, the position of the force machine is fixed, record the data at this
time and define it as zero. Then load to half of the range, record the load at this time, then load to full range,
record the load at this time, and then unload to half of the range, record the load at this time, and finally
completely unload, record the zero point after unloading again. Repeat three times and take the average value
of each data.
3. Remove the nickel sheet from the original position of the sensor, paste it to position 2, and repeat steps 1 and
2.

The state of the force machine used in the experiment and the loading of the sensor is shown in Fig. 15.
According to the data obtained in the experiment, the sensor lag value when the nickel plate is stuck in
different positions is calculated, and the experimental data are shown in the following table. In addition, the data
at position 1 is defined as the initial value, and the data at position 2 is compared with the data at position 1, and
the data obtained is shown in Table 5.
In the above table, “ppm” represents one millionth of the unit. The lag value at position 2 is unchanged
compared to position 1, and the full load value is reduced by 1.33 ppm, which has no impact on the performance
of the sensor.

Verify the influence of the sticking position of the nickel sheet on the internal temperature
field of the sensor
In Sect. 3.2, we conducted steady-state thermal analysis of the sensor, and the results of the obtained data
show that the root-mean-square error of △2 and △4 is smaller than that of △1 and △3 and under different
temperature conditions. As the distance between position 2 and position 4 and PCB board is equal, the finite
element analysis results show that the average strain of position 4 is slightly less than that of position 2, so
position 4 is tentatively the best position, so the pt1000 thermistor is affixed to position 1, position 3 and position
4 respectively. Since strain gauges are affixed to the position of strain gauge, the thermistor is affixed to the
position close to the strain gauge. Its resistance is 1000Ω at 0℃, as shown in Fig. 16.
After the pt1000 was pasted, in order to simulate the real working condition, it was sealed with a welding
bowl and tested in a high and low temperature test chamber. The resistance of the pt1000 thermistor will also
change with the change of temperature, and the change of temperature roughly presents a linear change, so as
long as the resistance value of each thermistor under different temperature conditions is measured, and then

Figure 14. Schematic diagram of the position of the original nickel sheet.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 9

www.nature.com/scientificreports/

Figure 15. Schematic diagram of force machine loading.

Type Location1 Location2

Full load value(g) 225544.10 225543.80
lag(ppm) 292.40 292.40
△Full load value(ppm) 0 -1.33
△lag(ppm) 0 0

Table 5. Hysteresis to full load difference for position 1 and position 2.

Figure 16. Schematic of pt1000 thermistor location.

according to the relationship between the pt1000 thermistor and temperature, the temperature at the position
can be calculated. The experimental diagram of the temperature box is shown in Fig. 17.
Based on this idea, the design of the experiment is as follows.
Experimental tools: 220 kg range weighing sensor, several pt1000 thermistors, high and low temperature test
chamber, resistance meter, several copper wires.
experimental procedure:

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 10

www.nature.com/scientificreports/

Figure 17. Experimental diagram of the temperature box.

Setting temperature(℃) Proximity gauge(Ω) location1(Ω) location2(Ω) location3(Ω)

40 1160.11373 1159.82247 1159.17790 1158.63627
20 1083.59201 1083.37411 1082.69202 1082.15112
-10 974.56442 974.51750 973.64883 972.94360

Table 6. Resistance at different positions at different ambient temperatures during sensor operation.

1. Attach a pt1000 thermistor to the sensor at position 1, position 3 and position 4 and near the strain gauge,
and connect each resistor to the resistance meter with copper wire, and the resistance meter is connected to
the computer sampling number.
2. put the sensor into the high and low temperature test chamber, set the temperature of the chamber to 20℃,
6 h after the temperature in the chamber is stable, use the computer to collect data.
3. put the sensor into the high and low temperature test chamber, set the temperature of the chamber to 40℃,
6 h after the temperature in the chamber is stable, use the computer to collect data.
4. put the sensor into the high and low temperature test chamber, set the temperature of the chamber to -10℃,
6 h after the temperature in the chamber is stable, use the computer to collect data.
5. After collecting the data, average the temperature at each position and obtain the following Table 6.

After the average temperature of each position is obtained, we calculate the average temperature of each place
through the relationship between pt1000 resistance and temperature. When − 200℃ < t < 0℃, the expression
between its resistance and temperature is:
 
Rt = R0 · 1 + A · t + B · t2 + C· t3 · (t − 100) (5)

When 0℃< t < 850℃, the expression between the resistance and temperature is:

Rt = R0 · (1 + A · t + B · t2)(6)

Where: is the resistance at temperature t, the resistance of pt1000 when R0 is 0℃, and its size is 1000Ω, A, B,C are
the temperature coefficient, and its size is: A = 3.9083 × 10−3,B =-5.775 × 10−7,C =-4.183 × 10−12, t is the actual
temperature at the position.
According to the above formula, we use the nonlinear solution function in Python and Scipy library: scipy.
optimize.root function to calculate the temperature of the thermistor at each position at each temperature, and
the obtained data are shown in Table 7.
After obtaining the above data, the temperature difference between the three pre-selected nickel plates at
different temperatures and the strain gauge is calculated, and the calculated results are shown in the following
Table 8.
In the table, △1, △3 and △4 are the difference between the average temperature of the strain gauge and
the average temperature at 1, 3 and 4 respectively. At different temperatures, the difference between 2 and 4 and

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 11

www.nature.com/scientificreports/

Setting temperature(℃) Proximity gauge(℃) Location1(℃) Location2(℃) Location3(℃)

40 41.21866 41.14322 40.97627 40.83599
20 21.45635 21.40031 21.22461 21.08534
-10 -6.50183 -6.51380 -6.73564 -6.91570

Table 7. Temperature at different positions at different ambient temperatures during sensor operation.

Setting temperature(℃) △1(℃) △2(℃) △3(℃)

40 -0.07544 -0.24239 -0.38267
20 -0.05604 -0.23174 -0.37101
-10 -0.01197 -0.23381 -0.41387

Table 8. Difference between each position and the temperature at the strain gauge at different ambient
temperatures during sensor operation.

Location Root-mean-square error×10−3

location1 26.55415
location2 4.61003
location3 18.10091

Table 9. Root mean square error of temperature difference at different ambient temperatures at each location.

Figure 18. Root mean square error of temperature difference at each position with respect to the strain gauge
at different temperatures.

the average temperature of the strain gauge is more stable than that at 1 and 3. The average value of each group
of data is calculated. The average value is the predicted value for that data set and the root-mean-square error is
calculated for each data set as shown in Table 9.
The comparison between experimental data and finite element simulation results is shown in Fig. 18.
The experimental results show that the root-mean-square error of the mean temperature difference between
position 2 and strain gauge is the smallest, that is, the mean temperature difference between position 2 and strain
gauge tends to be stable at different temperatures, which is consistent with the finite element simulation results.

Conclusion
CREO was used to model the sensor and other components. Through ANSYS/Workbench thermal coupling
simulation analysis, changing the position of the nickel sheet had no impact on the measurement performance
of the sensor. Then, through steady-state thermal analysis of each position, the optimal position of the patch was
found, and the authenticity of the simulation was verified through experiments. It provides a theoretical basis for
compensation of load cell. It is evident in the paper that:

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 12

www.nature.com/scientificreports/

1. because the experimental temperature of the sensor is -10℃, 20℃, 40℃. The span is 50 ℃, during this pe-
riod, the Young’s modulus of the sensor elastomer material is almost constant, so at these temperatures, the
deformation of the elastomer at full load is almost constant. At different temperatures, the nickel plates were
attached to the four pre-selected positions, and the thermal coupling was carried out. The results showed that
no matter where the nickel plates were attached, the measurement performance of the sensor was very small.
2. At -10℃, 20℃ and 40℃, through steady-state thermal analysis, we found that the temperature difference
between the nickel sheet and the strain gauge was the most stable when it was stuck at position 2 or position
4, which was conducive to the compensation of the sensor.
3. The experiment verifies that the temperature difference between the nickel sheet and the strain gauge is the
most stable when the nickel sheet is stuck at position 2, which is consistent with the simulation results.

This paper starts by changing the position of the nickel compensation resistor and, through a series of studies,
identifies the optimal compensation position for the nickel compensation resistor in the load cell. This method
is more direct and accurate compared to previous studies. In terms of cost, this method involves moving the
nickel sheet from its initial position to another position without incurring additional costs, demonstrating the
feasibility of this method from a cost perspective. This method aids in the temperature compensation of load
cells, allowing them to be used in more precise fields.

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon
reasonable request.

Received: 25 June 2024; Accepted: 16 October 2024

References
1. Doğan, B. K. & Karaçay, T. Experimental fault analysis of polymer hybrid bearing using accelerometer and load cell[J]. Tribol. Int.
185, 108498 (2023).
2. Abdeetedal, M. & Kermani, M. R. An open-source integration platform for multiple peripheral modules with Kuka robots[J]. CIRP
J. Manufact. Sci. Technol. 27, 46–55 (2019).
3. Projoth, T. N. & Nanthakumar, P. Analysis and prediction of cutting force through lathe tool dynamometer in CNC turning
process[J]. Mater Today Proc. 46, 4174–4179 (2021).
4. Clausen, I. & Sveen, O. Die separation and packaging of a surface micromachined piezoresistive pressure sensor[J]. Sens. Actuators
A: Phys. 133 (2), 457–466 (2007).
5. Smreczak, M., Rubbert, L. & Baur, C. Design of a compliant load cell with adjustable stiffness[J]. Precis. Eng. 72, 259–271 (2021).
6. Ramachandran, P. et al. In vivo strain alterations in mandibular molars after root canal treatment procedures[J]. J. Endod. 46 (12),
1849–1855 (2020).
7. Montoro-Bombú, R. et al. Validity and reliability of a load cell sensor-based device for Assessment of the isometric mid-thigh pull
Test[J]. Sensors 23 (13), 5832 (2023).
8. Byrne, D. T. et al. Sheep lameness detection from individual hoof load[J]. Comput. Electron. Agric. 158, 241–248 (2019).
9. Saedin, N. S., Muttalib, M. F. A. & Jusoh, M. F. Performance Evaluation of Bar Load Cell Sensing System for Soil Moisture
Measurement[C]. J. Phys. Conf. Ser. 2550 (1), 012013 (2023) (IOP Publishing).
10. Jakati, R. S., Balavalad, K. B. & Sheeparamatti, B. G. ensitivity enhancement in piezoresistive micro-pressure sensor using perforated
diaphragm[C]//2017 2nd IEEE International Conference on Recent Trends in Electronics, Information & Communication
Technology (RTEICT). IEEE 396–399 (2017).
11. Zhuang, S. et al. Analysis of Return-to-Zero Error after the First Load of Load Cell[J]. Sensors 23 (21), 8712 (2023).
12. Bartel, T. W. & Yaniv, S. L. Creep and creep recovery response of load cells tested according to US and international evaluation
procedures[J]. J. Res. Natl. Inst. Stand. Technol. 102 (3), 349 (1997).
13. Lin, H. et al. Zero-point fault detection of load cells in truck scale based on recursive principal component analysis and
comprehensive evaluation method[J]. Measurement 159, 107706 (2020).
14. Kanekal, D. & Jindal, S. K. Prefabrication Analysis and Numerical Modeling of Freely Supported MEMS Piezoresistive Pressure
Sensor Employing Square Shaped Silicon Diaphragm[C]//2023 IEEE 20th India Council International Conference (INDICON).
IEEE, 1259–1264 (2023).
15. Sabhapandit E, Jindal S K, Kanekal D, et al. Numerical simulation of a striated piezoresistive MEMS pressure sensor on circular
silicon diaphragm: a finite element method-based study[J]. Nano 18 (04), 2350023 (2023).
16. Kanekal D, Jindal S K. Investigation of MEMS piezoresistive pressure sensor with a freely supported rectangular silicon carbide
diaphragm as a primary sensing element for altitudinal applications[J]. Silicon 15 (4), 1947–1959 (2023).
17. Luo, L. X., Xu, B. T. & ,Bi, H. L. Research on compensation method of temperature drift in pressure Sensor using double wheatstone-
bridge Method[J]. Adv. Mater. Res. 1643 (459–459), 311–314 (2012).
18. Dadasikandar, K. et al. Formulation analysis and follow-up study of multi-turn configuration for performance enhancement
of MEMS piezoresistive pressure sensor utilized for low-pressure measurement applications[J]. Sensor Review 44 (4), 462–476
(2024).
19. Yi, J. H. Temperature dependence of zero point in force transducers: material properties of strain gages causing such dependence[J].
Exp. Tech. 38 (5), 64–69 (2014).
20. Yi, H. J. & Kim, H. J. Temperature dependence of Zero Point in Force Transducers II: how to Account for the effects of a Zero
Balancing. Resistor[J] Experimental Techniques 40 (1), 221–225 (2016).
21. Du, D. L. et al. Temp. Characteristic Platinum Piezoresistive Press. Sensor[J] Key Eng. Mater. 1304 (483–483), 180–184 (2011).
22. Huichao, S. & Wenlong, L. Temperature Compensation Methodology based on Artificial Neural Network in a MEMS Electro-
Thermal Excitation Resonant Pressure Sensor[J]. J. Residuals Sci. Technol. 13(8), 1884–1893 (2016).
23. Wang, S. et al. Temperature compensation for MEMS resonant accelerometer based on genetic algorithm optimized backpropagation
neural network[J]. Sens. Actuators A: Phys. 316, 112393 (2020).

Author contributions
S.Z. was in charge of researching progress, conceived the innovation, guided the data analysis, and wrote the
manuscript. W.Y. finished the theory exploring, calculated the simulation results, and contributed to the draft.

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 13

www.nature.com/scientificreports/

Yuxiang Zhou . and Ying. Zou. explored the structure model, made an initial beginning work. C.L. and L.Z.
collected the experimental conditions, and guided the literature review. M.T. and J.M. conceived the ideas and
guided the researching progress. All authors have read and agreed to the published version of the manuscript.

Declarations

Competing interests
The authors declare no competing interests.

Additional information
Correspondence and requests for materials should be addressed to W.Y.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives
4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in
any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide
a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have
permission under this licence to share adapted material derived from this article or parts of it. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and
your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain
permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/
licenses/by-nc-nd/4.0/.

© The Author(s) 2024

Scientific Reports | (2024) 14:26213 | https://doi.org/10.1038/s41598-024-76688-0 14