Which value have the results you get?

Are the three values identical?

What happens if you replace the strain gauge with fixed resistors (1.5 k, 1.8 k, 2.0 k)?

BTW, if the resistance of the strain gauge is about the same as that of the other three resistors in the bridge then the input voltage range at your difference amplifier will be located around zero. In this case, R6 should not be connected to GND but to about +2.5 V. Alternatively, you could replace R6 by two resistors of 3.6 k in series to GND, and add two resistors of 3.6 k in series between the +input and +5 V.

Nazanin Nafisi There are several probable explanations for why you aren't noticing any changes in output voltage when stretching the sensor:

1. The sensor might be out of calibration. To obtain reliable results, the sensor should be calibrated before use.

2. The excitation voltage may be insufficient. The excitation voltage is the voltage that is provided to the strain gauge in order for it to produce a quantifiable output. A strain gauge typically requires an excitation voltage of 5-10V.

3. The amplifier gain may not be properly set. The amplifier amplifies the modest voltage variations produced by the strain gauge. Check that the amplifier's gain is properly set.

4. It is possible that the connections between the sensor, amplifier, and Arduino are not secure. Check all connections to verify they are securely attached and making excellent contact.

5. The sensor might be linked to the wrong position. Ascertain that the sensor is properly linked to the point where the strain is to be measured.

6. The sensor might be broken. Check that the sensor is not damaged; if it is, the output will be erroneous.

Check the sensor datasheet for any special instructions for the sensor you're using, and speak with an expert if you can't fix the problem.

I see several issues in your circuit: It has a too low amplification, it loads the sensor significantly, and the differential amplifier is not able to operate with negative input signals. Also the ancient LM358 is not the proper IC for this task. I would recommend to follow the already published ideas, for instance, in this application note: https://www.ti.com/lit/pdf/sboa247. The presented circuit should be able to provide the right signal also for an Arduino board without the necessity of changing any value. For the reference (signal Vref), the Arduino's AREF terminal can be used.

Ivo Cermak , I agree except for AREF: It's an input of the Arduino board, so one still needs a voltage source resp. a voltage divider.

I am not an expert regarding the Arduino boards, but according to the data sheet of the used ATMEL MCU, the AREF pin is a reference input or output. For the discussed application, it needs to be configured as reference output or an external reference needs to be used.

In any case, the AREF has to be connected to the signal Vref of the external circuitry since the Arduino's ADC has to work with the same reference voltage. The reason is to perform a ratiometric conversion, where the reference voltage does not matter. Therefore, the AREF=Vref can be connected to any reasonable voltage that the MCU can accept or that it generates by itself.

Oops, sorry, I was wrong: If AVcc or the internally generated voltage of 1.1 V is selected as reference voltage for the ADC, the same voltage is indeed available at the pin AREF.

Unfortunately, a voltage of about 2.5 V is missing, and 1.1 V would not work as well.

There is a regulated 3.3 V output on the Arduino boards. Perhaps the simplest way to get 2.5 V without too much ripple is to feed this 3.3 V into a voltage divider, followed by a capacitor.

Regarding the reference voltage, the circuit runs with virtually any voltage, the 1.1/V reference of the Arduino board would be fully sufficient - for the reference voltage Vref as well as for the excitation voltage Vex. If the ADC can run with a gain of 1/2, as the data sheet of the MCU used in the Arduino boards suggests, the ADC input range is 0 to 2 x AREF. The design https://www.ti.com/lit/pdf/sboa247 provides with the presented values output voltages between 0 and an 2 x Vref, i.e. if you set Vref=Vex=AREF, the voltage output should match the ADC input almost perfectly.

I would recommend to replace the ICs by the LTC2055, 2x LTC2054 or any other similar type. These ICs are DC precise and would not introduce offset errors like the original TLV9002 does.

The TLV9002 has a GBW of 1 MHz, the LTC205x about 500 kHz, at an amplification of 1000, this results in a bandwidth of less than 1 kHz. Please keep this in mind, if a faster response should be achieved, other ICs would be required.

It may also be a good idea to insert an RC filter between the output of the second IC and the ADC input to reduce the (pickup) noise and limit the bandwidth to a value lower than the Nyquist frequency of the ADC.

There were many other improvements possible if the circuitry should have a fast response or if the noise level should be lowered to the physical level given by the strain gauge. This would be, however, out of the scope of the present communication.

An input range of the ADC from 0 to 2 x AREF would be convenient but unfortunately it is otherwise. The datasheet of the ATmega328, which is used in Arduino Uno boards, states:

"... The minimum value [of the ADC result] represents GND and the maximum value represents the voltage on the AREF pin minus 1 LSB. Optionally, AVcc or an internal 1.1 V reference voltage may be [internally] connected to the AREF pin ..."

Therefore, Vref = Vex = AREF would clip the positive half of the output voltage. AREF = 2 * Vref seems a good choice, with AREF = 5 V or AREF = 3.3 V, and accordingly Vref = 2.5 V or Vref = 1.65 V.

Ivo Cermak Thanks for your explanation.

I just have another question regarding this datasheet that you sent.

Our strain gauge has 2 120 ohm and one 350 ohm resistors. How can I change the circuit for that 350 ohm resistor?

Joerg Fricke Thanks for your response.

I usually get some results around 2.6 V and sometimes around 0.6 V and it is almost constant during the test. (I think it might be because of a shot circuit)

Qamar Ul Islam Thanks

I will check those too.

The resistance if the strain gauge does not have any direct influence on the function of the circuit. The output voltage of the gauge is fed to the positive inputs of the operational amplifiers - they have large input impedance and do no load the gauge in any measurable way. The gauge resistance has, however, influence on the current drawn from the voltage Vex, if you like to use a low-ohmic one (e.g. the 350 ohm), it would be advantageous to use lower reference and excitation voltage to reduce the current and also the power dissipated in the strain gauge. I think that the operation with the 1-V reference of the MCU should be fine. The only really important parameter of the gauge is the gauge factor, see, for instance, https://en.wikipedia.org/wiki/Strain_gauge. It determines which differential voltage the gauge produces at a certain mechanical stress if it is supplied by 1 V. You should use the one with a larger gauge factor to get a larger signal.

Regarding my last post about the voltage magnitude, I referenced to Arduino boards with the SAMD21 Cortex-M0+ ARM MCU, see https://content.arduino.cc/assets/mkr-microchip_samd21_family_full_datasheet-ds40001882d.pdf. For its ADC, see page 891, where the gain and other parameters are specified. With the gain of 1/2, the ADC is able to measure voltages between 0 and 2 x AREF, i.e. the operation with AREF as reference and excitation voltage of the circuit should be fine.

:|

If you put a nice multimeter across the points of V1 and V2 do you see a change in voltage as you stretch the gauge?

I'd unhook the op-amp and Arduino for now, and just see if the strain gauge is happy: although that all looks fine. On its own the bridge ought to behave like a standard 'quarter bridge'.

https://www.bestech.com.au/wp-content/uploads/2020/04/quarter-bridge.jpg