Baseline voltage not stabilizing

Thread Starter

kpooja

Joined Mar 20, 2023
4
Hello. I am trying to prepare a filter and amplifier circuit for a strain gauge (Wheatstone bridge in half-bridge configuration). Input power to the Wheatstone bridge is 1 VDC, and each output (+ve and -ve) of the Wheatstone bridge circuit goes the filter.

The filter is a second order active low pass filter with fc = ~16Hz using NE 5532 op-amp. The outputs of the filter go to the differential amplifier INA118. The output of the INA118 is connected to a DAQ. The power supply to NE 5532 and INA118 is +/- 15V with bypass capacitors of 100 nF. All the grounds are common. I have attached the schematic of the circuit diagram.

When I test the soldered circuit, the noise attenuation seems to be satisfactory (~20 mV), but the baseline is not stable at all. It fluctuates ~200 mV or more. Could someone please suggest what I should do to stabilize the baseline? Is there some fault with the circuit or am I missing some component?
Sample output.pngCircuit diagram.png
 

Audioguru again

Joined Oct 21, 2019
6,671
Your circuit biases the input on the very old NE5532 dual opamp with 2.1M ohms but its input current has a max of 1uA.
Then the offset voltage at the input can be as high as almost 2V and it will drift with temperature change.

The NE5532 is a dual so we expect the two opamps have fairly matched input currents but not perfect.
Why not use a modern dual opamp with Fet inputs that have almost no input current like an OPA2134?

I would never use 1M resistors in a 20Hz lowpass filter. Instead I would use 30k resistors and 330nF capacitors.
The 590 ohms of R5 is too low for most opamps to drive. Change R4 to 10k and change R5 to 5.9k.
 

Thread Starter

kpooja

Joined Mar 20, 2023
4
Hello. I am trying to prepare a filter and amplifier circuit for a strain gauge (Wheatstone bridge in half-bridge configuration). Input power to the Wheatstone bridge is 1 VDC, and each output (+ve and -ve) of the Wheatstone bridge circuit goes the filter.

The filter is a second order active low pass filter with fc = ~16Hz using NE 5532 op-amp. The outputs of the filter go to the differential amplifier INA118. The output of the INA118 is connected to a DAQ. The power supply to NE 5532 and INA118 is +/- 15V with bypass capacitors of 100 nF. All the grounds are common. I have attached the schematic of the circuit diagram.

When I test the soldered circuit, the noise attenuation seems to be satisfactory (~20 mV), but the baseline is not stable at all. It fluctuates ~200 mV or more. Could someone please suggest what I should do to stabilize the baseline? Is there some fault with the circuit or am I missing some component?
View attachment 290321View attachment 290322
The schematic shows a band pass filter, but I later changed it to a low pass filter. Sorry for the confusion. The revised schematic is attached herewith.
 

Attachments

Thread Starter

kpooja

Joined Mar 20, 2023
4
Your circuit biases the input on the very old NE5532 dual opamp with 2.1M ohms but its input current has a max of 1uA.
Then the offset voltage at the input can be as high as almost 2V and it will drift with temperature change.

The NE5532 is a dual so we expect the two opamps have fairly matched input currents but not perfect.
Why not use a modern dual opamp with Fet inputs that have almost no input current like an OPA2134?

I would never use 1M resistors in a 20Hz lowpass filter. Instead I would use 30k resistors and 330nF capacitors.
The 590 ohms of R5 is too low for most opamps to drive. Change R4 to 10k and change R5 to 5.9k.
Thank you for the suggestions. I will test them and get back to you soon.

One question - I was wondering what the difference between using 1M and 10 nF combination, as compared to 30k and 330 nF combination is..
 

Ian0

Joined Aug 7, 2020
9,667
An NE5532 is Low-voltage-noise op-amp.
It is NOT a low-current-noise op-amp.
With a 2MΩ input impedance you will increase the noise from 5nV/√Hz to 5μV/√Hz. Your choice of resistor has added 60dB extra noise.
Its input bias current of 1uA is also far too high for 2MΩ on the input. It will generate an offset of 2V. You need MUCH lower resistor values on your filter.


@Audioguru again Yes it is an old op-amp, but it is still one of the best in its class.
 

RPLaJeunesse

Joined Jul 29, 2018
252
Another thing to consider is the source signal voltage (probably under 1mV at most times) as compared to the offset voltage specification for the NE5532 - as much as +/-4mV. When the offset uncertainty so exceeds the signal the circuit will never do as you wish. Solutions are 1) drive the bridge with +5V and -5V to raise the output level, and 2) look for op-amps with input offset voltage specification well below 100uV. Or 3) use two simple passive filters between the sensor and the uV offset in-amp, then a single low-pass stage after the in-amp (which may allow for a mV offset amp to be used, depends on in-amp gain).
 

kaindub

Joined Oct 28, 2019
125
Would it be possible to use an HX711 strain guage amplifier ? It provides an I2C output suitable for micrprocessors. Then most of the work of interfacing has been done for you.
 
Top