Scaling the output voltage in opamp circuit

Thread Starter

Dever

Joined May 2, 2022
35
Hi,
I'm designing circuit for 4-wire PT100 temperature measurement (-30 ° C to 150 ° C). On the left, the current source (U3 with R8 and 5REF) gives aprox. 0.5mA on PT100. On opamp input there is around 44.11mV, on out ~4.42V @-30° C. I would like to have signal from 0 to 5V on "out" and connect out terminal into Sigma-Delta ADC.
1657136562939.png

Is it a proper solution to use offset pin on opamp to reduce input signal? Does it reduce accuracy of measurement?
 

Ian0

Joined Aug 7, 2020
9,671
You're nearly there - you just need a better op-amp.
The TL071 doesn't do "single supply". This circuit doesn't need to output negative values, so get a single-supply op-amp.
Most of the good ones are 5V supply, such as Microchip's MCP6022, so change the reference to 2.5V.
Then join 2 to 2 and 1 to 1, and you will get 100x the voltage across the sensor at the output, referred to ground.
No-one uses offset pins on op-amps these days! We just buy precision op-amps.
Don't forget to allow for the current in R1 and R2 - you will effectively have 9.4k in parallel with your sensor.
If that's a problem, connect three op-amps in an "instrumentation amplifier" configuration instead of your second op-amp.
 

Papabravo

Joined Feb 24, 2006
21,159
You have given us one data point. In order to map that data point to the desired range we need to know what temperature is represented by 44.11 mV. Absent that information it would be anybody's guess how to do it. We also do not know if your temperature range will map linearly onto the desired voltage range. If not, there may need to be corrections and/or calibration after the A/D conversion.
 

crutschow

Joined Mar 14, 2008
34,285
That circuit is not a good way to accurately measure the PT100 voltage.

First you are putting a 5V common-mode voltage on the signal which means the differential circuit will have to reject that large signal to accurately amplify the small voltage difference from the PT100 change in temperature.

Likely better is to use a 1 to 2.5 V reference for the current generator and change R8 to get the 0.5mA current

Then, for best accuracy, use a differential input instrumentation amplifier to detect the signal.
They have a large common-mode rejection value, the gain is controlled by the selection of one resistor, and the output has a reference input that can be used to provide the desired output DC offset you want.
 

Thread Starter

Dever

Joined May 2, 2022
35
Thank you for answers.
Yes, I plan to do corrections after A/D conversion and check linearly of sensor with Flukke's stuff.

At this point approx. values
-30° C (88 Ohm ) 44mV
150 ° C ( 156 Ohm ) 77.1mV

They have a large common-mode rejection value, the gain is controlled by the selection of one resistor, and the output has a reference input that can be used to provide the desired output DC offset you want.
Then, I will use INA or create one. Would you be kind to represent your thoughts on image?

89.0Ω is about -12°C
89.0Ω is about -30°C

Anyway, I will do precise calibration when I get those parts.
 
Last edited:

Thread Starter

Dever

Joined May 2, 2022
35
An implementation of reference input on output to provide desired DC offset.
@Ian0
In my case, I will look for single-supply instrumentation amplifier, such as AD8223.
 

Ian0

Joined Aug 7, 2020
9,671
An implementation of reference input on output to provide desired DC offset.
@Ian0
In my case, I will look for single-supply instrumentation amplifier, such as AD8223.
Just use three precision op-amps. A quad op-amp such as MCP6024 will do the whole job.
 

Ian0

Joined Aug 7, 2020
9,671
You need 0.1% resistors with that go get good common-mode rejection.
True, but they aren't as expensive as they used to be!
36p each from RS for 5 off.
10p each for 1000 off.
In the current economic state, I'd fancy my chances of getting a precision quad op-amp and four (or 6, depending on circuit) 0.1% resistors, rather than finding an instrumentation amplifier. Total cost is about the same.
 

Thread Starter

Dever

Joined May 2, 2022
35
I decided to use INA321, which is available in store.

I would like to make this INA works in the best range of output voltages for input differentail voltages from 44mV to 77mV.
According to datascheet I should have 100 V/V gain. It isn't true even with Vref = GND (I get G=65). I thought I have to put +3V on Vref to decrease 7.7V to 5V on output when Vin = 77mV.
1658057278871.png
And the full circuit. Currently doesn't work in simulation.
1658057304662.png
 

Ian0

Joined Aug 7, 2020
9,671
Constant current drive of 1.25V/2.5k = 500uA
500uA multipied by PT100 resistance of 100Ω = 50mV
Output is 50mV x gain of 100 = 5V,
but you only have a 5V supply, so it's not going to get higher than 4.98V.
Increasing the reference voltage will only make things worse, because the reference voltage is added to the output.
 

Thread Starter

Dever

Joined May 2, 2022
35
The only solution is to set Vref to GND and gain = 63?
Then maximal Output voltage is going to be 4.851V @ Vin = 77mV.
1658061679189.png
Now it works. Can I connect straight VF1 into MCP3422 without additional aliasing filters due to C1 on feedback?
 
Last edited:

dl324

Joined Mar 30, 2015
16,846
Your schematics would be more readable if you avoided unnecessary wire jogs, didn't draw wires over symbols (or text), and drew ground always point down (some schematic editors use an arrow pointing up for positive voltage supplies). I misread one of the arrow supplies for ground when it was used upside down in some Proteus schematic.

From Eagle's supply1 library:
1658085363780.png
 
Top