Instrumentation Amplifier Design using AD8544

Hello anirudharam,

this is most likely because the simulated AD8544 runs into a limitation that does not exist for the idealized model of "lib opamp.sub".

It is not easy to say what exactly this limitation is, because it is not clear which input signal you use for the instrumentation amplifier. What is hidden inside  the block "U4 TC_K" in the simulation, which signal does this block provide for the instrumentation amplifier?

I guess that the input voltage is such that a negative voltage should be generated at the output of U2 in your second simulation. The model of the AD8544, which is supplied only unipolar, can not provide this negative output voltage. The idealized OpAmp of "lib opamp.sub" knows no supply and can also deliver negative values at the output.

You can check for this assumption by probing the output of U1 in your fist simulation. Is it going into the negative range? This cannot be delivered by the AD8544 with single supply.

best regards

Achim

Hello Achim,

        Thank you for the reply. the block "TC_k" is a thermocouple simulator, you can find more details about it here: http://www.ecircuitcenter.com/Circuits/Thermocouple/Thermocouple1.htm#:~:text=The%20basic%20thermocouple%20is%20made,circuit%20at%20the%20reference%20junction. And I have simulated it only for unipolar operation.

        Please see the below simulation which does not make use of "TC_k" block:

Picture A: 

Picture B:

        Please let me know your thoughts.

Thanks and Regards

Anirudha Ram

Hello Anirudha,

thanks for clarification of the input signal.

AnirudhaRam said:

Please let me know your thoughts.

It's just as assumed yesterday: your instrumentation amplifier would need to create a negative voltage at the output of the lower AD8544, which cannot be done without negative supply voltage. The idealized opamp in your first simulation does not care about supply, it has no limitations at the supply rails.

For the given input signal your opamps need a negative supply. At least the "lower" AD8544 labeled U2 in your second simulation.

In the following simulation you see the instrumentaion amplifier working almost as expected. Here U2 has a negative supply voltage of -5V. Therefore the output of U2 can reach into the negative range (red curve V(down) in the simulation). You just get a small offset at the lower limit of the curve.

If you give all three AD8544 a negative supply, the amplifier works as desired over the full range of the input signal:

Another option would be to shift the common mode voltage at the input of your instrumentation amplifier. In the following simulation all AD8544 have a unipolar positive supply as in your original design. The common mode voltage of the input signal is shifted to the midlevel of the supply. So all voltage nodes of the circuit stay inbetween the unipolar rails and the amplifier is working as desired.

best regards

Achim

Hello Achim,

    I am extremely grateful for the very detailed explanation. I have implemented the suggested circuit with 2.5V shift in common mode voltage, but the experimental output did not come closer to simulation.

    However I was able to obtain good results with OP484. I guess it is because of low noise and low offset voltage characteristics of OP484. Please correct me if I am wrong. This experiment was for the differential input voltage range between 0 to 54mV.

Thanks and Regards

Anirudha Ram

Hello Anirudha Ram,

AnirudhaRam said:

I have implemented the suggested circuit with 2.5V shift in common mode voltage, but the experimental output did not come closer to simulation.

What result did you get instead, how did the voltage deviate from the expected value?

AnirudhaRam said:

  However I was able to obtain good results with OP484. I guess it is because of low noise and low offset voltage characteristics of OP484. Please correct me if I am wrong.

I don't think that the offset and noise performance explains the bad funcitonality of the circuit with AD8544 - unless the problem with the AD8544 circuit was that the offset at the output was too large for you. Up to 6mV input offset are indeed not very good for an input signal ranging just between 0 and 54mV. Here a general purpose opamp with CMOS-inputs will always be inferior to a precision opamp with bipolar input stage.

The ad8544 should work in principle in this circuit (albeit with a larger offset error). If it did not work at all, something in the circuit may have been broken.

Is it possible that one of the AD8544s was defective because it saw a supply voltage that was too high during the various tests? The AD8544 can only tolerate a maximum supply voltage of 6V in total (not +/-6V).

best regards

Achim

Hello Anirudha Ram,

AnirudhaRam said:

I have implemented the suggested circuit with 2.5V shift in common mode voltage, but the experimental output did not come closer to simulation.

What result did you get instead, how did the voltage deviate from the expected value?

AnirudhaRam said:

  However I was able to obtain good results with OP484. I guess it is because of low noise and low offset voltage characteristics of OP484. Please correct me if I am wrong.

I don't think that the offset and noise performance explains the bad funcitonality of the circuit with AD8544 - unless the problem with the AD8544 circuit was that the offset at the output was too large for you. Up to 6mV input offset are indeed not very good for an input signal ranging just between 0 and 54mV. Here a general purpose opamp with CMOS-inputs will always be inferior to a precision opamp with bipolar input stage.

The ad8544 should work in principle in this circuit (albeit with a larger offset error). If it did not work at all, something in the circuit may have been broken.

Is it possible that one of the AD8544s was defective because it saw a supply voltage that was too high during the various tests? The AD8544 can only tolerate a maximum supply voltage of 6V in total (not +/-6V).

best regards

Achim

Hi AnirudhaRam  , 

It looks like your concerns has been resolved.

Thank you for using Ltspice.

Regards,

PaulDaria