Pressure sensing conditioning circuit - Sim problems

[Alaezae]

I'm working on a project for a masters program involving a resisto-jet satellite propulsion system. Without going into too much detail, we have a pressure sensor on an expansion chamber (PN HKM375, datasheet attached) and I am trying to build an input conditioning circuit for it. I want to have a differential to single-ended conversion with adjustable gain and bias point, as we are unsure what the actual range of pressure is that we'll see on the device. I am going to want to be able to adjust the gain and bias of the amplifier as we test (I think, at least) which is why I am starting with this circuit. We're expecting a level between 0-100mV, and I think that the sense-able phenomenon probably will have a rise-time on the order of milliseconds (it is physical pressure changes, meaning it cannot happen that quickly). The op amp is driving a microcontroller's ADC.

I went to a book, "Op Amps for Everyone" by Ron Mancini, to look for a reference circuit for the design. I had picked an op amp for an earlier version of this design, the OPA375 from TI. I found this circuit in the book which is an adjustable-gain differential amplifier:



First off, I think that this circuit isn't correct. It looks like the non-inverting and inverting terminals are switched. I don't know why we'd have positive feedback in this case. I simulated it with what I'm thinking is the correct location for the non-inverting terminals. Additionally, I redid the analysis on the circuit, and I'm pretty sure that there should be some term on the bias that multiplies it by the gain of the feedback-path inverting amp. That's not in the expression that the book describes.

I started with ideal amplifer models (LTspice's UniversalOpAmp1 for anyone curious) just to play around with the circuit to see how it works. I was able to force a 100 Hz sinewave in and see the gain being applied as I expected it from my equations. Interestingly, I found that the overall gain is increased with Rg/Rf, which seemed weird but matched with what I calculated. Anyways, I managed to get the ideal op amp circuit working, and wanted to move to the model of the real op amp I was planning on using.



Now, moving to the real op amp circuit, I had trouble getting the simulation to work. The output seems to be stuck with a miniscule output at around 50mV. I don't know why. I was trying to play with the circuit gain to remove the possibility of swamping the output with the bias since I could see that the output was dependent on the gain and bias. That didn't help at all either. It was suggested to me that there could be a bias current problem on the op amps, but putting paths to ground in the loop didn't fix anything.



Unfortunately, the OPA375 model from TI is a PSpice model, which might cause problems in LTspice. Usually, you can try importing it in LTspice through the .LIB file and see if that works. For their unencrypted models, it usually does. This one happened to work, which is why I used it. That might be the source of the problem.

My questions:

• Is this something anyone else has used before? Am I missing a key element of this circuit anywhere?
• Is there something about the op amp that I picked that is obviously incorrect?
• Can you see anything in the model that might be causing the behavior that I'm seeing at the output?
• Are there reasons you might not want to use a circuit like this? Is there a better way to achieve what I want to achieve?

Thanks for your help!

NOTE: Export .zip contains the LTspice files

book: https://web.mit.edu/6.101/www/reference/op_amps_everyone.pdf

[Terry Bites]

Have you checked Vref?
I think your circuit sets the value of Vref to 0V not Vcc/2
That would crash your signal to the negative rail I think.

There are far better monolithic solutions known as instrumentation amplifiers.
You can get superb performance from these amplifiers in terms of gain accuracy, drift and offset.
The circuit relies heavily on closely matched low drift resistors.
The component cost will be higher than an instrumentation amplifier.

[mtwieg]

I believe that circuit is drawn correctly. Yes, the feedback for the upper opamp is going back to its noninverting input. But that feedback path goes through the lower opamp which is inverting. So the feedback is still negative overall. I'm pretty sure their equation is not correct though (Vout should depend on Vref).

It is a rather unconventional circuit though, and I don't see any advantage vs conventional instrumentation amplifiers.

[daisizhou]

You need an instrumentation amplifier circuit.

As shown in the image below, you can adjust the gain value by adjusting the bias voltage on the left or by adjusting the "r" resistor.

[Terry Bites]

I've redrawn it to make easier to understand. For me anyway.
Its a text book differential amplifier with DC offset added.
Ignoring the stuff in the box, the usual gain equations apply.
You have to change both R3 and R5 to alter the gain.
Or R1, R2 if you prefer.
The stuff in the box only adds DC offset and sets in the input common mode range.
There are some complications in applying this circuit in a real world implementation though.

Inst amp idea attached.