Project help design an op-amp

[spec]

Hi, spec,

It happens that I was born and live, since more than 6 decades, in Aleppo city (Syria). Thank you for asking.

I uploaded one of the possible solutions.

Hy Kerim,

We don't see many post from Syria on ETO.

As you say another solution to the requirement. I see you used an instrumentation amplifier approach (differential amplifier) to null the offset and scale the gain- very neat.

spec

 

[JoeJester]

How did you solve the minus ten pounds per square inch equals negaitve ten volts.

I'm having a difficult time going from zero lbs/in^2 to negative 10 lbs/in^2.

 

[spec]

How did you solve the minus ten pounds per square inch equals negaitve ten volts.

I'm having a difficult time going from zero lbs/in^2 to negative 10 lbs/in^2.

If you are asking about the circuit of post #13, -10lbs = a positive current into the virtual earth point, so the opamp output has to go negative to remove that current through the feebback resistor R1. By design the voltage is -10V.

spec

 

[JoeJester]

spec

No, I am asking how do you get negative 10 lbs per square inch of pressure in a physical world? I can see 10 lbs per square inch less that the lbs/square inch at sea level, but I can not see negative 10 lbs per square inch with respect to zero.

 

[spec]

spec

No, I am asking how do you get negative 10 lbs per square inch of pressure in a physical world? I can see 10 lbs per square inch less that the lbs/square inch at sea level, but I can not see negative 10 lbs per square inch with respect to zero.

Ahh, I see. Zero is zero ie a perfect vacuum which does not exist anywhere in the cosmos but it gets very close. You can't get any lower pressure than zero because zero pressure means the complete absence of any matter.

Similarly, you cant get any lower than absolute zero temperature (0K or -273.15C) because that is the complete absence of Brownian motion (hope I have that right).

I'm sure the scientist on ETO will correct me if that is not quite right.

spec

 

[MikeMl]

Depends entirely on the reference with respect to which the pressure is measured. I have several gauges around here that show PSIG wrt standard atmospheric pressure that are marked with negative and positive numbers...

 

[spec]

Here is another pressure transducer conditioning circuit. This one uses one opamp, in a parallel feedback configuration, and four resistors. The pressure transducer is 'the right way up'. In this case it is better to use a voltage based analysis rather than current. See if you can spot the mistakes.

spec

 

[MikeMl]

I find it ironic that we using LTSpice to validate various designs...

When I simulated Kerim's circuit of post #20, I noticed he has a correct circuit that solves the wrong problem. He incorrectly used V=1.31*P + 1.683 (instead of the stated V=0.131*P + 1.683) as the Sensor Voltage vs Pressure function. I redesigned his circuit (by solving simultaneous equations in multiple unknowns) to come up with a circuit that satisfies the goal.

I also simulated Spec's latest from post #27. Apart from some round-off error in his calculations, it satisfies the goal. I would however quibble with the fact that the problem statement gives you a reference of 1.00V, and he actually used a reference of -1.00V. If I was grading his effort, I would deduct some points...

I plot the deviation (error) from the desired output of the three respective circuits. The ideal would be a flat zero mV line. The green line (mike's) comes closest...

 

[spec]

I find it ironic that we using LTSpice to validate various designs...

When I simulated Kerim's circuit of post #20, I noticed he has a correct circuit that solves the wrong problem. He incorrectly used V=1.31*P + 1.683 (instead of the stated V=0.131*P + 1.683) as the Sensor Voltage vs Pressure function. I redesigned his circuit (by solving simultaneous equations in multiple unknowns) to come up with a circuit that satisfies the goal.

I also simulated Spec's latest from post #27. Apart from some round-off error in his calculations, it satisfies the goal. I would however quibble with the fact that the problem statement gives you a reference of 1.00V, and he actually used a reference of -1.00V. If I was grading his effort, I would deduct some points...

I plot the deviation (error) from the desired output of the three respective circuits. The ideal would be a flat zero mV line. The green line (mike's) comes closest...

View attachment 99517

I think you are under the impression you are some kind of examiner. Get off your high horse. And don't be so presumptuous.

I don't pull you circuits to pieces or any one elsse's. Would you like me to start?

Also, you could learn to apologize when you are wrong.

spec

 

[MikeMl]

I think you are under the impression you are some kind of examiner...

I am a retired University Professor with a PhD E.E. . I taught undergrad students like Wjs. That is why I don't just hand them a finished design without any explanation of where it came from, but try to get them to discover the solution on their own...

 

[spec]

I am a retired University Professor with a PhD E.E. . I taught undergrad students like Wjs. That is why I don't just hand them a finished design without any explanation of where it came from....

There is no doubt that you are a good designer- or I wouldn't bother replying to you. It is your attitude that is the problem.

I can't imagine a more through explanation than I gave for the circuit of post #13

I don't want to fall out with you.

spec

 

[JoeJester]

Mike,

I agree you "can" have a negative 10 psi with respect to a non-zero reference. Take for instance sea level. Negative 10 psi referenced to sea level represents between 20 and 30 thousand feet. We are not in a purely academic world.

 

[JimB]

Or, from a slightly different perspective, sea level is -10psi when you are 20 feet underwater.

I think. Imperial units for these measurements feel alien to me, can I re-state it in SI units?

Sea level is -1bar when you are 10 metres underwater.

JimB

 

[JoeJester]

JimB

You can state it in whatever units you wish, but the assignment was stated in PSI.

 

[Kerim]

I find it ironic that we using LTSpice to validate various designs...
When I simulated Kerim's circuit of post #20, I noticed he has a correct circuit that solves the wrong problem. He incorrectly used V=1.31*P + 1.683 (instead of the stated V=0.131*P + 1.683) as the Sensor Voltage vs Pressure function. I redesigned his circuit (by solving simultaneous equations in multiple unknowns) to come up with a circuit that satisfies the goal.

First, I like to thank you, Mike, for spotting my mistake.
And as you may have already noticed, if the right function is applied on my circuit, the output varies from -1V to +1V instead of -10V to +10V.
In this case, it should be followed with a non-inverting amplifier with a voltage gain of 10 to get the required result.

[Edited]
For instance, I usually solve equations using "solver" of Excel, I guess everyone has a sort of "solver" on these days, mainly for the non-linear ones.

Kerim

 

[Kerim]

I think the teacher, or perhaps the student, who gave the formula of the pressure sensor (on OP) can explain it, practically speaking.
It also looks rather strange to me, but I have a very limited experience in pressure sensors.

 

[spec]

I also simulated Spec's latest from post #27. Apart from some round-off error in his calculations, it satisfies the goal. I would however quibble with the fact that the problem statement gives you a reference of 1.00V, and he actually used a reference of -1.00V. If I was grading his effort, I would deduct some points...

I plot the deviation (error) from the desired output of the three respective circuits. The ideal would be a flat zero mV line. The green line (mike's) comes closest...

Mike, your simulation of circuit #27 is incorrect. For example, at the zero point the simulator shows an output error of 9mV when the nominal error can be very closely calculated to be +- 5uV * 20.482= +- 102.41 uV, +-5uV being the nominal input offset voltage for the OPA192. The worst case output offset of is +- 20uV * 20.482= 409.64uV. The opamp input current is +- 5pA so, in practical terms, that can be ignored. If a perfect opamp were assumed, there would be absolutely no offset error. How a simulator can say anything else throws doubt on the simulator or the implementation. And how you can't see this by a cursory inspection of the circuit is surprising.

It is not specified what the output impedance of the pressure transducer or voltage reference is, so how can you plot an error. Also, an error band giving +- deviations would be more meaningful.

The requirement says that the voltage reference is 1V. It says nothing about any orientation of the reference or the transducer for that matter. In the real world references can be a voltage cell or a shunt voltage reference (including Zener diode) which can be connected in any sense (not a serial reference though). This orientation of the voltage reference is not important to the exercise anyway, especially as changing the sense of any signal is trivial.

Your simulations are incomplete and difficult to read because the text is so small. I can't make head nor tail of the simulator circuit of post #9, and I have been designing similar circuits for years, so it would be a complete mystery to a newbee.

No opamp is specified, there seems to be two undefined outputs, and as far as I can tell that configuration will not work because it is a fundamental tenant of that architecture that the reference voltage must be higher than the transducer offset voltage. If you are fixing that by putting a potential divider across the transducer, as in Kerim's circuit, that contradicts your statement that your circuit of post #9 does not load the transducer. Also there is not a complete description of how the simulation was done- what assumptions etc.

You are inconsistent: On one hand you complained about me posting a circuit when you had already shown a circuit in post#9 and now you are analyzing the circuits to the Nth degree.

This is not about designing a functioning circuit, or demonstrating who is the best designer, it is about helping the OP understand how this type of circuit works and how to do the calculations. Besides which, you can make any of the circuits arbitrarily accurate, within reason by using an appropriate opamp, and by calculating the resistor values to the Nth decimal place. Even so an error of 9mV only corresponds to an accuracy of 9mv/20V= 0.045% of full scale range. Most, voltage references would not be that accurate and pressure transducers certainly are not.

The architecture of the circuit in post #13 is not only commonly used, but is a lot simpler to understand and design. As it is much easier to calculate the resistor values to change the offset and gain, because there is little interdependence, the circuit is also a lot easier to re-scale. There are other advantages too.

Please stick to technical matters, and also please show the basis of your claims with a circuit analysis and calculations and not simulations. Simulations are not the real world and the models are not verified as Texas Instruments and National Semiconductor point out. Also there is the old maxim: garbage in, garbage out.

spec

 

[spec]

When I simulated Kerim's circuit of post #20, I noticed he has a correct circuit that solves the wrong problem. He incorrectly used V=1.31*P + 1.683 (instead of the stated V=0.131*P + 1.683) as the Sensor Voltage vs Pressure function. I redesigned his circuit (by solving simultaneous equations in multiple unknowns) to come up with a circuit that satisfies the goal.

Kerim should be given credit for choosing a good circuit for this application. All he did was to make a copying error which anyone could do (even me). His offset nulling is spot on, better than your revision which has an unnecessary 6.895mV offset at the output.

I must admit that I didn't analyse Kerim's circuit, especially as the simulation showed the correct output, because I didn't think it important. It is the architecture and principal that counts.

I did notice though that the circuit would be way off due to the input current and input offset voltage of the LM324 opamp. This also applies to your corrected circuit. What I can't understand is why didn't the simulator pick that up. There is lesson here about the dangers of simulators in two cases.

spec

 

[Kerim]

On my side, I use a simulator as 'one' of many tools to analyse the response(s) of a circuit since it has its advantages and disadvantages, as it is the case for any other tool/method used in a design.
A circuit simulator is just one type of programmable calculators; it gives 'always' the right answer for the given data. But most of these given data are implemented in the component/device models which can never be made perfect (that is reflecting the exact responses/functions of the items, in all conditions). So the results of a simulator are as good as the given models (of the elements used in the circuit) are. And only by testing a real circuit, one could discover how close to the real world the simulator results, he got, are.

Kerim

 

[spec]

On my side, I use a simulator as 'one' of many tools to analyse the response(s) of a circuit since it has its advantages and disadvantages, as it is the case for any other tool/method used in a design.
A circuit simulator is just one type of programmable calculators; it gives 'always' the right answer for the given data. But most of these given data are implemented in the component/device models which can never be made perfect (that is reflecting the exact responses/functions of the items, in all conditions). So the results of a simulator are as good as the given models (of the elements used in the circuit) are. And only by testing a real circuit, one could discover how close to the real world the simulator results, he got, are.

Kerim

Hy Kerim,

Yes, that is the way to do it. As you say, simulators are just one tool and, like any tool, they should be used appropriately. Also, because of circumstances, many people have no other option than to use a simulator, and in some cases, a design would be practically impossible without a simulator: large logic arrays and Monte Carlo testing for example. In complex filter design and optimization simulators are essential too.

We have all got different ways of arriving at good designs, but my point is that it is important to understand the fundamental architecture and operation of a circuit and its critical areas, rather than slavishly relying on simulators. My other point is that if you are teaching someone how to design a circuit, as in this case, it is fundamentally important to be clear, concise, thorough, and focused.

It is also important to understand about errors and learn how to deal with them because errors are endemic. In your case you made a simple copying error which resulted in the wrong gain, but the architecture of your approach was spot on (so was the offset balancing) and that is the important thing. Even text books have errors, perversely often in the places where they are most confusing for someone trying to learn a technique. The books of one software author are famous for their errors, but his message about the techniques involved are crystal clear.

One thing that I would say though, I have been a design engineer for many years and have found that engineers, especially graduates, who are overly reliant on simulators could never do a good practical design. They also lacked innovation and intuition and were completely stumped by an unusual or new circuit. It is in the mindset. I could tell many stories about this.

spec