Op-Amp Insanity

[charmcity]

Er, my impending insanity that is - in not being able to get this Op-Amp circuit to function as I'd expect. Very simple. I've got 2 sensors putting out about 2.6V - each sensor goes into a voltage follower op-amp and then the output of each goes into a differential amp which -- I'm hoping will give me the difference between the 1st and 2nd sensor voltage. Unfortunately I get about 2V and even jacking up one of the sensors outputs to as much as 3V still results in a 2V output??? Something very wrong about what I'm doing. Please have a look!!

What I need is sensor 1 voltage - sensor 2 voltage on the output. I've checked the voltage followers and they are working as expected. I'm using UA741 op amps with +10.5V Vcc+ and 0V Vcc- I also have the bias pins grounded

Here's my schematic:

[drgeoff]

The gain of the final stage for signal from the top follower is 1. (Negative feedback.)

The gain of the final stage for signal from the bottom follower is 0.5. (3.3k and 3.3k divider, then voltage follower).

So you are trying to have 2.6 volts on -ve input and 1.3 on +ve input of the final op-amp.

Try removing the 3.3k resistor to ground from the +ve input of final op-amp. You may even be able to also dispense with the other 3.3k resistor on that +ve input.

Simples?

[PeterO]

The gain of the final stage for signal from the top follower is 1. (Negative feedback.)

The gain of the final stage for signal from the bottom follower is 0.5. (3.3k and 3.3k divider, then voltage follower).

So you are trying to have 2.6 volts on -ve input and 1.3 on +ve input of the final op-amp.

Try removing the 3.3k resistor to ground from the +ve input of final op-amp. You may even be able to also dispense with the other 3.3k resistor on that +ve input.

Simples?

Not as "Simples" as your analysis though !

See http://en.wikipedia.org/wiki/Operationa ... plifier.29 for the correct circuit analysis.

And for the "full fat" version see http://en.wikipedia.org/wiki/Instrumentation_amplifier

Charcity: The circuit looks ok so I expect it will be a wiring fault in your implementation. With grounded inputs every point in the circuit should be at zero volts as well.

PeterO

AARGGHH now I see the problem. This will not work with a unipolar power supply ! It needs +ve and -ve supplies,

PeterO

[drgeoff]

That will teach me to rush an answer just before going to bed after a very long day. I woke up this morning and my first conscious thought was that I'd got it wrong. (Yes, I need to get a life!)

The gain of the bottom bit is also 1; the follower having gain of 2 because the feedback on the other input also sees a 3.3k and 3.3k divider.

The +ve input will be at 1.3 volts so the feedback will try to make the - ve input be at 1.3 volts. That happens when the output is at zero. As PeterO says the amp needs powering from +ve and - ve rails to do that.

[Burngate]

A 741 is the wrong op-amp to use here. It may be the wrong one almost everywhere! (I didn't realise they still made it!)
Most places the 741 would fit, TL071 or TL081 would work better.
The 741 datasheet doesn't seem to mention minimum supply rails, only giving Large Signal Voltage Gain with ±5v rails - 10v total. I wouldn't trust it to do anything useful with just 5v.

Since you want the output to reach ground, and run from just 5v, you could consider OPA342 - or the quad-version OPA4342, which comes in a DIL package, especially as you want three op-amps.

And think about shifting the output to swing both sides of 2.5v.

[drgeoff]

A quick hack might be to return the currently grounded end of the 3k3 resistor to 2.6 volts instead. It needs to be low impedance so if you are dealing with DC or low frequency signals a decoupling capacitor may not be adequate and you should consider a voltage regulator with an additional load resistor.

[Burngate]

... return the currently grounded end of the 3k3 resistor to 2.6 volts instead.

I'd suggest two resistors of twice the value, one to ground and the other to 5v.
But they don't make 6k6 resistors ... which is irritating.

[DougieLawson]

But they don't make 6k6 resistors ... which is irritating.

Two 3K3 resistors in series makes a 6K6.

[r4049zt]

As the opamp on the right is expected to give 1x the difference between the other two, it really ought to have +- power supplies such as +-9V and not the single supply that you've done, so that its output can go quite negative when that is the right answer. The CMMR of the 741 might cope with diffing two inputs both near +2.6V but look it up on the datasheet, look up stuff near there, and decide from that. I'd be inclined to make both opamps on the left run off +- power supplies too, which can be either +-9V or a bit less if that would suit your sensor. Can you provide +-3V to your sensor instead of 0V,+5V? The principle of your circuit would be unchanged, but the wiring somewhat more complex. Also, you need to have a 0V for "signal 0" output midway between the +- supplies which is not always the same as a battery terminal if the batteries run down unequally.