Forums » CircuitLab Support Forum

Hi,

Wondering if you can help. When I build the circuit below the output doesn't drop below 1.9 - 2.0 V. It follows to simulated output when the input is above 3.5, but below this the output stays at 2v. I have rebuilt it a number of times using different op amps.

https://www.circuitlab.com/circuit/3q6x74/test-stage1/

thanks

Steve

@carbonarc,

Your circuit as drawn (@ 121106) simulates exactly as you would expect for a lightly loaded rail to rail input and output (riro) opamp.

However, the 741 is not a riro opamp, even though CL models it as one.

Opamp outputs can only swing to within H volts of the +ve and -ve supplies; where H is the headroom and may be anywhere between a few mV to a few volts, depending on device and load currents.

Check the datasheet for your real opamps for output voltage swing capability.

Other points:

Are you sure you've drawn it correctly in CL?

What is your real test bench circuit supposed to do?

What does your real test bench circuit actually do?

"I have rebuilt it a number of times using different op amps."

On the test bench or in CL?

Remember that all CL opamps are modelled as riro types.

If you're feeling adventurous, have a look at:

or

:)

Thanks heaps - That makes sense. I've since tested it with a LM3900 and the results seem better although it has different gain characteristics, so I'll need to tweak the resistance to get it right.
My goal is to build an interface unit that modifies the feedback signal (0 - 5v) from a throttle body position sensor so that it opens more quickly between 1 and 4 volts, but remains standard for the 0-1v and 4-5 volt sections. By adding a 1 volt offset to the output this circuit generates then using the lower voltage off this and the original signal, I can get the resultant signal I need .

Here's the full circuit - https://www.circuitlab.com/circuit/2d85br/sersor-interface/

If you have any ideas/better ways to do it, be my guest.

Thanks again

Provided that your circuit keeps all the opamps in their linear region (i.e. you don't rely on any of the opamp outputs hitting either supply rail for your circuit to work as desired) then I think you can do all of it in 1 opamp.

If you work out the overall transfer function I think you'll find that what you want can be described in terms of the simple linear function:

y=mx+c

And than means you can do the whole offset and scaling thing in a single opamp.

It might wreck your brain trying to see how to do it but if you have a tag search in CL for learning resources, you should be able to find some info and links to describe how to design a generic opamp based sum and difference with gain stage.

BTW: In fact, for what you are trying to do, you should do the transfer function equation anyway. That should be part of your design requirements specification.

I know, it's only a lab practical or a homer or a hobby project but it's still a design job and worth taking seriously. Especially if it's got anything to do with connecting the user to a means of injuring themselves^!

Besides, how else do you properly check you've got it right?

^ You did mention "throttle body position sensor" didn't you?

:)

You've got my attention now! 'One" Op Amp - that would be far better, I been follow the recommendations of another forum that I needed to break it up into different sections and then assemble to get the overall output. Totally agree with taking the design seriously. Thanks again - I shall research and let you know.

Almost there ...

I can't work out how the get the output to equal the input from 0 to 1 v, the rest is looking good. Any ideas?

https://www.circuitlab.com/circuit/a427c3/feedback-function/

If your transfer function does not fit the y=mx+c formula then you may need more than one device.

Have you derived the formula for the xfer function?

If you are looking for something that has a gain of G1 over part of an input range of Va to Vb and a different gain G2 from Vb to Vc then you may still be able to do it with one opamp but it will not be described by the y=mx+c linear formula and will require some more advanced techniques to create the breakpoint(s) needed to change the slope.

Probably something based on Precision Diodes or Precision Half Wave Rectifiers.

If you can't think how to write the formula as a simple mathematical expression, try using CL to draw you a graph of it. Like this:

This crude diode function generator circuit produces something like the PWL() graph shown above:

I haven't managed to find much on the web about diode function generators but here are some references:

http://www.analogmuseum.org/library/comdyna_functiongenerators.pdf

http://www.uni-bonn.de/~uzs159/diode_tutorial/index.html

http://www.kennethkuhn.com/students/ee431/piecewise_linear_circuits.pdf

:)

Thanks heaps - This is a big help - I'll let you know how I go

For precision diode clamping instead of using diodes with npn emitter followers, have a look at:

:)