Inverting amplifier

[MikeK]

I have this circuit wired up to invert a negative input voltage (-2V), but I get about 4V on the output no matter the input is.  When I simulate it I get the proper output, but my breadboarded circuit fails.  What am I doing wrong?

[amspire]

The TL071 needs a negative supply. You cannot take an input down to the same potential as the negative supply rail, as you have done.

All you are getting is a maximum output voltage, which must be the default for that chip when the input stage of the opamp is basically off.

If you replaced the TL071 with a LM324 (that can work with inputs on the negative rail) , it would work.

Richard.

[MikeK]

Crap.  I thought Vcc- just limits the negative swing?

[free_electron]

Ah yes, the common mode problem. welcome to 'real opamps' as opposed to 'simulated drab'

the TL071 does not really need a negative supply. No opamp does. An opamp has only 2 power supply pins. You'll ofteen see mentioning of symmetrical supplies , assymetrical supplies etc. that is all nonsense. An opamp has a power input and a power return pin (the rails : we call the power supply pins of an opamp the 'rails' to avoid confusion with + and - input).
The opamp amplifies the delta between its inputs. this amplification is very large by default and you can control this ( bring it down ) by applying external components.

In your case the output should indeed be 2 volts. Why ?
You have one input tied to a certain point ( in this case '0 volts' or 'ground' ) so the opamp would like to see the same potential on its other input. simple ohms lawa application teaches us that :
- if that input is indeed 0 , there will be -2 volts across the first resistor. this gives -2/10k = -0.2 milliampere. this current has to go somewhere. since the input of the opamp is high impedant ( but not infinite ! there is a bias current to take into account but with fet input opamps like the tl071 it is very small ) this current can only flow through the other resistor. so there we have a drop of 2 volts ( current sign changes ) . so the output sits at 2 volts.

now in reality this is only true if a number of other conditions are met:

1) the output can indeed swing to such level. if the output would have needed to go to 100 volt and the opamp is only powered with 5 volts on its rails ... it ain't gonna work....

2) you are inside the input operating range.... for a TL071 the input stops working when you come closer than 4 volts from its negative power rail. Since you tied this also to '0 volts' the opamp works only correctly if the inputs are 4 volts or more ... this is called the common mode range of an opamp and is given in the datasheets. Most simulators 'forget this'... because they suck (the models are incomplete) Really good simulators with correct and complete models are not cheap....
For a tl071 powered with 30 volts the operational range is 4 volt to 30 volt, this is given in the datasheet

The reason behind this has to do with the construction of the opamp. I wrote a whole explanation in another topic on this forum. Basically you want to see where the current pump sits. An opamp is made with to matched transsistors and a reference current generator (the current pump).  There is a voltage drop across this current generator. if this pump sits between -rail and the transistors then you cannot work lower than the drop across the pump of the -rail. if the pump sits topside, between +rail  and the transsitors then you will get into trouble if your input signals get closer to the +rail than the drop across the pump).

3) the ouput can deliver enough current. there is a limit to what the output can do ... overload it and all bets are off.

4) the bias currents do not mess up your circuitry. the higher resistance values you are using the more impact there will be because of bias currents. this current can be positive or negative depending on the construction of the opamp ( jfet, cmos , bipolar ) and screw up really bad your system

For AC signals things get even more complicated. the you get frequency dependent behaviour. the opamp has a reaction time. if you go too fast , it will react too slow and it will screw up again ....

So, to solve your problem : you can disconnect -rail from '0 volt and tie it to -4 volts or so and the beastie should behave. Or, you make a bias generator and 'lift' both inputs away from '0 volts' and tie them to the bias voltage.

[amspire]

Crap.  I thought Vcc- just limits the negative swing?

No.

If the Vcc- is -5V, then the inputs may be able to go down to -1V reliably. This is what you would expect from the junction FET input circuit.

It is an opamp with really excellent noise and distortion performance, but it is not a good low voltage or single supply opamp.

Opamps like the LM324 and many of the modern opamps are much better to use with, say, a single 5V supply rail.

[free_electron]

the 324 works becasue its common mode includes the -rail ( its pump sits topside ) you can see this in the datasheets .

http://www.ti.com/lit/ds/symlink/lm324.pdf page 4

On the left hand side you see the input stage with the transistors. above them sits the 6ua current pump to VCC ( the + rail ).
Now why do they use double transistors on the inputs ? to get 2 diode drops ... you see the 2 transsitors connected to GND the left one with base and collector shorted ? this makes a balanced current sink ( think of it as two identical resistors. but 'active' they match each other dynamically) but this thing in itself has a 0.6 volts drop ! so , you would not be able to really go down to 0 volt... you always have 0.7 volts across these guys and you drop another 0.7 in the input transsitors. so we apply a trick : we put two transistors in series to get 2x0.7 drop. So it is guaranteed you can work down to gnd ( actually this opamp will work down to 0.5..0.6 volts BELOW its -rail ..)

http://www.ti.com/lit/ds/symlink/tl071.pdf page 5 :
the tl071 has a problem because it uses jfets... those things need 3 to 4 volt before they really turn off completely... ( this is called the pinchoff voltage ) combine this with the voltage that sits across the 1080 ohms and the bipolar transistor just above it and there is your problem...

[MikeK]

I'm going to have to read all this a few more times to understand it.  But I did drop the LM324 in and got the expected results.

I found out that it can't get closer than about 1.5V from the rail (in my case 5V) so I'm going to do some playing around....I was planning on using this to invert (and scale) the negative variable output from my power supply so I can measure it with a PIC (and display it).  And the fixed output is only 5V, but I do have an unused 12.6V AC output from the transformer than I can rectify/regulate to suit the op-amp.

[amspire]

I'm going to have to read all this a few more times to understand it.  But I did drop the LM324 in and got the expected results.

I found out that it can't get closer than about 1.5V from the rail (in my case 5V) so I'm going to do some playing around....I was planning on using this to invert (and scale) the negative variable output from my power supply so I can measure it with a PIC (and display it).  And the fixed output is only 5V, but I do have an unused 12.6V AC output from the transformer than I can rectify/regulate to suit the op-amp.

You can get the LM324 to go pretty close to the 5V rail if you add a resistor (say a 1K resistor) from the output to the 5V rail - at the expense of a higher minimum output voltage.

There are "rail-to-rail" opamps - the term tends to encompass opamps with rail to rail inputs, rail to rail outputs, or in some cases, both.

Richard.