[SOLVED] Problems on Difference Amplifier

[xiahanzh]

Hi,

I have designed a difference amplifier. As shown below, V4 is my input signal and I wish to amplify it by 1000. And I suppose Vin is almost equal to Vip and offset can be ignored. However, Vip is following V4 as I expect but Vin always tends to 1mV no matter what Vip is. Therefore, Vout3 is always close to 1V so it does not work properly. The output swing of the opamp is 0.6V to 1.2V so I do not think it should be fixed to 1V. The gain of the opamp is 66dB. Is there any idea why opamp's inputs are not equal with each other? Thanks.

 

[LvW]

Where are the supply voltages? Single or dual?

 

[barry]

You say "offset can be ignored". What led you to that conclusion? What opamp are you using? If you are using something like, say, an LM324, that has an offset of several millivolts (maybe 9), which will give you an output of several VOLTS.

Don't forget that your offset is being multiplied by your gain, just like your signal.

 

[xiahanzh]

It is a single supply voltage of 1.8V.

 

[crutschow]

Sounds like you may be amplifying the offset of the op amp. What is its offset value?

 

[xiahanzh]

In my opinion, offset is constant and should not change with V4. But actually no matter how I change V4 (in a proper range 0.8mV -- 1.1mV), Vin always tends to close to 1mV without large change (only change by <0.05mV). I think Vin should follow Vip even with offset. But actually it does not.

 

[barry]

If you have overdriven your output with the offset then everything else is wrong. Why are you even bothering to measure the opamp inputs? They are extremely high impedance, and you could be messing everything up with your probe.

What opamp are you using?

And you say the output is "always close to 1V", and then you say "The output swing of the opamp is 0.6V to 1.2V". Which is it?

 

[xiahanzh]

I'm an IC designer and now it is just my start point. All tests are simulations, not real tests. I designed the opamp myself. It is a folded-cascode amplifier with output swing from 0.6V to 1.2V. Then I connected everything and simulated DC and found out the output is always close to 1V no matter how I change V4. Finally I found out it is because Vip is not equal to Vin and Vin is always close to 1mV. I cannot figure out why it happens because in my memory opamp input voltages should be equal (or at least relatively equal if offset occurs).

 

[barry]

And you know that your folded-cascode amplifier is designed properly because....

 

[LvW]

And what is the open-loop gain of your designed opamp?

 

[Analog Ground]

Your small input signal is referenced to ground. If you are using a single power supply, then the Vin and Vip input signals are very near one of the power supply voltages. Is the common mode input range of your op amp wide enough for this situation? If not, try biasing your input at 0.9V which is 1/2 of the power supply? Maybe it works better?

 

[LvW]

Finally I found out it is because Vip is not equal to Vin and Vin is always close to 1mV. I cannot figure out why it happens because in my memory opamp input voltages should be equal (or at least relatively equal if offset occurs).

Xiahanzha, there is a big problem in your design.
You have an amplifier with 66dB open-loop gain and you are trying to get a gain of 60 dB using feedback - under the assumption that the simplified gain formula is valid.
This cannot work.
More than that the input difference (Vip-Vin) is identical to the ouput voltage divided by the open-loop gain Aol of the amplifier (as long as the device is in linear operation).
For Vout=1V and Aol=60 dB the input difference is 1mV (which is identical to the applied signal input).

As another point: Nothing is known about the input and output impedances.

 

[xiahanzh]

Yes. I found this problem last night when I discussed with my group member. I adjusted the close-loop gain to 40dB and it worked much better than before. Thanks.