What is the correct way of simulating input offset voltage

[calculus_cuthbert]

Hi all,

I have a deadline to meet and I really need some help. What is the correct way of simulating input offset voltage.
I referred two books:
First Method says:
ground the inverting terminal and make a dc sweep on the non-inverting terminal and check the DC transfer characteristic. The point where the DC transfer curve intersects the sweep is the offset voltage.

Now is this the offset voltage times the open loop gain or the input referred offset voltage????

Second Method:
Connect the opamp in a unity gain configuration and make a DC sweep at the noninverting terminal. The difference between your input sweep and output curve is the input offset voltage.

Now I tried both methods and I get two different answers:

with the first method I got 7.11mV and with the second method I got 600uV

PLEASE help me out.. which answer is correct and why are the answers different ..

Thanks a lot!!!

 

[Milad-D]

I am worrying a little bit. We have 2 type of offset. 1. Systematic offset 2. Statistical offset.
If you design a circuit pretty well, you will not have systematic offset.
And about the second offset you Must do a monte-carlo simulation to see the second type of offset which I think you are intereseted in the second one.
So first tell me what do you really measure?

---------- Post added at 01:29 ---------- Previous post was at 01:26 ----------

If you are interested in systematic offset, then both ways are correct. But be careful about correct biasing of the input ports during these simulations.

 

[calculus_cuthbert]

Hi Milad-D,

Thanks for your reply. Right now I am measuring systematic offset. How do you say both ways are correct. My answers don't match up. With the first method do i need to divide it by the open loop gain? Even by doing that my answers don't match.. So I am not sure which method to use..

 

[Milad-D]

Vout=a(vin+voff_in) , if you do the first method, for vin=-voff_in yoy get vout=0 or for vin=0 you get vout=avoff_in.
In the second method the vout=(a/(a+1))vin_off. If the a is high then vout=vin_off

 

[calculus_cuthbert]

So according to your post:
With my first method I got 7.11mV and my gain is 1590 so my input referred offset will be (7.11m/1590) = 4.46uV But this does not match the 600uV I am getting with the second method.

This is why I am confused. I thought with the first method we get the output offset voltage i.e vout=avin_off and with the second method we get vout=vin_off.

 

[Milad-D]

Systematic error must be very low, as it is your first simulation , but the second simulation gives you very big input referred offset voltage. Therefore , this means that when you put your amplifier in unity gain feedback, something goes wrong. What type of amplifer you have? Telescopic or folded cascode. You should be careful about input common mode range voltage and output common mode voltages. They should be compatible. You may put one of the input transistors into triode. Please check you amplifer dc points.

 

[calculus_cuthbert]

I am actually simulating the input offset voltage of a comparator. Do you think I should use the first method for comparators because comparators are not connected in unity gain configuration?

thanks a lot!

 

[Milad-D]

Yes. For your case is logical.