Hi all.
I am looking for a way to utilize resistors/potentiometer to adjust the offset at the input of a fully differential op amp to be close to zero with little affect on gain. I have simulated a pot attached at the negative input before the input resistor but it affects gain. Anyone know of any ideas? Thanks.

 

I cannot see any attachment.

 

Hi all.
I am looking for a way to utilize resistors/potentiometer to adjust the offset at the input of a fully differential op amp to be close to zero with little affect on gain. I have simulated a pot attached at the negative input before the input resistor but it affects gain. Anyone know of any ideas? Thanks.

Which op amp?

 

I have an idea, Post your diagram!!!

 

Welcome to AAC!

It would be helpful if you posted the part number for your opamp.

Here is a general purpose offset null circuit from Nat Semi:

 

Hi all.
I am looking for a way to utilize resistors/potentiometer to adjust the offset at the input of a fully differential op amp to be close to zero with little affect on gain. I have simulated a pot attached at the negative input before the input resistor but it affects gain. Anyone know of any ideas? Thanks.

Use the offset null pins on your opamp. Since you didn't share your secret part number, I'm free to proclaim that your part has offset trim pins.

ak

 

Note that the solution in post #5 is practical but in circuits in which offset is critical, even the tiny effect the offset adjustment has on gain can require several iterations of adjusting gain then offset, then back to gain...

 

Hi,

Here's a quick plot of the output voltage with a differential input of 2v and a gain of 10 and varying the pot from 0 to full range. Near the center it's not bad but toward the end of range of the pot it gets worse.

 

Hi,

The main problem with the change in gain (although it is small) is that the impedance from the bottom of the 10 ohm resistor to ground changes as the pot is adjusted. The impedance would be the total resistance from that point to ground given by the parallel/series combination of resistors.
This means that at least in theory if another op amp was used with very low output impedance that would help eliminate the effect of changing the pot, or possibly a dual pot with the other section inserted into the feedback path. Probably not necessary in most cases i would bet.

 

Yet, adding a buffer is an appropriate solution if you have room for another opamp.

 

Here's a quick plot of the output voltage with a differential input of 2v and a gain of 10 and varying the pot from 0 to full range.

Are you sure you're not just measuring the offset voltage change?

 

The circuit is very efficient; you adjust the gain and the offset voltage with only one pot!

There is interaction between the two. is some applications this circuit works very well.

 

Are you sure you're not just measuring the offset voltage change?

Hi,

Yes, you are right. But i did check the gain and it is off by a little.
The actual gain change can be detected by computing the total resistance of the offset null part and adding that to the 10 ohm resistor, then recalculating the gain. I'll try to get to do this a little later today, unless you care to do it too.

 

The circuit is very efficient; you adjust the gain and the offset voltage with only one pot!

There is interaction between the two. is some applications this circuit works very well.

Hi there,

I do not understand what you mean here. When we use a circuit like this we usually want a certain gain, say 10, and the offset is maybe 5mv so we want to later adjust that, but we dont want the gain to change to 10.005 or 9.995 at the same time.
Oh maybe you were joking, which is cool too

Yes the circuit is not that bad really. I'd use it, and i did use a circuit similar to this except instead of a pot i used a fixed resistor to bias one of the inputs depending on if the offset was plus or minus. That circuit works about the same but no requried pot.

I'll try to get back here later with a plot of the actual gain change vs pot setting unless someone else wants to do it first.

 

Hello again,

Here is a graph of the gain change with offset adjustment when the original gain was set to 10.

As the graph shows the change does not show up well because it is so small so the numbers are printed below that for when the pot is set to either end or right in the center of travel.

The first set of values is for when the resistor R3 is set as R2-R4, and the second set is for when the resistor R3 is set for a value equal to R2-RT where RT is the total resistance of the pot plus 20k in parallel with the 10 ohm resistor R4.

As we can see, the gain changes very little with this setup. This is for when R3 is many times greater than R4 though. For smaller values it probably makes a bigger difference so watch out for that.


Someone may want to verify these numbers also.

 

Someone may want to verify these numbers also.

If R4 is 2 or more orders of magnitude larger than R3, for all practical purposes, the error will be too small to measure.

EDIT: correction below...

 

If R4 is 2 or more orders of magnitude larger than R3, for all practical purposes, the error will be too small to measure.

Hi,

Are you talking about that 10 ohm resistor?
Note that if we change that 10 ohm resistor to 100000 ohms (value of R3) then the gain changes by 2 times the original gain.
The reason it is so small like 10 ohms is so that it helps to swamp out the change in impedance when the pot is adjusted. If it says low then the 10 ohms dominates, and that keeps the gain almost the same.
If you dont agree, try a couple calculations with the pot at one end and with the pot in the middle, those are the two extremes.
Let me know what you find out and i'll check it again if you still dont agree.

 

Are you talking about that 10 ohm resistor?

My mistake. Meant R3 at least 2 orders of magnitude greater than R4.

 

My mistake. Meant R3 at least 2 orders of magnitude greater than R4.

Hi again,

Oh ok, that's good. I didnt check for that, but i guess what you mean by too small to measure is 1mv maybe?
I guess the main thing is that the gain does not CHANGE too much as we adjust the offset.

 

but i guess what you mean by too small to measure is 1mv maybe?

For the ratio I mentioned, the error would be less than 0.001%.