DC offset null in an opamp voltage follower..

Thread Starter

liquidair

Joined Oct 1, 2009
192
Hi all-

I am concerned about DC offset in a non-inverting op amp circuit I am working on. In this example, my input resistor (to gnd) is 25k. I was under the (apparently wrong) impression that you could null the DC offset at the input by simply inserting a resistor whose value matches your input resistor in the feedback loop from output to - input, just like we would do if we had a circuit with voltage gain where the input resistor would equal the feedback resistors in parallel. I just simulated the circuit and it was unstable until I removed the feedback resistor.

Is there a way to null this offset at the output without resorting to caps or DC servo circuits?

Thank you in advance for your help!
 

bountyhunter

Joined Sep 7, 2009
2,512
Hi all-

I am concerned about DC offset in a non-inverting op amp circuit I am working on. In this example, my input resistor (to gnd) is 25k. I was under the (apparently wrong) impression that you could null the DC offset at the input by simply inserting a resistor whose value matches your input resistor in the feedback loop from output to - input, just like we would do if we had a circuit with voltage gain where the input resistor would equal the feedback resistors in parallel. I just simulated the circuit and it was unstable until I removed the feedback resistor.

Is there a way to null this offset at the output without resorting to caps or DC servo circuits?

Thank you in advance for your help!
Yes, You use a pot connected across the +/- rails with a series resistor from the pot's wiper to the op-amp input. Adjusting the pot will compensate for offset voltage and current to zero the output.

Here is an example:
 

Attachments

Thread Starter

liquidair

Joined Oct 1, 2009
192
Al, thank you for the reply. I followed the link and sure enough, there's the equal value resistors in a follower. However, even with a series Rin I still get oscillations in a simulator (LTSPICE). I even decoupled which I'm sure isn't necessary in a sim.

BountyHunter that is a great idea which I am going to try, but doesn't that have a potential problem in a standard Opamp VF circuit where say the offset at + input is 1mV (just for an easy example), then 1mV should appear at the output and the - input. If there's 1mV at both inputs then the output should be 0Voff, in which case the - input can't have 1mV and thus the output should be 1mV....agh!

Dodgydave, I tried using both the LT1028 and LT1128 in the sim, but I'm between one of those, the opa2134, and the LME49723. I may try others for experimentation's sake.
 
In my experience, with op-amps at least, I found that I was struggling to simulate them simply because I didn't really understand all the newances.

I found that setting too with a breadboard and building some basic topologies whilst looking at what was going on with a little scope was far more rewarding.

Whenever a particular arrangement didn't work or was unstable I could tinker about and see which part of the circuit was causing the issues.

I am sure if I knew more about electronics in general and also LT Spice my simulation attempts would match real world applications but I fear there is some way to go for me at least.

I have built and used many op-amp circuits, in fact I love them, particularly single rail FET input varieties and transconductance amps.

My advice, get your hands dirty as well as reading.

Al
 

Ron H

Joined Apr 14, 2005
7,063
Yes, You use a pot connected across the +/- rails with a series resistor from the pot's wiper to the op-amp input. Adjusting the pot will compensate for offset voltage and current to zero the output.

Here is an example:
That is no help for a voltage follower.
 

Ron H

Joined Apr 14, 2005
7,063
Al, thank you for the reply. I followed the link and sure enough, there's the equal value resistors in a follower. However, even with a series Rin I still get oscillations in a simulator (LTSPICE). I even decoupled which I'm sure isn't necessary in a sim.

BountyHunter that is a great idea which I am going to try, but doesn't that have a potential problem in a standard Opamp VF circuit where say the offset at + input is 1mV (just for an easy example), then 1mV should appear at the output and the - input. If there's 1mV at both inputs then the output should be 0Voff, in which case the - input can't have 1mV and thus the output should be 1mV....agh!

Dodgydave, I tried using both the LT1028 and LT1128 in the sim, but I'm between one of those, the opa2134, and the LME49723. I may try others for experimentation's sake.
First of all, making the resistors in the + and - inputs equal only compensates for input bias current. They do nothing for input offset voltage or input offset current.
Secondly, why do you have a resistor to ground on your input? is the input AC-coupled?
Input offset voltage on a voltage follower is difficult to compensate for unless you choose an op amp that has separate offset adjustment pins.
 

bountyhunter

Joined Sep 7, 2009
2,512
That is no help for a voltage follower.
You could simply give it a gain of 2X (use R for feedback resistor and source resistor) and use an R-R divider at the output to get back to unity gain overall for the gain stage. As long as the inputs have some resistance you can use simple current injection with a pot across the rails to balance the output offset voltage.
 
Last edited:

Thread Starter

liquidair

Joined Oct 1, 2009
192
Dyslexicbloke,
I like to learn as much as I can before I build anything, as there is so much knowledge out there at our disposal. Plus, it may save a couple of $$ in the process. Then I simulate, learn more, then breadboard...and of course learn more when the breadboarded circuit doesn't work right.

Ron,
I sorta figured this was a lost cause after a while and thinking about things. Since this is an audio application (a monitor controller), I assume that my D/A converter is AC coupled on it's output, so the 25k (may use 10k to be more in line with standards) resistor drops the input impedance to the typical balanced line input impedance. But I began to realize that since this design won't have any gain, only loss, do I even need to worry about a tiny DC offset? As good design would dictate, I probably do need a set of caps on the very output of the device, or at least a servo.

Bountyhunter,
I was actually sort of thinking about this in some form or another, but the gain of 2 seems the most appropriate for ease of design. Thank you for the ideas!

Does anyone here design/has designed audio equipment that can verify if I need to worry about small amounts of DC throughout a design that is simply a buffered attenuator/switcher?
 

Ron H

Joined Apr 14, 2005
7,063
Dyslexicbloke,
I like to learn as much as I can before I build anything, as there is so much knowledge out there at our disposal. Plus, it may save a couple of $$ in the process. Then I simulate, learn more, then breadboard...and of course learn more when the breadboarded circuit doesn't work right.

Ron,
I sorta figured this was a lost cause after a while and thinking about things. Since this is an audio application (a monitor controller), I assume that my D/A converter is AC coupled on it's output, so the 25k (may use 10k to be more in line with standards) resistor drops the input impedance to the typical balanced line input impedance. But I began to realize that since this design won't have any gain, only loss, do I even need to worry about a tiny DC offset? As good design would dictate, I probably do need a set of caps on the very output of the device, or at least a servo.

Bountyhunter,
I was actually sort of thinking about this in some form or another, but the gain of 2 seems the most appropriate for ease of design. Thank you for the ideas!

Does anyone here design/has designed audio equipment that can verify if I need to worry about small amounts of DC throughout a design that is simply a buffered attenuator/switcher?
DC offset only affects the dynamic range by a tiny amount. You could probably AC-couple your DAC directly into the ADC input, DC biased by an R-R voltage divider between the ADC's Vref and GND.
If the DAC, by some miracle, has the same dynamic range as the ADC, just connect them together with a wire or a series resistor.
 

bountyhunter

Joined Sep 7, 2009
2,512
Bountyhunter,
I was actually sort of thinking about this in some form or another, but the gain of 2 seems the most appropriate for ease of design. Thank you for the ideas!

Does anyone here design/has designed audio equipment that can verify if I need to worry about small amounts of DC throughout a design that is simply a buffered attenuator/switcher?
No problem:

as for DC in audio amps, if the output is DC coupled to the speaker you must null the DC offset away or there will be a DC current running through the speaker all the time.

You can just capacitively couple the output to the speaker. Most "HI-FI" setups don't use caps because of degradation of damping factor but for decent sound it's OK to use capacitor coupling.
 

Thread Starter

liquidair

Joined Oct 1, 2009
192
Bountyhunter,
I thought about using single op amps (opa627) like this but given that it is a balanced line application, the number quickly gets out of hand (4 for balanced left/right on the input and 4 on the output, plus another 4 for mono mode, then another 4 for an aux out...then a headphone amp and a meter circuit.

I was wondering about the objection I had in my 2nd post on the thread. What does happen in an op amp circuit under the following conditions? Say a dc voltage of 1mV (just for example, I know this isn't practical) appears at the + input. In a voltage follower, slightly less than 1mV of DC appears at the output, which is direct coupled to the - input. If there is 1mV of DC at both inputs, then there should be 0 V DC at the output which can't happen because there has to be 1mV there. What am I missing here?
 

bountyhunter

Joined Sep 7, 2009
2,512
Bountyhunter,
I thought about using single op amps (opa627) like this but given that it is a balanced line application, the number quickly gets out of hand (4 for balanced left/right on the input and 4 on the output, plus another 4 for mono mode, then another 4 for an aux out...then a headphone amp and a meter circuit.

I was wondering about the objection I had in my 2nd post on the thread. What does happen in an op amp circuit under the following conditions? Say a dc voltage of 1mV (just for example, I know this isn't practical) appears at the + input. In a voltage follower, slightly less than 1mV of DC appears at the output, which is direct coupled to the - input. If there is 1mV of DC at both inputs, then there should be 0 V DC at the output which can't happen because there has to be 1mV there. What am I missing here?
Every op amp has a spec called "input offset voltage" which is defined as the difference in input voltage when the output is zero. Or conversely, if you force the pos input to zero in a voltage follower, there will be some voltage at the output because of the offset voltage between the inputs.
 
Last edited:

ramancini8

Joined Jul 18, 2012
473
Junk the program because it probably lies anyway, and build/test your circuit idea. When you get it working in real life use the sim. to find the workable range of component values.
 

Ron H

Joined Apr 14, 2005
7,063
Bountyhunter,
I thought about using single op amps (opa627) like this but given that it is a balanced line application, the number quickly gets out of hand (4 for balanced left/right on the input and 4 on the output, plus another 4 for mono mode, then another 4 for an aux out...then a headphone amp and a meter circuit.

I was wondering about the objection I had in my 2nd post on the thread. What does happen in an op amp circuit under the following conditions? Say a dc voltage of 1mV (just for example, I know this isn't practical) appears at the + input. In a voltage follower, slightly less than 1mV of DC appears at the output, which is direct coupled to the - input. If there is 1mV of DC at both inputs, then there should be 0 V DC at the output which can't happen because there has to be 1mV there. What am I missing here?
What you are missing is this: The op amp has a huge open loop gain, like 1 million. If the input voltage is 1mV, then the output will go to (1mV-1mV/1,000,000), which is damn close to 1mV. This assumes, of course, that the input offset voltage is zero. If the input ofset voltage is not zero, the same theory still applies. Apply a change of Δv, and the output will change by (Δv-Δv/Aol), where Aol is the open loop gain.
Another way of saying this is that, in a circuit with negative feedback, the differential input voltage will be 1/Aol times the output voltage. So, the differential input voltage isn't zero, but it's darned close.
 
Top