Inverting op amp extra resistor question.

Status
Not open for further replies.

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
I vaguely remember some books suggesting a resistor be added to an inverting amplifier design, but I never understood why. I remember it suggesting a value for R1 to be equal to the parallel resistance of the other two resistors. In retirement now I have the time to pursue answers to questions I never found an answer to. So what is R1 supposed to accomplish? I built a circuit with and without R1 and see no difference.
I used an LM741 with an offset pot if it makes a difference.
Is it not apparent in my circuit?
Is there an equivalent resistor in a non-inverting amplifier?
 

Attachments

crutschow

Joined Mar 14, 2008
24,115
The resistor is to cancel the differential voltage offset from the input bias current.
If the equivalent resistance at each input is the same then the voltages caused by the bias current will cancel, leaving no differential input offset voltage.
(This assumes that both bias currents are equal. In reality they usually have a small difference as stated in the bias offset value in the op amp data sheet. The voltage from this offset current is not cancelled by the equal value resistors).
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
What you want to look into is input bias current compensation in op-amp circuits.
So what effect should it have in my circuit? Why the same result with and without it?
Your answer sounds like a quote from a text book. My question deals with a specific application. What does your answer have to do with my example?
I deliberately used an op amp that should be a good example of a bad example of such bias problems ... yes?
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
The resistor is to cancel the differential voltage offset from the input bias current.
If the equivalent resistance at each input is the same then the voltages caused by the bias current will cancel, leaving no differential input offset voltage.
(This assumes that both bias currents are equal. In reality they usually have a small difference as stated in the bias offset value in the op amp data sheet. The voltage from this offset current is not cancelled by the equal value resistors).
Okay, so the LM741 is always hailed as the worse op amp on the market, If there is a bad characteristic the LM741 should have it. Why no difference between my two results?
Again, text book quotes. How does this apply to my circuit?
 

WBahn

Joined Mar 31, 2012
25,104
So what effect should it have in my circuit? Why the same result with and without it?
Your answer sounds like a quote from a text book. My question deals with a specific application. What does your answer have to do with my example?
I deliberately used an op amp that should be a good example of a bad example of such bias problems ... yes?
I am not going to spend a bunch of time walking you through the kind of errors that neglecting input bias compensation can have in a circuit and how to compensate for it. That material is very easy to find and I told you what to look for. So look for it and study it.

In lots of applications, the non-ideal features of real op-amps can be safely ignored and you won't see an appreciable difference. That does not mean that applications don't exist in which the non-idealities DO result in appreciable differences and that compensating for them is important.

If you don't want to compensate for them, fine, then don't.
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
I am not going to spend a bunch of time walking you through the kind of errors that neglecting input bias compensation can have in a circuit and how to compensate for it. That material is very easy to find and I told you what to look for. So look for it and study it.

In lots of applications, the non-ideal features of real op-amps can be safely ignored and you won't see an appreciable difference. That does not mean that applications don't exist in which the non-idealities DO result in appreciable differences and that compensating for them is important.

If you don't want to compensate for them, fine, then don't.
< :) >Ha, ha, I see. I appreciate your time so far. In other words you remember what the book said but can't apply it to a real circuit? </ :) >
Is this the answer you would give to a student in a classroom?
I can find thousands of words on the subject but none relate the concept to a real circuit. Just theoretical concepts I can't bend to apply to a real circuit.
I continue to try and apply. So far I have only tried a gain of 10 and a significant (tenths of a volt) input. Tomorrow I will try higher gains and lower input voltages. I can get down to a few millivolts reliably but if the difference is only in microvolts I will never see the result. Will it ever matter?
 

crutschow

Joined Mar 14, 2008
24,115
Okay, so the LM741 is always hailed as the worse op amp on the market, If there is a bad characteristic the LM741 should have it. Why no difference between my two results?
Again, text book quotes. How does this apply to my circuit?
I told you what it did. If the text book says the same thing, no surprise there. That's the answer to the question you asked. :rolleyes:

The 741 does have a relatively large bias current.
Apparently you didn't measure the output offset to a high enough precision to see the difference.
(Did you calculate the magnitude of the output offset voltage you would expect with the resistor and gain value you have? If not, then how can you say you there was no difference? :confused:)
Larger resistor values and higher circuit gain will generate more offset voltage and make it easier to see.

If the difference is too small that it doesn't affect your circuit application then you don't need the extra compensation resistor.
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
What you want to look into is input bias current compensation in op-amp circuits.
The LM741 has a typical input bias current of about 80 nA the data sheet says.
I can't measure stuff at that level.
Does this current flow through my 10K input resistor? that amounts to about 0.8 mV. barely within my ability to reliably measure.
Should this error be amplified by the gain of the circuit? I should then see an difference of 8 mV with and without the resistor. I don't. What did I miss?
Maybe I will see more tomorrow at higher gains. Or am I going about this all wrong in the first place?
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
I told you what it did. If the text book says the same thing, no surprise there. That's the answer to the question you asked. :rolleyes:

The 741 does have a relatively large bias current.
Apparently you didn't measure the output offset to a high enough precision to see the difference.
(Did you calculate the magnitude of the output offset voltage you would expect with the resistor and gain value you have? If not, then how can you say you there was no difference? :confused:)
Larger resistor values and higher circuit gain will generate more offset voltage and make it easier to see.

If the difference is too small that it doesn't affect your circuit application then you don't need the extra compensation resistor.
< :) > I am familiar with what the book says. What I am missing is what that means in a real circuit. Memory versus comprehension. The comprehension is what I lack. </ :) >
So far I have only tried it with a gain of 10. I expected to at least see a difference of some millivolts. I didn't. The LM741 should have enough bias to be apparent after taking gain into influence ... right?
It's too late tonight to continue. Tomorrow I will try a gain of 100 and 1000.
I guess the question boils down to, "at what point does that resistor make a difference?" Gain less than 10 with some hundreds of millivolt, or higher, signals. Maybe not. High gain with low millivolt level signals? Answer coming up tomorrow.
 

crutschow

Joined Mar 14, 2008
24,115
< :) > I am familiar with what the book says. What I am missing is what that means in a real circuit. Memory versus comprehension. The comprehension is what I lack. </ :) >
So far I have only tried it with a gain of 10. I expected to at least see a difference of some millivolts. I didn't. The LM741 should have enough bias to be apparent after taking gain into influence ... right?
It's too late tonight to continue. Tomorrow I will try a gain of 100 and 1000.
I guess the question boils down to, "at what point does that resistor make a difference?" Gain less than 10 with some hundreds of millivolt, or higher, signals. Maybe not. High gain with low millivolt level signals? Answer coming up tomorrow.
The current does flow through he input resistor and is amplified by the gain of the circuit.
Try larger resistors (say 1 megΩ or greater) and higher gains to see a greater effect.
Perhaps you have a 741 with an usually low bias current. o_O

The offset becomes a problem when it is a significant percentage of the minimum signal you want to amplify.
Note that the worst-case bias current of the 741 is much worse than the nominal and that is what you should design for.
 

Thread Starter

hp1729

Joined Nov 23, 2015
2,304
worst-case bias current
Designing for worst case conditions???? The only design hint I have run across is "equal to the parallel value ...". No consideration for specific values there. What should be done? With the input resistor grounded measure the voltage across it. Derive the level of bias current from that? How do you get from there to a resistor value for R1 in my schematic? Are we looking at nV here for a decent op amp?
Should the value of this resistor change if we change the op amp?

Thanks for the time and attention.
 

AnalogKid

Joined Aug 1, 2013
8,256
1. For someone asking for free advice you sure have a short fuse. Are you sure you're old?
2. I think you are confusing old with incapable, something that someone in your position in life should know a thing or two about.
3. I don't do emojis. That was my version of humor.

The 741 isn't the worst of anything, it just predates modern improvements. As noted above, there are many reasons why you are not seeing the results you expect. That does not mean that the results are wrong, or we don't understand your questions, or are not capable of original thought beyond regurgitating text book statements. Maybe things that sound to you like textbook statements perfectly capture the answers to your questions. Also, not all of the reasons apply all of the time.

"equal to the parallel value ..." is not a hint, it is the answer. It is exactly what an analog circuit designer needs to know to mitigate the effects of non-ideal components. If you consider the two inputs as being nearly identical error current sources, then the total equivalent DC resistances that each of those two sources see determine the error voltages developed at the inputs. If the DC resistances in the external circuits are nearly identical, so are the error voltages, which can be cancelled out by the common mode rejection capability of the input differential stage. If the resistances are not identical and you still need low error contributions and high circuit gain, the offset adjust pins can be used to compensate for the unequal input conditions. Also, the whole balanced input bias current thing is not perfect either; note the word "nearly" above. Input bias currents are not perfectly balanced, so exactly matched DC impedances do not guarantee zero error, they merely help.

In many countries the 741 is used in schools because it is the only part they can get. I contend that it is the perfect teaching component precisely because of its...personality.

And don't get all cranked up about my age comment above. That really is my version of humor. I'm 64.833, and I think Wally is near that. If he isn't, he should be.

ak
 

WBahn

Joined Mar 31, 2012
25,104
< :) >Ha, ha, I see. I appreciate your time so far. In other words you remember what the book said but can't apply it to a real circuit? </ :) >
Resorting to the tactics used by petulant children isn't going to help your cause. Given that I've designed many mixed-signal integrated circuits including circuits that count individual photons or that stare straight into a laser and lock onto signals that 0.01 ppm of the laser's intensity, I'm willing to bet that I very much can apply these concepts to real circuits.

Is this the answer you would give to a student in a classroom?
No. I work through the theory and work examples. The students have paid tuition and I am being paid to present this material. This is not a classroom. You are not paying tuition. I am not being paid to hold your hand. You are responsible for the bulk of the effort to learn the concepts and we are willing to assist. I told you the what the topic is that you need to research. There's little to be gained complaining that I didn't devote hours to walking you through it step by step.

I can get down to a few millivolts reliably but if the difference is only in microvolts I will never see the result. Will it ever matter?
That depend on the application. As I and others have stated, many applications are not sensitive to these issues at all and you can completely ignore them. Many of the op-amps I used in chip designs had open-loop gains of 10 to 100 and yet, for many circuits, I could ignore that and treat it as an ideal op-amp. Not ignoring it would have been an unjustifiable waste of the customer's money. For other circuits ignoring the finite gain would have been disastrous.

Try increasing your R3 to 100 kΩ and your R2 to 10 MΩ and see what that does. Try making both resistors 10 MΩ. Make your input voltage in the 10 mV range (which is a large output signal for some sensors).
 

PeterCoxSmith

Joined Feb 23, 2015
148
< :) >Ha, ha, I see. I appreciate your time so far. In other words you remember what the book said but can't apply it to a real circuit? </ :) >
Is this the answer you would give to a student in a classroom?
I can find thousands of words on the subject but none relate the concept to a real circuit. Just theoretical concepts I can't bend to apply to a real circuit.
I continue to try and apply. So far I have only tried a gain of 10 and a significant (tenths of a volt) input. Tomorrow I will try higher gains and lower input voltages. I can get down to a few millivolts reliably but if the difference is only in microvolts I will never see the result. Will it ever matter?
I've been running gains of x5000; I would not use a 741 even with balanced resistor paths.
 

WBahn

Joined Mar 31, 2012
25,104
I've been running gains of x5000; I would not use a 741 even with balanced resistor paths.
Hmmm. Open-loop gain as low as 10k, input offset voltage in the 5 mV range, input bias currents as high as 800 nA and offset currents as high as 400 nA, gain-bandwidth product of as low as 500 kHz.

Just guessing, but I'm thinking you don't lose a lot of sleep worrying about having made that decision, huh? :D
 
Status
Not open for further replies.
Top