Hi all. I've got what it seems to be a simple circuit. I've got different voltage on outputs (~100mV difference). I tried swapping all matching resistors (high with low) with no result. What could the reason for this behavior be?

 

The input offset voltage is guaranteed to be less than 0.007 volts per amplifier. 7 mv multiplied by a gain of 20 is 140 millivolts (each). Maximum error in this configuration is 280 millivolts. You measured 100 millivolts. That is within spec.

ps, please include the power supplies for the amplifiers in future posts.

 

I was assuming offset voltage on inputs of a dual opamp would match. Wrong I guess?

 

Your measurements just proved the guess was wrong.

 

Thanks, I guess it was it...

 

I was assuming offset voltage on inputs of a dual opamp would match. Wrong I guess?

If they could do that they should to avoid offset error in first place

 

If they could do that they should to avoid offset error in first place

That is a good observation. A dual op-amp with equal offsets could be arranged to eliminate the offset!
I don't think those are available...or is that how it is done when making very precise op-amps?

Or did I just ask the wrong person?

 

I mean that if practical to make offset predictable then practical to make offset exactly as wanted
Sorry that my posts are deleted because of topic

 

if practical to make offset predictable then practical to make offset exactly as wanted

It is possible. Here is a page of single and dual op-amps with 1 micro-volt of input offset voltage, in the range of $6.50. Whether that is practical depends on the circuit requirements.

http://www.mouser.com/Semiconductor...tional-Amplifiers-Op-Amps/_/N-6j73m?P=1yvcej2

 

It is possible.

FWIW: 'My money' says no! --- Nothing, but nothing is perfect and hence, "exact"!

TTFN
HP

 

Opamp input error theory says that equalizing the DC impedances seen by the inputs will reduce the untrimmed offset error. Add a 10K resistor (9.53 K if you're picky) in series with each + input to see.

It is safe to assume that whatever errors there are in dual and quad devices, they will track with temperature. Not perfectly, but if one amp's offset error increases with temperature, the other one will also. The LM358 was an amazing feat of engineering in its day, but that was 16,000 days ago. And even then, it was more robust than accurate.

ak

 

The LM358 was an amazing feat of engineering in its day, but that was 16,000 days ago

So... The LM358 'hit the market' on Sept 26 1971 (a Sunday?!) --- Aye! aye!! aye!!! -- I'll bet you play one mean game of Trivial Pursuit!

Best regards
HP

 

Opamp input error theory says that equalizing the DC impedances seen by the inputs will reduce the untrimmed offset error. Add a 10K resistor (9.53 K if you're picky) in series with each + input to see.
......................

It's perhaps obvious but that's for any output offset error due to the input bias currents. It has no significant effect on the output offset due to the input voltage offset.

 

So... The LM358 'hit the market' on Sept 26 1971 (a Sunday?!)

OK, 16,000-ish.

 

I mean that if practical to make offset predictable then practical to make offset exactly as wanted

It's possible.

'My money' says no! --- Nothing, but nothing is exact!

The blonde says, "exactly". The pragmatist says, "It's possible." but forgets to say, "to about 1 part per million", and the logophile objects. Entirely predictable.

Then the logophile actually counts out 16,000.0 days on a calendar and the pragmatist thinks, "Academic is a nice word for pedant".

 

Just so much matching can be achieved with op amp design. Input transistors can be played with until they achieve good results or matching, but production variables always accumulate to ruin the performance. Precision op amps and instrumentation amps use active trim to achieve their excellent performance, and active trim costs time and silicon area (money). Anybody who uses LM358s expecting any degree of matching will be sorely disappointed. Adding a 10K resistor in the positive lead of the op amp reduces the bias current offset, not the voltage offset, and it introduces a small but real common mode voltage that can introduce other problems. Conclusion: buy the precision you desire and be willing to pay for it---written by an ex semiconductor company employee.

 

I found this in the Jokes Thread. It seems appropriate.

A mathematician, a physicist, and an engineer were asked to review this mathematical problem. In a high school gym, all the girls in the class were lined up against one wall, and all the boys against the opposite wall. Then, every ten seconds, they walked toward each other until they were half the previous distance apart. The mathematician, physicist, and engineer were asked, “When will the girls and boys meet?”

The mathematician said, “Never.”

The physicist said, “In an infinite amount of time.”

The engineer said, “Well... in about two minutes, they'll be close enough for all practical purposes.”

 

Then the logophile actually counts out 16,000.0 days on a calendar and the pragmatist thinks, "Academic is a nice word for pedant".

More a (failed?) attempt at humor than pedanticism -- but what the hey! Its a fair cop!

I found this in the Jokes Thread. It seems appropriate.

A mathematician, a physicist, and an engineer were asked to review this mathematical problem. In a high school gym, all the girls in the class were lined up against one wall, and all the boys against the opposite wall. Then, every ten seconds, they walked toward each other until they were half the previous distance apart. The mathematician, physicist, and engineer were asked, “When will the girls and boys meet?”

The mathematician said, “Never.”

The physicist said, “In an infinite amount of time.”

The engineer said, “Well... in about two minutes, they'll be close enough for all practical purposes.”

"Degenerate motion" is (or would be) an interesting, and, I daresay, even inspiring concept were it not for that "pesky' 'plank length issue' fairly ripping away the 'mist illusion' of purity -- Wadda world

But -- to the topic:

Just so much matching can be achieved with op amp design. Input transistors can be played with until they achieve good results or matching, but production variables always accumulate to ruin the performance. Precision op amps and instrumentation amps use active trim to achieve their excellent performance, and active trim costs time and silicon area (money). Anybody who uses LM358s expecting any degree of matching will be sorely disappointed. Adding a 10K resistor in the positive lead of the op amp reduces the bias current offset, not the voltage offset, and it introduces a small but real common mode voltage that can introduce other problems. Conclusion: buy the precision you desire and be willing to pay for it---written by an ex semiconductor company employee.

Incredibly, I find the ('dirt cheap') LMC6482 to be possessed of remarkably consistent characteristics even lot-to-lot? --- Despite the prescribed protocol Re: 'gift horses', I am, nonetheless, keen to understand this

Best regards
HP

 

More a (failed?) attempt at humor than pedanticism --

Thank you. You aren't the only person who makes, "jokes" that I don't understand. I will try to remember there are levels of humor that completely escape me the next time I feel an urge to mail a flaming paper bag to a certain person.

Even with this information, I fear it will take me some time to expand my horizons in that field of endeavor.

 

Thank you. You aren't the only person who makes, "jokes" that I don't understand. I will try to remember there are levels of humor that completely escape me the next time I feel an urge to mail a flaming paper bag to a certain person.

Even with this information, I fear it will take me some time to expand my horizons in that field of endeavor.

No problems! Someday maybe I'll embrace the sad fact that 'dry' humor and textual communication just don't 'mix' -- then again, maybe not...

Best regards
HP