Can op-amp offset error be calibrated out?

Go To Last Post
5 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi - I'm working on a design that hopefully will hit mass production (millions of units/year). To save cost, it'd be nice to use a high input offset op-amp. This will introduce an error in some analog readings. Can I calibrate out this error at the factory? The temperature coefficient of the offset voltage is not bad at all. I'm just worried that for whatever reason this offset will drift over the life of the part.

edit: also, can you count on offset error always having the same polarity? (ie all chips from one manufacturer would have +offset error)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well, if the goal is to shave a few cents off the production costs yuo won't like my answer...

I recall one design where there was an extensive analog front end to the (biomed) insrument. I ended up putting an analog switch early in the signal chain and the uC could inject a known calibration signal, and then back calculate the correction factors.

It worked fine, hit the cal button and then use the instrument.

Of course I built one of them, not one million.

Perhaps you could hire me to calibrate them for you ;)

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Giorgos_K wrote:
The input offset can either be positive of negative. Most old-fashioned op-amps have an extra couple of pins to null their input offset voltage externally, using a trimmer. But, the long term and the temperature drifts cannot be avoided.

A possible solution is the use of "Auto-Zero," "Zero-Drift" or "Chopper-stabilized" amplifiers, which have extra internal circuitry to constantly measure and compensate the input offset, minimising the drift over time and temperature and the effect of 1/f noise. Their Vos can be as little as a few micro-volts, while there are parts with sub-μV input offset.

Of course, this line of op-amps is more expensive and, the worst part is the extra time the amplifiers will need to re-balance themselves when their input saturates and/or their output overloads. The recovery time penalty you will have to pay is in the order of a few tens of milliseconds. Analog Devices stands out with some of their Zero-Drift amplifiers that have a recovery time of a few microseconds only.

I guess that either solution (an extra trimmer plus the time needed to null the offset, or a costlier part) has its cost...

-George


George - another solution for us is to just use more expensive OAs with better offsets. Choppers are definitely too spendy. I'm worried about long term drift of the cheaper OAs with high Vos - how much of a problem is that?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If its an old (and cheap) op amp, the bias currents are all the same direction, but the offset voltage can be either polarity. The input offset current can also be either polarity. Input offset voltage CAN shift over time, though its not likely to, particularly if the temperature coefficient is reasonably small.

I have a hard time believing that calibration costs will be lower than the incremental cost for an op-amp with the offset voltage you need.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tere are two possible things here.
1) you want to limit the initial amount of offset as much as possible and then the offset changes over time will be there, but they are well within the limits you have set

2) you want to always have 0 offset.

In case 2 you will need to always callibrate your system before use as you do not know the temperature at the time of use, nor the supply voltage, nor the drift of the opamp itself.

in the first case from the head you can use some additional resistors to reduce the offset from the opamp. This will mean that you need some potmeters to get things exact and then thus need more time in the factory for testing and callibrating.
I think it might be worth checking what the factory calculates for that extra needed test time and then perhaps decide to buy a more expensive opamp as this might in the end be cheaper then the workers needed. not mentioning the quality of the callibration that will dependent on how a workers mood is and how much pressure there is to produce enough units at a given day.