Modern opamps with null offset

[RES]

Hello EEVworld

Q: Why do old opamps (LM741, TL081) have output offset null adjustment pins and it seems modern opamps don't?
Which modern opamp has an offset adjust, or has offset adjust pins? Or how or where to find them.
This is the problem; Can't adjust a digital potentiometer to zero Volt because the wiper resistance is 50 Ohm when at lowest position (GND). Try to make an adjustable voltage reference to zero Volt from a 256 digipot by using an opamp with offset null adjustment.
Thanks.

[c4757p]

Because having to manually tweak the offset isn't really compatible with modern production methods. It's expensive to pay someone to twiddle the knob. Much cheaper to just buy an opamp that comes from the fab with a sufficiently low offset already.

[RES]

okay I see.

[dom0]

It is also quite easy to get wrong, because you need to null the offset of the op amp and only the op amp itself. The offset voltage drift is (first order wise) proportional to static offset voltage, so if you null the circuit, and not the op amp, you can get tempcos that are far worse than almost any op amp with no nulling. After a certain point this also has an impact on linearity errors, i.e. distortion increases.
(The exact implications of the offset null depends on the nulling scheme in the op amp: most use a 741-style adjustment)

Last but not least what c4757p said: besides the fact that a correct offset null can be difficult or impossible in some circuits it is also a manual labor step that can take a good while and is simply not compatible with time + cost requirements in modern fabrication.

'Modern' fab processes are much improved, so you get much better precision and a much lower price point: The fabrication is better, so better un-trimmed accuracy, plus faster (=cheaper) and more accurate wafer trimming.

A completely different story are compensation pins. These can be quite handy, because they can also be used as a strobe input in many op amps. These mostly went out the window because most op amps are dual in SO-8 or smaller packages and there just ain't enough pins for that.

[Paul Price]

There are many models of single op-amps packaged in 8-pin packages and so have room for and have offset adj. pins.

Since most circuits require precision are now using MCU's, it is easy to simply measure any op-amp offset and subtract this reading from all measurements, so correction is accomplished by a simple subtraction operation and so low cost op-amps can be used to replace more expensive precision op-amps for most tasks.

[Kleinstein]

Temperature coefficient scaling with the offset setting is only true for BJT based OPs. With FET OPs the two parameters don't interact that much.

Modern precision OPs (e.g. OP277) in single 8 pin package usually still have the offset adjust pins. Today also Auto zero OPs are not that special any more. So if needed a AZ OP could be cheaper and definitely smaller than a precision OP with trimming.

[uncle_bob]

Hi

The simple answers are:

1) It didn't work very well.

2) Cheap chopper stabilized amps have made it an un-needed feature in about 99.99% of the cases.

3) If you *do* need that level of nulling, it is easy enough to drop into the middle of the circuit.

That said, indeed there are still a very few amps out there with a null input. You have to dig pretty far and often they are (as you have seen) new spins of old designs.

Bob

[RES]

Found one; the LT1636, an over-the-top opamp.

Input Offset Nulling
The input offset voltage can be nulled by placing a
potentiometer between Pins 1 and 8 with its wiper to
(see Figure 1). The null range will be at least ±1mV.

[David Hess]

'Modern' fab processes are much improved, so you get much better precision and a much lower price point: The fabrication is better, so better un-trimmed accuracy, plus faster (=cheaper) and more accurate wafer trimming.

I am not so sure about this; I suspect the difference is the wide use of cross coupled quads for the input transistors so they match better.  Maybe some patents ran out?

A completely different story are compensation pins. These can be quite handy, because they can also be used as a strobe input in many op amps. These mostly went out the window because most op amps are dual in SO-8 or smaller packages and there just ain't enough pins for that.

Offset pins have their own non-offset uses.  My favorite is to drive the offset pins with a chopper stabilized amplifier to correct the drift and remove the 1/f noise from the main amplifier gaining the drift and 1/f noise of a chopper and the wideband noise of a conventional amplifier.  Another use is that some operational amplifiers allow the input stage to be bypassed and replaced with a discrete stage which drives the offset pins to get to the second stage.

Temperature coefficient scaling with the offset setting is only true for BJT based OPs. With FET OPs the two parameters don't interact that much.

And it is not even true for all bipolar based operational amplifiers.  It was a pretty neat feature though when it was first introduced (LM121? LM725?  OP-05?  LM308A?).  I think the National LM121 datasheets and application notes have the best discussion about it.  PMI also mentioned it in some of their datasheets.

Modern precision OPs (e.g. OP277) in single 8 pin package usually still have the offset adjust pins. Today also Auto zero OPs are not that special any more. So if needed a AZ OP could be cheaper and definitely smaller than a precision OP with trimming.

Chopper stabilized amplifiers are not suitable for all applications unfortunately.  They suffer from long overload recovery and high wideband noise.

What about the offset pins on comparators?  Comparator offset is sure something you are not going to calibrate out after the fact.

[Kalvin]

There are some op amp families which are factory trimmed giving very good offset voltage characteristics, in range of only tens microvolts and very little drift.

For example the TI's e-Trim op amps OPA192 and OPA2192 achieve max. 25uV offset voltage, typ. drift 0.15uV/Celcius, RRIO, GBW 10MHz, Power supply 4.5V ... 36V: http://www.ti.com/product/opa192

Other manufacturers have similar products.

It might be very difficult to achieve similar and consistent characteristics with external offset trimming.

[David Hess]

There are some op amp families which are factory trimmed giving very good offset voltage characteristics, in range of only tens microvolts and very little drift.

There are lots of them going back to at least the OP-07 which is a trimmed OP-05.  They and their modern descendants still include external offset trim capability.

For example the TI's e-Trim op amps OPA192 and OPA2192 achieve max. 25uV offset voltage, typ. drift 0.15uV/Celcius, RRIO, GBW 10MHz, Power supply 4.5V ... 36V: http://www.ti.com/product/opa192

Other manufacturers have similar products.

It might be very difficult to achieve similar and consistent characteristics with external offset trimming

The OPA192 is a good example showing why offset trim is not as common as it used to be; its rail-to-rail input stage cannot be externally trimmed with one adjustment!

The OPA192 like almost all (*) rail-to-rail input amplifiers has dual input stages which are active depending on the common mode input voltage; they have to be trimmed separately.  Even doing this, over its full common mode voltage range precision suffers because the two differential pairs will not be well matched to each other.  Notice that two separate specifications are given for offset voltage, CMMR, and noise with them all being significantly worse (2 to 5 times) when the common mode voltage is close to the positive rail where its n-channel input stage becomes active.

(*) Exceptions to this include some National LMC series operational amplifiers which use input MOSFETs which shift from enhancement to depletion and were some of the first rail-to-rail input operational amplifiers (these have horrible precision and the OPA192 and similar are a huge improvement over them) and at least one chopper stabilized amplifier (LTC1152) which includes an internal charge pump to provide bias for the input stage.