LM358 from LT? (crossreference for an LTSpice simulation)

[RoGeorge]

What would be a Linear Technology OpAmp equivalent for LM358, please?  I am not familiar with their ICs.

I need to simulate a LM358 filter (1st OpAmp) + some logarithmic amplifier (2nd OpAmp), at less than 10Hz but with some capacitive loads.  For now, the physical LM358 circuit tends to self-oscilate, but only sometimes.

A model of OpAmp already existing in the default LTSpice libraries would be great, but I wouldn't mind a 3rd party LM358 model as long as it's free.

[Ian.M]

For a first cut at it, use 'Opamps\UniversalOpamp2' with appropriate instance parameters to match your OPAMP's characteristics - see "LTspiceXVII\examples\Educational\UniversalOpamp2.asc" for details.

There used to be a LM358.mod model file on the LTspice Yahoo group files area (now defunkt due to Yahoo), you could try looking for. Use it with the 'Opamps\opamp2' symbol.  If you cant find it anywhere else, its probably on the replacement group: https://groups.io/g/LTspice (registration required).

[T3sl4co1l]

LT1013 or thereabouts, I suppose.

Obviously, it significantly outperforms the LM358...

Tim

[RoGeorge]

Will try, thank you.

For now I am using LT1006, and it doesn't oscillate like the physical circuit does.  I'll just redo the existing shoddy design to get rid of the capacitive load.

[nrxnrx]

The ltspice yahoo group moved to https://groups.io/g/LTspice (along with the models).

https://groups.io/g/LTspice/filessearch?q=358 brings up a lot of models, so I won't attach. You need to make an account and join the group to access the files, but now it's instant - you don't need to wait a day for a person to approve you. You can disable emails, so you can just join for the files.

[David Hess]

Will try, thank you.

For now I am using LT1006, and it doesn't oscillate like the physical circuit does.  I'll just redo the existing shoddy design to get rid of the capacitive load.

The LT1006 is essentially the single version of the dual LT1013 or quad LT1014 but with added flexibility since it supports offset null and programmable supply current.  The LT1013/LT1014 are improved replacements for the LM358/LM324 with much improved precision, an output which actively sinks current to the negative supply, class-AB output stage for no crossover distortion, and no phase reversal.

Linear Technology makes some other dual parts which could also replace the LM358 but the LT1013 is now sourced by multiple manufacturers making it the lowest cost improved alternative.  The single LT1077, dual LT1078, and quad LT1079 are 65 microamp micropower alternatives.  The dual LT1178 and quad LT1179 are 17 microamp micropower alternatives.

[Kleinstein]

The output cross over distortion of the LM358 can responsible for some odd errors. Not all models will include this. To add a crude approximation of it, one can use two anti-parallel diodes at the Ops output.

[David Hess]

The output cross over distortion of the LM358 can responsible for some odd errors. Not all models will include this. To add a crude approximation of it, one can use two anti-parallel diodes at the Ops output.

I am not sure why the LM358/LM324 did not use a class-AB output stage to reduce or eliminate crossover distortion.  At the time they were released, Fairchild had their own versions with different part numbers which did use a class-AB output stage.

[Kleinstein]

The original LM358 has a class B output stage. The cross over error is clearly visible in many cases. AFAIR the dead range is more like 1 diode drop, so the 2 extra diodes give more cross over distortion.

[Zero999]

What would be a Linear Technology OpAmp equivalent for LM358, please?  I am not familiar with their ICs.

I need to simulate a LM358 filter (1st OpAmp) + some logarithmic amplifier (2nd OpAmp), at less than 10Hz but with some capacitive loads.  For now, the physical LM358 circuit tends to self-oscilate, but only sometimes.

A model of OpAmp already existing in the default LTSpice libraries would be great, but I wouldn't mind a 3rd party LM358 model as long as it's free.

There are plenty of third party SPICE models. It's easy to include them in a schematic. Open the file in a text editor and copy it to the clipboard. In LTSPICE, insert a SPICE directive into the schematic, by clicking the .op button on the toolbar. Paste the model into the text box and click OK. Add the op-amp symbol opamp2 to the schematic and change the value to LM358. Now it should work.

Attached is an example with the National Semiconductor LM358 model, which doesn't have any crossover distortion, so isn't very accurate.

[Ian.M]

Although that works, its a real mess on the schematic for all except short model files, and you cant include custom symbols, so it makes more sense to keep the files separate, make a folder and put them in it with the sim file then in the sim, .lib the model file to use it.

[magic]

I am not sure why the LM358/LM324 did not use a class-AB output stage to reduce or eliminate crossover distortion.  At the time they were released, Fairchild had their own versions with different part numbers which did use a class-AB output stage.

I think I found it: µA798, and I suppose there is no particular mystery here, 4x higher supply current was the reason.

The original LM358 has a class B output stage. The cross over error is clearly visible in many cases. AFAIR the dead range is more like 1 diode drop, so the 2 extra diodes give more cross over distortion.

The output uses a darlington on one side so it's 3 diode drops, actually. Given the 0.6V/µs slew rate, theory predicts some 3µs of Hi-Z before the output switches polarity.

edit
It's even worse, more like 10µs in (presumably) TI version here:
https://e2e.ti.com/support/amplifiers/f/14/t/698220

[Zero999]

Although that works, its a real mess on the schematic for all except short model files, and you cant include custom symbols, so it makes more sense to keep the files separate, make a folder and put them in it with the sim file then in the sim, .lib the model file to use it.

I agree, it is messy, although that can somewhat be mitigated by choosing the smallest font size. It does have the advantage of making it easier to post here, because some people are wary about opening a zip file, which can contain just about anything.

[David Hess]

I am not sure why the LM358/LM324 did not use a class-AB output stage to reduce or eliminate crossover distortion.  At the time they were released, Fairchild had their own versions with different part numbers which did use a class-AB output stage.

I think I found it: µA798, and I suppose there is no particular mystery here, 4x higher supply current was the reason.

There is quite a bit of variation between manufacturers of the LM358 and I doubt designers cared.  Later Fairchild manufactured both the quad LM324 and uA3403 and the supply current difference was 1.5 and 2.8 milliamps typical.

[David Hess]

Something else which is odd about the Fairchild parts is that they specify a single supply output however their unusually detailed schematics show only a PNP emitter follower for sinking current.  Yet they specify 1 milliamp of sink current at 400 millivolts which is *better* than the LM358/LM324.  Maybe the schematics are incomplete but there is also no detailed graph of sink current versus voltage.  The outputs will work to ground as shown but an external pull-down would be required.

I might suspect that they never went into production however the part was listed in multiple editions of Fairchild's databooks.  It might be interesting to find some on Ebay for testing.

[magic]

Check AliExpress, there is plenty of recycled chips there. One thing though, you have to type full part numbers with all the -C and -N suffixes because their search engine is dumb as rock.

Of course they could be fakes, but I doubt anyone would bother faking some obscure quasi-LM358 from 40 years ago. And you can test if it's the original chip or some painted over LM358 by the presence of xover distortion.