Trivia - LM193

[S185149]

When was this comparator introduced and who made it first? The TI datasheet for this comparator dates back to 1979. Were they the first or did it come out before 79?

S18

[amspire]

When was this comparator introduced and who made it first? The TI datasheet for this comparator dates back to 1979. Were they the first or did it come out before 79?

S18

I am fairly sure that the LM393 family of comparators was designed by National Semiconductors (bought by TI).

1979 sounds like a plausible date. It could have come out a few years earlier.

[S185149]

Thanks for the lead. There is an entry in wikipedia regarding the origin of the LM series and it was attributed to National Semiconductor. Amazing that these parts are still made and in use after all these years.

[floobydust]

The comparator is listed in my 1978 National Semiconductor Linear Databook as low power low offset voltage dual comparators -
LM193, LM293, LM393, LM193A, LM293A, LM393A, LM2903.
Just different packaging, offset and bias current ranks.

[S185149]

The comparator is listed in my 1978 National Semiconductor Linear Databook as low power low offset voltage dual comparators -
LM193, LM293, LM393, LM193A, LM293A, LM393A, LM2903.
Just different packaging, offset and bias current ranks.

What can I say but WOW! You still have databooks from 1978 and I am thankful for that. Thanks for looking it up.

S18

[amspire]

The earliest reference I can find for the LM393 family is 1973. They did an application note AN-74 on the chips dated January 1973. So they must have released it 1972 or earlier.

It would make sense as the LM324 opamp family came out in 1972. The LM393 used the same input stage as the LM324 opamps so they were probably designed by the same team at the same time. The LM393 is basically a LM324 with an open collector output stage.

They are impressive designs for the time as this is only 9 years after the very first op-amp IC ( the Fairchild uA702A) and only 7 years after the first reliable opamp (the uA 709A). The uA702 was amazing with a bandwidth of 20Mhz from memory, but it did have the quirk that it could self-destruct very easily if voltages on the pins weren't perfectly managed. The uA 702 was designed by Bob Wildar who left a few years later to start up National Semiconductors.

[S185149]

Excellent information. I found a copy of AN-74 by National Semiconductor in what appear to be a Polish web site.

Thanks,

[Zero999]

The earliest reference I can find for the LM393 family is 1973. They did an application note AN-74 on the chips dated January 1973. So they must have released it 1972 or earlier.

It would make sense as the LM324 opamp family came out in 1972. The LM393 used the same input stage as the LM324 opamps so they were probably designed by the same team at the same time. The LM393 is basically a LM324 with an open collector output stage.

Also no frequency compensation on the comparator, otherwise it would be very slow.

[David Hess]

The 1976 Fairchild and Signetics databooks had parts with this design but not the 1973 Fairchild or 1972 Signetics databooks so it does indeed look like it was originally a National design.  The same part made by a second source may or may not use the same prefix or part number.  For instance the Fairchild version of the LM339 is uA775.

It would make sense as the LM324 opamp family came out in 1972. The LM393 used the same input stage as the LM324 opamps so they were probably designed by the same team at the same time. The LM393 is basically a LM324 with an open collector output stage.

They share the idea of using a Darlington PNP input to gain wide differential input range and operation to ground but there is a major difference.  The operational amplifier version also has input stage transconductance reduction, a major innovation, which is not shown on the "simplified" schematics and allows the compensation capacitor to be much smaller yielding a much smaller die and lower cost.  The comparator not needing compensation would leave this out.

The thing which puzzled me about the LM324/LM358 is that Fairchild's version was much better including a class-AB output stage and a single version with offset nulling but they did not last.  I wonder if that was caused by a patent issue.

[floobydust]

The earliest reference I can find for the LM393 family is 1973. They did an application note AN-74 on the chips dated January 1973. So they must have released it 1972 or earlier.

My copy of NS 1973 Linear Applications covers LM339, not the LM393 in AN-74.
Same internal schematic and specs, just different offset values i.e 2mV vs 5mV.

It's like they made a new part number for every grade of part?
I'm not sure how IP agreements worked back then, as the LM710 I thought Fairchild created- but it has NPN diff amp.

[David Hess]

The earliest reference I can find for the LM393 family is 1973. They did an application note AN-74 on the chips dated January 1973. So they must have released it 1972 or earlier.

My copy of NS 1973 Linear Applications covers LM339, not the LM393 in AN-74.
Same internal schematic and specs, just different offset values i.e 2mV vs 5mV.

It's like they made a new part number for every grade of part?

The LM393 is the dual comparator and the LM339 is the quad comparator; otherwise they are the same.  There is also the LM392 which is a single 324 type operational amplifier and a single 339 type comparator.  I do not remember if there was a quad version of the LM392.  There were some other mix and match parts which included references.

For the National LM series, the first digit indicates the temperature range.  Some of the parts had a 4th grade which was had a lower maximum voltage but almost military temperature range (automotive?) like the LM2902 for the LM324 and the LM2904 for the LM358.

[Zero999]

The 1976 Fairchild and Signetics databooks had parts with this design but not the 1973 Fairchild or 1972 Signetics databooks so it does indeed look like it was originally a National design.  The same part made by a second source may or may not use the same prefix or part number.  For instance the Fairchild version of the LM339 is uA775.

It would make sense as the LM324 opamp family came out in 1972. The LM393 used the same input stage as the LM324 opamps so they were probably designed by the same team at the same time. The LM393 is basically a LM324 with an open collector output stage.

They share the idea of using a Darlington PNP input to gain wide differential input range and operation to ground but there is a major difference. 

That's a common misconception. Neither the LM393 & LM339 nor the LM358 & LM324 have a Darlington input stage. They both have a normal PNP input stage, followed by an emitter follower on each input. This is an important subtle difference. Using a Darlington pair differential amplifier would result in lower transconductance, poorer bandwidth and stability, than an ordinary differential pair & emitter followers.

LM393 & LM339


LM358 & LM324

The thing which puzzled me about the LM324/LM358 is that Fairchild's version was much better including a class-AB output stage and a single version with offset nulling but they did not last.  I wonder if that was caused by a patent issue.

I didn't know about Fairchild's version having a class AB stage.

There is a single version of the LM358, the LM321, but it's not trimmable. It came out much later, which is interesting because there are much better op-amps in an SMT package. Perhaps engineers wanted something cheaper than the LM358, in a much smaller package?
http://www.ti.com/lit/ds/symlink/lm321.pdf

[David Hess]

That's a common misconception. Neither the LM393 & LM339 nor the LM358 & LM324 have a Darlington input stage. They both have a normal PNP input stage, followed by an emitter follower on each input. This is an important subtle difference. Using a Darlington pair differential amplifier would result in lower transconductance, poorer bandwidth and stability, than an ordinary differential pair & emitter followers.

You are right about that; even I did not look closely enough.

But it is funny that you say "result in lower transconductance" when that is exactly what the LM324 design is most obscurely famous for although simplified schematics do not show it.  The article written by the LM324 designers in 1972 is locked away behind a paywall and unavailable however National Semiconductor application note A discusses exactly what is going on and the On Semiconductor schematic below shows both the differential pair configuration which deliberately reduces transconductance without emitter degeneration and the fixed 50 microamp current sink on the output.

I assume the LM393 and LM339 comparators lack transconductance reduction and the ON Semiconductor schematic shows this.  Fairchild made a quad comparator equivalent to the LM339 and its schematic also shows a lack of transconductance reduction.

The thing which puzzled me about the LM324/LM358 is that Fairchild's version was much better including a class-AB output stage and a single version with offset nulling but they did not last.  I wonder if that was caused by a patent issue.

I didn't know about Fairchild's version having a class AB stage.

The Fairchild improved parts were the single uA799, dual uA798, and quad uA3303/uA3403/uA3503.  Later Fairchild second sourced the LM324 and LM358 under the National Semiconductor part numbers.

They look more complicated than the simplified LM324 and LM358 schematics because like the On Semiconductor schematic, the Fairchild schematics show the current mirrors, transconductance reduction, and bias circuit.  What they do not show is a 50 microamp current mirror pulling the output down to the negative supply but the datasheets say the output goes down to the negative supply.  Maybe it got left out of the schematics.