Historical question about op amps

[Lee Leduc]

Why do quad op amps have their power pins reversed from single and dual packages?
Just curious.

[Paul Rose]

The single pinout goes back before the 741.  It just is what it is.  It was widely copied for compatibility.

The double pinout is more logical, with one amp per side, following the corner pin standard for most chips of the time.

The quad uses center power pins, which is nice for bypass cap layout, and has 4 nice logical areas for each amp. 

There were some (not so common) DIP logic gate packages that had center pins ( I think TI promoted the idea ).  If I recall correctly, they had VCC on the "left", just like the dual quad op-amp.

[Zero999]

The pin-out for the quad op-amp, makes perfect sense.

I'm more interested as to, why did they choose a totally different pin-out for the quad comparator?

http://www.learningaboutelectronics.com/Articles/LM339-quad-voltage-comparator-circuit.php

http://www.ti.com/lit/ds/symlink/lm2901.pdf

[Jwillis]

Just a guess but wouldn't pin outs be placed according the the chip layout to take the shortest possible route minimizing the wiring inside the plastic case.

[danadak]

Many factors in pinout and chip layout. Stuff like
getting power out thru the leadframe, the parasitics
associated with routing, thermoelectric effects, die
hot spots, crosstalk.....

A witches brew of conflicting needs.


Regards, Dana.

[mikerj]

The pin-out for the quad op-amp, makes perfect sense.

I'm more interested as to, why did they choose a totally different pin-out for the quad comparator?

A total guess, but it does provide separation from fast edges on the outputs with and the inputs i.e. minimising crosstalk issues.

[Circlotron]

For the quad comparator, having the output pins all together would make it easy to strap them together for a sort of wired OR output. For the dual and quad op amp, having the (-) next to the output makes board layout easy for a unity gain buffer. There may be other reasons of course.

[Zero999]

For the quad comparator, having the output pins all together would make it easy to strap them together for a sort of wired OR output. For the dual and quad op amp, having the (-) next to the output makes board layout easy for a unity gain buffer. There may be other reasons of course.

Sounds reasonable. If that's the case, do you have any idea why they didn't do the same with dual comparators, such as the LM393?

http://www.learningaboutelectronics.com/Articles/LM393-comparator-circuit.php

[David Hess]

The double pinout is more logical, with one amp per side, following the corner pin standard for most chips of the time.

There was a second dual operational amplifier pinout which can be seen in the LT1013 datasheet which other early precision parts also used.  I have no idea how it came about.  I once ordered a whole tube of these for prototype production before noticing the difference.

There were some (not so common) DIP logic gate packages that had center pins ( I think TI promoted the idea ).  If I recall correctly, they had VCC on the "left", just like the dual quad op-amp.

Centered ground and power pins are used to lower ground bounce by lowering parasitic inductance.  TTL FAST and AS and CMOS AC logic are examples where corner power and ground pins limited performance.  Fast bus oriented logic devices which have lots of simultaneously switching outputs also used centered ground and power pins for the same reason.

I'm more interested as to, why did they choose a totally different pin-out for the quad comparator?

This isolates the fast output edge from the input signals.  Coupling to the inverting input is especially problematical as it will make the comparator "chatter".  Note that JFET input comparators were never popular because of the same problem; high input impedance is not compatible with the fast switched output of a comparator.

For the quad comparator, having the output pins all together would make it easy to strap them together for a sort of wired OR output. For the dual and quad op amp, having the (-) next to the output makes board layout easy for a unity gain buffer.

For an operational amplifier, a small amount of coupling between the output and inverting input is actually desirable to cancel capacitance to ground at that point which would otherwise lower phase margin.

Sounds reasonable. If that's the case, do you have any idea why they didn't do the same with dual comparators, such as the LM393?

That is a good question.  The LM393 is not really fast enough to require the extra isolation and they may have wanted to keep the LM358 pinout for easy substitution since it is commonly used as an even slower comparator as well.  High speed dual comparators use a different pinout.

I like Circlotron's explanation for why the equally slow LM339 didn't use the standard quad operational amplifier pinout; bridging four open collector outputs would be more difficult than two.

[T3sl4co1l]

Just a guess but wouldn't pin outs be placed according the the chip layout to take the shortest possible route minimizing the wiring inside the plastic case.

Back when silicon was expensive, this was probably a big motivation.

I suspect a lot of CD4000 pinouts were designed this way.  What a cluster!

It was later realized that silicon is cheap, pins are huge, and PCBs are even huger; there is much savings to be had by doing all the hard routing work on die.

Not that this is a universal method.  A lot of MCUs have shitty pinouts.  Comparing an orderly ATMEGA in QFP to a messed up MSP430 in QFN, I literally waste more space on routing signals all the way around the chip, than the package size difference alone!  (Given the same 2 or 4 layer design, that is.  Of course, the QFN unconditionally saves space when more layers are available, especially if blind vias or HDI are used.  But, those are expensive, too.)

Arguably, some of that can be "fixed in software", but it's a big pain shuffling around port bits.  One option costs development time, the other costs physical space...

On a related note, you sometimes see chips in different SMT packages, and one has a rotated pinout, or something weird like that.  Usually this happens when the die is rectangular, and just barely fits inside the package in question -- but only in one direction, so the pins have to be bonded out differently.

Tim

[David Hess]

On a related note, you sometimes see chips in different SMT packages, and one has a rotated pinout, or something weird like that.  Usually this happens when the die is rectangular, and just barely fits inside the package in question -- but only in one direction, so the pins have to be bonded out differently.

Oh hey!  That explains the pinout of early precision dual operational amplifiers in SO8 packages.  Mystery solved.

Later parts must have used an SO8 friendly layout which would have worked fine in a DIP package to avoid this problem.

[IanMacdonald]

I dunno, but it always seems that once someone has established a nice regular standard that everyone can work to, someone else just HAS to ruin it.