Op amp (and other devices) pinout rant

[daqq]

I'm doing an analog design and I'm getting pretty annoyed with a small thing - the pinout on most SOT23-5 op amps. Basically, I can't figure out the reasoning behind this becoming the industry standard pinout. The only thing that's kinda advantageous is that you can put many of them above one another with a single power bus. The problem is that in most configurations you are tying IN- to OUT via some kind of feedback network (FB), generally a few passive components.

In the classic layout (displayed is the LT1800, but most others are the same) the OUT pin is on the other side of the IN- pin - the farthest possible place it can be. There's no nice way to route the FB network - you can:
 - go to the other side of the board via vias
 - go under the chip, which will disrupt your power routing
 - do something else completely, but still, not a very elegant solution
Basically, you are limited by the pinout and anything you add to the op amp is unreasonably complicated.

But, by simply swapping OUT with VS+ you get a very nice solution - the OUT is right across the IN- pin, the power pins are close together, allowing a very convenient place to put a bypass capacitor, the IN+ is still close to the VS- (it's common to tie it to GND). The power can still be routed 'under' the device, or use vias, or something, no problems there.


There is another common pinout, such as the MAX4490 ( https://datasheets.maximintegrated.com/en/ds/MAX4490-MAX4492.pdf ), which is an improvement, but still, the power pins could be next to one another :-(

I noticed many devices where a simple swapping of a few pins would greatly simplify layout (I noticed a step up converter, can't think of the specific device at the moment ).
Now, this is not a major issue, just a minor annoyance, I can easily work with it. I just wanted to know, whether or not there's any particular reason that that kind of layout was chosen as an industry standard?

[exe]

I think microchip produces sot-23 opamps in several pinout configurations. Check MCP6001U in http://ww1.microchip.com/downloads/en/DeviceDoc/20001733K.pdf for example.

[David Hess]

There is really only one pinout mistake which is possible with an operational amplifier; the non-inverting input should either be adjacent to the inverting input with space on the other side for a trace or should have space on both sides for traces.

The reason for this is guard shielding of the non-inverting input in low leakage applications.  The guard is held at a voltage equal to the inverting input.  It may even include the signal at the inverting input but this is not required and it may be driven seperately.  Without this capability, some operational amplifier applications are not possible because leakage cannot be controlled.

Your layout fails this because the non-inverting input is adjacent to a power pin.  The standard layout for DIP-8 parts works because the physical size is large enough to allow a trace between the non-inverting input and negative power pin however it fails for SO-8 and smaller 8 pin parts and some circuits cannot be built with those packages because of it.

Modern precision low input bias current parts use a 10 pin DIP package in surface mount for this reason.  In the distant past, some parts used a special 10 pin TO-99 package.

[magic]

V- position is probably for convenience: in SOT23 the middle pin is the most obvious one to glue the die to (maybe the only one possible?) and therefore it has to be V-, at least for basic bipolar processes which keep the bulk of the die at V-.

That being said, it's still possible to do so and move both inputs to the opposite side for isolation.

[David Hess]

V- position is probably for convenience: in SOT23 the middle pin is the most obvious one to glue the die to (maybe the only one possible?) and therefore it has to be V-, at least for basic bipolar processes which keep the bulk of the die at V-.

That also applies to common CMOS processes because they are junction isolated.

[exe]

Is it possible change the pinout with different wirebounding and/or orienting the die differently?

[magic]

Actually it is

I found LM358 dies with only one channel bonded out to single channel pinout and sold as OP-07 on AliExpress.
But it isn't common in legit products AFAIK, perhaps to use less bonding wire or to reduce parasitics or both. The bonding pads are laid out in the vicinity of the pins they connect to.

And you can't move the negative supply / ground pin in packages which connect the bottom of the die to one of the pins. At least not without modifying the package to connect the die to a different pin. Such modifications were available in TO92, but I have never seen it in TO220 for example. Not sure if anyone bothers with SOT23. You will notice that most SOT23 ICs have V-/GND on the middle pin.

SO/DIP packages don't connect the die carrier to any pin so the limitation doesn't apply to them.

[David Hess]

Is it possible change the pinout with different wirebounding and/or orienting the die differently?

Yes but with some limitations because the pad arrangement on the die constrains routing the wire bonds.

Check out the LT1013 datasheet where the SO-8 packaged dual has a pinout which is rotated 90 degrees from the standard 8 pin DIP pinout.  This was because the rectangular LT1013 die had to be rotated to fit in the smaller package.  Later dual parts , including a replacement for the LT1013, had smaller dies so this was not a problem and the SO-8 and smaller pinouts were standard.