Re: [sdiy] Rail to rail op-amp?

[jhaible at debitel.net]

>I have not seen such an op amp for +/-15V, but new ones pop up every
>day. Most manuf have parametric searches on their web sites so it might
>not be more than a days work to hit the major players.
>
>Maybe it would be better to rethink your design; perhaps use larger
>supplies to get the desired output swing.

You can go very close to the rails with a FET output stage.
If you'd need +/-7.5V and only low current, I'd say use a 4000 series 
inverter in linear operation.
Now, for +/-15V, you can build the same configuration with
discrete FETs. It's important that your output stage has voltage
gain - otherwise you just shift the problem to the driver stage.
So your output would be the *drains* of the FETs connected
together, and a only  feedback network will bring the output
impedance down to useful values.
And then there is the current. If you need a lot of current,
this will require too much bias current thru the two FETs
(if your circuit should be self-biasing as the 4000 inverters),
or in requires some complex control circuitry for the two gates,
to get a sort of class B operation.

I hope this gives some idea at least - it depends a lot on your
application. I think there were CMOS opamps which had such an
output stage (CA3130 ? - not sure), but I also think they didn't
work on +/-15V.

I would be surprised if you'd find a modern rail-to-rail opamp
for +/-15V. 5V (or less) single supply demand was the driving force
behind the development of modern rail-to-rail opamps, after all.
And no one uses a high voltage process for chip design unless he 
really has a good reason to do so.

After all my babble - what is th eapplication?

JH.