[sdiy] Mind blown -- TL074 saturated outputs different

David G Dixon dixon at mail.ubc.ca
Wed Nov 28 21:01:24 CET 2018

> Have I posted my "READ THE DATASHEET" rant before? I recall 
> Bob Pease saying (both in print and in person when National 
> put him on tour circa 2001) very emphatically that you never 
> use "typical" values for design, that you should always use 
> Min and Max to calculate the WORST CASE values, as these are 
> the only guaranteed numbers. "Typical" means it MIGHT be that 
> value on a good, sunny day, but it's certainly not guaranteed.
> Here's the datasheet:
> http://www.ti.com/lit/ds/symlink/tl072a.pdf
> Looking on page 13, the "Maximum peak output voltage swing" 
> shows (with +/-15V supplies, temp at 25C and a 10k ohm load) 
> minimum of
> +/-12V and typical of +/-13.5V. There's no maximum given, but even
> between the minimum and typical, there's a 1.5V variation.
> Furthermore, there's no mention that these are or should be 
> symmetrical, so the output could well have a negative peak of 
> 12V and a positive peak of at least 13.5V, or vice versa.

I found that datasheet to be completely worthless for this.  To me, the
notation "+/-" implies symmetry.  If the output was definitely asymmetrical,
then there are better ways to express the voltage range than "+/-".  If the
person who wrote that datasheet was a student of mine, he'd get a solid C-
for that effort.

Indeed, I hate it when people say "read the datasheet" like I'm some kind of
fucking idiot.  Of course I read the datasheet, after I discovered the
difference, and just as I predicted, I found absolutely nothing of value in
the datasheet.  It doesn't address this issue at all.  I also found nothing
on the internet.  I've always been told that the opamp's output is "two
diode drops below the rail voltage".  Well, I guess these are some pretty
variable diodes!  Not only that, but the positive output is always
significantly larger than the negative output, but this is not part of the
model for TL074 that TI presumably gave to NI for Multisim, so again, the
eggheads got it wrong.  If this difference is always the case, than I would
expect it to be expressed in the spice model.

Whatever.  Now I know, and I learned it the only way I ever learn
anything... by doing.  I don't believe anything anybody says until I prove
it to myself, and this episode is just another validation of that attitude.

More information about the Synth-diy mailing list