[sdiy] The TL074 has changed!

[mskala at northcoastsynthesis.com]

On Sun, 3 Aug 2025, Mike Beauchamp wrote:

> Sure. Here's the section of the schematic:
> https://mikebeauchamp.com/dump/tl074-prob.png
...
> However, with the new TL074 it does something really different. With the
> resonance potentiometer at full, adding the additional CV source works for a
> bit and then totally turns the resonance off after about 1V. As though the
> op-amp went into phase reversal, but that's not something I'd expect from the
> specs of the new TL074.

Doing the Ohm's Law calculation:  with the resonance pot at maximum
(slider at 15V) and the op amp inputs at zero, you get 15V across the 47k
resistor R249, which will put 319uA current through it.  Then with the
resonance CV at +1V you also get 56uA coming in through R248 for a total
of 375uA.  In order to balance that out with current through R253 and R329
(total resistance 42k) the op amp will need to bring the emitter of Q14 to
about -15.7V, and its own output to something like -16.4V.  Which, of
course, it can't really do on a +-15V power supply.

What will really happen is that the negative input of the op amp will be
driven to a voltage a little higher than zero, reducing the current
through R248 and R249 and increasing the current through R253 and R329,
until these current balance out - and that balance point will depend on
exactly how negative the op amp is able to drive its output.  If its
drive capability changed because of the new version of the op amp, it
could explain the difference in behaviour.  Very likely the older TL074
wouldn't really go rail-to-rail, so you'd get some backing off or
soft-knee behaviour as you approached this point, and the newer TL074H can
go closer to the rail, so the knee is sharper with the change.

I don't think there is really any phase reversal going on, because the op
amp (either version) can surely go at least as far as about -12V output,
at which point under the described circuit input conditions the op amp
inputs will still be within a couple of volts of zero, nowhere near the
point where they'd trigger phase reversal.  I think the real issue is
clipping:  the newer version can drive its output closer to the rails, and
so the circuit which depends on clipping behaviour, ends up clipping
differently.

-- 
Matthew Skala
North Coast Synthesis Ltd.