Opinions: On op amp replacement

[Rene Schmitz]

At 03:24 30.01.99 +0100, you wrote:
>> I find the greatest reductions in circuit (audio) noise often come from
>> replacing Carbon Comp. resistors with Metal Film. The metal films are
>> usually 1%, which makes summing and diff amps happier at the same time.
>> Also, faster op-amps somtimes do weird things. I built a phase shifter
>with
>> quad op amps (LM324) and replaced them with the TL084 (it oscillated at a
>> near ultrasonic frequency). It drove my mother's dog nuts (but she
>couldn't
>> hear it). Made me want to not fix it (I hated that Dog...... :-)
>
>I can confirm that. Phasers are aparently quite sensitive. No HF
>attenuation
>in a chain of all pass filters, opamps driving capacitive loads with
>variable
>resistors in series ...
>
>Built a VACTROL-based 12-stage with TL072 and ran into problems. Works
>fine with TL062's and 6 stages, though.
>
>Replaced the dual opamps in my HiFli Clone with NE5532's and ran into
>stability problems at extreme modulation settings.

That being, I am puzzled on what opamp to choose! Phasers usually have so
many stages that It would be uneconomic to use singles, and I think this
problems 
arise from the way they are usually constructed. There are however opamps
specially (?) designed to drive capacitive loads, like the LF356.
Or perhaps it would help to put a 10pF cap over the resistor from the
output to the inverting input, to delimit the bandwidth to some 50kHz ?!

>Sometimes 741's and 1458's are not *that* bad !

Yes, but not for instance in a fet-based phaser where the signal is already
down in the mud. I'd say these are fine when bandwidth (LFO) and gain are
limited. 
The circuit where I had the trouble with noise was an audio circuit, here
we have 
usually input levels of 200mV and I would not use a 741 to process this.
Though there are audio circuits from the 1980s where 741s where used in EQs
and the like, but phono preamps where usually discrete at that time. 

>
>BTW, I think one can avoid slew rate distortion on slow opamps by using
>very small feedback (or load) resistors. They would clip the signal with
>the
>output current limit feature of the opamp, which should sound much more
>pleasant than intermodulation, or than getting stuck at the supply rails.
>Note the words "think" and "should" - I have not really made tests so far.
>It's
>just some speculation about existing designs.

Signals faster than the slewrate should be processed by a faster circuit.
If you build things from scratch a TL071 wont set you back more than a 741.
But to answer the question of the original poster, what to do with the 741s,
I'd suggest using them in non critical CV applicatons, where offset and
drift doesn't matter. For instance in the current source for an OTA VCA. Or
as a buffer for an ADSR, or LFO. However I'd not put these into the audio
path. 
For all new designs I use now TL08x /TL07x, and where more speed is needed 
I plan to use the TL05x (which has twice the slewrate!)

There is one thing to look out for: The circuits from the 70s and 80s where
the 741 is used thoughout, usually have lots of FET buffers in them, where
low leakage is needed. It would pay to change the circuit as well, since
the FET-resistor buffer has a nonlinearity and offset drift! (I have the
formant, and the SEM-1a and even the TB303 in mind here). When changing the
circuits to FET-opamps then these buffers become obsolete.


Bye
 Rene



             ,   : (uzs159 at uni-bonn.de)
|") [" |\ | ["   : (http://www.uni-bonn.de/~uzs159)
|"\ [_ | \| [_   : (http://members.xoom.com/Rene_Schmitz)