[sdiy] Dual 914 diode in a SOT-23 package?

[Neil Johnson]

Hi Justin,

I'm at work now and my op-amp cook book is at home so I can't comment
on specifics.  However make sure that you're not confusing inverting
and non-inverting op-amp configurations.

> Not disagreeing with you - just asking for clarification. The last time we had this thread it did kind of end up as a three way tie so I know there's some different opinions on this matter.

So your circuit, I ran it up in LTSpice (before production I'd
actually build one and characterise it).  If we just consider
excessive positive inputs, diode D8 is leaking about 2nA during the
0-9V inputs.  If this was in the audio path you have just introduced
something similar to cross-over distortion.  The leakage current is
fairly constant, so that 2nA effectively adds a 0.4mVp-p bipolar
square wave to the output.

As the input at the jack reaches 9V then D8 starts turning on
properly.  It slightly affects the rise of the -ve op-amp input until
about 10V at the jack, at which point the op-amp can no longer balance
itself by negative feedback as it has hit the rail.  The op-amp's -ve
input then steadily climbs until the jack input voltage reaches about
60V....

... At which point D10 starts conducting, and limits the -ve input pin
to about 12.6V.  As we're looking at a positive over-voltage input and
you're using a TL072 then you're OK.  However, for a negative
over-voltage input, which would see the op-amp's -ve pin go to -12.6V,
this has exceeded the common-mode input range, the TL072 will go into
phase reversal, and you have locked up your input circuit until you
power-cycle it.

If you just want to protect your input, then I'd have a 1k0 to 3k3
input resistor, reverse diodes to the rails, and not bother with
anti-parallel across the op-amp inputs.  OR just put low-leakage
anti-parallel diodes across the op-amp inputs with no offset
cancelling resistor (R10 in your schematic).

Note that both these simplified diode protection circuits will
introduce some audio distortion.  How much is acceptable to you is
down to your own requirements.  Designing a low-noise protected input
is not easy!  If you want a good example check out the schematics of
HP's 339A distortion analyser, and note the interesting combination of
fuse, diodes, zeners, FETs, capacitors, bootstrapping, etc, used to
minimise distortion at the critical input stage.

Cheers,
Neil
-- 
Modules and more: http://www.cesyg.com
Homepage: http://www.njohnson.co.uk