Joe's op-amp design takes advantage of the low incremental resistance of D1 (especially versus the 100k resistor feeding current to it) to drive/couple signal to the Base terminal of T2 without too much transfer loss ... by playing with R2 and R4 we can adjust the output offset voltage somewhat; in the case presented (with +/- 9v rails) this output voltage is left at around +500mV DC // which is acceptable in applications where absolute precision is not required ... YMMV
(http://www.lynx.net/~jc/DavissonCompressionOpAmp-OperatingPoint.jpg)
two sets of open-Loop gain curves are provided with no special set-up or care thanks to stable DC conditions (which usually cannot exist with modern op-amps, even in simulation) ... the first (stock version) shows the role of T2's emitter bypass capacitor in setting the lower cutoff ... a mid-band gain of 80 db means a signal gain of 1000v/v (indeed, modest in comparison to modern 1e6/1e7 gain op-amps - nonetheless potentially useful part. in audio) ... the second shows what happens when that cap is removed and Re,Rc left the same (@ unity gain) ...
(http://www.lynx.net/~jc/DavissonCompressionOpAmp-ACopenLoop.jpg)
(http://www.lynx.net/~jc/DavissonCompressionOpAmp-DCopenLoop.jpg)
when operating Joe's op-amp as an inverting gain amplifier, say aiming for unity gain using two external 100k resistors, it would be interesting to note what the output impedance of the op-amp is like ... this might tell us, for example, what to expect from the circuit when put to use in well known (musical) clipping applications (here shown with the 100k feedback resistor locking the loop)
(http://www.lynx.net/~jc/DavissonCompressionOpAmp-Zout.jpg)
one last thing ... TRANSIENT analysis using a stimulus voltage source of 1vpp at 500Hz shows why the op-amp is called "Compressing" ... the diode string inside the op-amp causes anti-symmetric clipping when clipping does hit // this produces a bias shift, and gain shift ... and the perception of (dynamic) compression
(http://www.lynx.net/~jc/DavissonCompressionOpAmp-Sine1vpp500hz.jpg)
(http://www.lynx.net/~jc/DavissonCompressionOpAmp-oneSidedClipping.jpg)
you may ask yourself, what can I do with this thing ?! ... or, why use it even ...
well, almost anything you'd want to do with a std. modern (high-gain) op-amp ... the kind you get in IC packages ... you might want to try with Joe's
except, you wouldn't expect Joe's op-amp to make the circuits the other op-amp live in sound or work the same
case in point, the test characteristics posted above // though not performed with pedantic regard for canonical representation // can be used to give broad traits to this circuit:
open Loop gain of 1000 when operating in high-gain mode, open loop of 100 in low-gain mode ... Zout (in lock) ~1.3k ... linearity/accuracy: especially poor (passive biasing, diode string) ...
the low-drive aspect of this op-amp should catch the eye of people trying to get diodes to clip (softly) like tubes
something to be said about low-gain analogue
---
(simulation platform http://www.ti.com/tool/tina-ti (http://www.ti.com/tool/tina-ti))
Very interesting thanks for sharing your insights.
-Joe
what about feedback loop diodes as in a TS? don't they interfere with the circuit?
Clipping looks like a triode. ;)
Interesting. Have to breadboard it.
Cheers.
Feedback diodes should be fine. I remember testing ts9/dod250 type stages, and the first two stages of the Shredmaster. Probably should stick with regular opamps for other things (tone controls, etc).