To make it practical, I wanted it to retain the 3M reverse-audio panel control, and the function of the foot pedal, in particular, absence of the pedal leaves the effect off. Thirdly, I want to power it from some source already within the amplifier's circuitry and easily fit into the space in the chassis.
I've seen R.G. Keen's solid-state replacement for Fender Tremolo circuit.
http://www.geofex.com/Article_Folders/s ... remolo.htm
Clearly, a lot of very hard work went into this design, a product Keen's innumerable talents.
I decided against implementing this for a few reasons, complexity, parts count, cost and inertia on my part. Thus a project undone for a while now.
I happened upon an unexpected inspiration from Peavey, the mid-1990's Transtube circuit, patented by Sondemeyer et-al. The circuit is the Peavy Transtube 212. It is very curious in the use of multiple transistor preamp stages, implemented as discrete transistor pairs in an unusual common-emitter amplifier configuration. Only after reading the patent declaration, did I grasp what they were thinking.
patent link:
http://redirect.viglink.com/?key=71fe21 ... 2752B1.pdf
Schematic link at item #62:
http://redirect.viglink.com/?key=71fe21 ... matics.htm
I was curious enough about this topology to breadboard out an example of the stage to play with. It turns out to be a really interesting circuit. The darlington pair presents a high gain, high input impedance amplifier, which mimics those characteristics of a tube amplification stage.
Getting back to the tremolo idea, the high impedance feature is important, if you want to retain the same LFO range and the same 3M reverse audio taper pot used in most blackpanel and later Fender amp circuits. This circuit also functions on a relatively high Vcc, of about 30v and doesn't require much current to function. This supply level is easily derived from the "C" winding (bias supply) tap of the standard Fender style power transformer.
The presented circuit oscillates reliably in the range of 2.5hz - 9hz, using a 3M-RA pot to control it. The waveform output is sinusoidal and with a Vcc of 30 volts, produces approximately 27 volts peak-to-peak output. It's also small enough to easily render on a small circuit board. It isn't particularly picky about transistors. I built it out of "junk box" small NPN. I find it works best with moderate gain (hfe ~250) small signal transistors, but substituting random matched and unmatched transistor pairs works fine with no difficulty. It is pretty robust in other words.
Now, we're in business!
With this level of output, configuring the oscillator with a suitable output stage, it could drive an LED, or the existing neon bulb in the tremolo opto-coupler, or coupled to the bias voltage to implement a bias-wiggle tremolo.
My plan is to implement this on a small circuit board, to replace the usual bias-supply tag board. It would supply the negative bias supply normally produced on the tag board, along with the complete transistorized oscillator circuit and whatever output stage is desired.
I think a tremolo roach with a red LED in it would be ideal, but a small open-drain MOSFET of suitable voltage can easily drive the existing neon bulb in the opto-coupler. Similarly, an emitter follower or some such should work well for bias modulation, as found in earlier Fender amps.
I haven't worked out the footswitch control method entirely, but connecting/unconnecting ground from the legs of the phase-shift network does have the desired function. I think I'd like something cleaner than that.
The goofy thing is that I can't seem to get ngspice (in KiCad) to properly simulate the circuit. The simulation doesn't at all match the breadboard. Probably something dumb on my part, but I'm not a SPICE guru.
Here's the basic circuit:
