Green Ringer Q2 Signal Splitting

[Rayman]

Could the PNP Q2 that splits the signal in the Green Ringer be replaced with an opamp?

[Mark Hammer]

Not a single one.  Q2 provides two equal-amplitude opposite-phase outputs at the emitter and collector.  When people DO use op-amps to derive two such signals they do it in a manner similar to what Tim Escobedo does in his Octup circuit.

You can see here that one output is provided by the first op-amp and a second equal-amplitude opposite-phase output is provided by feeding the first stage to a unity-gain inverting op-amp.  That is clearly more complicated that simply subbing ONE opamp for Q2.

[Rayman]

Could we just have two opamps in parrallel, on inverting, the other non-inverting, then combine with a third opamp?  Then rectify so the bottom half of signal is cut off?

[Rayman]

Mark,

is there a good explanation of the Green Ringer like a "technology of the..."?  It seems like a bare bones circuit for an octave, I'm particularily interested in the Q2 section.

Ray

[Mark Hammer]

What there needs to be is a generic explanation of a bunch of similar octave-up units: Superfuzz, Fender Blender, Foxx Tone machine, and a bunch of others I'm too tired to remind myself about.  What you will see in common with all of these is a transistor somewhere with equal-value emitter and collector resistors and an output from both the collector and emitter of that transistor.  That transistor is a "phase-splitter" that provides two equal-amplitude, but opposite-phase versions of the signal.  Those two signals form the raw material of the octaving.

[R.G.]

Could we just have two opamps in parrallel, on inverting, the other non-inverting, then combine with a third opamp?  Then rectify so the bottom half of signal is cut off?

No. The problem is that if you combine the two opamps, the out-of-phase voltages add to zero. You must take advantage of the out of phase signals to select one half when it's positive (or negative) and then select the other half when it finally goes positive (negative). Otherwise you get nothing by adding them together. It's this selection of only the momentarily positive-going (or negative-going) one of two different signals that is the essence of full wave rectification.

You could replace the diodes with an active switch. Have a circuit which detects when the output of one of the phases was positive (negative) and turn on its switch for the time it's positive (negative), the same for the other phase. The sum of those is then full wave rectified. This is, by the way, what diodes do.

The only difficulty with diodes, and the reason for the plethora of octaving circuits, is that we want to rectify very small signals with them, smaller than the necessary forward drop to turn on a silicon diode. The Green Ringer and the Fox Tone Machine (and its clone/copies) use two ordinary diodes, but bias them with a bare trickle of current so they are just at the edge of conduction and only a tiny signal is needed to turn them on to pass their appointed phase. The Blender is similar. The Univox Super Fuzz uses the base-emitter of bipolar transistors in a similar manner, but adds the quirk that the rectified signal is amplified at the same time. Even the Tychobrahe Octavia does much the same thing; it transforms the signal to a much higher voltage level so it's rectifying diodes are presented a signal with a larger voltage, making the diode voltage much less significant.

The keynote of all of these is a phase splitter to make first one half, then the other half of the incoming signal be the one which causes the rectifiers to conduct, and then two rectifiers, with some means to make the forward drop of the rectifiers be less significant.

It is possible to do this same thing with opamps and rectifiers. If you use opamps to enclose the rectifying diodes inside the feedback loop, the diode forward drop is reduced by the open loop gain of the opamp, and you get a semi-ideal diode. A phase splitter and two of these makes a very accurate FWR. Even more significantly, there is an opamp FWR circuit using only two opamps and one active diode that does the same job as a phase splitter and FWR with only the two amps. For some reason, possibly unfamiliarity, this circuit never made it into the early octave up effects, and now that it's well known, everybody is too busy building clones of the oldest ones.

[Rayman]

Thanks Mark and RG.

"Even more significantly, there is an opamp FWR circuit using only two opamps and one active diode that does the same job as a phase splitter and FWR with only the two amps. For some reason, possibly unfamiliarity, this circuit never made it into the early octave up effects, and now that it's well known, everybody is too busy building clones of the oldest ones."

Does anyone have a schematic of the opamp design RG mentions above?

[Mark Hammer]

Try this: http://www.lynx.bc.ca/~jc/superFullWave.html

[R.G.]

I was going to suggest John Hollis' Omnidrive, but it's the same circuit, and has been around since the late 1960s that I know of, probably before that in the Philbrick tube-based opamps.

[Rayman]

Thanks again Mark and RG. 

I've been searching the threads and came across the MOS Doubler.  Is that a dead issue now?  From what I can tell in the threads RG ended up with the Mu Doubler.  Just for my info, on the MOS Doubler, can't you just ground the sections of the CMOS not used like a multistage opamp?  That looked like a sweet design.

Ray

[R.G.]

I've been searching the threads and came across the MOS Doubler.  Is that a dead issue now?  From what I can tell in the threads RG ended up with the Mu Doubler.  Just for my info, on the MOS Doubler, can't you just ground the sections of the CMOS not used like a multistage opamp?  That looked like a sweet design.

Thanks, Ray, I think it is a good design, as well as unique as far as I can tell. The MOS Doubler has two issues, the biggest one being reproduceability - the degree to which one works just like the others. The other is the availability of matched MOSFET pairs; this is really a variety of the first issue.

What causes this is the way the Doubler circuits work. The circuit function is specifically to have two devices amplify a signal, but out of phase. Then the two outputs are added. The out-of-phase-ness makes the main signal cancel. But second order (and fourth, sixth, etc.) distortion products are always in phase, so they add together. The output then theoretically is a signal that's the square of the input signal, but only two times the distortion ratio of the devices big. MOSFETs without distortion give about 1-2% distortion, so the output signal is about 2-4% as big as the original signal fed to the doubler pair, which may have been amplified from the input signal. Generally you have to amplify the Doubler signal more.

The reproduceability issues lie in the fact that the two Doubler transistors need to be within about 1% of matched to get this to come out right. That's either a selection process, a tweaking process, or buying monolithic transistors. I thought I had it with using the CD4007, but it turns out that some makers' versions of the CD4007 are different from others, and Murhpy's Law being what it is, the unsuitable makers' versions were the only ones the beginners here bought. 

Actually, there is some progress there. Advanced Linear Devices (ALD) now offers matched monolithic pairs of MOSFETs  at Mouser. These should be a great device for this. You'll still have to amplify up the result.

Some people have been successful with this and have written to me about liking it very much.

[Rayman]

RG,

I'm going to try to use the max1044 from the article on your site to provide +12/-12v to the AD633 and set it up as a frequency doubler.  Just experiement with it a little.