The idea for this project came from the fabled middle section of the Serge wave multipliers. At the time I designed this, I had never seen or heard one of these units, nor had I seen its schematic diagrams, but armed with descriptions and suppositions by various people who had seen them, and a couple of photos of CRO traces from the output of the module, I decided to design myself one.

The result as displayed on the CRO is very close to those of the Serge, with the exception of the final fold, where, in my design, the wave maintains more of its original shape. It can produce the most amazing, harmonically rich, filter-like sweeps.

It took me weeks of experimentation before I came up with the final design. I tried various configurations (usually mixing the outputs of successive rectifiers) as suggested on websites devoted to the original, and while I did have success, there were elements of the designs that I felt were poor.

It wasn't until I approached the problem from a different angle that I was happy with the result. The circuit I came up with I would describe as a "reflector", and is ingeniously simple.

The day after I released the first version of this module, I was emailed the schematic of the Serge version, and apart from the value of the resistors, the "folder" was the same "reflector" I had designed. There were some other differences, such as in how I drove the VCA, and how the final fold was dealt with. Initially I decided to release a four-stage unit, having experimented with up to 7 stages. This version re-instates two of the folding stages.

In addition to this multiplier, there are two more simple multipliers, one created by adding lag to the feedback path of an op-amp, the other being the "Nonselective Frequency Tripler" by R. Lockhart. Its functionality is not unlike Moog's single transistor sawtooth to triangle wave converter. It's intended purpose is to convert a signal into another of three times the frequency. To do this it expects a +/-1.2 volt triangle or sine wave, and outputs a +/-0.4 volt complex waveform. Unfortunately, due to the uneven spacing of the frequency multiplied waveform, it does not sound like it is a fifth above the incoming signal. This of course in of no particular concern in this application, as the purpose of this module is to create a complex harmonic structure from a simple input, not to triple the frequency. See the Simple Wave Folder for further details on this circuit.

Input range of signals is +/- 5V. Greater signals will not damage it, but may cause distortion.

Changes in this version (VER2.0):

Two extra folding stages have been added to the main multiplier. The prototyping area has been modified to allow easy substitution of an LM13600/LM13700/NE5517N for the LM3080/CA3080, which is now obsolete. TL07x series chips have been substituted for the LM324. Resistors affecting levels have been marked. In addition to this, there are some suggested modifications to improve it further, though it will work without the modifications.

How to use this module:

For the primary multiplier, the "folder", connect the input to the triangle wave output of a VCO. Connect a LFO, envelope generator or even a DC voltage to the folds input. The result will be a harmonically rich signal at the "F out" output. A second input based on a lag circuit allows square waves and other hard-edged waveforms to be used as the signal source.

There is also a "squared" output available, with "pwm" inputs to further vary the possibilities.

To use the second multiplier, the "grinder", feed the input from the output of a VCO. Adjusting the "drive" and "lag" pots will give variation to the output signal.

Obviously both multipliers can also be used to mangle control voltages, the outputs from LFOs etc.

A little on how it works:

Click here for the schematic.

The "grinder" is simply an inverting amplifier with a lag circuit in its feedback path. Depending on the time constant of the lag pot and its associated capacitor, the op-amp will take longer to settle. In trying to maintain the virtual ground at pin 2 of the first op-amp the output will overshoot, then as the virtual ground settles, then passes the optimal point, the output will again try to compensate, with the same results. The result is a "ringing" that is imposed on the input waveform around the points it changes direction. This introduces a lot of high frequency hash. The op-amp best suited to this circuit is the LM358. The characteristics of the grinder will differ significantly if another chip type is substituted. You may even find you need to try several LM358 variants and select the best.