Moog Synthesizer Patents

Possibly the most famous electronic music patent ever, this is the guts of the circuitry for the Moog 904A voltage-controlled low-pass filter, the Moog 904B voltage-controlled high-pass filter and the Moog 904C filter coupler.  And all later Moog VCFs are based on the 904A.

This filter is certainly well known for its characteristic sound, but I want to point out some of the breakthrough concepts behind it.

First, voltage-tuning a filter over a wide range is a difficult electrical engineering problem.  How do you do it?  Electrical circuits don't normally work that way.  Bob Moog looked to the emitter resistance of balanced pairs of transistors -- the emitter resistance can be varied exponentially over a wide range and the symmetry of the balanced circuit cancels out most of the control signal and even-order harmonic distortion.

Bob Moog probably noted that a filter for musical applications should have a steep slope (24 dB/oct in this case), should have the resonance adjustable without affecting the tuned frequency (and vice-versa), and that the resonance should be adjustable to oscillation.  The four-single-pole-low-pass-sections-with-feedback approach was not typically used for filter designs at the time, but it addresses these issues perfectly.  (For comparison, the peaking frequency for the traditional 2-pole biquad filter is not the same as the filter's tuned frequency, although the two approach each other toward oscillation.)

And because the filter can be run at or near oscillation, it will likely be used that way patched with 1 volt/octave voltage-controlled oscillators.  So it is important that the tuning of the VCF accurately track the tuning of the VCO.
The transfer functions of the low-pass and high-pass filters are:

Without feedback, all four poles are real and sit on top of each other at x = -ω.  As feedback is applied, the four poles split out, literally in the shape of an "X" and move apart 90^o from each other.  At a feedback factor of 4.0 the two rightmost poles hit the the y axis and the filter oscillates.

Bob Moog's design does all this with a remarkably simple circuit that performs well, and any inaccuracies it does have provide a lot of character to the sound. (The filter can be overdriven in a musically pleasing way as all the transistor stages clip gradually.)

So this is a killer design from many angles.

The 904B high-pass filter is the electrical brother of the low-pass filter, although the schematics look very different from a distance.  The high-pass filter does not have a feedback path and thus no resonance control; perhaps there are stability issues here.   (On the other hand, the Moog AES article below mentions employing a small amount of feedback to sharpen up the response curve.)

The 904C filter coupler is the obvious way of doubling up a low-pass VCF and a high-pass VCF in series for a bandpass filter or in parallel for a band reject filter with individual controls of center frequency and bandwidth.  The advantage here is that the bandwidth control voltage is calibrated in volts/octave, again something you won't see in more traditional filters.

Roland and other companies have produced diode ladder filters using diodes as the tuning elements instead of transistors, presumably as a "way around the patent".  The diodes do not separate the capacitor sections from each other like the transistors do, so the diode-based ladder filters have their poles scattered along the x axis instead of all at the same location.  They sound different, their cutoff slope is much more gradual, and they require a lot more feedback to oscillate.

This is the patent for the Bode Frequency Shifter manufactured by Moog and available as module 1630.  Full schematics are included in the Moog modular service manual.  The patent schematic and the sevice manual schematic are very similiar although the patent has, as expected, far less circuit detail.

The basic idea is an implementation of the trigonometric identity for the cosine of the sum of two angles to shift the frequency up:

cos(x + y) = cos(x) cos(y) - sin(x) sin(y)

(Remember this from high school math?)  And this variation will shift the frequency down:

cos(x - y) = cos(x) cos(y) + sin(x) sin(y)

Bode's approach uses:

• a "Dome Filter", two all-pass networks tuned 90-degrees apart.  The input signal goes through these in parallel, effectively providing cosx and sinx signals (relatively).
• two multipliers.
• a quadrature variable frequency oscillator

The latter item, the quadrature variable frequency oscillator, is tricky.  Bode uses a fixed-frequency quadrature sine wave oscillator running at 20kHz, a voltage controlled oscillator with a range of 15kHz to 25KHz, and with two multipliers beats the oscillators to get a range of -5kHz to +5kHz.  It sounds like a lot of work to go through for a local oscillator, but it does get you quadrature sine waves, over a wide range, and through-zero operation.

It's important to note that this effect is a frequency shift and not the pitch shift effect currently available in modern digital effects boxes.  In a frequency shifter all frequency components of the input signal are offset by a specific number of Hz while in a pitch shifter all frequency components are multiplied by a specific factor.  The former sounds something like a ring modulator while the latter sound something like playing back a tape at a different speed.

This is a linear voltage-controlled triangle wave generator with a potentiometer adjustment of the waveform shape (duty cycle) from sawtooth to triangle to reverse sawtooth.  Also, the circuit has a gating arrangement for single cycles, so it is suited for LFO/envelope generator applications.

This wasn't used in any Moog product, but it was used for the second LFO ("Shaper Y") in the Crumar Spririt synthesizer designed by Bob Moog, Jim Scott and Tom Rhea and released in 1983.

It's a PWM-modulated version of the classic Moog ladder filter!

In the early days of hobbyist electronic music a VCA or VCF was often implemented on the cheap, avoiding transconductance amps and matched transistors, by pulse-width-modulating a CMOS gate at a high frequency to simulate a voltage controlled resistor.  That is what is happening here.

There is an interesting approach in this patent where the PWM source is not a constant frequency but is instead a high-frequency VCO driving a multistage monostable multivibrator to provide constant duration pulses for each of the gates in succession.  This avoids the really thin pulses you would need at the low end of the tuning range, and the PWM oscillator frequency doesn't need to be a very high frequency when the highs are going to be filtered out anyway.  And the multiple monostables separate the control of the simulated voltage controlled resistors a bit so there are fewer funny distortions from them interacting.  And you can use an exponential VCO for the high-frequency oscillator to provide a proper exponential voltage control for the filter.

This was never used in a Moog product, and not surprising as it really isn't their style, so I'm guessing this patent was mostly to keep other manufacturers from building PWM filters.

I wonder what this thing sounds like.  More specifically, I wonder how objectionable the aliasing effects are, especially with more complex program material.

These are the patents for the Moog Taurus Bass Pedals.

The first is a vague block diagram of a synthesizer device that can be built so that each functional block can switch between four preset parameters.  Included in this patent is the schematic for the Taurus' four-way flip-flop (with LEDs) and the circuitry for selecting one of four mixes of two VCOs.

The second patent concerns the circuitry for a synthesizer with linear VCOs and a keyboard that generates an exponential control voltage.  The keyboard uses a special thick-film resistor array for accurate intonation and, in case more than one key is depressed, double-throw switches.  The VCO's trip point and the keyboard tuning voltage are both referenced off the same power supply voltage for tuning accuracy.

The third patent covers the mechanical and operational workings of the Taurus.

There is apparently some prior art for ribbon controllers built from a conductive ribbon and a resistive element.  This patent is for a more rugged variation and features a resistive element this is constructed with resistance wired wrapped around a nonconductive bar and sunk into a groove in the base piece.

These three patents are closely related.  And they were all filed on the same day.

The first patent is for a keyboard with a pressure sensor under each key.  Each pressure sensor works by creating capacitance between a flexible conductive sheet and a curved prong under each key.  The capacitance/force curve can be modified by changing the shape of the curved prong.

The second patent is for a keyboard scanner, multiplexer and demultiplexer for a keyboard with individual capacitive pressure sensor similiar to the above.  A ring oscillator is built with one stage per key, and each stage consists of one transistor, one resistor and the capacitive sensor.  The pulse width of each stage is then dependent on the key pressure of that stage.

The decoder is more complex, and either there's something wrong in the explanation here, or I'm missing something, but it doesn't seem workable.

The third patent is an analog demultiplexer built from a shift register and a number of sample-and-holds.  I would think this would be too obvious to patent.  It may have been holdover from the Chicago Musical Instrument Co.

This is a chorus circuit similar to the one used in the ARP Omni.  This one came first.

While the ARP chorus uses three delay lines driven from three separate LFOs, this one uses three delay lines driven from a mix of three phases of one LFO and three phases of second LFO.  ("Three" is the example number; neither patent restricts the number of delay lines.)

This is the patent for the MinitMoog and Moog Satellite preset synths. These are something like the Arp Soloist; 13 preset sounds where each preset controls 12 parameters through a resistor matrix.  1 LFO, 1 VCO, 1 BP VCF, 1 LP VCF, 1 VCA.  (The MinitMoog has a second VCO, literally piggybacked on the circuit board.)

These are the patents for the PolyMoog, a fully polyphonic synthesizer with a 71-note keyboard.

The pitches come from top octave generator ICs and flip-flop dividers, and there is a custom 16-pin chip for each key that performs various envelope, modulation, waveform mixing and 2-pole ladder VCF duties.  The chip's envelope generator is sensitive to the key velocity and also to the length of time since the key was last played.

Besides the 71 custom chips there is also a lot of additional control and filter circuitry.  And no microprocessors.

The first patent describes the custom keyer chip.  The biggest problem with a fully polyphonic synthesizer is the amount of circuitry required and the biggest problem with a custom chip is being able to manufacture it in sufficient volume to make it worth the development expense.  So packing all the circuitry for a single key onto a single custom chip is a reasonable approach.

The keyer chip circuit is very simple by today's standards; maybe 25 transistors total.  It takes two sawtooth inputs and performs clipping modulation on one and pulse-width modulation on the other, pans between them, handles the envelopes, and implements a two-pole ladder filter.

The second patent describes a chorus system based on top-octave generator chips.  The major problem with TOG-based instruments is that the chromatic pitch outputs are unnaturally locked in sync.  In the PolyMoog there are two signal sources based on two separate TOG chips.  The first TOG is clocked at pitch but the second TOG is clocked one semitone sharp with its outputs are shifted down one semitone to compensate.  Thus the pitch errors of the chromatic outputs of the two chips will be different.  Very clever.  Phase-locked loop circuits for driving the TOG chips are described with chorus options including variable phase, variable frequency and octave modes.

The third patent is nearly identical to the second, and provides some additional claims.

The first patent also describes the mechanics of a keyboard that uses two magnets, one mounted on the key and one on the chassis, aligned so as to provide the keyboard with a "breakaway" feel.

This is a block-level design for a phase shifter effects device with these features:

• Stereo outputs, phase+in and phase-in.
• A third output that can be set with a 360^o four-tap pot panning between
  
  input
  input + phase shifted
  phase shifted alone
  phase shifted - input
• A voltage controlled LFO; a footswitch selects between a slow speed (0.1 Hz) and a fast speed (unspecified), and the risetime here (0.5 second) is faster than the fall time (2 to 4 seconds) to simulate the mechanical inertia of a Leslie speaker motor.
• The LFO control voltage is subtracted from the LFO triangle waveform before driving the phase shift stages, so the overall tuning of the phase shift network sinks down for higher LFO rates.

This patent does not really apply to the Moog 12-Stage Phase Shifter.  While the 12-Stage does have stereo outputs, they are in+phase and in-phase.  And while the 12-Stage does have motor speed up and slow down times, they are both the same, and the 12-Stage fades in the degree of phase shifting with motor speed up instead of tuning the phase shift stages.

Mix the sum of six resonant filters in with a guitar signal to "minimize aural fatigue". That's what it says in the patent, but I'll suggest that this circuit adds musical resonances.  The filters are at frequencies of 700, 960, 1333, 1839, 2539 and 3500 Hz.

This circuit was used on the Norlin/Moog Lab Series guitar amps models L5, L7, L9 and L11.  At least according to the service manual; I've never heard or even seen these amps.  Interestingly the amps used frequencies of 1000, 1370, 1900, 2630, 3630 and 5000 Hz (again according to the service manual).  In both cases there is a constant ratio between adjacent frequencies of the fifth root of 5.

Here is a plot from the Technical Service Manual for the Moog Lab Series Amplifiers (Norlin Music).

A vocoder has two input sources.  A set of filters and amplitude detectors measure the spectrum of the first source, and a second set of filters, each with a VCA, apply that spectrum to the second source.  Typically you would connect a microphone to the first input and a synthesizer to the second input, and it will make your synth talk somewhat convincingly.

This is the patent for the Moog 16-channel Vocoder.

This is the patent for the circuitry in the Moog 304A Three Band Parameteric Equalizer.  Each band has controls for frequency, width (Q) and height.

Bob Moog puts a guitar signal through a compressor, then through a CA3080 OTA.
This gives you a selection of:

1. soft odd-order distortion from the OTA
2. hard odd-order distortion from severely overdriving the OTA
3. soft even-order distortion from running some of the signal into the bias pin of the OTA, and
4. hard even-order distortion off the full-wave rectifier in the compressor

Distortion circuits operating in modes 1 and 3 were used in some of the Gibson Lab series amplifiers.

Take a guitar with two pickups, put the neck pickup through a compressor and the bridge pickup through an expander.

This circuit was used on the Gibson RD Artist guitar, an instrument that proved to be unpopular most likely because it was shaped like a melted candy bar.

There is a schematic of the Gibson RD Artist guitar on Leper's Schematic Archive, but since that version is a little inconvenient to view (it's a zip of a PostScript file) I've made a more accessable copy here.

This keyboard pressure sensor is a piece of deformable material with conductive layers above and below it.  The sensor is placed underneath the keyboard so that pressure on a key changes the capacitance between the conductive layers.  This capacitance is measured by detecting the amplitude of a high frequency signal.  Also the base capacitance (for all keys up) is noted by a sample-and-hold circuit and subtracted from the measured capacitance.

Used on the MinitMoog.

Some processing to detect the positive and negative peaks of an input signal, a period-to-voltage converter and some processing to ignore away far-off readings.

Each peak detector consists of three "peak pickers" in a row, each with a differentiator on its output.  (I have no idea how that is supposed to work.)

The details of the period-to-voltage converter aren't given.  (Why a period-to-voltage converter instead of a frequency-to-voltage converter?)

There are a three sample-and-hold units after the converter; the first holds the converter output (it could even be considered part of the converter), and the second and third hold the output of the first.  The third S/H provides our control voltage output, but it is only gated when the outputs of the first and second S/H's are close.  This means that individual far-off samples will be thrown out, but a sudden change in frequency can be tracked within one additional cycle.

Address the lack of polyphony in a basic VCO/VCF/VCA monophonic synthesizer by adding a bare bones top-octave-generator single waveform organ circuit, and mix the VCO and the organ signals at the input to the VCF.  You'll need a two bus keyboard; one bus for the pitch control voltage and one bus for the organ signal.  And the presence of the organ signal is used to trigger the envelope generator.

Used on the Moog Liberation and the Moog / Radio Shack Concertmate MG-1.

An improvement on several of Mr. Pepper's previous patents for two-dimensional touch sensors.

Might this be related to the touch panel on the Moog Voyager?