BogaudioModules

README.md (127360B)

---

 # BogaudioModules
 
 ![build badge](https://img.shields.io/github/actions/workflow/status/bogaudio/BogaudioModules/buildDevelop.yml?label=build) ![version badge](https://img.shields.io/github/v/release/bogaudio/BogaudioModules)
 
 Modules for [VCV Rack](https://github.com/VCVRack/Rack), an open-source Eurorack-style virtual modular synthesizer:
 
   - [Oscillators](#oscillators)
   - [LFOs](#lfos)
   - [Filters](#filters)
   - [Envelopes and Envelope Utilities](#envelopes)
   - [Mixers, Panners and VCAs](#mixers)
   - [VCAs and Dynamics](#dynamics)
   - [Noise/Random, Sample and Hold](#random)
   - [Sequential Switches and Sequencers](#sequencers)
   - [Visualizers](#visualizers)
   - [Polyphony Utilities](#poly)
   - [Pitch CV Utilities](#pitch)
   - [Other Utilities](#utilities)
   - [Miscellaneous](#misc)
 
 Most Bogaudio modules support VCV Rack's polyphony feature; see <a href="#polyphony">Note on Polyphony</a>.
 
 Some modules have expanders; see <a href="#expanders">Note on Expanders</a>.
 
 Many modules support bypassing in a non-default way; see <a href="#bypassing">Note on Bypassing</a>.
 
 ![modules screenshot 1](./doc/www/modules1.png)
 
 ![modules screenshot 2](./doc/www/modules2.png)
 
 ![modules screenshot 3](./doc/www/modules3.png)
 
 ![modules screenshot 4](./doc/www/modules4.png)
 
 ![modules screenshot 5](./doc/www/modules5.png)
 
 The modules have two <a href="#alternate_panels">alternate panel designs</a>, "Dark":
 
 ![Dark alternate panels screenshot](./doc/www/skin-dark.png)
 
 And "Dark (low-contrast)":
 
 ![Dark (low-contrast) modules screenshot](./doc/www/skin-lowcontrast.png)
 
 
 ## Builds/Releases
 
 Mac, Linux and Windows builds of the latest version are available through the [VCV Rack Library](https://library.vcvrack.com).  Find release notes on the [releases page](https://github.com/bogaudio/BogaudioModules/releases).
 
 
 ## Building
 
 You'll need to be set up to build [VCV Rack](https://github.com/VCVRack/Rack) itself.  Under the Rack build directory, switch to `plugins/`, and then:
 
   ```
   git clone https://github.com/bogaudio/BogaudioModules.git
   cd BogaudioModules
   make
   ```
 
 The master branch of this module currently builds against Rack 2.0.x.
 
 
 ## Modules
 
 ### <a name="oscillators"></a> Oscillators
 
 ![Oscillators screenshot 1](doc/www/oscillators1.png)
 
 #### <a name="vco"></a> VCO
 
 A standard VCO featuring:
   - Simultaneous square, saw, triangle and sine wave outputs.
   - Traditional exponential and linear through-zero FM.
   - Pulse width modulation of the square wave.
   - Hard sync.
   - Slow (LFO) mode.
   - Antialiasing by a CPU-efficient combination of band limiting and oversampling.
 
 The main frequency knob is calibrated in volts, from -4 to +6, corresponding to notes from C0 to C6.  The default "0V" position corresponds to C4 (261.63HZ).  Any pitch CV input at the V/OCT port is added to the knob value to determine the oscillator frequency.  With CV input, the pitch can be driven as high as 95% of the Nyquist frequency (so, over 20KHZ at Rack's default sample rate).  The FINE knob allows an additional adjustment of up to +/-1 semitone (100 cents, 1/12 volt).  In slow mode, the output frequency is 7 octaves lower than in normal mode with the same knob/CV values.
 
 In linear mode, the frequency 1000HZ times the pitch voltage (as determined by the knob plus V/OCT CV) -- at 0V, the frequency is zero, and the oscillator stops.  In slow mode, it tracks at 1HZ times the pitch voltage.  Negative voltages will realize the same output frequency as the corresponding positive voltage (the oscillator runs backwards).  Use with with an FM input to create strange waveforms.
 
 The context menu option "DC offset correction", on by default, removes DC offset from the outputs.  Presently, this only affects the square output when the pulse width is set to anything besides 50%.  When this is enabled, and viewing the output on a scope, the waveform will move up and down relative to 0V as the pulse width is changed -- this is the DC offset removal in action.  When disabled, the waveform stays centered on 0V, which is useful if using the output as CV.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.  The poly port can be changed to the FM input on the context menu.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="lvco"></a> LVCO
 
 A 3HP subset of VCO, designed as a compact general-purpose oscillator.  The waveform is selectable between sine, triangle, saw, square and 25% and 10% duty-cycle pulses.  FM and linear modes are selectable on the context menu.
 
 Context menu option "Reset phase on wave change", if enabled, causes the waveform phase to be set to zero when the waveform is changed.  By default the continues to advance from wherever it was.
 
 _Polyphony:_ Same as VCO.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="sine"></a> SINE
 
 A 3HP subset of VCO, designed in particular for use making synth drums.  The waveform defaults to sine but is selectable on the context menu, with the same options as LVCO, with the addition of a ramp (inverse saw) wave.
 
 Additionally, there is a phase control with CV borrowed from XCO (if CV is used, the input is attenuverted by the PHASE knob).  This sets the initial position of the wave when the module is synced (if you're not using sync, changing the phase won't meaningfully alter the output).  This can be used to alter the harmonic content of sync sounds, and for transient shaping for drum synthesis.
 
 _Polyphony:_ Same as VCO.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="pulse"></a> PULSE
 
 A 3HP subset of VCO, oriented toward pulse-width modulation.  The only output waveform is square/pulse, and there is no FM.  However, the PWM CV input has a dedicated attenuverter, and unlike VCO, the result CV value is summed with the PW knob position.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="xco"></a> XCO
 
 Includes all the features of VCO, adding:
   - An onboard wave mixer with output at the MIX port.
   - For each wave type:
     - A wave modifier (pulse width for square; saturation for saw; a sample-and-hold/step-function effect for triangle; FM feedback for sine).
     - A phase knob/CV controlling the phase of the wave in the mix.
     - A mix knob/CV to control the level of the wave in the mix (waves are output at full level at their individual outputs).  The mix knob/CV responses are linear in amplitude.
   - A CV input for FM depth.
 
 The context menu option "DC offset correction" works as documented on <a href="#vco">VCO</a>.
 
 The context menu option "Mix output processing" sets how the mix output is scaled or clipped.  VCV Rack modules generally [should not output signals exceeding +/-12 volts](https://vcvrack.com/manual/VoltageStandards#Output-Saturation), and oscillators typically output +/-5 volts.  However, XCO's mix output combines four oscillators, and can easily exceed these values.  "Mix output processing" sets how to deal with this:
   - "Scaled to +/-5V" (default): the output is directly scaled down to +/-5V (like a regular oscillator) when it would otherwise exceed that range (it will not scale up; it is possible to reduce the output below this range with the MIX controls).  The scaling is updated once per oscillator cycle, and abrupt parameter changes can confuse it; in this case hard clipping takes over.
   - "Saturated": a saturator (soft clipper) keeps the output within +/-12V.
   - "Hard clipped": the output is simply clipped at +/-12V.
   - "None": no scaling or limiting is applied to the output, ignoring the voltage standards.  This was how XCO always behaved prior to version 2.*.41.  Older patches should restore this mode if affected by the addition of these options.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="additator"></a> ADDITATOR
 
 ![Oscillators screenshot 2](doc/www/oscillators2.png)
 
 An additive oscillator, where the output is the sum of up to 100 individual sine/cosine waves (partials).  Various parameter knobs/CVs allow control over the number, frequencies and amplitudes of the partials:
   - PARTIALS: sets the partial count.
   - WIDTH: sets the spacing of partials in frequency; at the default position each successive partial is pitched an octave higher than the one before.
   - O-SKEW: adjusts the spacing of odd-numbered partials up or down relative to WIDTH.
   - E-SKEW: adjusts the spacing of even-numbered partials up or down relative to WIDTH.
   - GAIN: Sets the level of the output by adjusting an internal amplitude normalization parameter.
   - DECAY: applies a positive or negative tilt to the amplitude decay of the partials; at the default position, amplitudes decrease proportionally with increasing frequency.
   - BALANCE: cuts the amplitudes of the odd or even partials.
   - FILTER: manipulates the partial amplitudes to simulate low or high-pass filter effects.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="fmop"></a> FM-OP
 
 A sine-wave oscillator and simple synth voice designed to allow patching up the classic FM algorithms (using multiple instances).  Features:
   - Linear through-zero FM response.
   - CV-controllable FM depth.
   - CV-controllable FM self-feedback.
   - CV-controllable output level.  The LEVEL knob and CV have a linear-in-decibels (exponential in amplitude) response; a context-menu setting makes this linear in amplitude.
   - An on-board ADSR, controlled by the GATE input, with selectable routing to output level, feedback and depth, with CV control over the sustain level.
   - A main frequency knob calibrated for setting the frequency as a ratio of the frequency dictated by the V/OCT input - assuming a single V/OCT CV is routed to multiple FM-OPs, this allows the relative frequency of each operator to be set via ratios.
 
 Anti-aliasing techniques are applied when feedback or external FM are in use.  Either condition for anti-aliasing can be disabled on the context menu.  Prior to version 1.1.36, due to a long-standing bug, there was no anti-aliasing for external FM, unless feedback was also on.  To get that behavior back, **the true vintage FM-OP sound**, disable "Anti-alias external FM" on the menu.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="chirp"></a> CHIRP
 
 Part VCO and part sound effect, CHIRP produces swept sine waves.  On each cycle, the output sine wave sweeps in frequency from FREQ1 to FREQ2, over the course of TIME.  If LOOP is enabled, the cycle repeats, otherwise it may be triggered manually with the TRIGGER button, or by CV at the TRIG input.  A pulse is emitted at output EOC each time a cycle ends.
 
 TIME takes a unipolar CV (0-10V), which is attenuated by the knob if in use.  FREQ1 and FREQ2 take bipolar (+/-5V) 1V/octave inputs at ports V/O1 and V/O2, which CVs are added to the corresponding knobs.
 
 The frequency sweep can be linear in time, or exponential, under the control of the EXP toggle.
 
 The sweep is upwards in frequency if FREQ1 is less than FREQ2, and downwards otherwise.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/O1 input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="lfos"></a> LFOs
 
 ![LFOs screenshot](doc/www/lfos.png)
 
 #### <a name="lfo"></a> LFO
 
 A standard LFO featuring:
   - Simultaneous ramp-down, ramp-up (saw), triangle, stepped random, square and sine wave outputs.
   - Knob and CV control of the pulse width of the square wave.
   - A CV-controllable "sample" modifier, which turns the output into a step function.
   - Onboard CV-controllable offset and scale of the output voltages.
   - CV-controllable output smoothing (the CV input is shared with offset, see below).
   - Reset (hard sync) input.
   - Slow mode.
 
 LFO tracks pitch CVs at the V/OCT input seven octaves lower than a normal oscillator: with a 0V input, the output frequency is note C-3 (2.04HZ).  The frequency knob is calibrated in linear volts (the small ticks), and its value is added to the input V/OCT.  With no input, the frequency range is from approximately 0.1 to 400HZ; with CV the frequency may be driven up to 2000HZ or down to arbitrarily low values.  In slow mode, the output frequency tracks the controls four octaves lower than otherwise (11 octaves below a normal oscillator).  Output is `(lfo * scale / 100) + offset`.
 
 The stepped random output selects a new random value in the range +/-5V once each cycle, each time the oscillator phase crosses 0.  Triggering RESET will select a new value.
 
 The sampling feature is not used with the square and stepped outputs, but applies to the others.
 
 By default OFFSET varies from -5 to +5V.  The "Offset range" context-menu item allows this range to be set to +/-10V.  Note that the output is clipped to +/-12V.  SCALE is applied to the output before the offset is added.
 
 Output smoothing is applied to the signal last, after offset and scale.  Smoothing is implemented with a slew limiter (see <a href="#slew">SLEW</a>), where the rise/fall times are a function of both the LFO rate and the smoothing amount.  The effect of smoothing varies radically with the amount, the wave selected, and with use of sampling and pulse width.  Some examples:
   - Smoothing will turn the stepped random "wave" into a random walk (similar to, but distinct from, the output of <a href="#walk">WALK</a>).
   - The triangle output is unaffected by smoothing even at its maximum (unless sampling is turned up).
   - Square waves can be rounded off to shark-fins.
   - Asymmetric waves (saw, ramp, pulse) will seem to get a positive or negative offset with increased smoothing (which can be countered with the offset setting); sampling reduces this effect with saw and ramp.
 
 To save space, offset and smoothing share a CV input port.  By default this will route CV to offset.  A context-menu option allows the CV to be routed to smoothing instead.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="llfo"></a> LLFO
 
 A 3HP LFO, with selectable waveform.  The features are a subset of LFO.  There are separate square and pulse waveform options: square is fixed at a 50% duty cycle, while pulse defaults to 10% but may be adjusted with the "Pulse width" option on the context menu.
 
 Sampling and smoothing functions are available on the context menu.
 
 Context menu option "Reset phase on wave change", if enabled, causes the waveform phase to be set to zero when the waveform is changed.  By default the continues to advance from wherever it was.  In either case, the output will typically jump to a new value, which may cause clicks or other undesirable effects depending on how the output is used; adding a bit of smoothing may help.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="fourfo"></a> 4FO
 
 A quadrature LFO, with outputs at 4 different phases.  The phases may be set by knobs and CVs (marked PHS); by default they are 0, 90, 180 and 270 degrees from the fundamental.  Otherwise, functionality is the same as with LFO, except that:
   - The wave shape is selectable, and all four outputs are of the same (phase-shifted) wave.
   - The sampling and pulse width knobs and CVs are combined, with their function depending on the selected wave.
 
 Note that with the stepped random output, each output will update its output when its phase crosses 0 degrees.  Each draws from the same random sequence, rather than separate ones -- the outputs only vary in phase.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="eightfo"></a> 8FO
 
 An "octature" LFO, like 4FO, but with outputs at 8 phases.  By default the phases are 0, 45, 90, etc, degrees from the fundamental.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="filters"></a> Filters
 
 ![Filters screenshot](doc/www/filters.png)
 
 #### <a name="vcf"></a> VCF
 
 A general-purpose filter with a selectable lowpass, highpass, bandpass or bandreject (notch) output.  Being based on pure DSP theory, rather than a model of an analog filter design, it has a transparent character (which to say, no "character" at all, or not much -- which character is often very nice).  However, it achieves interesting features such as:
   - A slope (or transition rate at the cutoff) that can be smoothly modulated from very shallow (1 pole) to very sharp (12 poles).
   - Smoothly modulatable bandwidth in bandpass and bandreject modes.
   - Very accurate V/OCT tracking.
 
 The large knob sets the filter's base cutoff frequency (for lowpass and highpass modes) or center frequency (for bandpass and bandreject).  There are three CV inputs that affect the cutoff, in this order:
   - CV: a general, linear input expecting bipolar +/-5V signals. The input is attenuverted by the CV knob.
   - V/OCT: an input here is interpreted as a pitch CV; if connected, the input voltage is interpreted as a frequency and added to the cutoff.  Use this and set the main knob to 0HZ for accurate key tracking.
   - FM: an exponential FM input; the input here is attenuated by the FM knob.  It is implemented by converting the cutoff to a V/Octave pitch CV, adding the attenuated FM signal to that, and converting back to frequency.
 
 The RES/BW knob has two functions, depending on mode:
   - In lowpass and highpass modes, it controls the resonance of the filter at the cutoff frequency.  The filter does not self-resonate.
   - In bandpass and bandreject modes, it sets the bandwidth of the filter -- the width in HZ around the center frequency of the passband (bandpass mode) or stopband (bandreject mode).  Note that at the minimum setting, the bandwidth can be quite small, even inaudible, depending on the center frequency (see below).
 
 The R/BW CV input expects a unipolar 0-10V input; when in use this input is attenuated by the RES/BW knob.
 
 SLOPE modulates the rate of transition between the pass and stop bands in each filter mode.  (For example, in lowpass mode, a higher slope means a faster/sharper change, at the cutoff frequency, from frequency passing through the filter to being suppressed by the filter.)  This ranges from a slow (at 1) to fast (at 12) transition.  It may be modulated by the SLP CV input, which expects 0-10V, and which is attenuated by the SLOPE knob.
 
 The context menu option "Bandwidth mode" controls how the bandwidth is calculated for a given center frequency and RES/BW setting:
   - "Pitched" (the default): the upper and lower frequencies of the band are equally distant from the center frequency in pitch (octaves and semitones), with a maximum of 2 octaves.
   - "Linear": the upper and lower frequencies are equally distant in HZ from the center frequency, with a maximum of 2000HZ.
 
 Note: due to limitations in the filter's implementation, it has a couple workarounds in place to avoid problems:
   - The cutoff frequency uses a slew limiter (it has limit on how fast it can change), such that it takes approximately 100ms to move the cutoff from fully closed to fully open (or the opposite).
   - There is a fixed two-pole highpass filter on the filter output, at a cutoff of 80hz.
   - While the module's frequency knob goes to zero, the filter's cutoff won't actually go below 3hz.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="lvcf"></a> LVCF
 
 A compact version of VCF.  The filter slope may be set on the context menu.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="ffb"></a> FFB
 
 A fixed filter bank comprised of 12 bandpass filters, with low- and high-pass filters on each end.  The band frequencies are those used on the classic [Moog 914](https://modularsynthesis.com/moog/914/914.htm).
 
 Each knobs sets the attenuation of the output of its corresponding filter, down to -60db, before those outputs are mixed back together and sent to the outputs.
 
 The FREQ knob adjusts the center frequency of each band up to an octave in either direction.  It takes a bipolar (+/-5V) CV at FCV, which is attenuverted by the knob if in use.
 
 There are three outputs:
   - ALL: a mix of the outputs of all the filters.
   - ODD: a mix of the LP, HP, and odd-numbered band filters (125HZ, 250 HZ, etc).
   - EVEN: a mix of the LP, HP, and even-numbered band filters (175HZ, 350 HZ, etc).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to each of ALL, ODD and EVEN.
 
 #### <a name="eq"></a> EQ
 
 A basic low/mid/high three-band equalizer.  Each knob sets the gain of its corresponding filter from -36db, through unity (0db) to +12db.
 
 The cutoff/center frequencies of the three filters are:
   - LOW: 100Hz
   - MID: 350HZ
   - HIGH: 1000HZ
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="eqs"></a> EQS
 
 A stereo version of EQ.  The left and right inputs are processed by separate filters, but the filter parameters are set by the shared three knobs.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the L input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs.
 
 #### <a name="lpg"></a> LPG
 
 ![LPGs screenshot](doc/www/lpgs.png)
 
 LPG is a "low-pass gate", where an envelope generator, low-pass filter (LPF) and amplifier (VCA) are combined into a single sound-shaping unit.  It lends itself to percussive or plucked sounds, though longer notes are possible.
 
 LPG's envelope is a version of <a href="#vish">VISH</a>, with simplified controls.  On LPG, the single RESPONSE control (taking a unipolar 0-10V CV at RESP) sets the length and basic shape of the envelope (turning up RESPONSE is equivalent to turning up each of MIN GATE, RISE time and FALL time, proportionally, on VISH).  Enabling LONG rescales the knob for longer envelopes.  The RISE and FALL knobs set the shapes (curves) of the rising and falling envelope segments, as on VISH.  The envelope is triggered whenever LPG receives a rising edge at the TRIG input.
 
 The signal received at IN is processed by an LPF and VCA in series before being sent to OUT.  Each has a BIAS knob, which sets the base level of the cutoff of the LPF, and level of the VCA, when the envelope is off.  The ENV attenuverters control how the envelope affects the LPF and VCA respectively.  Each bias knob also has a CV input, marked LPF and VCA, expecting unipolar (0-10V) signals.
 
 If LINEAR is enabled, the VCA has a linear-in-amplitude response to its BIAS, CV and the envelope; otherwise it has a linear-in-decibels response.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIG input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="llpg"></a> LLPG
 
 LLPG is a compact, simplified version of <a href="#LPG">LPG</a>, where:
   - There is only one shape knob, which affects both the rise and fall shapes simultaneously.
   - There are no ENV knobs; the envelope applies in full to both the filter and VCA.
   - There are no CV inputs, and no linear VCA mode.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIG input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="megagate"></a> MEGAGATE
 
 MEGAGATE generalizes the idea of a low-pass gate, not least by adding a high-pass filter (HPF).  It is also stereo, optionally velocity-sensitive, and has full CV inputs over the filter and VCA biases and envelope routing.  
 
 The envelope circuit is the complete "vactrol-ish" envelope available independently as <a href="#vish">VISH</a> -- see VISH for a description of the envelope controls and CVs on left side of MEGAGATE, aside from VELO and TILT, which further process the envelope before it is sent on to control the filters and VCA.
 
 The VELO input provides control of the level of the envelope before it is processed by TILT.  It is designed to be used to implement velocity from a MIDI controller, but may be used as a general CV.  It expects a 0-10V input voltage: when this 0, the envelope gain is -6dB (it is cut in half); when VELO is 10V the output is unchanged (a 0dB gain). Intermediate values will set the output gain linearly to values between -6dB and 0dB.
 
 The velocity-zero gain of -6dB can be set to other values on the context menu, to make the velocity response stronger (-12dB, -24dB) or weaker (-3dB).  It can also be set to -60dB, such that a 0V velocity input will completely cut the output -- allowing the input to be used as a full-range, though inverted, CV.
 
 TILT is essentially a panner for envelope, turning it into left and right channel envelopes.  Turning the knob towards L will reduce the gain of the right channel envelope, or vice versa.  TILT accepts a bipolar (+/-5V) CV, which attenuated by the knob when in use.
 
 The left and right signal inputs are processed by the LPF, HPF and VCA sections before being sent to their corresponding outputs.  There are independents sets of filters and VCAs for the left and right channels, each set controlled by the same knobs and CVs, but receiving different envelopes as set by TILT.
 
 By default the LPF and HPF filters process the input (whether left or right) in parallel, and their outputs are mixed together to send to the VCA.  If FILTER SER is enabled, the filters are processed serially, with the LPF's output being fed to the HPF, and HPF to the VCA.
 
 Each of the LPF, HPF and VCA sections has a BIAS knob, which sets the base cutoff (for the filters) or level (for the VCA).  Each bias control accepts a bipolar (+/-5V) CV, and each CV has an attenuverter.  Each CV, subject to its attenuverter, is added to its knob's position.
 
 Each section also has an ENV setting, controlling how much of the envelope is applied to the final cutoff or level setting.  Again, each ENV accepts a bipolar CV, with attenuversion, and the CV is added to the knob.
 
 The final cutoff and level settings are determined by scaling (multiplying) the envelope by the ENV value (including CV), and adding that to the BIAS value (including CV).  In no case can the envelope and CVs drive a parameter outside of its knob range.
 
 If LIN VCA is enabled, the VCA has a linear-in-amplitude response to its BIAS, CV and the evelope; otherwise it has a linear-in-decibels response.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the GATE input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 ![Parameteric EQs screenshot 1](doc/www/parametric_eqs1.png)
 
 #### <a name="peq"></a> PEQ
 
 A three-channel parametric EQ, which is a filter bank where the band frequencies are controllable.  Each channel gets the input from IN and applies a bandpass filter with center frequency set by FREQ, where the filter's output is sent through a VCA controlled by LEVEL.  The outputs of each channel are mixed to OUT.
 
 The first channel may be configured as a lowpass filter, if the LP button is on.  The last channel can be highpass.  (And they are by default.)
 
 Each channel has a CV input for level; this is a unipolar (0-10V) CV, and corresponding knob attenuates the CV if the CV is in use.
 
 Likewise each channel has an FCV input for frequency modulation, and additionally there is a global FCV input, which voltage effects all channels.  For each channel, the channel FCV and global FCV are summed, then attenuverted by the channel's FCV knob, and the result is added to the FREQ knob setting.  These CVs are bipolar (+/-5V), where +5V will send the frequency from 0 to its max value.
 
 Finally, each channel has a BW (bandwidth) setting, that applies if the channel is a bandpass filter, and controls the width of the filter's frequency response.  These have unipolar CVs per channel.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="peq6"></a> PEQ6
 
 A six channel parametric EQ.  It generally works as PEQ does, except:
   - The global FCV control has a knob, which becomes an attenuverter if a CV is provided.
   - There is a global BW parameter with CV, rather than per-channel bandwidth controls.
   - The "FCV RNG" option controls the scaling of the frequency CV inputs; if set to OCTV, the full CV range will alter the band frequencies up to an octave in either direction; if set to FULL, the CV can run the frequencies over the full range, as on PEQ.
   - On the context (right-click) menu, there is an option "Exclude direct-output bands from mix"; if this is enabled, any band that has its direct output in use (if something is patched to its output) does not get its output mixed into the main output.  Usually, the main output includes all bands.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="peq6xf"></a> PEQ6XF
 
 Expands PEQ6 with envelope follower outputs, per band.  The DAMP and GAIN knobs control the followers; see the description of the <a href="#follow">FOLLOW</a> module for a description of how these work.
 
 ![Parameteric EQs screenshot 2](doc/www/parametric_eqs2.png)
 
 #### <a name="peq14"></a> PEQ14
 
 A fourteen channel parametric EQ.  The control scheme is as with PEQ6.
 
 It adds ODD and EVEN outputs: these are mixes of the odd and even channels, respectively.  If the low channel is set to LOWPASS, its output will go to both ODD and EVEN.  Same for the high channel, if its set to HIGHPASS.
 
 The "Exclude direct-output bands..." option, if enabled, applies to ODD and EVEN as well as the main OUT.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to each of OUT, ODD and EVEN.
 
 #### <a name="peq14xf"></a> PEQ14XF
 
 Expands PEQ14 with envelope follower outputs, per band.  The DAMP and GAIN knobs control the followers; see the description of the <a href="#follow">FOLLOW</a> module for a description of how these work.  DAMP has a unipolar (0-10V) CV; GAIN has a bipolar (+/-5V) CV.
 
 
 ### <a name="envelopes"></a> Envelopes and Envelope Utilities
 
 #### <a name="dadsrh"></a> DADSRH
 
 DADSRH (Delay, Attack, Decay, Sustain, Release, Hold) is an envelope generator, augmenting a standard ADSR design with a delay stage and a self-gating (hold) mode.
 
 ![Envelopes screenshot 1](doc/www/envelopes1.png)
 
 Features:
   - When the MODE switch is set to GATE, DADSRH is a more-or-less standard ADSR envelope generator, with an additional pre-attack delay stage.  The envelope is controlled by a gate CV at the trigger port, or by holding the TRIGGER button.
   - When MODE is TRIG, a trigger CV or press of the TRIGGER button will start a normal DADSR cycle, but controlled by an internal gate CV.  The internal gate persists for the time set by the HOLD knob.
   - The envelope is output as a 0-10 signal at port ENV.  Its inverse (10V - ENV) is output at INV.  When a release stage completes, a trigger is emitted at END.
   - When MODE is TRIGGER, the CYCLE switch controls whether the envelope loops or not upon completion of a release stage.
   - Toggles allow selection of linear, exponential or inverse-exponential ("logarithmic", more or less) shapes for the attack, decay and release stages.
   - The SPEED switch slows down the timing of segments by 10x.
   - The RETRIG switch controls the retrigger behavior (when a new gate or trigger happens while the envelope is running): ATT immediately attacks from the current envelope value (this is the typical behavior with many ADSRs), while RST causes a full reset of the envelope (restarting it at the delay stage).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIGGER input.  Pressing the trigger button will trigger all channels.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="dadsrhplus"></a> DADSRH+
 
 DADSRH+ is a DADSRH, with the addition of CV inputs for each knob, and gate outputs for each stage (a stage's gate output will be high for the duration of the stage).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIGGER input.  Pressing the trigger button will trigger all channels.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="shaper"></a> SHAPER
 
 SHAPER emulates the function of the Envelope Generator section of the classic [EMS VC3](https://en.wikipedia.org/wiki/EMS_VCS_3) and related synths.  It combines an envelope with a VCA.  Unlike an ADSR, the envelope stages are attack, on, decay and off, producing a signature trapezoidal envelope shape.
 
 ![Envelopes screenshot 2](doc/www/envelopes2.png)
 
 Features:
   - The ATTACK, ON, DECAY and OFF knobs specify times from nearly zero to 10 seconds.  The SPEED switch allows these times to be multiplied by 10.
   - The trapezoid envelope is output as a 0-10V control signal at port ENV, subject to attenuation by the ENV knob.  (INV outputs 10V - ENV.)
   - Audio input at port IN is sent through the internal VCA -- controlled by knob SIGNAL and the envelope -- to port OUT.  Turning up the SIGNAL knob can dramatically amplify the signal.  [This is actually a design error, but enough people have used it this way that we leave it be.]
   - A trigger CV at the TRIGGER port, or a press of the TRIGGER button, will start the envelope cycle.  When the off stage completes, a trigger is emitted at port END.  If the CYCLE switch is set to LOOP, the envelope restarts immediately.
   - The SPEED switch slows down the timing of segments by 10x.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIGGER input.  Pressing the trigger button will trigger all channels.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="shaperplus"></a> SHAPER+
 
 SHAPER+ is a SHAPER, with the addition of CV inputs for each knob, and gate outputs for each stage (a stage's gate output will be high for the duration of the stage).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIGGER input.  Pressing the trigger button will trigger all channels.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="ad"></a> AD
 
 An AD (Attack, Decay) envelope generator in 3HP.  The attack and decay stages have durations up to 10 seconds, and CV inputs expecting 0-10V inputs; if a CV is present, the corresponding knob attenuates it.
 
 When a trigger or gate is received at the TRIG input, the envelope cycle begins and runs to its end.  At the end of the cycle, a pulse is emitted at EOC.
 
 If the RT (retrigger) toggle is enabled, if TRIG receives a new trigger while the envelope is decaying, it reenters the attack stage at whatever level the envelope is currently at.  If the cycle ends and the TRIG voltage is high, the cycle restarts.
 
 If the LP (loop) toggle is enabled, the envelope cycles continuously (it doesn't need a trigger to start it).  If RT is also enabled, triggers at TRIG will restart the cycle (this is similar to syncing an LFO).
 
 By default, the attack and decay envelope segments have a logarithmic (more or less) curve -- in linear mode (the LIN toggle), the segments are linear.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIG input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="asr"></a> ASR
 
 ASR is AR (if triggered) or ASR (Attack, Sustain, Release -- if gated) envelope generator.  It has CV inputs for the attack and release times (if CVs are used, they are attenuated by the corresponding knob values).  The sustain level may be set by the small knob marked S.  The attack and release segments are curved by default, but can be made linear with the LIN toggle.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIG input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="adsr"></a> ADSR
 
 A standard ADSR (Attack, Decay, Sustain, Release) envelope generator in 3HP.  The attack, decay and release knobs are exponentially scaled with durations up to 10 seconds.  The sustain knob is linearly scaled, setting the sustain level from 0 to 10 volts.  Lights below each stage knob indicate which stage is active.
 
 By default, the attack, decay and release envelope segments have a curve -- in linear mode (the LIN button), the segments are linear.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the GATE input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="vish"></a> VISH
 
 VISH ("vactrol-ish") is an envelope generator designed to simulate the voltage shapes produced by the [vactrol](https://en.wikipedia.org/wiki/Opto-isolator) control circuits of hardware LPGs (low pass gates).  It lends itself to percussive envelopes, though long envelopes are possible.  It is also used as the envelope circuit in <a href="#megagate">MEGAGATE</a>, and in simplified form in <a href="#lpg">LPG</a> and <a href="#llpg">LLPG</a>.
 
 It operates by generating an internal 10V square envelope, which is fed through a slew limiter.
 
 When a trigger is received at the GATE input, or if any rising edge is received and GT TO TRIG is enabled, the internal envelop triggers and runs for the time set by the MIN GATE knob.  This parameter may be controlled by CV at the MIN input, expecting a unipolar (0-10V) input, which is scaled by the knob position.
 
 When GT TO TRIG is not enabled, then a rising edge at GATE will trigger the internal envelope in the same way, but when the envelope completes, whatever voltage is at GATE is fed to the slew limiter.  This can be used several ways:
   1. If the input is a full 10V, the output will sustain for as long as the GATE input remains high, with a minimum time set by MIN GATE.
   2. If the input is less than 10V, there is an ADSR-like effect, where the output, subject to the slew, will go to 10V for the duration of MIN GATE, then fall to whatever the input is.
   3. If MIN GATE is zero, the input is simply fed to the slew limiter as-is.
 
 The slew limiter is controlled by the RISE and FALL controls, which behave exactly like they do on the <a href="#slew">SLEW</a> module.  The time knobs control how much the rise and fall of the internal envelope are slewed, while the shape knob control the shapes of each segment.  The time knobs have unipolar (0-10V) CVs, while the shapes can be modulated by the single SHAPE CV input, which takes a bipolar CV (+/-5V), which is added to the value of the rise and fall shape knobs.  Context-menu settings allow the CV to be disabled, or applied inverted, for rise and fall separately.
 
 If the TIMES 10X option is enabled, the values of the rise and fall times, and MIN GATE, subject to their CVs, are multiplied by 10.  Thus fall and MIN GATE can go to a maximum of 10 seconds, and rise to 3 seconds.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the GATE input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="dgate"></a> DGATE
 
 ![Envelopes screenshot 3](doc/www/envelopes3.png)
 
 A trigger-to-gate utility, with gate duration up to 10 seconds, and an optional pre-delay of up to 10 seconds.  A trigger pulse is emitted at END when a delay/gate cycle ends.  If the STOP/LOOP switch is set to LOOP, or if the trigger is high when the cycle ends, the cycle repeats.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the TRIG input.  Pressing the trigger button will trigger all channels.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="rgate"></a> RGATE
 
 RGATE is a "clock-relative" gate generator, which outputs gates that have a length that is a ratio of the period of the incoming clock.  It can also be used as a clock divider/multiplier.
 
 The LENGTH control sets the length of the output gate relative to the incoming clock period (time between two clock pulses).  The length may be varied from a minimum of 1ms to the full clock period, at maximum.  With the length at maximum, the output gate will simply stay high.  A LEN unipolar (0-10V) CV may be supplied, which if in use is attenuated by the LENGTH knob.
 
 CLK DIV and CLK MUL alter the frequency and length of gate outputs.  Increasing CLK DIV will set the number of clock pulses the clock period extends over, while CLK MUL will set how many gates will be emitted in that period.  For example, setting CLK DIV to 3 and CLK MUL to 2 yield an output of 2 gates for every three incoming clock pulses, with a maximum gate length of half of 3 times a single clock period.
 
 DIV and MUL are CV inputs for CLK DIV and CLK MUL respectively, each expecting a unipolar (0-10V) CV, and each attenuated by its corresponding knob.  For example, if CLK DIV is set to 4, then a 0-2.5V input at DIV will select a division of 1, an input of 2.5-5V will select a division of 2, and so on.
 
 *About determining the clock period:* the module continuously updates its measurement of the clock period on each clock pulse received, setting it to the time since the last clock was received.  With a steady and continuous incoming clock, this works just fine; otherwise there are some issues to consider:
   1. When the module loads, it has seen no clocks yet, and needs to see two to establish the clock period.  To work around this, there is a default clock period which applies only after the first clock is received and until the second is.  This is configurable on the context menu, defaulting to 500ms (or 120 BPM).  
   1. If the clock stops, the module plays out the current (divided, multiplied) clock period, and then output will stop.  When the clock starts again, RGate will have measured a very long clock period, and will output a long gate.  The RESET function described below can help with this.
   1. An irregular or varying clock may cause odd or unpredictable behaviors.
 
 The RESET port allows resetting the state of RGate, with two modes, configurable on the context (right-click) menu:
   - HARD, the default, resets the calculation of the clock period, and the internal counter that drives the clock divider.  On receipt of the next clock after a hard reset, the default clock period applies.
   - SOFT resets only the clock divider.
 
 The output range of the module may be set on the context menu; it defaults to unipolar 0-10V.  It may be set to 0-5V, +/-10V or +/-5V.
 
 The module is usable as a general clock divider/multiplier; in this case it's advisable to set LENGTH to the minimum, as clocks usually output short trigger pulses.
 
 The module can also be used to generate pulse waves from incoming audio, where the output pitch is some ratio of the input, according to CLK DIV and CLK MUL.  For example, with CLK DIV set to 2, and CLK MUL to 1, and with a square wave input, the output will be a pulse wave tracking an octave below the input.  LENGTH becomes a pulse-width control in this case.  CLK MUL will multiply the increasing frequency to a point; at some point the output frequency would be faster than the internals of RGate update, and nothing happens.  If using the module this way, it makes sense to set the output to bipolar.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the CLOCK input, by default, or the LENGTH input, if so set on the context menu.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="edge"></a> EDGE
 
 A trigger-to-gate utility, comparator and rising/falling edge detector.  RISE and FALL set voltage levels: when the input goes above RISE, the module switches to "high" state and outputs a voltage at the HIGH output.  HOLD sets a minimum time that the module stays in the high state; this can be used to avoid jitter on the output if using high-frequency inputs.  1ms trigger pulses are output at RISE and FALL on the corresponding changes (note that if you switch the module state at audio rates, these will essentially always be high).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="follow"></a> FOLLOW
 
 An envelope follower (a utility that converts its input to a CV proportional to the level of the input's amplitude).  The DAMP knob and CV (0-10V) affect how quickly the output responds to changes in the input -- higher DAMP values effectively slow down and smooth out the response.
 
 The GAIN knob and CV (+/-5V) can attenuate or amplify the output.  Turning the knob counter-clockwise form center will cut the output up to -36dB; turning it clockwise will amplify it up to +12dB.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="mixers"></a> Mixers, Panners and VCAs
 
 #### <a name="mix8"></a> MIX8
 
 An eight-channel mixer/panner with mutes.
 
 ![Mixers screenshot 1](doc/www/mixers1.png)
 
 Features:
   - Eight input channels with decibel-calibrated level faders.
   - Level fader for the output mix.
   - CV control over channel and output levels; expects 0-10V CV; CV is attenuated by the corresponding slider when in use.
   - CV-controlled stereo panners; expects +/-5V CV; CV is attenuverted by the corresponding knob when in use.
   - Stereo outputs: if only one is patched, the output mix is mono.
   - Mutes per channel.
   - Right-clicking a mute buttons solos that channel (un-mutes that channel and temporarily mutes all others).  Right or left click will un-solo, restoring the old state.  Multiple channels can be "soloed" at once.
   - The fader handles contain lights indicating the signal level out of that channel.
   - The master output has MUTE and DIM controls (DIM is a partial mute, with a value configurable on the context menu; it defaults to -12dB).
   - When context menu option "Linear level CV response" is enabled, level CV inputs on each channel and on the master output affect the corresponding level linearly in amplitude (that is, a 5V input would cut the level set by the slider by half); by default they respond in decibels.
   - The output saturates (soft clips) at +/-12 volts.
 
 _Polyphony:_ The module is monophonic: if a polyphonic cable is present at an input, its channels will be summed.
 
 However, there is a non-standard polyphonic feature: on the context (right-click) menu, there are options to "spread" a polyphonic input connected to input channel 1 (only) across the mixer's inputs, as if the poly input had been split into eight mono inputs and each connected to the mixer.  This can be applied to channels 1-8 of the input, or channels 9-16.  Any input patched to an input other than input 1 will override the spread signal.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="mix8x"></a> MIX8X
 
 An expander for MIX8, adding an EQ for each mixer channel, and two sends and stereo returns.
 
 Each EQ section is based on the <a href="#eq">EQ</a> module, with same bands and gains.
 
 Each mixer channel can be routed to send A or send B by knob and CV.  The CVs expect a 0-10V signal, and are attenuated by the corresponding knob.  The knob/CV response is exponential in amplitude, linear in decibels.
 
 Below each send knob is a PRE switch (for "pre-fader"); if on, the send receives the unaltered input into its corresponding mixer channel; otherwise it gets the signal subject to the mixer channel's level slider, mute button and EQ.
 
 Return A and B are each stereo, with the right input being normalled to the left.  Each has a LEVEL knob. Return A (only) has a CV input for LEVEL (the CV works the same as with the sends).  Each return's inputs, subject to the LEVEL processing, are injected directly into the mixer's final stage, subject to the master level slider and mute.  Note that you don't have to use the sends to use the returns.
 
 MIX8X must be positioned to the right of, and ajacent to, the base MIX8 module it will pair with.  See <a href="#expanders">notes on expanders</a>.
 
 ![Mixers screenshot 2](doc/www/mixers2.png)
 
 #### <a name="mix4"></a> MIX4
 
 A four-channel version of MIX8 with the same features.
 
 _Polyphony:_ As with MIX8, this is a monophonic module, but with the same non-standard "spread" feature (in groups of four channels).
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="mix4x"></a> MIX4X
 
 An expander for MIX4, with functionality identical to what MIX8X adds to MIX8.
 
 ![Mixers screenshot 3](doc/www/mixers3.png)
 
 #### <a name="mix1"></a> MIX1
 
 A 3HP fader/VCA, with mute.  
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="mix2"></a> MIX2
 
 A stereo version of MIX1.  The left input is normalled to the right.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with channels defined by the L input.  However, as on MIX8 there is a poly spread feature -- if this is enabled, the module becomes monophonic, and will get the left and right inputs from a pair of polyphonic channels (channels 1 and 2, 3 and 4, etc) on the input L.
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs; left is passed to right if the right input is unpatched.
 
 #### <a name="umix"></a> UMIX
 
 A 3HP unity mixer, usable with audio or CV (e.g. to combine triggers).  Up to 8 inputs are summed to the output.  The output is limited to +/-12V (with clipping modes as below).
 
 The context (right-click) menu has a few options:
  - "Input gain" allows the input gain to be reduced up to -12dB.
  - "Output clipping" sets the manner of output clipping: "Soft" applies saturation or soft clipping, which is better for audio, and which is the default; "Hard" simply clips the output at +/-12V.  "Hard" is better for CVs, as this mode will achieve precise summing of CVs; otherwise the saturator will slightly affect (reduce) the sums at all levels.
  - "Average" causes the module to average, rather than sum, its inputs.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by the first/topmost input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="mumix"></a> MUMIX
 
 Essentially identical to UMIX, but with mute buttons for each input.
 
 If averaging mode is enabled, note that the divisor for the average is the count of how many inputs are connected.  For example, if three inputs are connected, and one is muted, the output will be the sum of the two unmuted channels, divided by three.
 
 See also <a href="switch81">SWITCH81</a>, which is similar to this, with options to attenuate or invert the inputs.
 
 _Polyphony:_ same as UMIX.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="mute8"></a> MUTE8
 
 MUTE8 provides 8 independent manual or CV-controlled mutes. Each channel is muted if its button is toggled on or if there is a positive voltage at its CV input.  Otherwise the input is passed to the output.
 
 As with MIX4 and MIX8, a right-click on a mute button will solo that channel (pass that channel through while muting all others).  Right or left click clears this.
 
 If context menu option "Latching CV triggers" is enabled, triggers on the CV inputs toggle muting on and off.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, where each of the 8 channels may be independently polyphonic, as defined by the cable at the channel's input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes each input unmodified to the corresponding output.
 
 #### <a name="vcm"></a> VCM
 
 A four-channel mixer in 10HP.
 
 Features:
   - Four input channels and mono mix output with knob and CV control over level.  CVs expect 0-10V; when CV is in use, it is attenuated by the knob.
   - Linear mode makes the knob/CV response linear in amplitude (this is good dialing in a CV mix); otherwise, and by default, the response is linear in decibels (and therefore exponential in amplitude).
 
 By default, the output is hard clipped at +/-12V (this is a standard in Rack).  A context menu option allows this limit to be disabled.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by the first IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="pan"></a> PAN
 
 A stereo panner with dual input channels.  Each channel's panner may be controlled with a +/-5 volt CV; when CV is in use, it is attenuverted by the corresponding knob.  The output saturates (soft clips) to +/-12 volts.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the first/top IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="xfade"></a> XFADE
 
 A crossfader (or two-channel mixer, or way to patch a dry/wet knob into any signal chain).  The MIX knob sets the relative strength of inputs A and B.  MIX may be controlled with a +/-5 volt CV; when CV is in use, it is attenuverted the knob.
 
 The SHAPE knob affects the attenuation curves of the two channels as MIX changes:
   - At the center position, SHAPE at produces a standard crossfader behavior.  A and B are attenuated by half; moving MIX to A or B brings that channel to full input level while cutting the opposite channel.  The attenuation is such that if the same signal is patched to both A and B, the same signal is produced at the output regardless of the setting of MIX.
   - With SHAPE at the full counter-clockwise (left) position, there is no output when MIX is centered; moving MIX to A or B brings that channel to full level.
   - With SHAPE at full clockwise (right) position, both channels are at full (unattenuated) level when MIX is centered; moving to A or B cuts the opposite channel.
 
 Linear mode (the LIN button) makes the level attenuation response of MIX linear in amplitude (useful when processing CV); otherwise and by default the response is linear in decibels (and therefore exponential in amplitude).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the A input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="matrix81"></a> MATRIX81
 
 ![Mixers screenshot 4](doc/www/mixers4.png)
 
 An eight input, one output version of <a href="#matrix44">MATRIX44</a>, below.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, as on MATRIX44.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="matrix18"></a> MATRIX18
 
 A one input, eight output version of <a href="#matrix44">MATRIX44</a>, below.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, as on MATRIX44.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="matrix44"></a> MATRIX44
 
 A 4x4 channel matrix mixer.  Each input can be routed with an independent level to each of the eight output mixes.  MATRIX44 is expandable with <a href="matrix44cvm">MX44CVM</a>.
 
 *Note that the matrix knobs are attenuverters, and default to zero.*  That means there will be no output, regardless of the inputs, until some knobs are changed to non-zero values.  The knobs can be set to unipolar mode, as below; they still default to zero.
 
 Saturation (soft clipping) limits each output to +/-12V.  This can be changed to a hard clip at +/-12V on the context menu ("Output clipping") -- as described on UMIX, this is mode is better if you need to precisely sum CVs.  Clipping may also be disabled entirely by setting "Output clipping" to "None".
 
 Another context menu option allows the input gains to be reduced by up to 12db.
 
 Option "Average" sets the output to be the average of its inputs.  The divisor for the average is the number of inputs in use; for example, if three inputs are connected, each output will be the sum of those inputs, scaled by the corresponding knobs, and divided by three.
 
 The knobs visually indicate their values with green/orange colors.  This can be disabled on the context menu.
 
 Option "Unipolar" sets the knobs to travel from zero to 100% over their full travel (which is to say that they can no longer be set to invert the input).  The panel does not update; the tick at noon for each knob indicates 50% in this mode.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by input 1.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="matrix44cvm"></a> MX44CVM
 
 An expander for MATRIX44, adding CVs and mutes for each point in the mix matrix.  The CVs are bipolar (+/-5v), and each is attenuverted by its corresponding knob when in use.  The mute buttons will mute the corresponding mix point, overriding the knob and CV.  As on <a href="mix8">MIX8</a> and others, the mute buttons can be right-clicked to solo that mix point -- all others will be muted.  A subsequent click on a soloed button restores the previous state, and other muted buttons retake effect.
 
 If the context menu option "Solo mutes by column" is enabled, the solo feature applies within the columns of mutes, which is to say soloing a mix point will only affect the output for that mix point, rather than all outputs.
 
 MX44CVM must be positioned to the right of, and adjacent to, the MATRIX44 module it will expand.  See <a href="#expanders">notes on expanders</a>.
 
 #### <a name="matrix88"></a> MATRIX88
 
 ![Mixers screenshot 5](doc/www/mixers5.png)
 
 An 8x8 version of MATRIX44.  It is expanable with <a href="matrix88cv">MX88CV</a> and <a href="matrix88m">MX88M</a>.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by the first/topmost input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="matrix88cv"></a> MX88CV
 
 An expander for MATRIX88, adding CVs for each point in the mix matrix.  The CVs are bipolar (+/-5v), and each is attenuverted by its corresponding knob when in use.
 
 MX88CV must be positioned to the right of, and adjacent to, the MATRIX88 module it will expand, or a MX88M module that is itself expanding a MATRIX88.  See <a href="#expanders">notes on expanders</a>.
 
 #### <a name="matrix88m"></a> MX88M
 
 An expander for MATRIX88, adding mutes for each point in the mix matrix. The mute buttons will mute the corresponding mix point, overriding the knob (and CV if a MX88CV is in use).  As on <a href="mix8">MIX8</a> and others, the mute buttons can be right-clicked to solo that mix point -- all others will be muted.  A subsequent click on a soloed button restores the previous state, and other muted buttons retake effect.
 
 If the context menu option "Solo mutes by column" is enabled, the solo feature applies within the columns of mutes, which is to say soloing a mix point will only affect the output for that mix point, rather than all outputs.
 
 MX88M must be positioned to the right of, and adjacent to, the MATRIX88 module it will expand, or a MX88CV module that is itself expanding a MATRIX88.  See <a href="#expanders">notes on expanders</a>.
 
 #### <a name="switch81"></a> SWITCH81
 
 ![Mixers screenshot 6](doc/www/mixers6.png)
 
 An eight input, one output version of <a href="#switch44">SWITCH44</a>, below.
 
 SWITCH81 is related to <a href="#mumix">MUMIX</a>, with the difference that a switch must be turned on to pass an input to the output, where on MUMIX the switches are mutes (they pass by default).  Also, this module has options for attenuating and inverting the inputs, as on SWITCH44.
 
 If the option "Exclusive switching" is enabled, on the context menu, then only one switch may be active at once (only one input will be routed to the output).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, as on SWITCH44.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="switch18"></a> SWITCH18
 
 A one input, eight output version of <a href="#switch44">SWITCH44</a>, below.
 
 If the option "Exclusive switching" is enabled, on the context menu, then only one switch may be active at once (the input is only routed to one of the outputs).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, as on SWITCH44.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="switch44"></a> SWITCH44
 
 Identical to MATRIX44, but with switches instead of knobs.  All switches default to off, passing no signal.  Clicking a switch sets it to pass voltage with unity gain.  Another click sets the switch off (unless second-click inverting is on, as below).
 
 Note that you can pass attenuated values, by use of Rack's arbitrary parameter-entry feature: right-click a switch, and set its value from -100 to 100% (fractional percentages are allowed).  If inverting is disabled, as below, the entry is from 0 to 100%.
 
 The signal inverting behavior may be set with the "Inverting" context menu options:
   - "Disabled" disables inverting entirely.  This option is the default.  It is best if you want to map MIDI controller buttons/pads to switches.
   - "By param entry" allows negative scale values to be set for a switch by the parameter-entry method, but clicks on a switch will just toggle between on and off.
   - "On second click" causes a click on a non-inverting but enabled switch to change to inverting; another click turns it off.
 
 Option "Average" sets the output to be the average of its inputs.  The divisor for the average is the number of inputs in use; for example, if three inputs are connected, each output will be the sum of those inputs, scaled by the corresponding switch values, and divided by three.
 
 Two other options, "Exclusive switching by row" and "Exclusive switching by column", if enabled, allow only one switch to be enabled in a row, or column, respectively.  Both options may be enabled at once.  (These options do not work well with MIDI mapping via Rack's MIDI-MAP module; this is a known issue for which there is no good solution; but see the discussion [here](https://github.com/bogaudio/BogaudioModules/issues/112) for a potential workaround.  The same problem may apply to other parameter-mapping methods.)
 
 When randomizing the module from the context menu, the behavior changes based on the options:
   - If neither "Exclusive" option is set, then randomizing the module sets each switch to 0% or 100% with equal probability (0%, 100%, and -100%, if inverting is enabled).
   - If one "Exclusive" option is set, but not both, then on randomization exactly one switch per row (or column) is set to 100% (or -100% if inverting is enabled), all others 0%.
   - If both exclusive options are set, on randomization one switch on the entire module is set to 100% (or -100%, if inverting is enabled), all others 0%.
 
 Every switch applies a bit of slew limitation when it changes values, as an anti-popping measure.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by input 1.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="switch88"></a> SWITCH88
 
 An 8x8 version of SWITCH44.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by the first/topmost input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="switch1616"></a> SWITCH1616
 
 A 16x16 version of SWITCH44.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphonic channels defined by input 1.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="dynamics"></a> VCAs and Dynamics
 
 #### <a name="vca"></a> VCA
 
 ![VCAs screenshot](doc/www/vcas.png)
 
 A two-channel voltage-controlled attenuator.  (An attenuator can only reduce a signal.)
 
 Each channel's level may be controlled with a 0-10V CV; when CV is in use, it is attenuated by the corresponding knob.
 
 In linear mode (the LIN button), the knob/CV response is linear in amplitude (useful when processing CV); otherwise and by default the response is linear in decibels (and therefore exponential in amplitude).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the IN input, independently for the top and bottom sections of the module.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT independently in each channel.
 
 #### <a name="vcamp"></a> VCAMP
 
 A voltage-controlled amplifier, capable of adding 12 decibels gain to the input.  (Twelve decibels gain is the same as multiplying the input by 4.)
 
 The level may be controlled with a 0-10V CV -- when CV is in use, it is attenuated by the corresponding slider.  The slider's toggle has a light indicating the output signal level.  The output saturates (soft clips) to +/-12V.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="velo"></a> VELO
 
 A voltage-controlled amplifier with three CV inputs of three different types: a regular one, a bipolar one (good for tremolo), and one that is inverted and scaled (designed for implementing MIDI velocity, and the reason for the module's name).  
 
 The LEVEL knob and CV words as on <a href="#VCA">VCA</a>; the knob sets the base VCA level if the LEVEL input is not patched; if it is patched, it expects a unipolar (0-10V) signal, which sets the LEVEL up to the position of the knob (which is to say that the knob attenuates the CV).  The LEVEL input is the place to patch in an envelope.
 
 The CV input takes a bipolar (+/-5V) signal, subject to attenuversion by the smaller knob below the LEVEL knob.  The resulting value is scaled by whatever value was set by LEVEL (knob and CV) and then added to the LEVEL value (this coupling of CV to LEVEL is enabled is by default, but see below).  The CV input is a good place to patch in an LFO, to achieve tremolo.
 
 The VELO input, if in use, takes a unipolar (0-10V) signal, scales it according to the "V. R." (Velocity Range) knob, inverts it, and adds it to the value produced by the previous two knobs and CVs.  Thus, if the VELO input is 0V, whatever level the VCA would open to according to the values of the previous inputs is reduced by the number of decibels set by the V. R. knob.  When VELO is a full 10V, there is no reduction.  If V. R. is fully clockwise, a 0V VELO input will fully close the VCA.  The VELO input is designed to be patched to the VELO output of Rack's MIDI-CV module.
 
 The suggested patch inputs are meant as examples; of course any inputs can be used.
 
 The LIN toggle, if on, sets the VCA's to a linear-in-amplitude response; when off, the VCA's response to the controls and CVs is linear-in-decibles.
 
 The bipolar CV is optionally coupled to the LEVEL knob/input, but this coupling can be broken by disabling option "Level knob/CV scales bipolar CV" on the context menu.  When the option is enabled, the value of the bipolar CV is scaled by the value of LEVEL (knob and CV) before being added to LEVEL.  Otherwise, it is simply added to level.  In a patch with an envelope to LEVEL and an LFO to CV, having this option on means the LFO's effect will be proportional to the envelope, which is to say the LFO won't open the VCA when the envelope is off.
 
 Using the bipolar CV, it's possible to drive the VCA past unity gain, up to +12db (4x amplitude).  The module soft-clips at +/-12V, and driving the signal into the clipper will result in distortion.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 ![Dynamics processors screenshot](doc/www/dynamics.png)
 
 #### <a name="amrm"></a> AM/RM
 
 AM/RM is a ring- and amplitude-modulation effect and CV-controllable variable wave rectifier.
 
   - With the default knob settings, the unit is a proper ring modulator: the MOD (modulator) and CAR (carrier) inputs are multiplied and the resulting wave is sent to the output.  MOD is passed unchanged to the RECT output.
   - As RECTIFY goes from 0 to fully clockwise, the negative portion of the MOD input is progressively folded to positive, until at the full value the wave is fully rectified.  The modified input is output at RECT; the modified input is multiplied with the carrier input and sent to the output.
   - The DRY/WET control mixes the output of the mod/carrier multiplication with the unmodified carrier input.
 
 The RECT inputs expects a bipolar (+/-5V) CV, which is added to the RECTIFY knob.  The D/W input works the same for the DRY/WET knob.
 
 Note: AM/RM is calibrated to expect +/-5V, 10V peak-to-peak signals (i.e. the output of a typical oscillator).  A modulator signal with a negative value in excess of -5V will be affected by the rectifier circuit even if the RECTIFY is zero.  To avoid this effect, you may need to attenuate a hot signal you wish to use as the modulator.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the CAR input.  The polyphony input can be switched to MOD on the context menu.
 
 _When <a href="#bypassing">bypassed</a>:_ passes the CAR (carrier) input unmodified to OUT; the RECT output is 0V.
 
 #### <a name="pressor"></a> PRESSOR
 
 PRESSOR is a stereo [compressor](https://en.wikipedia.org/wiki/Dynamic_range_compression) and [noise gate](https://en.wikipedia.org/wiki/Noise_gate) with many controls and a sidechain input.  A compressor attenuates signals louder than a threshold level; a noise gate attenuates signals quieter than a threshold level.
 
 The module's signal path has two main components: a detector, and the compressor/gate.  The detector -- effectively an envelope follower -- analyzes the inputs (including the sidechain input, if in use), producing a control signal which is used to control the compressor/gate.  The detector signal is also emitted at ENV.
 
 The various controls and ports work as follows:
   - The MODE switch sets whether the module works as a compressor (COMP) or noise gate (GATE).
   - THRESHOLD sets the threshold in decibels.  The default 0dB setting corresponds to the 10V peak-to-peak output level of a standard oscillator.  The TRSH input expects a unipolar (+10V) input; if in use this is attenuated by the knob.  The knob's range is -24dB to +6dB; menu option "Threshold range" allows this to be doubled to -48dB to 12dB.
   - RATIO sets the degree of attenuation applied to a signal.  In compressor mode, higher settings attenuate the signal more as the detector output goes above the threshold; at the maximum setting, the compressor becomes a limiter.  In noise gate mode, higher ratios more completely attenuate inputs below the threshold.  The RATIO CV input is unipolar (0-10V), attenuated by the knob
   - The COMPRESSION meter provides a visual indication of the amount of attenuation being applied to the input at any given moment, and is proportional to the CV output at ENV.
   - ATACK and DECAY control lag times in the the movement of the detector signal as the input changes.  Each has a corresponding unipolar (+10V) CV attenuated by the corresponding knob.
   - The DECTECT switch toggles the detector between RMS and peak level analyzers; RMS averages the input over a short window whereas PEAK uses the instantaneous level.
   - KNEE toggles between a slightly softer or harder response as the attenuation turns on or off as the signal crosses the threshold.
   - IN GAIN attenuates (up to -12db) or amplifies (up to +12db) the left and right inputs.  (It does not apply to the sidechain input.)  The modified input is sent to the detector and the compressor/gate.  The IGN CV input expects a bipolar (+/-5V) signal which is added to the knob position.
   - OUT GAIN applies up to 24db of amplification to the left and right outputs, after the compressor/gate circuit does its work.  The outputs are subject to saturation (soft limiting at +/-12V).  The OGN CV input expects a bipolar (+/-5V) signal which is added to the knob position.
   - IN/SIDE controls the input to the detector, as a mix/crossfade between the left/right inputs (which are processed by IN GAIN, then summed), and the sidechain input.
 
 Several of the settings can take fairly extreme values (e.g. OUT GAIN); this allows the module to be used as a distortion effect.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the L input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs.
 
 #### <a name="clpr"></a> CLPR
 
 CLPR is a compact (6HP) [clipper](https://en.wikipedia.org/wiki/Clipping_%28audio%29).  Its controls behave the same as the corresponding controls on PRESSOR.
 
 In contrast to LMTR, CLPR chops a signal at a voltage threshold corresponding to the selected amplitude, significantly distorting the signal.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the L input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs.
 
 #### <a name="lmtr"></a> LMTR
 
 LMTR is a compact (6HP) [limiter](https://en.wikipedia.org/wiki/Dynamic_range_compression).  Its controls behave the same as the corresponding controls on PRESSOR.  Controls for attack and release times are on the context menu.
 
 In contrast to CLPR, LMTR does not distort the signal very much; it just reduces the amplitude of the signal to keep it below the threshold.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the L input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs.
 
 #### <a name="nsgt"></a> NSGT
 
 NSGT is a compact (6HP) [noise gate](https://en.wikipedia.org/wiki/Noise_gate).  Its controls behave the same as the corresponding controls on PRESSOR.  Controls for attack and release times are on the context menu.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the L input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs.
 
 #### <a name="cmpdist"></a> CMPDIST
 
 A distortion effect based on a window comparator.  One or two input signals are used, at inputs A and B (if you only want to use one input, note that input B is normalled to +5V).  The inputs are scaled by the A and B knobs and their corresponding bipolar (+/-5V) CV inputs.  At the same time, a "window" voltage is set by the WINDOW knob and unipolar (0-10V) CV.
 
 The comparator calculates the following:
   - If the scaled A voltage is greater than B by at least the amount of the WINDOW voltage, then we're in the GT (greater than) state.
   - If A less than B by at least the WINDOW voltage, we're in LT (less than) state.
   - Otherwise, A and B are within WINDOW volts of each other, and we're in EQ state.
 
 The outputs GT, LT and EQ follow the state:
   - If GT, the GT output is +5V, and -5V otherwise.
   - If LT, the LT output is -5V, and +5V otherwise (which is inverted from what it should logically be; this makes the output interesting).
   - If EQ, the EQ output is +5V, and -5V otherwise.
 
 The MIX output combines the other outputs according to the GT MIX, EQ MIX and LT MIX knobs, which are simply attenuverters (rather than proper VCAs).  GT MIX and LT mix have bipolar CVs.
 
 The MIX output is also subject to the DRY/WET setting, where the dry signal is comprised of the A and B scaled inputs, each processed by the A DRY and B DRY VCAs.  DRY/WET has a bipolar CV.
 
 **Note:** this module is intended for audio-rate use, as a distortion effect; if you want a proper window comparator for use with CV, take a look at <a href="#cmp">CMP</a>.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the A input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="random"></a> Noise/Random, Sample and Hold
 
 ![Noise/random screenshot](doc/www/noise.png)
 
 #### <a name="noise"></a> NOISE
 
 A noise source, in types blue (f), white, pink (1/f), red (aka brown, 1/f^2) and Gauss (normal with mean 0 and variance 1).
 
 Additionally, NOISE has an absolute value circuit.  Patch audio into ABS to get positive CV.  For example, patch white noise into ABS to get uniform values in the range 0-10V.
 
 _Polyphony:_ For the noise outputs, the number of polyphonic channels is set on the context (right-click) menu.  Independently, the ABS circuit is <a href="#polyphony">polyphonic</a>, with polyphony defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="samplehold"></a> S&H
 
 A dual sample-and-hold and trigger-and-hold.  Sampling may be triggered by CV (on the rising edge of a trigger or gate) or button press.
 
 If nothing is connected to an IN port, sampling for that channel is normalled to an internal white noise source with range 0-10V. Alternative options for the normal source noise type and range are available on the context (right-click) menu.  The normal source selection applies to both channels.
 
 Each channel can be independently toggled into track-and-hold mode with the corresponding TRK button.  In this mode, when the input at GATE is high, or the button is held, the input is copied to the output.  When the gate goes low, the input is sampled and held until the next gate.
 
 Each channel may also be have its output inverted with the INV button.
 
 The GATE input on the lower section is normalled to GATE in the top section (but a press on the top button does not trigger the lower section).
 
 The GLIDE context menu option applies linear glide (slew limitation, smoothing) to the outputs.  The time value, which defaults to 0, determines how long the output will take to change (slew) 10V.  This option is ignored in track-and-hold mode.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the GATE input in each section.  If the bottom GATE is patched, the two sections will independently take their channels from their respective GATE inputs; if only the top GATE is patched, the bottom section normals to the top input and both sections have the same number of channels.  The polyphony port can be changed to IN on the context menu (this change applies to both top and bottom sections of the module; IN does not normal to the bottom section, and both sections will set their channels independently, from whatever is patched to their own IN).
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="walk2"></a> WALK2
 
 WALK2 provides two channels of chaotic output, where the output voltage moves as a random walk.  The two outputs are drawn as a trace, in X and Y, on the display.  It may also be configured as an X/Y controller.
 
 For each channel:
  - RATE (knob and CV) controls how sedately, or wildly, the CV moves around.  If CV is in use, it is attenuated by the knob.
  - OFFSET (knob and CV) adds or subtracts up to 5V from the base +/-5V output.  If the offset CV is in use, it is attenuverted by the knob.
  - SCALE (knob and CV) attenuates the output; if CV is in use, it is attenuated by the knob.
 
 DIST outputs a third CV, ranging over 0-10V, derived from X and Y channel outputs.  
 
 The TRIG input can be set to one of three actions by the TRIG selector below it:
   - JUMP: on each trigger at TRIG, both channels will jump to a random value.
   - S&H: on each trigger at TRIG, the value of the random generator on each channel is sampled and held.
   - T&H: while there is a high gate voltage at TRIG, the random generator output is passed to each channel's output; when the gate drops, the last value is held.
 
 The display can be interacted with directly:
   - A click on the display will jump the outputs to the corresponding value.
   - Clicking and dragging will force the outputs to track the mouse.
 Using these in combination with the S&H mode, where the values will hold until something changes them, makes WALK2 to into a handy X/Y controller.
 
 All discontinuous output jumps, caused by any of these means, are subject to a small amount of slew limitation, to avoid pops.
 
 Various options on the context (right-click) menu allow customization of the display (set the range of the display to +/-10V instead of the default +/-5V; hide the grid dots; set the color of the trace).
 
 _Polyphony:_ The module is monophonic (note that WALK is polyphonic).
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="walk"></a> WALK
 
 WALK is a single-channel random walk, identical to one channel of WALK2, in 3HP.  It has a JUMP input rather than a TRIG input, but the same S&H and T&H modes are available on the context menu.  
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the RATE input.  The polyphony port can be changed to OFFSET, SCALE or JUMP on the context menu.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="sequencers"></a> Sequential Switches and Sequencers
 
 ![Sequencers screenshot 1](doc/www/sequencers1.png)
 
 #### <a name="eightone"></a> 8:1
 
 8:1 is a sequential switch and voltage-addressed switch (multiplexer) at once -- it routes 8 inputs to 1 output according to either a clock or CV input (or both).
 
 As a sequential switch, a trigger at the clock input advances the input selection -- which input is routed to the output.  Like a sequencer, it can be reset with a trigger at RESET; the number of inputs to cycle through may be set with the STEPS knob; and the direction is set with the FWD/REV switch.
 
 As a multiplexer, it routes an input to the output under control of the SELECT knob and CV.  A -10-10V CV, divided into 16 equal divisions of 1.25V, controls the input selection.  A CV between +/-1.25V does nothing; a voltage of 1.25-2.49V will add 1 step to the selection, a voltage between -1.25V and -2.49V will subtract one step, and so on.  This value is summed with the knob setting; for example, setting the knob to 4 and inputting a 2.6V CV will send input 6 to the output.  When the knob-plus-CV value exceeds 8, it wraps around.
 
 Both functions may be used simultaneously: the SELECT+CV value is added to the sequential/clocked value, wrapping around.  Note that by default the STEPS value only affects the sequential value; for example, using a clock input and setting STEPS to 2 will yield an alternation between two adjacent inputs, but this pair can be selected with the SELECT knob or CV.  The context (right-click) menu option "Wrap select at steps" changes this, such that the SELECT+CV value wraps at the value of STEPS.  In this case, for example, if STEPS is 2 and SELECT+CV is 1, the sequence will alternate between steps 1 and 2, but on a reset, the sequence will reset to step 2.
 
 If option "Reverse step on negative clock" is enabled, negative or inverted clock pulses (e.g. a pulse from 0V to -5V) will step backwards.  This is still affected by the FWD/REV switch; if the switch is at REV, then a positive clock steps backwards and a negative clock forwards.  This negative-clock behavior can be used to achieve voltage control over the sequence direction (the utility module <a href="#inv">INV</a> can help here).
 
 If option "Select on clock mode" is selected, then the select value (knob and CV) is checked and used to modify the active step only when a clock is received, rather than continuously.
 
 Option "Triggered select mode" changes how the SELECT feature works, replacing the continuous voltage selection with a second internal sequence that offsets (adds to) the primary sequential switch step.  In this mode, the SELECT input expects trigger pulses, which advance the secondary sequence, while the SELECT knob sets the length of the secondary sequence (and a trigger at RESET will reset it). Thus different clocks and step lengths can be used to create complex output step patterns. "Select on clock mode" has no effect if "Triggered select mode" is enabled.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the CLOCK input.  This can be set to the SELECT CV input on the context menu.
 
 _When <a href="#bypassing">bypassed</a>:_ passes input 1 unmodified to the output.
 
 #### <a name="oneeight"></a> 1:8
 
 1:8 is the opposite of 8:1 -- it routes a single input to 1 of 8 outputs.  The control circuit behavior (CLOCK, SELECT, etc) is the same.
 
 _Polyphony:_ Same as 8:1.
 
 _When <a href="#bypassing">bypassed</a>:_ passes the input unmodified to output 1.
 
 #### <a name="addrseq"></a> ADDR-SEQ
 
 ADDR-SEQ is an 8-step sequencer where the step values are set by 8 knobs (with default output range of +/-10V).  It has the same clocked or voltage-addressed control circuit as 8:1 and 1:8.  It can be expanded to more steps, 8 at a time, with <a href="#addrseqx">ASX</a>.
 
 The output range of the knobs may be set on the context (right-click) menu to a variety of bipolar (e.g. +/-5V) and unipolar ranges (e.g. 0-5V).
 
 _Polyphony:_ Same as 8:1.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="addrseqx"></a> ASX
 
 ASX is a chainable expander for <a href="addrseq">ADDR-SEQ</a>, adding 8 steps to the base sequence.
 
 When ASXs are added to an ADDR-SEQ, ADDR-SEQ's STEPS and SELECT knobs (and select CV input) work over the total number of steps, including the expanders.  The knob dials will still read 1-8, but the knob will set the step length (or step selection) over the full count of steps.  The parameter tooltips, if enabled, will show the real values.
 
 Each ASX in a chain must be positioned to the right of, and adjacent to, the previous ASX in the chain, or the base ADDR-SEQ module.  See <a href="#expanders">notes on expanders</a>.
 
 ![Sequencers screenshot 2](doc/www/sequencers2.png)
 
 #### <a name="pgmr"></a> PGMR
 
 PGMR is a four-step programmer, or sequencer with the ability to select the current step manually or by CV.  It is expandable with PGMRX, to add four more steps.  Multiple PGMRXs can be chained on, to add arbitrarily many steps, four at a time.
 
 For each step, four knobs A, B, C, D control the voltage that will go to the corresponding output when that step is selected.  As with ADDR-SEQ, the output range of the knobs can be set on the context menu.
 
 The current step can be selected many ways:
   - By pressing the button, or sending a trigger to the SELECT input, for a given step.
   - By inputs to CLOCK and/or SELECT, subject to the FWD and S.O.C. ("Select On Clock") toggles.  The behavior of these is the same as it is on <a href="#eightone">8:1</a> (and 1:8 and ADDR-SEQ), with the exception that the voltage range to the SELECT input is divided by the number of steps present on PGMR and all its connected PGMRX instances (where the division is always by 8 -- 16 if you consider negative voltages -- on 8:1).
 
 The leftmost bottom output emits a trigger whenever the step changes.  The outputs below each channel selector emit a trigger when that step is selected.
 
 If context-menu option "Save last selected step to patch" is enabled, PGMR will remember the last selected step in the patch, and restore it on patch load.  This is the last step selected by pressing a step button, or triggering a step -- any effect from CLOCK or SELECT always follows the current state of the patch as it runs.
 
 Any PGMRX expanders must be positioned to the right of, and ajacent to, the base PGMR module, or the previous PGMRX in the chain.  See <a href="#expanders">notes on expanders</a>.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the CLOCK input.  This can be set to the SELECT CV input on the context menu.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="pgmrx"></a> PGMRX
 
 A chainable expander for PGMR.  Each instance adds four more steps to the base sequence.
 
 Each PGMRX in a chain must be positioned to the right of, and ajacent to, the previous PGMRX in the chain, or the base PGMR module.  See <a href="#expanders">notes on expanders</a>.
 
 
 ### <a name="visualizers"></a> Visualizers
 
 ![ANALYZER screenshot](doc/www/visualizers.png)
 
 #### <a name="vu"></a> VU
 
 A stereo signal level visualizer/meter.  The L channel is sent to both displays if if nothing is patched to R.  Inputs to L and R are copied to the L and R outputs.
 
 _Polyphony:_ Monophonic, but if an input is polyphonic, its channels are summed, and summed value is used to compute the level displayed (independently for the left and right inputs).
 
 _When <a href="#bypassing">bypassed</a>:_ passes left and right inputs unmodified to the corresponding outputs.
 
 #### <a name="analyzer"></a> ANALYZER
 
 A four-channel spectrum analyzer.
 
 Features:
   - Range setting: smoothly scrolls the displayed frequency range, from just the lower 10% at full counter-clockwise, to the entire range (up to half the sampling rate) at noon, to the upper 20% at full clockwise.
   - Smooth setting: controls how many analysis frames will be averaged to drive the display.  A higher setting reduces jitter, at the expense of time lag.  For convenience, the knob setting is time-valued, from zero to half a second (internally this is converted to an integer averaging factor based on the sample rate and other settings).
   - Quality setting: changes the FFT window size.  Higher settings yield finer frequency resolution, at a higher CPU cost.  The levels and sizes are: GOOD (1024 samples), HIGH (2048 samples) and ULTRA (4096 samples).  If Rack's sample rate is 96khz or higher, the sizes are doubled.
   - Window setting: sets the window function applied to the input to the FFT.  The options are Kaiser, Hamming and none/square.  The default, Kaiser, is probably best for most purposes.
   - Each channel has a THRU output, which passes the corresponding input through unchanged.
   - On the context (right-click) menu, the display vertical (amplitude) range can be set to extend down to -120dB (the default is -60dB); alternately it can be set to to display linearly, with the amplitude expressed as a percentage, where 100% is equivalent to 0dB.
   - By default the frequency axis has a logarithmic plot; this can be set to linear on the context menu.
   - When one clicks and holds on the display, the display freezes, and:
     - The frequency analysis bin under the mouse pointer is highlighted.
     - An overlay box is displayed, with details about the bin number and frequency range, and the level in decibels for each signal at that frequency range.
     - Dragging the mouse left and right will update the highlight and overlay.
     - While the mouse is held, the left and right keyboard keys can be used to change the analysis being displayed, up or down.  Moving the mouse resets any bin offset introduced this way.  This can be used to get to a specific bin in cases where the mouse tracking resolution is larger than the bin width (which can happen at higher frequencies).
 
 _Polyphony:_ Monophonic, with two exceptions:
   - If an input is polyphonic, its channels are summed, and the spectra of the summed signal is displayed.
   - A polyphonic input is copied unchanged (channels intact) to THRU.
 
 #### <a name="analyzerxl"></a> ANALYZER-XL
 
 ![ANALYZER-XL screenshot](doc/www/visualizers2.png)
 
 An eight-channel, 42HP version of ANALYZER, with edge-to-edge-screen design.  Options corresponding to ANALYZER's panel controls are available on the context (right-click) menu.  An extra "Quality" setting, "Ultra+" is available; this uses an FFT size of 16384 (or 32768 if Rack's sample rate is 96khz or higher).
 
 **Note:** Most of the surface of ANALYZER-XL is its display, and clicking the mouse on the display will trigger the freeze function described in the notes on <a href="analyzer">ANALYZER</a>.  This may be confusing if you're trying to click and drag on the module to move it.  To move the module, click near the left edge, for example on the module's name.
 
 _Polyphony:_ Monophonic, but if an input is polyphonic, its channels are summed, and the spectra of the summed signal is displayed.
 
 #### <a name="ranalyzer"></a> RANALYZER
 
 ![RANALYZER screenshot](doc/www/visualizers3.png)
 
 RANALYZER is a frequency response analyzer: it passes a test signal to another module, expecting the output of that module to be patched back, and then displays the frequency spectrum of the response relative to the test signal.
 
 **This module is primarily useful to plugin developers**, especially when developing filters.  Of course anyone may use it, to investigate the response of some module, or to tune a filter bank, or what have you.
 
 By default, will produce a one-shot test signal, emitted at SEND, upon receipt of a trigger (manual or CV) at the TRIG inputs.  The duration of the test signal is always 16384 samples (32768 if Rack's sample rate is 96K or higher) -- the duration in time will depend on Rack's current sample rate.  The same number of samples is collected from RETURN, if it is patched.  Both signals are converted to the frequency domain and displayed (test in green, response in magenta), along with an analysis trace (orange), which shows the difference between the response and the test signals, in the frequency domain, in decibels.  The context-menu option "Display traces" controls which of the traces are displayed.
 
 A window function is optionally applied to the signals as they are converted to the frequency domain; this cleans up noise that otherwise shows up in the signal plots as an artifact of the conversion.  It also slightly distorts the displayed signals.  The default "taper" window minimizes the distortion; Hamming and Kaiser windows are also available.  The window can be selected or disabled on the context menu.  The window is not applied to the test signal before it is emitted at SEND.
 
 The test signal is a swept sine wave (or "chirp", see also <a href="#chirp">CHIRP</a>), with an exponential (if the EXP toggle is on) or linear sweep.  The start and end frequencies for sine sweep are set by the FREQ1 and FREQ2 knobs, each with a range in hertz from 1 to a bit less than the Nyquist rate (half the sampling rate); if FREQ1 is less than FREQ2 the sweep is upwards in frequency, downwards otherwise.
 
 Patching an input to TEST overrides the swept sine generator; the TEST input is used as the test signal.
 
 If the LOOP toggle is enabled, the module continuously outputs, collects and displays the test, response and analysis signals.  Regardless of whether the collection cycle is triggered or looped, a pulse is emitted at the TRIG output when the cycle begins, and at the EOC output when it ends.
 
 The R. DELAY (response delay) control allows sample-accurate alignment of the test and response signals for analysis.  When SEND is patched directly to the module to be analyzed, and that module's output is patched directly back to RETURN, and the analyzed module does not impose an internal sample delay, then the returned signal will be received by RANALYZER two samples later, relative to the test sample RANALYZER emits.  More complicated patches between SEND and RETURN, or modules under test which have internal sample delays, can increase this, in which case R. DELAY may be set to whatever this actual sample delay is.  In practice, getting this exactly right will likely not be very important.
 
 The context-menu option "Trigger on load", if enabled, will auto-trigger the test cycle when the module loads or when the patch loads.  This has no effect if LOOP is enabled when the patch loads.
 
 The display's frequency and amplitude ranges and plot types (logarithmic vs linear) can be set to a few different values on the context menu.
 
 **Note:** Most of the surface of RANALYZER is its display, and clicking the mouse on the display will trigger the freeze function described in the notes on <a href="analyzer">ANALYZER</a>.  This may be confusing if you're trying to click and drag on the module to move it.  To move the module, click near the left edge, for example on the module's name.
 
 _Polyphony:_ Monophonic.
 
 
 ### <a name="poly"></a> Polyphony Utilities
 
 ![Polyphony utilities screenshot 1](doc/www/poly1.png)
 
 #### <a name="mono"></a> MONO
 
 MONO mixes down the channels of a polyphonic cable to a single-channel mono output.  It is an alternative to the SUM module that comes with VCV Rack, but adds a basic onboard compressor, to even out the level differences when only a few, or many, channels on the input are sounding.  The COMP control sets the amount of compression; at zero there is no effect and the behavior of this module is essentially equivalent to Rack's SUM.  As with SUM, the LEVEL control simply attenuates the output.  The output, post-LEVEL, is saturated (soft-clipped) at +/-12V.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="arp"></a> ARP
 
 ARP is a performance-oriented arpeggiator, where the arpeggiated notes come from polyphonic pitch and gate inputs.  It is designed to patched directly to Rack's MIDI-CV (with polyphony enabled) and be played with a MIDI keyboard.  As below, it can be controlled by other modules as well, though this needs to be done carefully.
 
 When a gate goes high on a poly channel of the GATE input, the pitch of the corresponding poly channel from the V/OCT input is added as a note to the arpeggio. (Thus, if ARP is patched to MIDI-CV, pressing a key on your MIDI controller adds a note to the arpeggio.)
 
 The arpeggio is played back under control of the input at CLOCK, which is required for the module to do anything, at the rate of one note per clock pulse.
 
 MODE controls how the arpeggio plays back:
   - UP: play the current notes in ascending pitch order.
   - DN (down): play the notes in descending pitch order.
   - PD (pendulum): play the notes up, then down, without repeating the minimum and maximum notes.
   - PR (pendulum repeat): play the notes up, then down, repeating the min and max notes.
   - IO (in order): play the notes in the order they were played into the module.
   - RD (random): play random notes from the arpeggio; notes may repeat.
   - SH (shuffle): on each arpeggio, play each note once, but in random order.  Repeated notes may still occur, but only when the last note of a sequence happens to be the same as the randomly-selected first note of the next sequence.
 
 The GATE knob sets the output gate length for each played note, as a proportion of the time between the last two clock pulses.  The minimum gate pulse is 1ms; if the knob is turned all the way up, the gate does not drop between notes.  Using an irregular clock, or starting and stopping the clock, will confuse the calculation of the clock rate and produce odd results.  To avoid these problems, if they come up, the context menu option "Max gate length" may be set to "Fixed": in this mode, the gate length is set by the GATE knob to a value from 1ms up to 500ms, without reference to the clock.
 
 HOLD latches the arpeggio, such that it keeps playing even when all input gates are low.  Once all gates are low, a new gate will start adding notes to a new arpeggio.  Notes will be added to the current arpeggio as long as any gate is high.
 
 A trigger at the RESET input will reset the playback of the current arpeggio on the next received clock.  
 
 By default, new notes recorded by the module are incorporated into the playing arpeggio only when the arpeggio restarts on its first note.  On the context (right click) menu, the "Use new notes" setting can be changed such that notes have immediate effect.  
 
 The outputs are always monophonic -- patch ARP into ASSIGN to play an arpeggio through a poly voice.
 
 **To control ARP from modules other than MIDI-CV**, you'll at minimum need to create a polyphonic pitch signal (with Rack's MERGE or something else) and connect it to V/OCT.  Then you'll need gates:
   - With a monophonic gate input, whenever the gate goes high, the pitches of the arpeggio will be set per the poly channels on the V/OCT input, reading in order from channel 1.  So if the gate goes high while V/OCT is getting a four-channel poly input with voltages corresponding to C3, E3, G3 and B3, the arpeggio will be set to those four notes.
   - With a polyphonic gate input, the effect is like pressing a key on a MIDI controller to trigger a gate on a poly channel.  ARP will see a new note using the pitch of the corresponding poly channel on V/OCT.  *You will get strange results if the V/OCT and GATE inputs have a different number of poly channels.*
 
 Note that the HOLD toggle and "Use new notes" menu setting still apply however you control ARP.  If not playing from a keyboard, it will usually make sense to toggle HOLD on and set "Use new notes" to "Immediately".  
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="assign"></a> ASSIGN
 
 ASSIGN is a mono-to-poly voice assigner and poly-to-poly voice reassigner (where the in and out channel counts may be different).  Mono vs poly operation is defined the poly channel count on the GATE input.
 
 One use of the module is to play a (monophonic) sequence through a polyphonic voice, such that envelope tails may ring out.
 
 With mono inputs, on each gate received at GATE, the current pitch at V/OCT is assigned to a poly channel at the V/OCT output, and a gate is output on the same channel at the GATE output.  (The output gate will drop when the input gate does.)  Voices are assigned by increasing channel number, up to the number of channels set by the CHAN knob, at which point channels are reused in order starting with channel 1.  A trigger at the RESET input resets the next assignment to channel 1.
 
 With polyphonic inputs, when a gate goes high on an input channel, the gate and corresponding pitch are assigned to an output channel.  Up to the number of channels defined by the CHAN knob may be output at once.  The module attempts to reuse free channels; if none are free the oldest assignment is replaced.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="unison"></a> UNISON
 
 UNISON allows for unison-mode play of a polyphonic voice, where monophonic pitch and gate inputs are copied out to multiple polyphonic channels, as set by the CHAN knob.
 
 DETUNE allows detuning the pitches sent out to the poly channels up to half a semitone (50 cents).  DETUNE may take a CV, which if present is attenuated by the knob position.
 
 The behavior of DETUNE depends on the number of output channels:
   - With one channel, the unaltered input pitch is sent to the one output channel.
   - With two channels, one channel is tuned up by the full detune amount, while the other is tuned down the full amount.
   - With three channels, one channel gets the input pitch, while the other two are tuned up and down the full detune amount.
   - With four channels, a channel gets each of: up and down the full amount and up and down half the amount.
   - And so on, such that the detune amounts are evenly spread, and one channel gets the unaltered pitch if and only if the channel count is odd.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 ![Polyphony utilities screenshot 2](doc/www/poly2.png)
 
 #### <a name="polycon"></a> POLYCON16
 
 POLYCON16 allows fixed voltages to be sent directly to some number of channels, by channel number, of a polyphonic output.  The number of output channels is set by the CHAN knob, unless an input is present at the CHAN input, in which case the channel count is taken from that input, and the knob is ignored.
 
 This can be used to introduce a bit of fixed variation across the channels of a poly voice.
 
 The context menu option range allows the output voltages to be set from several bipolar and unipolar ranges.  Note that when a unipolar range is used, 0V will be output when the knob is fully counter-clockwise, even though the knobs are drawn with a bipolar dial (such that 0V is usually at noon).
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="polycon8"></a> POLYCON8
 
 A compact version of POLYCON16, that only works with polyphony channels 1-8.  The channel count must be set on the context (right-click) menu.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="polyoff16"></a> POLYOFF16
 
 POLYOFF16 allows for the independent offset and scaling of voltages on each channel of a polyphonic signal.  It can also be used as a replacement for Rack's MERGE, which combines monophonic signals into a polyphonic signal, but here with per-channel offset and scale controls.
 
 The mode of operation is determined by the presence of an input at IN.  With an input:
   - The number of output polyphony channels is set equal to the number of channels on the input, and the CHAN knob is ignored.
   - The voltage of each polyphony channel on the input is processed by the corresponding OFFSET and SCALE knobs.
   - The offset for each channel may be CV-controlled by a (monophonic) input at at its corresponding CV port; the CV inputs expect +/-5V and are attenuverted by the OFFSET knob.
 
 With no input at IN:
   - The output channels are set by the CHAN knob.
   - A monophonic voltage may be provided at each channel's IN port; this is processed by the channel's OFFSET and SCALE before being merged into the polyphonic output.
   - With no input to a channel, the output voltage is just set by the channel's knobs.
 
 The module has the same "Range" context-menu options as POLYCON16, here altering the OFFSET voltages.
 
 The context-menu option "Order of operations" sets the order in which OFFSET and SCALE are applied to each channel.  The default is to scale, then offset.  This behavior was the opposite prior to version 1.1.36.  See <a href="#offset">OFFSET</a> for more detail on this; the behavior is the same here.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="polyoff8"></a> POLYOFF8
 
 A half-width version of POLYCON16, that only works with polyphony channels 1-8.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="polymult"></a> POLYMULT
 
 POLYMULT will turn a mono signal into a polyphonic signal, with a given number of channels, where each channel gets a copy of the input voltage.  The number of channels is set by the CHAN knob, unless an input is present at the CHAN input, in which case:
   - If the signal at CHAN is polyphonic, the channel count is taken from that signal, and the knob is ignored.
   - If the signal at CHAN is monophonic, it acts as a CV (0-10V), selecting the channel count up to the maximum set by the knob.
 
 Each OUT output is identical.  To simply make copies of an already-polyphonic signal, use the regular <a href="#mult">MULT</a> module.  If the input is a polyphonic singal, only the first channel is used to produce the outputs.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="pitch"></a> Pitch CV Utilities
 
 Utilities related to processing pitch CVs (1 volt / octave CVs, for controlling the pitch of oscillators).
 
 ![Pitch utilities screenshot](doc/www/pitch.png)
 
 #### <a name="detune"></a> DETUNE
 
 A 1V/octave pitch processor, for controlling a detuned oscillator pair.  A reference pitch in raised and lowered by the number of cents (hundredths of a semitone) specified by the knob and CV, and emitted at OUT+ and OUT-.  The input pitch is emitted at THRU.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes V/OCT unmodified to each of the three outputs.
 
 #### <a name="stack"></a> STACK
 
 A 1V/octave pitch processor for stacking oscillators.  The SEMIS, OCTAVE and FINE knobs determine an interval (up or down) to apply to the input pitch and send to OUT.  The input pitch is sent unmodified to THRU, for ease of chaining multiple STACKs to control multiple oscillators, e.g. to create chords.
 
 The CV input expects +/-5V; the value modifies the interval set by the knobs in the amount of one semitone per tenth volt.  If QZ (quantize) is active, the CV-controlled interval is quantized to the nearest semitone.  This specialized CV is output at the THRU port, with a value set by the knobs and CV in, when there is no input pitch.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the V/OCT input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes V/OCT unmodified to both THRU and OUT.
 
 #### <a name="reftone"></a> REFTONE
 
 A tuner that outputs a selectable (Western, chromatic) pitch as CV (1V/octave, for controlling an oscillator) or as a pure sine tone.  The base pitch is selected with the PITCH and OCTAVE knobs, while the FINE knob allows the output to be fine-tuned up or down a full semitone.  The LED-style display indicates the selected pitch, octave and fine tuning (in cents), and the corresponding frequency (in hertz).
 
 _Polyphony:_ Monophonic.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 
 ### <a name="utilities"></a> Utilities
 
 ![Utilities screenshot](doc/www/utilities.png)
 
 #### <a name="bool"></a> BOOL
 
 A boolean logic utility.  Inputs are considered true if the input voltage is greater than 1V.  The top section takes two inputs and computes AND, OR and XOR at the outputs.  The lower section computes the negation of its input.  Output is 5V if an output is true, 0V otherwise.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the first/topmost input. The NOT circuit is independently polyphonic based on its input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="cmp"></a> CMP
 
 CMP is a window comparator.  It takes two inputs, A and B, which normal to 0V.  Each is summed with the value of its corresponding offset knob and clipped to +/-12V.  The four outputs indicate the relative values of A and B:
 
  - A>=B will output high if A is greater than or exactly equal to B.
  - A<B will output high if A is less than B.
  - EQ will output high if the difference between A and B are less than or equal to the window voltage.
  - NOT will output high if EQ is low.
 
 The WINDOW knob specifies the window voltage.  LAG specifies a time of up to one second by which a change in the output will lag a change in the inputs states; if the input state switches back before the lag expires, the output does not change. WINDOW and LAG each take a unipolar (0-10V) voltage, each of which is attenuated by the corresponding knob.
 
 The OUTPUT switch sets the high and low voltage values for the outputs: 0V low/+10V high, or -5V low/+5V high.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the A input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="cvd"></a> CVD
 
 A simple delay designed for use with CV (though it works fine with audio).  Use it to delay triggers or gates, create a flip-flop that resets itself after a time, make a sequence run for a while then stop, double up an envelope, or what have you.
 
 The large TIME knob sets the delay time, in seconds, as scaled by the small knob (up to 0.1, 1 or 10 seconds); TIME takes a 0-10V CV, attenuated by the knob.  Reducing time truncates the internal delay buffer.  The DRY/WET knob sets the mix of the original and delayed signals at the output, with a +/-5V CV input.
 
 CVD may be used for lag/latency correction; using parameter entry (by right-clicking the TIME knob) is handy here, to set a precise delay.  You can get sample-accurate delays by dividing the desired number of delay samples by Rack's current sample rate, and setting TIME to that number (leave the time scale knob at 1).
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="flipflop"></a> FLIPFLOP
 
 A boolean memory utility with two independent channels.  A high voltage at TRIGGER will cause the state of a channel to change from A to B.  A subsequent trigger will flip it back.  Output is 5V at whichever of A and B is selected, 0V at the other.  A trigger voltage at RESET sets the channel back to state A.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the channels at the TRIG input, independently for the two sections of the module.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="inv"></a> INV
 
 A dual signal inverter, with CV or manual control, and optional latching.  In each section separately, the signal at IN is inverted and sent to OUT when the button is held an input voltage at GATE is high, and passed to OUT unchanged otherwise.
 
 If LATCH is enabled, a button press or high voltage toggles the state of the inverter.  A context menu option "Save latched state to patch" will, if latching is enabled, save the latched state to the patch and restore it on patch load.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the channels at the IN input, independently for the two sections of the module.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT independently in each channel.
 
 #### <a name="manual"></a> MANUAL
 
 A manual trigger/gate with 8 outputs.  A constant +5V is sent from each output for as long as the TRIG button is held; 0V is output otherwise.  The high output voltage may be set to +10V on the context menu.
 
 MANUAL may be set to output a trigger pulse (the high output voltage for 10ms) on patch load (akin to a Max/Msp loadbang).  This is off by default; enable clicking "Trigger on load" on the module's context (right-click) menu.  The pulse is emitted 100ms after the patch starts processing samples.
 
 _Polyphony:_ Monophonic.
 
 _When <a href="#bypassing">bypassed</a>:_ no output; disables "Trigger on load".
 
 #### <a name="fourman"></a> 4MAN
 
 A version of MANUAL with four independent trigger buttons with separate outputs.
 
 The "Trigger on load" and "Output" options, as on MANUAL, apply to all four outputs if enabled.
 
 _Polyphony:_ Monophonic.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="mult"></a> MULT
 
 A 3 HP multiple (signal splitter or duplicator).  There are two 1-to-3 channels.  There is also a 1-to-6 mode: if nothing is patched to the second channel's input, the input to the first channel is copied to all six outputs.
 
 _Polyphony:_ Polyphonic inputs are duplicated (channels intact) at their corresponding outputs.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="offset"></a> OFFSET
 
 An offset and scaler.  The OFFSET and SCALE knobs have CV inputs (unipolar, 0-10V).  There are two operating modes, as set by the "Order of operations" context (right-click) menu option:
 
   - Scale, then offset (the default): the output is `(input * scale) + output`.  With no input connected, the output is just `offset`.
 
   - Offset, then scale: the output is `(input + offset) * scale`.  With no input connected, the output is `offset * scale`.
 
 Note that prior to version 1.1.36, there was no mode setting, and the behavior was to offset, then scale.  This behavior change might affect older patches.
 
 By default, the output is capped at +/-12 volts (this is a standard in Rack).  A context menu option allows this limit to be disabled.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the channels of the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="slew"></a> SLEW
 
 A slew limiter - when the input changes rapidly, the output changes less rapidly, lagging the input.
 
 The rising and falling slew rates and shapes are set independently.  The RISE and FALL time knobs are calibrated to set the time the output would need to catch up to a 10V change in the input.
 
 The RISE and FALL shape knobs affect the movement of the output as it catches up to the input (the shape it would draw on a scope). The shapes vary between log, linear and exponential curves.
 
 RISE and FALL each have a unipolar (0-10V) CV input affecting the corresponding slew rate.
 
 If SLOW is enabled, the slew rates for both RISE and FALL are 10x slower what they would otherwise be, based on the knobs and CVs.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the channels of the IN input.
 
 _When <a href="#bypassing">bypassed</a>:_ passes IN unmodified to OUT.
 
 #### <a name="sums"></a> SUMS
 
 An arithmetic logic utility.  The top section outputs the sum, difference, maximum and minimum of its input signals (unpatched inputs send a 0V signal into each computation).  The lower section negates (reverses the sign of) its input.
 
 By default, the output is capped at +/-12 volts (this is a standard in Rack).  A context menu option allows this limit to be disabled.
 
 _Polyphony:_ <a href="#polyphony">polyphonic</a>, with polyphony defined by the first/topmost input.  The NEG circuit is independently polyphonic based on its input.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 #### <a name="switch"></a> SWITCH
 
 A signal-routing module with two through channels.  If the button is held or the GATE input is high, the HIGH input for each channel is routed to the corresponding OUT.  Otherwise, each LOW input is routed to each OUT.
 
 If LATCH is enabled, a button click or trigger pulse at GATE will toggle the output to HIGH; a second click or trigger resets it to LOW.
 
 If the context menu option "Save latched state to patch" is enabled, and latching is on, the latched state will be preserved in the patch and restored on patch load.  Otherwise the module be in LOW state on patch load.
 
 _Polyphony:_ If polyphonic input is present at GATE, then the module is polyphonic in the standard way, independently switching the independent polyphonic channels on the high/low inputs (the button will switch all channels).  Additionally, if the input at GATE is not present or monophonic, but polyphonic cables are are present at any high/low inputs, and such an input is switched to the output, it is duplicated to the output with channels intact.
 
 _When <a href="#bypassing">bypassed</a>:_  passes LOW unmodified to OUT independently in each channel.
 
 #### <a name="lgsw"></a> LGSW
 
 LGSW is a version of <a href="#switch">SWITCH</a> with two gate inputs and onboard logic.
 
 If nothing is patched to either of the gate inputs, or if just one is patched (either one), the behavior of the module is identical to SWITCH (the logic section has no effect).
 
 If both gate inputs are in use, then the logic will evaluate to true according to the gate states and the logic mode:
   - OR: true if either gate is high.
   - AND: true if both gates are high.
   - XOR: true if exactly one of the gates is high.
   - NOR: true if neither of the gates is high.
   - NAND: true if both gates are low.
 
 When LATCH is off, and both gates are in use, the module outputs the HIGH input whenever the logic is true or the button is held.
 
 When LATCH is on, and both gates are in use, the latch will toggle when the logic state changes, or when the button is pressed.
 
 The logic mode can be set by CV.  Using a CV for the logic overrides whatever selection was made with the small mode-select button.  A O-5V CV will select the mode as:
   - OR: less than 1V.
   - AND: 1V but less than 2V.
   - XOR: 2V but less than 3V.
   - NOR: 3V but less than 4V.
   - NAND: 4V or more.
 
 _Polyphony:_ Same as <a href="#switch">SWITCH</a>, except that the polyphony channel count is set by the topmost gate input only.
 
 _When <a href="#bypassing">bypassed</a>:_ no output.
 
 ### <a name="misc"></a> Miscellaneous
 
 ![Miscellaneous modules screenshot](doc/www/misc.png)
 
 #### <a name="blank3"></a> BLANK3
 
 A 3HP blank panel.
 
 #### <a name="blank6"></a> BLANK6
 
 A 6HP blank panel.
 
 
 ## <a name="polyphony"></a> Note on Polyphony
 
 Most modules in this collection support VCV Rack's polyphony feature according to Rack's <a href="https://vcvrack.com/manual/VoltageStandards.html#polyphony">polyphony standards</a>.
 
 Generally this means that a module with polyphonic support can run up to 16 independent internal "channels".  The number of channels a module will run is defined by the polyphony of the signal (cable) present at a particular, designated input port.  (This will be the pitch input of an oscillator, or the gate input of an envelope, and so on.)  When the input at that port carries more than one channel, the module's outputs will carry the same number of channels.
 
 With secondary inputs (CV inputs), the behavior is usually:
   - If the input signal is monophonic, that single value is used by each channel in the module.
   - If the input is signal is polyphonic, each processing channel will use the value of the corresponding channel of the CV input.  If the CV input does not have as many channels as the module is processing, higher channels will see a zero input from the CV.
 
 Knobs, faders and switches apply equally to all modules channels.  Where a module has indicator lights, these usually display an average value across the active module channels.  (This indicator behavior isn't specified by the standard.)
 
 A few modules are strictly monophonic.  In this case, a polyphonic input will be handled in one of two ways:
   - The polyphonic channels will be summed (typically if the input is designed for audio).
   - The value of the first channel will be used, ignoring others (typically if the input is designed for CV).
 
 See the documentation for each module for notes on any divergence from these basic rules.  Each module's documentation will also indicate which input port defines the module's polyphony.
 
 Other notes:
   - With some modules, it's not obvious which input should define the module's polyphony -- but it is always the case that the channels will be taken from just one input.  For example, with stereo modules with left and right inputs, the left input defines the polyphony of the module.  With mixers like UMIX and MATRIX88, the first input is used.  With some modules, the input to use can be set on the context (right-click) menu.  Each module's documentation will describe which port or optional ports are used for polyphony.
   - Modules with no inputs may still support polyphonic outputs; in this case the polyphony is set on the context (right-click) menu.  NOISE is an example.
   - Some modules are not polyphonic in a strict sense, but have non-standard polyphonic features (see the channel spread feature on MIX4 and MIX8).
 
 Finally, please note that with the addition of polyphony, the term "channels" can have several meanings in this documentation.  Polyphonic modules have up to 16 internal processing "channels", defined by the polyphonic "channels" of their inputs, and carrying over to the polyphonic "channels" of their outputs.  But, elsewhere, MIX8 is an "8-channel" mixer; stereo modules will have left and right "channels"; VCA is a dual module with two independent "channels" (each of which may be polyphonic); ANALYZER has four input "channels" (in that it will display the spectra of four different inputs simultaneously); and so on.
 
 
 ## <a name="expanders"></a> Note on Expanders
 
 A few modules in this collection have expanders.  These will generally do nothing unless they are paired with the base module they're meant to expand.
 
 To pair an expander with its base, the expander **must** be positioned **to the right** of the base module, and **adjacent to** (touching) the base module.  That's all that's required; there is no button or context menu switch to pair modules.  Conversely, there's no button or option that disables the pairing of a base and expander if they're correctly positioned.
 
 Rack allows left expanders as well as right, but to simplify things we've chosen not to use them, allowing only right expansion.  Rack doesn't support expanders above or below the expanded module.
 
 Some expanders may be chainable: multiple instances can be added to continue expanding the base.  In this case, the expanders must all be to the right of the base, and all touching.
 
 
 ## <a name="bypassing"></a> Note on Bypassing
 
 Rack adds a "Bypass" option to the context (right-click) menu on every module.  By default, this fully disables the module (while allowing it to remain patched), such that the module produces no output (or more precisely, constantly outputs 0V).  For some modules, however, it makes more sense to pass the module's unmodified input to the output.
 
 Most Bogaudio modules implement bypassing according these general rules:
   - Modules that generate a signal (VCOs, LFOs, noise/random generators, envelope generators), or combine signals (mixers, logic), typically produce no output.
   - Modules that modify a signal (filters, effects, VCAs, attenuators) typically pass their input to the output.
 
 There are various exceptions and special or ambiguous cases; see the documentation for each module for details on that module's behavior.
 
 
 ## <a name="alternate_panels"></a> Alternate Panels
 
 ![alternate panels image](doc/www/alternate_panels.png)
 
 There are three sets of alternate panels for the modules:
   - Light
   - Dark
   - Dark (low-contrast)
 
 For nearly all modules (ANALYZER-XL is an exception), the panel to use can be set on the module's "Panel" context menu.  Also, a global default can be set for the plugin, with the "Default to..." panel menu options.
 
 When new modules are added to a patch, they are set to use the default panel design (which is Light, unless the default has been changed to something else).  When the default is changed, all modules set to use the default panels will update.  
 
 The default panel set is used when the modules are displayed in Rack's module browser (though, the module browser may not update immediately when the default changes; restarting Rack will force it to).
 
 When a new default panel set is selected, the module places a file named `Bogaudio.json` in your Rack user directory (e.g. `~/Documents/Rack` on Mac), to save the preference.
 
 ## Other Notes
 
 #### <a name="resumeloop"></a> Loop Behavior on Patch Load
 
 Modules Shaper, Shaper+, DADSRH, DADSHR+ and DGATE can each be set to loop.  These modules will automatically being running (looping) when the patch loads if they were looping when the patch was saved.
 
 This behavior can be disabled on a per-module basis by right-clicking the module and unchecking "Resume Loop on Load".
 
 ![Resume Loop on Load menu screenshot](doc/www/resumeloop.png)
 
 
 ## Issues and Feedback
 
 Bug reports and feedback are welcome: please use the [issue tracker](https://github.com/bogaudio/BogaudioModules/issues).
 
 
 ## Acknowledgements
 
 Uses [FFTReal](https://github.com/cyrilcode/fft-real) under the DWTFYWT Public License.
 
 The Github Actions auto-build scripts under [.github/](.github) are originally from [SubmarineFree](https://github.com/david-c14/SubmarineFree) -- thanks @david-c14 and @dewb -- find a detailed explanation [here](https://github.com/david-c14/SubmarineFree/wiki/Cross-compiling-using-github-actions).
 
 Thanks to [@Eurikon](https://github.com/Eurikon) for the low-contrast panel design, and all who contributed to the alternate-panels development on [#47](https://github.com/bogaudio/BogaudioModules/issues/47).