Skip to content
Switch branches/tags
Go to file
Cannot retrieve contributors at this time
945 lines (572 sloc) 64.1 KB

COLD MAC: A Technical Map

COLD MAC is a swiss army knife or a centerpiece, a jack of all trades or a master of one, depending how you patch it. It can be the hub around which an entire patch revolves, or the basic utilities you need in the moment. It is a tool for mixing, processing, reconfiguring, slicing, transforming, and generating voltages at control and audiorates, for macroscopic control of an entire patch or microscopic control over a single wave. Whether using its discrete functional blocks as independent entities or weaving them into a self-patched web, COLD MAC is sure to find a home in any patch!



Addition and Subtraction
Dry/Wet Control
Pre-FADEr Aux Send
Quad Panner / Vector Mixer


Or / And / Not / Nor / Nand / Xor
Min / Max
Halfwave Rectification
Logical Mixing
Logical Crossfading
Signal Splitting/Routing
Bernoulli Gates and Event Probability


Envelope Following


Waveshaping & Timbre Control
Fullwave Rectification
Slew Limiting / Portamento
Phase-Related LFOs
Mid/Side Processing
Sequence Harmonizer

Table of Contents

  1. Specifications
  2. Overview
  3. Block Diagrams
  4. SURVEY: Macro Control
  5. LEFT/RIGHT/FADE: Crossfader/Panner
  6. OR/AND: Analog Max/Min and Digital Logic
  7. SLOPE: Fullwave Rectification
  8. FOLLOW: Envelope Follower and Slew Limiting
  9. CREASE: Waveslicer
  10. LOCATION: Integrator
  11. MAC: AC-Coupled Audio Mixer
  12. Phase-Related LFOs
  13. Dynamics Control: Compression and Expansion
  14. Extended Waveshaping
  15. Feedback Sounds
  16. Mid-Side Processing
  17. Bernoulli Gates and Event Probability
  18. Quadraphonic Panning and Vector Mixing
  19. Extended Logic
  20. Further Reading
  21. Warranty


  • 8HP
  • 30mm depth
  • 65mA @ +12V
  • 62mA @ -12V


COLD MAC is made up of six DC-coupled, analog voltage processing blocks and an AC-coupled audio mixer. These blocks can be patch-programmed in a modular fashion to build CV and audio processing networks. The SURVEY control is connected (or 'normalled') to a number of inputs, enabling a single macro-control of all 8 outputs. Breaking these normals by inserting a cable enables each block to function independently.

The graphs on the panel next to each output jack represent the transfer function of each block when nothing is patched into that block. Input SURVEY voltage is on the X-axis. SURVEY=0V at the horizontal center of the icon. Output voltage is on the Y-axis. The height of the diagonal line indicates the output voltage for any given SURVEY voltage. So, as SURVEY sweeps from -5V to +5V, the graphs show the output voltage of each block when nothing is patched. Note that the LEFT output graphic is a copy of SURVEY, so this is the 'reference' function to compare against. You can view larger copies of these transfer functions here.


All jacks which are normalled are indicated on the panel by curved lines entering the jack from the top left. Inserting a cable into the jack will 'break the normal', i.e. disconnect the default connection going into that jack. The SURVEY voltage (sum of the SURVEY knob and CV) is normalled into several jacks, each indicated by an S above the jack. The SURVEY voltage will feed these jacks, provided the jacks are unpatched. -5V is normalled into LEFT, indicated by a negative sign above the jack, while +5V is normalled into RIGHT, indicated by a positive sign above the jack. When two jacks are connected by a curved line, the upper jack is normalled into the lower jack: inserting a cable into the upper jack will feed the signal into both jacks, provided the lower jack is unpatched.

-5V into LEFT
+5V into RIGHT
OR1 into AND1

Naming Conventions

All input jacks will be referred to by their label (e.g. LEFT, RIGHT, FADE, OFFSET, etc.). The OR and AND input jacks will be called OR1/OR2 and AND1/AND2. OR1 and AND1 will be the inputs in the left-most column, while OR2 and AND2 will be the corresponding jacks in the middle column.

The output of the MAC audio mixer, envelope follower and integrator will be referred to using their panel labels (MAC, FOLLOW, LOCATION, respectively). The outputs in the right-most column labeled only by transfer function graphs will be referred to by their associated input jack's name and '(OUT)': LEFT(OUT), RIGHT(OUT), OR(OUT), AND(OUT), SLOPE(OUT), CREASE(OUT).

Block Diagrams


OR/AND block



MAC block

SURVEY: Macro Control

The SURVEY knob sweeps from -5V to +5V. The value of the SURVEY CV input is added to the value of the knob. This CV input is DC-coupled and capable of handling audio-rate modulation!

The summed SURVEY value is then normalled into these jacks:

  • LEFT/RIGHT Crossfader Block: FADE Input
  • OR/AND: OR2 Input (which is normalled to AND2 Input)
  • SLOPE/CREASE: SLOPE Input (which is normalled to CREASE Input)

This normalling acts as if there is a phantom patch cable which patches SURVEY into each of the above jacks. Patching a dummy cable into any of these jacks breaks the corresponding normal, and the functional block will no longer be affected by the SURVEY control.

The summed SURVEY value is also used as the MAC Audio Mixer's volume control, with -5V corresponding to the silence, and +5V corresponding to the max volume.

LEFT/RIGHT/FADE: Crossfader/Panner

The FADE block is used to control the routing of two inputs to two outputs by panning the inputs between the outputs. An optional OFFSET voltage can be added to both outputs. This block can be used to mix up to three audio or CV signals, pan an input to two destinations, crossfade two inputs to one destination, swap two inputs between two outputs, perform addition/subtraction/inversion, bias an audio signal for wavefolding, act as a VCA and...

FADE is the crossfade/pan/level control. With nothing patched to this jack, the SURVEY voltage will control FADE. When FADE is -5V, the LEFT input is directly passed to the LEFT(OUT) jack and the RIGHT input is directly passed to the RIGHT(OUT) jack. When FADE is 0V, both inputs appear at both outputs equally at half of their amplitude (-6dB). When FADE reaches +5V, the two inputs have swapped sides: the LEFT input appears at the RIGHT(OUT) jack and the RIGHT input appears at the LEFT(OUT) jack.

The OFFSET input is added to each of the two outputs equally, regardless of the FADE level.

This demo will help give you a feel for how the panning and crossfading works! It demonstrates three sine waves at different frequencies plugged into LEFT, RIGHT and OFFSET. The slider labeled F_ade controls FADE. L_on, R_on and O_on act as switches which will connect or disconnect the sine waves into LEFT, RIGHT and OFFSET (respectively) so that you can see the difference between panning and crossfading! When set to 0, it is as if 0V is patched to the corresponding input. When set to 1, it is as if the corresponding sine wave is patched in.

For the mathematically inclined, the crossfader block can be expressed by the following transfer functions (FADE has been mapped from [-5V,+5V] to [0,1] to simplify the equation):

It may be helpful for some to conceptualize this crossfading action as sending each input through two VCAs controlled by signals with opposite polarities. This can be illustrated by the following diagram.

FADE block as VCAs

Crossfading and Swapping

LEFT(OUT) or RIGHT(OUT) can be used on their own to create a crossfade between the two input sources.

When the FADE voltage is -5V, LEFT(OUT) will pass a copy of the LEFT input. As FADE increases, the LEFT(OUT) signal will crossfade from the LEFT to RIGHT inputs. At 0V into FADE, LEFT(OUT) will be a balanced mix of the LEFT and RIGHT inputs at half their amplitude (-6dB). Once FADE reaches +5V, the crossfade is completed and LEFT(OUT) passes a copy of the RIGHT input.

The same procedure applies to the RIGHT(OUT) with the polarity of the FADE voltage reversed.

Try using the crossfader to slowly blend between two different audio sources (for instance a dry/wet mix), or to morph between two different modulation sources (for instance two LFOs at the same frequency but with different shapes, or between a stepped random and a smoothed random source).

Using both outputs simultaneously creates two outputs with crossfades inverted relative to the other. This enables swapping two sources back and forth between two destinations.

Signal Duplication and Inversion

LEFT is normalled to a -5V reference and RIGHT is normalled to a +5V reference. If nothing is patched to LEFT, RIGHT, and OFFSET, then LEFT(OUT) will present a copy of the FADE input. RIGHT(OUT) will be an inverted copy of the FADE input.

With nothing patched to the FADE input, the LEFT(OUT) and RIGHT(OUT) respectively become duplicated and inverted copies of SURVEY, due to the normalling of SURVEY into FADE. This is indicated by the transfer function icon next to the output jack.

Addition and Subtraction

With nothing patched to LEFT and RIGHT, patching an additional signal into OFFSET simply adds to both of these, resulting in the following outputs:

This property will be explored in some of the extended techniques described at the end of this document.

Panning and VCA Functionality

Using one input along with both outputs enables traditional panning controlled by FADE. Setting FADE to +/-5V will hard pan the input signal to one side. 0V will place the input signal in the middle of LEFT/RIGHT at half its amplitude (-6dB).

If a signal is patched to LEFT and nothing is patched to RIGHT, sweeping FADE will fade the LEFT signal out of LEFT(OUT); however, it will also fade the +5V signal normalled to RIGHT into the LEFT(OUT) output signal, this means an unwanted DC-offset will be introduced at LEFT(OUT). Similarly, the DC-offset will be faded out of RIGHT(OUT) as the input signal at LEFT fades into RIGHT(OUT). To eliminate this offset from the output jacks, patch a dummy cable into RIGHT (leave the other end of the cable unpatched). This demo demonstrates using a dummy cable to eliminate the DC offset when panning (r_dummy will connect a dummy cable to the right input when set to 1).

  • At -5V into FADE, the signal will be panned hard to LEFT(OUT), with nothing at RIGHT(OUT).
  • At 0V into FADE, the signal will be panned equally between the two outputs at half amplitude.
  • At +5V into FADE, the signal will be panned hard RIGHT(OUT).

An equivalent panner or pair of VCAs is created by patching the signal to be panned into RIGHT and dummy-cabling LEFT. The only difference is the polarity of the FADE control.

NB: The LEFT to RIGHT(OUT) path and the RIGHT to LEFT(OUT) will not entirely close for high frequency signals.

Dry/Wet Control and Feedback

The crossfader can be used to mix between a dry signal and a wet (i.e. processed signal) by multing the dry signal to LEFT and the processing chain. The processing chain's output (wet signal) should be fed into RIGHT. However, even if you do not have a mult or stackable, the following patch can be used to achieve the same effect!

Input Signal > LEFT

RIGHT(OUT) > to processing

from processing > RIGHT

LEFT(OUT) > Output Signal

SURVEY/FADE act as the dry-wet crossfade control.

When FADE is at -5V, the dry signal passes directly from LEFT to LEFT(OUT). When FADE is at +5V, the dry signal is sent to RIGHT(OUT) where it then goes to the processing chain. It returns from the processing chain at RIGHT, which is then passed to the final output at LEFT(OUT).

When FADE is between the two extremes, there will be a feedback loop from RIGHT(OUT) to the processing chain to RIGHT and back to RIGHT(OUT). If necessary, try using the FOLLOW to tame the feedback!

Pre-FADEr Aux Send/Return Loop

The panning functionality can be paired with the MAC mixer to set up a pre-fader send/return loop without the need for additional VCAs or stackable cables!

Input Signal > LEFT

Dummy Cable > RIGHT

Dummy Cable (or dry level CV) > FADE

MAC > Send

Return > OFFSET

LEFT(OUT) > Output Signal

SURVEY controls Aux Send level (pre-FADE)

FADE controls dry signal level

You will want to make sure FADE has been decoupled by SURVEY by either using a dummy cable or using CV to control the level of the dry signal. SURVEY will control the level of the input signal sent to the processing chain (pre-FADE) via the MAC mixer. The processed signal returns and is added back in via OFFSET. Additional signal can be added to the send loop via the other inputs in the left row.

OR/AND: Analog Max/Min and Digital Logic

OR implements analog maximum while AND implements analog minimum. The first input of the OR block is normalled into the first input of the AND block to easily find the min and max of the same signal. SURVEY is normalled into the second input of both blocks to set the comparison level for both functions together.

Logic block

OR/AND: Analog Logic

You can visually explore it by using this demo, which shows the two outputs when sine waves with varying frequencies are used as inputs. The sliders labeled "f1" and "f2" control the frequencies of the sine waves, while "spread" is used to visually compare the input waves and the output waves.

OR: Analog Maximum / Boolean OR

OR(OUT) will always output the greater of the two OR inputs, OR1 and OR2. This can be defined:

With 0V patched into the one of the inputs, OR(OUT) will be the halfwave rectified version of the other input.

When using binary (Boolean or two-state) signals like gates, this acts as the Boolean logic OR operator. The output will be high whenever either of the two inputs are high. For instance, when two rhythmic gate streams are patched to the OR inputs, OR(OUT) will output a gate whenever either of the two input gates hits.

OR1 OR2 OR(OUT) = OR1 | OR2

NB: the HIGH output voltage will be the same as the input gates' HIGH voltage.

AND: Analog Minimum / Boolean AND

AND(OUT) will always output the lesser of the two AND inputs, AND1 and AND2. This can be defined:

When using binary (Boolean or two-state) signals like gates, this acts as the Boolean logic AND operator. The output will only be high when both of the two inputs are high. For instance, when two rhythmic gate streams are patched to the AND inputs, AND(OUT) will only output a gate when both of the input gates hit simultaneously.


Other Boolean logic operators can be implemented via combination with the OR block and the Crossfader block acting as logical NOT. See the section on Extended Logic.

Control Signal Splitting

Because of the normalled connections, it is convenient to create a CV splitter which sends CV values above a threshold to one destination while CV values below the threshold are sent to another destination. With nothing patched, this is what happens to the SURVEY knob coming out of AND(OUT) and OR(OUT).

SURVEY is normalled into both OR2 and AND2. When nothing is patched to OR1 or AND1, both jacks are held at 0V. As such, OR(OUT) will output SURVEY when SURVEY is greater than 0V, otherwise it will output 0V. Similarly, AND(OUT) will output SURVEY if SURVEY is less than 0V, otherwise it will output 0V. Patch these two signals to two separate parameters in your synthesizer to control both parameters from the same source while only ever affecting one at a time! If you want both parameters to be swept from 0 to +5V, you might patch AND(OUT) to FADE instead of the second parameter; then patch RIGHT(OUT) to the second parameter. Now, the negative AND(OUT) signal will be inverted so that it goes from 0V to +5V instead of 0V to -5V.

To split an external CV source between two outputs, patch it to OR1 (it will cascade to AND1 due to normalling). SURVEY will now set the threshold comparison level at OR2 and AND2. When the input signal is greater than SURVEY, it will come out of OR1. When it is less than SURVEY it will come out of OR2. If you would like the control-splitting block to be independent of the SURVEY control, mult a threshold control signal into both OR2 and AND2.

Halfwave Rectification

By patching a signal into OR1 (and thus AND1 via the normalled connection as well), the SURVEY knob (via normalling into OR2 and AND2) becomes a threshold control for halfwave rectification - the "positive" half of the wave will appear at OR1, while the "negative" half of the wave will appear at AND1 (where "positive" is taken to mean exceeding the threshold and "negative" is taken to mean falling below the threshold). This technique can be useful when an input expects unipolar modulation or gates.

When used with audio, slicing the audio signal has the effect of adding harmonics due to the flat edges created. Adjusting the slicing threshold will affect both the timbre and the amplitude of the heard sound. If the input audio is a square wave, these blocks can act as simple VCAs with opposite polarities. Try a variety of sound sources and see how adjusting the threshold affects the timbre and loudness.

If the threshold is set to 0V, adding AND(OUT) and OR(OUT) back together will recreate the original signal. However if AND(OUT) and OR(OUT) are processed independently, interesting spectral effects will occur. Try filtering or distorting the "top" and "bottom" of the waves before recombining!

Logical Mixing

Patching two separate audio signals to OR1 & OR2 or AND1 & AND2 provides a unique way of mixing two audio signals with an additional harmonic distortion effect. The output will always track one of the two signals (the greater signal for OR, the lesser signal for AND), however two audio rate signals will constantly cross over each other, switching which one is the greater and which one is the lesser very frequently, resulting in the output switching between the two signals very rapidly. The effect is a kind of cross between halfwave rectification and using the signals to amplitude modulate each other.

The best way to understand it sonically is of course to try it out!

You can visually explore it by using this demo, which shows the two outputs when sine waves with varying frequencies are used as inputs. The sliders labeled "f1" and "f2" control the frequencies of the sine waves, while "spread" is used to visually compare the input waves and the output waves.

Logical Crossfading

OR(OUT) and AND(OUT) can be combined to create spectrally enriched crossfades between two audio inputs.

Patch one audio input to OR1 and the other to AND1. Patch AND(OUT) and OR(OUT) to a unity mixer/summing utility (e.g. patch into FADE and OFFSET then listen to LEFT(OUT)). SURVEY will control the "crossfade" between the two input signals (via OR2 and AND2).

When SURVEY is at -5V, OR(OUT) will be a copy of OR1, as OR1 is greater than SURVEY. Meanwhile, AND(OUT) will be -5V, as SURVEY is less than AND1. The sum of OR(OUT) and AND(OUT) will then be OR1 with a -5V offset.

Increasing SURVEY will allow more of AND1 to pass through to AND(OUT) and less of OR2 to pass through to OR(OUT). When SURVEY reaches 5V, the sum of OR(OUT) and AND(OUT) will be AND1 with a DC offset. Use this online demo to explore the waveforms generated by logically crossfading two sine waves with varying frequencies ("f1" and "f2" sliders control frequency).

This can be used for crossfades with spectrally rich centerpoints. Using an audio-rate signal to control SURVEY provides a new twist on the 'AM' sound.

SLOPE: Fullwave Rectification

SLOPE(OUT) is a fullwave-rectified version of the SLOPE input: all negative voltages are inverted, while positive voltages are left unchanged. Mathematically, the output is the absolute value of the input.

You can use this demo to see fullwave rectification in action! Here, a bipolar sine wave is used. Notice how it is reflected above the x-axis whenever it goes negative. Try using the "spread" slider to see how SLOPE(OUT) matches and differs from the SLOPE input.

SLOPE: Fullwave Rectification

Fullwave rectification is useful when only positive voltages are desired, e.g. for control of MANGROVE's FM INDEX.

Beyond CV conditioning, SLOPE can be used as a timbral effect on an audio signal, adding harmonics. This technique is at the core of the classic Hendrix 'Octavia' sound! With many waveforms, it acts as a frequency doubler. eg. sawtooth and triangle waves will be converted to triangle waves at double the frequency. Try using it with simple sound sources like sine waves as well as harmonically rich sources.

Biasing an audio signal (i.e. adding a constant DC offset to it) before sending it through SLOPE will provide additional control over the timbral effect of fullwave rectification. Try sweeping the DC offset to hear the changing timbre. The simplest approach is to patch a sound source to SURVEY, and listen to SLOPE. The SURVEY knob controls DC bias. Alternatively, patch a sound source to OFFSET, and patch LEFT(OUT) to SLOPE - now you can CV the bias via SURVEY. This demo shows the effect of biasing a signal before processing it through SLOPE. Notice how the symmetry of the two halves changes - this is what transforms the timbre... You are on your way to west-coast synthesis!

FOLLOW: Envelope Follower and Slew Limiting

The FOLLOW block takes the output of the SLOPE fullwave rectifier and passes it through a slew limiter. The slew limiter's time-constant is held constant for traditional envelope following duties.


FOLLOW: Slew Limiting

Whenever the input voltage changes, the output will slew to the absolute value of the input voltage (i.e. if it is negative, it will slew to the positive voltage with the same magnitude).

It is as if the output is laggy version of the input: it FOLLOWs the input as best it can and catches up to the input whenever the input remains constant.

Fun fact: It is the same envelope follower from the ARP2600!

Some use cases for CV inputs to SLOPE include adding portamento/glide to pitch or modulation sequences, smoothing out random sequences into undulating waves, converting triggers/gates to AR/ASR envelope, creating side-chain control voltage or a signal for taming feedback.

Envelope Following

FOLLOW is primarily designed for use as an [envelope follower](< An envelope follower generates a control voltage which tracks the loudness of an input audio signal: the louder the input, the higher the control voltage. Patch an audio signal to the SLOPE input to get started. The output will be a contour which tracks the volume of the the input SLOPE signal over time. The louder the input, the farther FOLLOW moves from 0V. It's like a control voltage VU meter!

This phenomenon allows you to use the loudness of one signal to control synthesis parameters anywhere in your system.

This is especially effective when using an external acoustic signal that has been patched into your system via a pre-amplifying module. FOLLOW converts events in the real world, like striking a drum head, into events in your synthesizer.

You can also try feeding the envelope follower back into the signal-being-followed's processing path for feedback. In this way it can be useful for taming feedback by sending an inverted copy to attenuate a feedback-amount.

When paired with a video synthesis system, this is a great processing block for audio-visualization techniques - for instance, making the colorization change whenever a kick drum occurs.

CREASE: Waveslicer

The CREASE block performs a form of wavefolding on the CREASE input; it is not a fold in the same sense as the classic Buchla and Serge wavefolders, but similarly it can be used for harmonic effects. Negative voltages are shifted up by +5V, while positive voltages are shifted down by -5V. This can be mathematically expressed as:

Or, more succinctly:

NB: Due to the implementation of this circuit, the amount added won't be precisely +/- 5V.


CV Processing

CREASE can be used to re-arrange your input modulation. All modulation will still move in the same direction, but the range of the modulation will be shifted depending on whether the input modulation is positive or negative. It also introduces dramatic discontinuities in the modulation as the input crosses 0V.

Audio Processing

When used with audio signals, CREASE adds harmonics to your input audio by introducing sharp discontinuities in the waveform. Contrary to most distortions, CREASE is most aggressive with quieter sounds, decreasing added harmonics as volume increases.

  1. Patch the same audio source (start with a sine wave) to the LEFT input of COLD MAC's crossfader section as well as to the CREASE input.
  2. Patch CREASE(OUT) to the RIGHT input of the crossfader section.
  3. Patch one of either LEFT(OUT) or RIGHT(OUT) to your synthesizer's audio output.
  4. Sweep FADE (e.g. by turning SURVEY with nothing patched into FADE), and hear the final output crossfade between the original sound source and the wavefolded sound source. Try patching an LFO or envelope to FADE!

Additional timbral control over the CREASE block can be added by biasing the audio signal (i.e. adding a DC offset to it) before sending it in to CREASE. As the bias voltage changes and the audio signal is translated up and down, the point where the audio signal crosses over 0V will change. This has a great timbral effect. When the bias voltage exceeds +/- 5V, the audio signal will always be above or below 0V, resulting in no wavefolding at all!

  1. Set SURVEY to 0V (knob at noon, no CV).
  2. Patch an audio source to the crossfader block's OFFSET input.
  3. Leave LEFT, RIGHT and FADE unpatched.
  4. Patch LEFT(OUT) to CREASE (LEFT(OUT) will be the sum of FADE and OFFSET).
  5. Patch CREASE(OUT) to your synthesizer's audio output.
  6. Sweep the SURVEY knob slowly to hear the effect of changing the bias via FADE.
  7. Patch an LFO or envelope to FADE to break the normal and give control of biasing to your CV source.

You can also use this demo to see the effect of biasing a signal before processing it through the various inputs!

Phase-Shifted Phasors

Patching a sawtooth or ramp wave into the CREASE input will result in a 180-degree phase-shifted copy of the input at the output. Beware (or exploit!) the 'glitch' that occurs in the middle of the shifted phasor.

LOCATION: Integrator

The CREASE input jack controls the LOCATION output. LOCATION outputs a voltage with a rate of change proportional to the input voltage. In mathematical or electrical terms, the output voltage is the integral of the input voltage. The output signal is clipped at max/min of +/- 5V.

The sign of the input voltage determines the direction in which the output voltage moves: positive inputs make the output voltage rise, negative inputs make the output voltage fall. 0V stops the output voltage at the current LOCATION.

The magnitude of the input determines how fast the voltage moves. The farther away from 0V, the faster the output voltage will move.

LOCATION: Integrator

Imagine an ant moving along a string. The string is marked with numbers to indicate its location. A tag labeled "0V" is hanging from from the midpoint of the string. Each end of the string is attached to a wall to stop the ant from moving beyond the boundaries; one wall is labeled "+5V", the other "-5V". There are additional tags hanging from the string indicating various LOCATIONs (e.g. "-4V", "-3V", "-2V"...)

The input voltage determines the speed (velocity) of the ant: which direction it is walking, and how fast it is walking in that direction. The output voltage is the LOCATION of the ant along the line!

If the input voltage stays positive for long enough, the ant will eventually hit the +5V wall and stay there as this ant can't climb walls. If the input voltage stays negative for long enough, the ant will eventually bump into the -5V wall and stay there.

If the input voltage is 0V, the ant's velocity is 0, and so it simply stays in place; the output voltage does not change from its current value.

Creating Gestures

LOCATION can be used especially effectively to create smooth gestures which affect an entire patch. Try patching LOCATION to multiple parameters around your synthesizer (including self-patching it to COLD MAC). With nothing patched to CREASE, LOCATION will be controlled by SURVEY. Try changing SURVEY from positive to negative to create a slow animation across your entire patch. Set the SURVEY knob to noon to freeze the gesture in an intermediate state. The speed of the gesture will depend on how positive or negative you set SURVEY, so dramatic transitions can be executed by quickly changing it from CW to CCW (or vice versa).

MAC: AC-Coupled Audio Mixer

MAC mixer

MAC is a utility for mixing audio signals patched to the 6 jacks on the left side of the module (LEFT, RIGHT, OR1, AND1, SLOPE, CREASE).

MAC creates an even sum of all inputs and then passes this sum through a VCA. The SURVEY voltage controls this VCA: -5V is fully closed, while +5V is fully open.

Because the MAC summing circuit is AC-coupled, any CV (i.e. slowly moving voltages) at the MAC inputs will be blocked from entering the MAC mix. This allows you to use some of the inputs for audio mixing through the MAC output, while using other inputs for CV processing using COLD MAC's standard utility blocks. For instance, you might patch audio signals to SLOPE and CREASE to mix into MAC while patching CV signals to LEFT and RIGHT to process using the crossfader. The CV signals would be blocked from entering the MAC mix.

OR1 is normalled into AND1. So, if AND1 is unpatched, any audio signal patched into OR1 will also be patched into AND1 and be added into the MAC mix twice. As such, it will be subjected to 2x gain relative to the other inputs. For OR1 to be mixed at the same level as the other inputs, patch a dummy cable (or another audio signal) to AND1 to prevent the OR1 signal from being added in twice. The same 2x-gain effect occurs with SLOPE and CREASE, since SLOPE is normalled into CREASE.

When you want to avoid this effect, patch your audio signals for the MAC mix into AND1 or CREASE before OR1 or SLOPE (respectively). However, you may also wish to exploit this effect! For instance, it can be very useful for quiet sound sources (e.g. sine waves) or for a sound which you want to sit higher in the mix (e.g. a kick drum).

Phase-Related LFOs

COLD MAC can be used to turn a single LFO into 8 related LFOs with different shapes, phases, or even frequencies derived from the parameters of the input LFO. Patch a source LFO into SURVEY while leaving all other inputs unpatched. Set the SURVEY knob to noon. Each output will apply the transfer function indicated on the panel to the SURVEY voltage, generating 8 related LFOs, one for each output.

SURVEY LFOs, ramp wave input

Use this link to explore what the 6 primary output LFOs would look like (FOLLOW and LOCATION are excluded from the demo). Use the "waveform" slider to select one of 3 input waveforms patched to SURVEY(sawtooth, triangle, sine). When nothing is patched to the crossfader block, LEFT(OUT) becomes a copy of SURVEY, so you can see the input waveform at LEFT(OUT). Use the "spread" slider to spread out the waveforms so that you can see them individually or overlaid on top of each other.

The FOLLOW and LOCATION LFO shapes will depend not only on the shape of the input LFO but also the frequency of the input LFO.

Adjusting SURVEY will bias the input LFO (offset it up or down) before it is processed by each block. For most of the blocks, this will have the effect of changing the symmetry. You can see this effect in the linked demo by adjusting the "bias" slider.

More complex waveshapes can be generated by modulating parameters of the input LFO or by using more alternative LFO waveshapes. Sending a sample+held LFO or random voltage into SURVEY is also a great way to get six related staircase sequences! Try patching them through quantizers for evolving harmonization.

Dynamics Control: Compression and Expansion

FOLLOW's loudness contour output can be used to implement dynamics control techniques like compression, expansion, side-chaining and ducking.


Imagine playing your favorite record through a stereo system, but you don't want it to get too loud. Whenever the record gets louder than a certain threshold, you turn the volume down. Once the record is quieter than the threshold, you turn the volume back to its original level. Compression is the process of automating this gain control. The amount the volume changes when the threshold is crossed is called the ratio (ratio = input/output). Turning the volume down above the threshold is downward compression: the dynamic range is made smaller by making the loud sounds quieter. The dynamic range can also be made smaller by making the quiet sounds below the threshold louder. This is known as upward compression.

Expansion is the process of increasing dynamic range: either making loud sounds above the threshold louder, or quiet sounds below the threshold even quieter.

Side-chaining means using the loudness of one signal (the "side-chain") to affect the loudness of another signal. This technique is commonly used in electronic music to create a pulsing space that follows the kick drum.

FOLLOW outputs a CV contour which tracks the loudness of the SLOPE input signal; this contour can be used to construct a compressor when paired with COLD MAC's VCA functionality and CV processing utilities.

Expansion/Noise Gate

First, unpatch all inputs. Patch the audio signal you want to expand into SLOPE. Patch FOLLOW to SURVEY. Set the SURVEY knob around 9 or 10 o'clock. Use MAC as the final audio output. As the input signal gets louder, FOLLOW will get larger. This will turn SURVEY up and increase the volume of the input being sent to the MAC output. This is known as a "noise gate" because it allows loud sounds to pass through by turning the gain up while leaving the gain very low when the input is quiet. This can be useful for making input sounds more "staccato" or for processing an external signal from a microphone which may be picking up unwanted ambient noise at a low level.

Adding in other signals to the MAC mix via the left column input jacks will subject them to the same expansion using the original SLOPE input as the control signal (a.k.a. side-chain control). This can be especially effective when you want to use an external instrument like a guitar to "pluck" or "gate" several elements in your mix.


The only change necessary from the patch above is to invert the FOLLOW envelope by patching it to FADE first; RIGHT(OUT) will now be an inverted copy of FOLLOW and should be connected to SURVEY. Set SURVEY between noon and 3 o'clock. Now, as the SLOPE input gets louder, FOLLOW becomes more negative after it passes through the inverter. This will close the MAC mix instead of opening it.

NB: Make sure LEFT and RIGHT are unpatched so that the FADE inverter works correctly.

Adjusting the Ratio

To adjust the ratio, or the amount the sound changes based off the loudness of the input, patch the FOLLOW signal through an attenuator. Patching it through an attenuverter will allow you to switch between compression and expansion.

FADE: Compression, Expansion, and Sidechaining

The FADE block can also be used simultaneously for both compression and expansion. Patch the signal to process to both LEFT and SLOPE. Alternatively, patch the signal to process to LEFT and the side-chain control signal (like a kick drum) to SLOPE.

Patch a dummy cable to RIGHT. Patch the FOLLOW output to SURVEY, and leave FADE unpatched. Set SURVEY around noon. Use LEFT(OUT) as the audio output.

Now, as the SLOPE signal gets louder, FOLLOW will increase. Since it is patched to SURVEY and FADE is unpatched, FADE will increase as well. As FADE increases, the LEFT signal will get quieter at LEFT(OUT) (compressed) and louder at RIGHT(OUT) (expanded). SURVEY sets the default volume (or makeup gain) when the SLOPE input is silent. If you are looking to side-chain a sound to a kick drum, patch the sound to LEFT, the kick to SLOPE, and take the output from LEFT(OUT).

If you are sidechaining, you may wish to to mult the side-chain control signal into OFFSET in order to mix it with the sidechained signal feeding LEFT.

OR/AND: Threshold Control

The logic blocks can be used to add a threshold control. Patch the FOLLOW output to OR1 and leave AND1 unpatched (OR1 will remain normalled into AND1). Now, SURVEY will set the threshold via OR2/AND2. When the SLOPE input signal is louder than the threshold, FOLLOW will come out of OR(OUT). When it is quieter, FOLLOW will come out of AND(OUT). If you are looking to make loud sounds quieter (downward compression) or loud sounds louder (upward compression), use OR(OUT) as your gain control. If you are looking to make quiet sounds louder (upward compression) or quiet sounds quieter (downward expansion), use AND(OUT) as your gain control. You will likely want to offset and attenuvert the signal before sending it to the VCA control input. These will adjust makeup gain and ratio respectively.

If you wish to use SURVEY to offset your gain control and control the FADE or MAC VCA (as in the previous examples), set the threshold at OR2/AND2 using an external CV offset rather than the normalled SURVEY control.

Extended Waveshaping

Complex wavefolding networks are possible by self-patching COLD MAC.

Here are some key techniques which can be combined in a modular fashion to create more new timbres. Try using LFOs, envelopes, staircase random sequences, and even audiorate modulation to control any of the CV parameters detailed below.

Although the language below references audio rate input signals for folding, try using control rate signals to slice into new modulation shapes. Try sending these complex modulations to a sample+hold/quantizer to create harmonizing and evolving pitch sequences from elementary input sources.


Although this does not change the quality of the wavefolding, blending is a simple way of adjusting the intensity of the harmonic distortion.

Use the crossfading block to blend between a wavefolding sound and the dry sound! For instance:

Audio Input > LEFT & CREASE (multed)


LEFT(OUT) > Audio Output


Since wavefolded sounds often lose the fundamental, it can be useful to mult the audio input to OFFSET so that it is still present in LEFT(OUT) alongside the folded version when FADE is at +5V.


Connecting multiple waveshaping blocks in series (one feeding the next) can create rich new waveforms. Here's an example which combines SLOPE and CREASE:

Audio Input > CREASE


SLOPE(OUT) > Audio Output


Patch two waveshaping blocks in parallel and mix their outputs together - or use one of the outputs to modify the waveshaping of the other!

This patch simply takes the sum of AND and CREASE processing the same input. SURVEY will affect the nature of the AND block wavehsaping.

Audio Input > CREASE & AND1



LEFT(OUT) > Audio Output

LEFT, RIGHT, and AND2 must remain unpatched.

SURVEY adjusts timbre.


Biasing the signal will alter the effect of wavefolding as it changes where a signal lies in relationship to the various folding thresholds.

Use the crossfading block to bias (add a constant voltage) to a signal before sending it to a wavefolding network! For instance:

Audio Input > OFFSET


SLOPE(OUT) > Audio Output

Bias CV > SURVEY/FADE controls biasing

LEFT and RIGHT must remain unpatched.

The FADE block can be used to combine crossfading and biasing as well by setting up a dry/wet loop:

Audio Input > LEFT



LEFT(OUT) > Audio Output



Here, when FADE is at -5V, the audio output is just the audio input + the bias CV as LEFT passes directly to LEFT(OUT). When FADE is at +5V though, LEFT is first first routed through the crossfader block to RIGHT(OUT) and mixed with the bias CV before being sent to SLOPE. SLOPE(OUT) then return the folded signal to SLOPE, which gets routed to the final audio output at SLOPE(OUT).

This patch demonstrates using one parallel waveshaped signal to bias the other waveshaping path.

Audio Input > FADE and AND1



CREASE(OUT) > Audio Output


SURVEY can be used to control multiple waveshaping parameters simultaneously, like biasing and blending:

Audio Input > LEFT



LEFT(OUT) > Audio Output


OR1, OR2, and FADE must remain unpatched.

SURVEY will affect both the blending (via FADE) and biasing (via OR(OUT) > OFFSET). From -5V to 0V, SURVEY will blend from the clean sound to an even mix of the clean and folded sounds. OR(OUT) will be stuck at 0V since OR1 is unpatched, so no biasing occurs. As SURVEY sweeps from 0V to +5V, it completes the crossfade to the wavefolded sound. However since SURVEY is now greater than OR1, OR(OUT) will track SURVEY and begin biasing the signal as it passes through the wavefolding loop. This has the effect of transforming the nature of the fold as the blending progresses, resulting in a multi-dimensional timbral change with just a single knob.

The following patch extends the above macro by introducing the AND block in series.

Audio Input > LEFT




LEFT(OUT) > Audio Output


OR1, OR2, AND2, and FADE must remain unpatched.

Now, in addition to controlling biasing and blending, SURVEY also will change the wavefolding by setting a comparison threshold for the AND block. You should also try overwriting SURVEY's normals with different LFOs, envelopes, and sequences to hear what happens when the parameters are controlled independently!

Feedback Sounds

Self-patching COLD MAC can lead to many unexpected feedback results. FOLLOW can be used to track the loudness of feedback; pairing this contour with FADE can help keep feedback manageable.


LEFT > CREASE (yes, input to input!)


LEFT(OUT) > Audio Output (Square Wave)

MAC > Audio Output (Sine Wave)

SURVEY or FADE controls frequency of oscillation.

Try using a stackable mult to connect LEFT & CREASE together. Patch a cable into one end of one of the stackables - try touching the end of the cable to change the timbre of the sound.

Try patching CREASE into other inputs, or MAC/RIGHT/LEFT into other inputs simultaneously and the timbre of the VCO will change.

Good luck explaining why this one works!

Mid-Side Processing and Width Control

Mid-Side Processing is a technique that allows you to re-organize how you process sounds in a stereo field. Rather than processing the left and right sides of the field separately, sounds in the middle of the field are placed in one channel and sounds on the outside of the field are placed in the other channel. This allows the user to apply processing to sounds depending on how far they are from the middle. For instance, it allows the user to apply delay to sounds that exist at the edges of the stereo field while compressing sounds in the center of the field. One particularly effective use of M/S processing is to convert a mono effect into a pseudo-stereo effect by creating a sense of space: by applying the mono effect to the Side signal and adjusting the gain of the side (before or after processing, depending on the use case) allows the mono effect to appear more strongly at the edges of the stereo field and less intensely for sounds in the middle of the field, giving a sense of width.

The Mid signal is created ("encoded") by summing (mixing) the Left and Right signals:

The Side signal is created ("encoded") by taking the difference of the Left and Right signals. If a sound is in the middle, it appears equally in both L & R, and thus will be cancelled out when creating the Side signal:

To recover ("decode") the Left channel, add the (post-processed) Mid and Side signals:

To recover ("decode") the Right channel, subtract the (post-processed) Mid and Side signals:

Both the L & R signals have been perfectly recovered without any loss of information! Notice that the gain of the Left and Right signals has doubled. Setting different attenuations/boosts for the Mid and Side signal will result in different stereo width perceptions.

To realize a full M/S processor requires two COLD MACs, however it can also be done with just one COLD MAC plus another external summing/inverting utility.

CM1 refers to the first COLD MAC, while CM2 refers to the second COLD MAC.


With nothing patched to CM1's LEFT and RIGHT inputs, LEFT(OUT) will be OFFSET+FADE while RIGHT(OUT) will be OFFSET-FADE. This is used to encode the L/R signals to M/S by assigning the L signal to OFFSET and the R signal to FADE.

Left channel > CM1.OFFSET

Right channel > CM1.FADE

CM1.LEFT(OUT) = Mid = Left + Right

CM1.RIGHT(OUT) = Side = Left - Right


Send CM1.LEFT(OUT) and CM1.RIGHT(OUT) to your independent Mid and Side processing chains respectively. For instance, perhaps you want sounds at the edges of your stereo field to have reverb added while sounds in the middle should have delay added. These will then be recombined to recover the resultant Left and Right channels.

Adding Width Control:

If you want to use COLD MAC to control the sense of space, patch the Mid or Side signal to one of the four unused inputs and take the output from MAC. SURVEY will now control your Mid or Side gain, depending on which signal you chose.

When used with the SIDE signal, SURVEY becomes a stereo width control as it adjusts the MAC gain. You might choose to do this before processing the M/S signal or after. -5V corresponds to no width, while +5V corresponds to full-width.

You can also patch your pre- or post-processed Side signal to any other VCA for stereo-width control.



Side > CM2.FADE

CM2.LEFT(OUT) = Left channel output

CM2.RIGHT(OUT) = Right channel output

You can also implement the decoding with mixers and inverters if you do not have another COLD MAC. Use the previously outlined method for stereo-width control by patching your side signal through the MAC VCA (either pre- or post-processing).

Decoding variation:

This method allows you to free the other COLD MAC inputs and lets you control stereo width using the FADE input of the second COLD MAC. It does however require an additional inverter... perhaps from a third COLD MAC!


Side > CM2.LEFT

Side > Inverter > CM2.RIGHT

Stereo Width CV > FADE (or SURVEY, normalled into FADE)

CM2.LEFT(OUT) = Left channel output

CM2.RIGHT(OUT) = Right channel output

The side signal will cancel with its inverted copy when the Stereo Width CV into FADE is 0V, while full stereo width will occur with a CV of -5V or +5V. This method will result in phase cancellation when summed to mono, so be careful if you are playing out or your track is being summed to mono without your knowledge!

Bernoulli Gates and Event Probability

COLD MAC can be combined with a noise source to create a pair of complementary events. A Bernoulli trial is a test which has exactly two possible outcomes, like flipping a coin. A probability p determines the likelihood that one outcome will occur, while 1-p determines the probability that the other outcome will occur. The events never occur simultaneously, but one of the two events always occurs, so they are "complementary".

In the modular synthesis context, a "Bernoulli gate" determines which of two synthesis events will occur based off a probability control voltage. Examples of possible events including advancing different sequencers, striking drum modules, and exciting envelope generators. By changing the probability, you can slowly morph from one event dominating to the other. This can be useful for creating evolving hocketing patterns or stochastic variation within a patch.

This patch uses the crossfader block to calculate the difference between a randomly generated voltage and a probability CV threshold. If the random value is greater than the threshold, the difference will be positive. If the random value is less than the threshold, the difference will be negative. This difference is patched into both logic blocks with a comparison voltage set to 0V; when the random value is greater than the threshold, a positive voltage will come out of OR(OUT), otherwise, 0V will come out. When the random value is less than the threshold, a negative voltage will come out of AND(OUT), otherwise, 0V. AND(OUT) is patched to SLOPE for rectification.

So, whenever the random voltage updates, one of the two outputs will output a positive voltage while the other outputs 0V. SURVEY/FADE controls the probability balance between which output will be positive and which will be 0V. While a positive voltage is sent into SURVEY/FADE, the balance is weighted towards SLOPE(OUT) being positive, while a negative voltage into SURVEY/FADE will weight the balance towards OR(OUT).


Noise source (e.g. a S+H white noise sequence) > OFFSET

RIGHT(OUT) > OR2 & AND2 (stackable)

OR(OUT) > Gate input of Envelope Generator 1 or Sequencer 1


SLOPE(OUT) > Gate input of Envelope Generator 2 or Sequencer 2

SURVEY (or FADE) controls probability balance.

NB: If you do not have a stackable, you can patch RIGHT(OUT) to OR1 and use a cable to patch OR2 and AND2 together, breaking the normal from SURVEY.

You may wish to offset the two outputs by a small positive voltage (just less than the target input's gate/trigger threshold sensitivity). This will ensure that whenever the random voltage crosses the probability threshold, the final output voltage is high enough to cross the target's sensitivity threshold.

If you are using a sample & hold source, it may help to multiply the two outputs (OR(OUT) & SLOPE(OUT)) by multed copies of your clock source. This will force both outputs to go to the low state between every clock pulse. This is important for being able to retrigger AR envelope generators, but is unnecessary for ASR generators.

If you are using a white noise source (as opposed to sample & hold), you will want envelopes with instant attack, and gate-sensitive ASR envelopes or retriggerable AR envelopes.

Quadraphonic Panning and Vector Mixing


This panner allows you to pan a signal between 4 destinations positioned at the corners of a square, according to a cartesian <x,y> location in the 2-D plane. It requires 3 COLD MACs.

This is useful for spatializing sound in a quadraphonic speaker setup, but it is not limited to this interpretation of the 4 outputs. Maybe the destinations are not speakers in real space, but rather four different processing chains.

Quad Panner

COLDMAC1 pans the signal forward/backward. COLDMAC2&3 works together to pan the signal left/right while maintaining the forward/rear balance. You will want to use a dummy cable for the RIGHT inputs on each COLD MAC to eliminate DC offsets.

When <x,y> is <-5V,-5V>, the signal should be panned hard left and hard rear - i.e. almost entirely out of the Rear Left speaker. Similarly, <+5V,+5V> corresponds to the Front Right.

Voltage <x,y> Quad Panning Positions

Vector Mixer

Alternatively, you can flip the three COLD MAC's to create a monophonic mix of four signals arranged in 2D space - a vector mixer like the Prophet VS.

Vector Mixer

COLDMAC1 & COLDMAC2 select the weighting of the inputs along one axis, while COLDMAC3 selects the weighting of the inputs along the other axis.

This is useful for additive synthesis and harmonic oscillator (e.g. tuned sine wave) techniques as well as evolving pads and chords. It can also be useful for creating new timbres by blending together four different elementary waveshapes in different amounts but at the same frequency. Other use-cases of the vector mixer include sending variable amounts of four sound sources to an FX processing chain, or blending between different CV modulation sources for a single destination parameter.

Voltage <x,y> Vector Mixes

Extended Logic

AND and OR are useful for gate logic. They can be used to "filter" & combine gate patterns and create accenting gates. You cannot create "new" gates though, only filtering or accenting gates which already occur. By adding in the crossfader block, you can create new gates where none previously occurred. This can be used to create hocketing patterns as well unique, hybridized digital rhythms. Try using JUST FRIENDS as a clock divider (or generator) with variable pulse width and combining different divisions with the following logic blocks! These videos will give a great idea of just some of the possibilities available with logic!

NOT Operator

The NOT operator logically inverts all gates: HIGH goes to LOW and LOW goes to HIGH.

With nothing patched to LEFT or RIGHT, patch a +5V signal to OFFSET. This creates the following transfer function for FADE:

When FADE is 0V, RIGHT(OUT) is +5V, and when FADE is +5V, RIGHT(OUT) = 0V. This then is logical inversion, or the Boolean NOT operator. When a GATE is now patched to FADE, RIGHT(OUT) will be the result of applying the NOT operator to the GATE input.

0V +5V
+5V 0V

If you do not have a +5V reference at hand, you can turn SURVEY fully CCW or CW and patch SLOPE(OUT) to OFFSET.

NB: If 2.5V is sufficiently high for your gate input sensitivities, then you can place a dummy cable into RIGHT instead of sending a +5V offset into OFFSET. When the FADE gate is at 0V, RIGHT(OUT) will be 2.5V, and when the FADE gate is at +5V, RIGHT(OUT) will be 0V.

NAND Operator

The NAND operator outputs a gate as long as both inputs are not simultaneously high - in other words as long as the result of applying the AND operator to both inputs is not true: A !& B = !(A & B)

This combines the AND section the NOT patch to hybridize two gate patterns into a third, new pattern.

GATE1 and GATE2 are patched to AND1 and AND2. AND(OUT) is used as the gate source for the NOT patch, which will now output GATE1 NAND GATE2.

Since any Boolean expression can be built with a sufficient number of NAND operators, you could build an entire digital computer capable of executing programs and storing data in RAM using COLD MACs. Let us know if you would like to do this, and perhaps we could provide you with a few of the hundreds of millions of needed COLD MACs for free.

0V 0V +5V
0V +5V +5V
+5V 0V +5V
+5V +5V 0V

NOR Operator

The NOR operator only outputs a gate if both inputs are low - in other words as long as the result of applying the OR operator to both inputs is not true: A !| B = !(A|B)

This combines the OR section with the NOT patch to create a new gate pattern out of two component gate patterns.

GATE1 and GATE2 are patched to OR1 and OR2. OR(OUT) is used as the gate source for the NOT patch, which will now output GATE1 NOR GATE2

0V 0V +5V
0V +5V 0V
+5V 0V 0V
+5V +5V 0V

XOR Operator

The XOR operator only outputs a gate if exactly one of the two inputs are high, but not both. A XOR B is equivalent to the following expression: (A|B) & !(A & B) = (A|B) & (A !& B)

GATE1 > OR1, AND1 (normalled, no mult needed)

GATE2 > OR2, AND2 (mult/stackable needed, or patch to SURVEY CV with SURVEY knob at 0V)




Whenever both gates are 0V, FADE and OFFSET are 0V since both AND(OUT) and OR(OUT) are 0V. This results in RIGHT(OUT) = OFFSET-FADE = 0V.

When one of the two gates is +5V, OR(OUT) and OFFSET will be +5V but AND(OUT) and FADE will still be 0V, resulting in RIGHT(OUT) = OFFSET-FADE = +5V.

When both gates are +5V, OR(OUT) and OFFSET will be +5V, but so will AND(OUT) and FADE, resulting in RIGHT(OUT) = OFFSET-FADE = 0V.

0V 0V 0V
0V +5V +5V
+5V 0V +5V
+5V +5V 0V

Further Reading

For an alternate take on COLD MAC, we strongly recommend Martin Doudoroff's Patching Cold Mac guide, along with a great YouTube tutorial. A number of ideas in this Map borrow from Martin's work!


No matter if you just bought your first module, or found a grimey old relic, we're committed to keeping it running in top form.

Formally, Whimsical Raps warrants each product to be free of defects in materials or construction for a period of one year from the date of original purchase (proof of purchase required). Malfunction caused by abuse of the product, or any other cause that we determine to be the user's fault, will not be covered by this warranty.

Formalities aside, if you've got an issue we absolutely want to help sort it out. Send us an email and let us know how we can help. Our service fees are only our direct costs, and we typically ask you to pay shipping one-way. Here's to keeping synths out of landfill!