Analog Modular

PugixWelcome to my DIY synthesizer website. The name Pugix (and the graphic toon) belonged to an online gaming character that I used to play and has nothing to do with synthesizers.

My most recent activity:


Here’s a Cocoquantus piece.  It is a single sample from the Sidrax, made to circulate, with the coco time delays being controlled from the Quantussy.  I intervened a few times to touch the nacel to change the lights (the internal interactions).  The noise is part of the piece.

Rollz-5 Experiments

First a dark drone that uses AVDogs and Ultrasounds

Next a rhythmic section created by Gongs with an external digital echo.

Next a combination of external processing with E560 and E580, plus Cocoquantus.

Rollz-5 Completed

Here’s the assembled Rollz-5 project!
rollz-5-patched-runningThe whole assembly can operate outside of its box.  Here it is being tested.  To the right you can see the power inlet, ground jack, and left and right quarter-inch outputs.  Below those, suspended beneath the clear panel, is the hand-wired board containing the voltage regulator (Recom R-78C9.0-1.0) and output mixer.  The single green jack goes to an LED driver and LED on that board for test monitoring.

I described the panel layout in more detail on this previous post.   On the left are the eight Rollz LFO square wave oscillators, each with a separate fixed frequency.  From top (red) to bottom (green) they go from fastest to slowest at non-harmonic intervals.  Next to the right are the four AVDogs with their associated square wave audio oscillator.  The oscillator output is directly available on the panel.  Next come the Ultrasound modules, and finally the four Gongs.  The six gray output mixer knobs are on the right.

Frequency Ranges

The measured frequency range of the manual square wave oscillators is 50 Hz to 2.1 KHz.  The Ultrasound oscillators are 750 Hz to 30 Khz.  And (a bonus from adding inputs and outputs to the Gong) when the Gong output is patched to its own input, it becomes an oscillator with a range of 95 Hz up to 2.1 Khz.  The upper frequency of the Gong is limited by a series resistor installed on the PC board, which I made different values from 47K to 470K.  I will be changing the 470K to a smaller value, since it limits the upper frequency on that Gong to 220 Hz.


The added inputs and panel outputs have really enhanced the patch flexibility of this board.  By accident I discovered that patching the output of an AVDog to the Gong aux input and the Gong out to the AVDog aux input makes an interesting gated oscillator.  The Gong begins oscillating as the AVDog VCA opens up and then stops as it closes.  Since the AVDog can have a very long envelope (several seconds long) that repeats, it results in a repeatedly rising and falling drone.  And you can patch four of them all at different frequencies.  All the pitches are manually tuned, so it can sound melodious if desired.

The AVDog, Ultrasound, and Gong can be patched in series in any combination with the Oscillator inserted at whatever starting point you like.  For example I can route an Oscillator through a Gong then through AVDog.  The switch on the AVDog input selects between the Oscillator and whatever is patch to the violet aux input jack.  Being able to switch this input is useful during a performance without re-patching.  A 10K ohm series resistor on each output allows for stacking outputs together to one input.  I found when using the Gong aux input that if the Gong is also being triggered via the green jack, the aux input tends to overwhelm the ringing, but I can patch in one of my in-line attenuators to remedy this, if needed.  I think I’ll probably use a Gong in either the ringing bongo mode or the filter mode.

The two brown Node jacks on the Rollz LFO allow ample opportunity to link it to other LFOs.  The brown jacks generate short, negative-going (below zero volts) pulses that can be used to trigger AVDog and Gong and also jog Ultrasound.  The orange output is a positive-going square wave, like the Rollz on Plumbutter. Using it as a trigger has subtly different effects on the AVDog and Gong.

Needless to say, I have plenty of experimenting to do with all this patching variety.

More Photos

A closeup of the CGS LED driver board.  These were perfect to use here.  Notice the notch, cut with a Dremel tool, to straddle the solder lug on a banana jack.  I used water-clear high efficiency LEDs, the same ones as used on the Quantisise, because of their low power consumption.  I matched the current resistor with each color to get a more or less uniform brightness across the four colors.


Here are some photos of the assembly, prior to folding it together.  This first one is just after wiring up the Rollz circuits.


This one shows all the wiring.  I estimate there is more than 200 feet of wire.  It took several days of work to wire it all up.  I made one wiring mistake, reversing the connections to the frequency and Q pots on two of the Gongs, because I didn’t pay close enough attention to the asymmetric layout on the board.  But this took only swapping six wires to correct.  This pic shows the incorrect wiring!


And here is how it looks secured into the travel case, a standard tool/gun case.  It pops in and out of the case with just four large screws.  There is room left to store cables and the wall wart power supply for carrying.  It’s waterproof, too!


Naturally I will be posting some recordings.  I also plan to get a video camera soon, so I can make some demonstration videos of this and other projects.

Rollz-5 First Tryout

The Rollz-5 is finished!  Full update to come.  Here is a little tryout recording.

Rollz-5 Assembly Progress

The Meng Qi Rollz-5 PC board project in mid-assembly stage.
rollz-5-prewiringThe image above shows the PC board mounted behind a clear plexiglass panel that holds pots, jacks, switches, etc., sitting inside of the yellow case that will hold everything.  No panel wiring yet!   All wires will be visible behind the panel, going to the PC board.  An additional hand-built board will mount to the right and hold a voltage regulator and output mixer.  Here’s a pic of just the panel.

rollz-5-panelI’ve enhanced the PC board to provide auxiliary inputs and outputs, more panel controls, plus an output mixer, to make it into a small, self-contained synthesizer.  To the left are the eight 2-Rollz LFOs, each with two brown jacks for the nodes, an LED and an orange jack for the positive going output pulse.  Each LFO has a different, fixed rate.  Next to the right are four doubles rows of jacks and controls, consisting of 1 AVDog, 1 Ultrasound, and 1 Gong for 4 total of each module.


  • Green jack – Input pulse to trigger the envelope
  • White jack – VCA out
  • Violet jack – Aux VCA in
  • Yellow jack – Oscillator out
  • Red knob – Envelope rate
  • Blue knob – Envelope Q
  • Yellow switch – selects internal oscillator or aux input to the VCA
  • Yellow knob – Oscillator frequency


  • Green jack – In
  • White jack – Out
  • Yellow jack – Oscillator out
  • Yellow knob – Oscillator frequency


  • Green jack – Input pulse to ring the filter
  • White jack – Filter out
  • Violet jack – Aux filter in
  • Green knob – Pulse delay time
  • Red knob – Filter frequency
  • Blue knob – Filter Q

To the far right are six gray knobs.  These are the output mixer.  Each knob controls the level of two devices, for example the top knob sends the topmost AVDog to the left output and the second AVDog to the right output.  Ultrasounds and Gongs are also paired.  This is like the output mixer on Plumbutter.  Note that the outputs are independently available at full level on the white jacks for patching.  This is also similar to Plumbutter, but the Ultrasounds also have white jacks here.  Note the lack of CV inputs.  Rollz-5 is not voltage controlled (except for the internal VCA in AVDog).

rollz-5-whole-board-unwiredAbove is a pic of the mostly assembled PC board.  Chips not inserted yet.  If you look closely (click the image) you can see that the traces from the 12 outputs to the internal mixing buss have been cut.  The black/red wire to the right goes to a connector that brings in regulated 9V power.  Power starvation feature is unused. Here are more details of each section.

rollz-5-rollz-detailShows six of the eight 2-Rollz oscillators.  Jumper wires shorten the nodes to be only two, since I only care about simple LFOs with two nodes each.  Capacitors range from 0.33 uf through 4.7 uf to get the rates I wanted.  These are faster than the Rollzer from Ciat-Lonbarde.  External LED drivers will be mounted behind the panel.  I also provide a positive going output, like on the Plumbutter Rollz.

rollz-5-avdog-detailHere is one of the four AVDog circuits.  The ‘x’ resistors will be wired to the rate pot on the panel.  The 4k7 resistor seems missing, but isn’t.  I couldn’t get 1/8 watt 4k7 resistors, so 1/4 watt resistors are soldered on the back, because the leads are too big to fit the holes in the PC board.  A 100K is missing, because that is where the input to the VCA will be switched in.

rollz-5-ultrasound-detailAbove is the Ultrasound detail.  Two circuits share one 4066 CMOS chip.

rollz-5-gong-detailAbove is the Gong detail.  Two circuits here share a 4013 CMOS chip.  Trimpots will be replaced by panel pots for adjusting the filter frequency.  The aux input to the filter will mix with the pulse, just like the aux input to Gong on Plumbutter.


Inline Attenuators

When patching to CV inputs on Ciat-Lonbarde instruments (Cocoquantus, Plumbutter, Sidrax) it can be useful to add some attenuating ability.  For example, if I want to control the frequency of a Sidrax oscillator from a Castle output of Cocoquantus, I want to vary the depth of modulation.  There’s no attenuator on the CV input.  Hence:  Inline Attenuator in a small box with two banana jacks.



The little box comes from Small Bear Electronics, part number 1090NS.

This is not a potentiometer, i.e. there is no ground connection.  It is simply an inline variable resistor.  It also works well for patching a CV coming out of the modular into a Ciat-Lonbarde instrument.  The CV inputs on Ciat-Lonbarde are very high impedance (often 1M ohms).  And they expect a relatively high output impedance (10K ohms) to drive them.  If you patch a very low impedance CV (like the 1K standard for modulars) or a pot with a grounded end, the current is pulled down so much that control can be ineffective.  So here I have used a 1M audio taper pot as a series resistor.  It can go down to zero resistance.  It’s wired so that as the knob is turned clockwise, the resistance decreases.  Full anti-clockwise does not shut off the CV, but only maximizes the inline resistance to 1M.