Three years

Three years ago, when I started this blog, I wanted to share my experience of making a modular synthesizer. First in English only, the blog contained electrical schematics and technical explanations. A few months later, I decided to create a French version to share this journey in the world of modular synthesizers with more people around me. The French version contains less schematics and more sonic illustrations. The articles are pretty much the same, but the two blogs are not 100% identical.



This is the 36th article on this blog. I kept on with my goal to publish one article per month on average. Some months were more prolific than others though.
The blog has now more than 19000 views and 10 comments. 21% of the audience comes from the USA.  The most popular article is the one on the Output module with 1200 views.


Since August 2018, Bob, a little Lego figure, accompanies me on every article about building something.  Initially relegated to cleaning tasks, Bob was able to show some skills in drilling or module tuning.  His small size makes him a valuable partner for delicate work.
 

There were so many evolutions between the first family photo in September 2017 where my first modules are finally assembled in a 3U Eurorack synthesizer row beginning and now with almost two finished rows of 84HP.  The synth includes, from top to bottom from left to right, One ADSR envelope generator, one CV quantizer, two oscillators, two filters, four VCA, one phaser effect, one output module, one sample & hold with noise generator, one slew limiter, two attenuators, a mixer, a voltage reference, a double multiple, one chaotic modulation generator, a signal multiplier with CV offset, a double modulation generator, one random generator and a resonator.


First family photograph, September 2017 :

And nearly 3 years after....

Some of the big steps are detailed here below :

- my first module (August 2017 - built on November 2016) : you have to start somewhere

















- the case (February 2018) : modules have a home now









- first kits (May 2018) : I now have nice modulation sources

- repairing my VCO (July 2018) : I finally have a tuned sound source












- Doepfer DIY (September 2018) : comes with my first filter









- Replacing the Doepfer DIY with separated elements (early 2020) : new ADSR, new VCO, new filter.








I learned a lot on this journey into the world of do-it-yourself synthesizers.
First of all, I improved my brazing technique.  I can now make denser builds than before.
I've learned how to design module panels.  I also tested and approved new suppliers, such as Thonk.
But more importantly I learned how to use the synthesizer, how to use it to create generative music, how to create different types of sounds.

I used the synthesizer to make 21 tracks.  Mostly as illustrations of the function of a module for the blog, but not only :
- 11 sound examples of the module at hand;
- 7 sound illustrations of the state of the synthesizer;
- 2 participations to the monthly KVR Music Cafe;
- 1 participation to Audiofanzine Inspired Composition (Compos Inspirées).

Jay F. · Modular Suitcase

And here is the track of the day :

Jay F. · Comet Dust

All sound from the modular synthesizer.  Too many tracks to detail them all. 
Most tracks were performed in a very traditional way for me : first writing and recording a MIDI track, then programming the synth sound and recording the audio using the Korg SQ-1 sequencer as a MIDI to CV converter.  Exceptions : drums were made using the SQ-1 as the sequencer and percussion were generated by the Turing Machine.
DAW : Reaper

Sloth Chaos

Ready to be mounted in synth

In a chaotic system, tiny variations of the initial conditions lead to apparently random states of disorder.  A chaotic circuit in a modular synthesizer is the promise of slowly changing, never repeating ambient soundscapes.

This module produces two different, very slow, chaotic CV signals. It is my implementation of the Non Linear Circuits Sloth Chaos.  I chose the regular version with 1 cycle every 15-20 seconds.  Depending on component values, other versions include Apathy (about 1 minute), Super sloth (15-20 minutes) and Stasis (about 1.5 hour !).

I needed to accommodate the values to match what I have on my drawers.  I tested on a breadboard to see if little changes on some components values would make a difference.  It turned out that yes, it does make a big difference.  It looks like there is a pretty good balance between the feedback and the small signal, going to the integrator for the big signal.  The torpor potentiometer plays with that balance.  But go with 20% more or 20% less resistance on that path and you end up with either two synchronous, though out of phase and saturated signals, or two chaotic but very small signals.

Experimenting

The butterfly pattern on the oscilloscope is called a strange attractor (or Lorentz attractor) and is the plot of the two signals out of the circuit, each one on a different axis.  When the signals are roughly in synch they form an ellipsis.  Two wells of pseudo equilibrium are visible.  The signals spend a bit of time there before drifting apart.  None of this is predictable and highly depend on initial conditions.  Hence the chaotic nature of the circuit.

I did not have 91k resistors, 82k + 10k  would do the trick.  100k was too high, 82k too low.  Same with the 4.7M, replaced with 2x 2.2M.

I also made a small change in the original schematics.  I wanted to have the LED be less than a negative/positive indicator and follow the signal a bit more.  So I chose the output of the buffer after the small signal output instead of the output of the following opamp.

Layout

Adapted schematcis

Almost there

The main board is really cramped.  Those 1uF capacitors were bigger than expected.   I really had to be very cautious on some connections.  It's a little bit of a mess at times, but I'm quite happy with the result.

The two boards before assembly

Victory !

After some minor corrections to both the layout and the wiring of the boards, I finally got the familiar chaos attractor pattern on the oscilloscope. 

Torpor potentiometer does not have a lot of effect.  Tough it seems to me that the signals are spending more time around a stable state (i.e well) when the pot is maxed out.

Marking holes before drilling

Panel design is inspired by Clarke Robinson's panel with the small butterfly reminding of the popular view that even the minuscule disturbance of the air due to a butterfly flapping its wings in China can cause a hurricane in Texas

Bob cutting the hole for the potentiometer

And now a bit of sound.
We have here two saw waves originally in tune.  One of them detuned by Sloth small signal.
The filter cutt-off frequency is modulated by Sloth big signal.
Low Pass output goes to the phaser.
Band Pass output goes to Rings to Rings, modified by Sloth and output from sample & hold.

Rythmic gate from TAL Filter 2 and reverb from Voxengo OldSkollVerb in Reaper.

No VCA was harmed during the making of this piece.

Jay F. · Sloth Chaos

This seems to put a close to the bottom row of the synth suitcase.

Synth suitcase bottom row

Slew Limiter and Passive Attenuator

Next in line is a simple slew limiter associated to a passive attenuator.

Nothing fancy.  No CV control.  No choice of slope.

Finished module

When I removed my DIY-101 module, I lost the portamento on the oscillator.  Mind you, I seldom used it, because the VCO was not good enough. 
Nevertheless, it is an effect I fancy on an analog oscillator.

Bob drills the panel.

The slew limiter circuit is inspired by Yves Usson own Dual Gated Slew and the simple schematics from Synovatron.  Components value were constrained by what I had on my drawers at the time.

The attenuator is a simple passive one.

Layout and schematics

I changed the way I wired the panel.  In fact, I don't wire it anymore : pots and jacks are soldered on their own PCB, with the main board attached via spacers.  I'm not sure it spares connection wire but it is cleaner on the panel side : no more screw.  On the other hand, I have to be very accurate with my designs and my drilling.  I did three paper prototypes before finding the correct positioning of things.

Thonk proved a good resource to find the appropriate parts.

Assembled

Finally, here is a small sonic example.  The effect of the slew limiter on the pitch CV of the oscillators can be heard after 16 seconds.

Jay F. · Slew Limiter on a Bass Pattern

Inside the machine

Here is the current state of my suitcase modular synthesizer.

With only 2 rows of 84 HP (plus a couple of modules on the side), it does not look very impressive.  Anyway, with two oscillators, two filters, four VCA and the bottom row full of CV modulators, it can still be put into good use.

Here is my first take on what is called a Krell patch.  You'll find the patch detail below.

It is a single mono track, directly recorded into Reaper with some delay and reverb to give a bit more volume.  It's an assembly of two parts of a longer session.

It is self generated.  There is no sequencer, just a bit of manual parameter change here and there.

The sound is reminiscent of the soundtrack from Forbidden Planet.  The Krell are the extinct race of advanced beings of the planet Altair IV.  Hence the name of that style of auto-generated music.




There a two voices.

First a drone voice : saw wave from Mammouth VCO into Ripples, modulated by Rampage envelope and the output of the sample and hold.
Second : square, triangle and sine from Doepfer VCO mixed into Erica Synth VCF; Band Pass output goes the VCA, then Rings; resonance and cut-off frequency of the filter are modulated by a mix of the two Rampage envelope.
Both oscillators receive pitch CV from 2hp tune fed with the output of the Turing Machine.

The two envelope of Rampage are in cycling (i.e. LFO) mode.  First is relatively fast, the second reasonably slow.  The End of Cycle of the first gives the pulse of the system and the clock to the Turing Machine.
The apparent random behavior is provided by the modulation of the Rampage parameters : fall CV of first envelope is the second envelope and rise CV of the second envelope is the noise output of the Turing Machine.

Here is my first approach to document that specific patch.

New filter

Now that I got rid of my DIY-101 based on Doepfer DIY, I'm back to a one filter configuration.   And that filter, Ripples, does not have a high pass output.  I was in need of a new multimode voltage controlled filter and I chose Erica Synth Multimode VCF kit.

Multimode VCF

This is another straightforward kit.  It is rather dense, despite being comprised of two boards.  The kit comes with three bags of parts : one for each board then there is the traditional potentiometers and jacks.

The kit

Sorting parts

Bob cuts resistors legs

The documentation (online, once again) is heavily illustrated and pretty complete with schematics and placement files.  No step by step instructions though.   Just photos of boards at various steps.

Boards are dense but perfectly labeled.  I noticed a single mistake : boards label called for 51k resistors and I received 56k (or was it the other way around ?).  I reckon a 10% difference is no problem.

This went like a breeze.

Main board, chilling on the bench

Odd capacitor placement

One oddity though is the fact that the boards are so heavily packed, they felt the need to solder the electrolytic capacitors upside down on the control board.  Two of which have to be bended parallel to the PCB because the main board will be above them. I put some insulation tape between the board and the capacitors despite the fact there is no indication to do so.  Better safe than sorry.

These capacitors have to be soldered the last according to the manual.  But I would rather put them after the standard parts and before the jacks and pots : once they are put in place, it is quite full.



Anyway, here it is, mounted in the synth.  That first row is now nearly set.

One thing I found a bit odd with this module is the placement of the potentiometer.  I would have put the c-off cv and reso cv close to the corresponding jacks, same with the audio level.  So I would have preferred an inversion of the two rows of potentiometer.  I would have found it more logical.

VCF in place

I do not have any 2 HP blank.  A clear oversight from my side.  I now have that small 1cm gap between the Mammouth VCO and the Erica Synth VCF.  It is supposed to be filled later with a 2 HP multiple.

So, here is a small piece of sound from the synth as it is.





Two tracks recorded live from the modular. Each track is sequenced by one channel of the Korg SQ-1.

DAW : Reaper

Usual effects : TAL Dub II, NI Supercharger, Voxengo OldSkoolVerb, Thomas Mundt LoudMax.

Bass voice : saw and square from Doepfer VCO to Erica Synth VCF Low Pass directly to VCA. Modulated by envelope from Befaco Rampage. Very standard.

Drum voice : mix of two parts.

1. Ripples pinged by Rampage, modulated by second envelope of Rampage and sequencer.

2. Rings, same trigger, same sequence.

Doepfer on Acid

Close-up

So now that I decided to relegate my DIY-101 module based on Doepfer DIY in a box, it is time to say farewell.  I previously wrote that I will certainly regret the filter.  This is why.

The following track is entirely based on the Doepfer DIY filter.  There is no other filter involved.  All sounds are produced by the modular, except the drum part (909 samples) and an overdub on the pads (Korg Polysix emulation).
The base of the track is made of a single PWM output from the Doepfer A-110-1 VCO, into the Doepfer DIY filter.  Each of the filter output (LP, BP, HP) is treated in a different path and recorder separately.
Low pass takes the role of the bass, band pass gives some overtone and high pass gives the high pitch percussions.

The modular synth suitcase as featured in this track

The Korg SQ-1 sequences the lot.  There is a single sequence in two modes : zip of 16 steps on intro, and bridge, 2 lines of 8 steps (one for low pass, one for high pass) for the verses and outro.  High pitched percussions are driven by the Turing Machine, hence their random aspect.  Filter frequency is driven by two envelopes and manipulated by hand as well.

Wind effects are pink noise from the Random module to Doepfer DIY filter, mixing HP and LP before going to the VCA.

Lead synth is a mix of saw and square wave from Doepfer A110-1 VCO through Doepfer DIY filter.  Portamento courtesy of Befaco Rampage.

Pads are a mix of the modular and the addition of pads from Polysix emulation from Korg Legacy Collection, with heavy use of chorus.

Drums : 909 samples in NI Kontakt 5

DAW: Reaper

Effects : TAL Dub II, TAL Chorus-60, TAL Tube, TAL Reverb II, NI SuperCharger, Thomas Mundt LoudMax

Synth in action

Mammouth VCO

Here is my second module from Mammouth Machine : a VCO (Voltage Controlled Oscillator) to take the place of the oscillator of the DIY Doepfer based module that I plan to relegate in a box.  Having two oscillators will allow either to enlarge the sound palette (via mixing or FM), or to have two separate synthesis voices.

Finished Module

Not much more to add than for the ADSR assembly from the same producer.  Same workmanship, same (splendid) wooden, laser-cut face.  I applied the same 2 coats of transparent matte varnish.
Small detail: this kit included the flat cable to connect the module to the bus board.  Curiously, this was not the case for the ADSR.

On the bench

I did have two little problems, though.

Ongoing calibration.

First was with the kit: I received two electrolytic capacitors of 4.7uF instead of 47uF.   With another type of module, I would have smiled and soldered the 4.7uF capacitors, but a VCO needs all the power supply stability it can get.  Maybe 4.7uF is enough.  Nevertheless, I decided to replace these capacitors with the closest value I had in stock.  No 47uF in the drawer, the 100uF are too big.  Remain two 22uF capacitors which fit like a glove.

Then, a small omission in the manual: I couldn't find out how to use the tuning trimpot to calibrate the module, when it's obviously the V/oct scale adjustment, which is critical and not that easy to adjust.
Luckily I had a little help from Bob to do the calibration.

First rows.

And now, as promised, a little bit of sound :



There are two synth voices in this track.
For the first one, the two, slightly detuned, square waves from the VCO are mixed, then sent to Ripples, whose frequency is modulated by the ADSR output.
For the second voice, the sine of the Doepfer oscillator is directed to Rings, then the phaser.
The whole is mixed in the output module.

Rampage generates the two main envelopes. One triggers the other and vice-versa.  The random outputs of the Sample&Hold and the Turing Machine are used to control the rise and fall times as well as the Rings parameters.
Notes are randomly drawn by the Turing Machine and calibrated by the 2hp Tune module.

DAW: Reaper
Usual Suspects : TAL Reverb II, TAL Chorus-60

Mammouth ADSR

ADSR module

I finally decided to replace the Doepfer DIY-based module with separate elements. First, because the quality of the electronics is not up to my (modest) expectations: the oscillator is quite poor and keeps detuning, the VCA does not close completely and is influenced by the LFO, the envelope generator is not very reactive. I'll regret the filter though. Separate elements in kit will also save me some space: my realizations are not very dense.

Since I'm lucky enough to have a local producer, I went to Mammouth Boutique, in our good old town of Liège.

This ADSR module is the first module from Mammouth Machine I built.

Kit is of quality : bulk components in bags, super clear PCB indications, online manual but very well done.  Only caveat for english speaking people : manuals are in french.

The kit

Bob helps to fix the switch

The components are loose in the bags and are not identified other than by their marking, whereas other suppliers offer specific labeling.
But there is nothing wrong with that.  The information on the PCB and in the manual is very clear. A small cheat sheet comes in the kit with resistor values color coding and capacitor indications. With a little care, everything goes well. Nevertheless, I tested every resistance with an ohmmeter, but that's what my perfectionist side is all about.




The face is made of wood, laser cut. The wood is untreated, I decided to delicately apply two thin coats of transparent matte varnish to ensure a certain durability over time.

Ongoing

No particular remark on this kit, everything went perfectly.

[edit Feb 29, 2020] After my first tests, I noticed a residual voltage (about 60mV) after the end of the envelope.  This leaves a VCA a bit open and I could hear a faint noise.  It does not look to be problematic in real conditions though.