Back to JH. homepage
 

My latest CD
The Secret Life of Trees
 

JH. Matrix FX

The idea:

Use the concept of the EMS VCS3 (Matrix patching, a single output attenuator for each function, Joystick, Vernier Dials, monitor speakers ...) for a collection of Effect (FX) devices.

Keep enough of the VCS3's original functions in the Matrix FX to be able to sell my Synthi Clone some time.


Function Implementation Applications
3 Oscillators
Basically providing all the waveforms of the VCS3 for _each_  Oscillator.
All oscillators can be audio range or LFO range.
Different implementation, more features.
Osc. 1 has a quadrature sin / cos core, and is also used as modulator for the SSB modulators.
Osc. 2 can be "Shape modulated" by Osc. 3.
Osc. 3 also has random S&H waveform with variable probability.
Audio Oscillators
Modulation Oscillators (LFO and Audio Range)
Random Voltage Source
Filter Original VCS-3 LPF plus 4-pole HPF

Trapezoid Generator with VCA
Same function as, but different implementation than , VCS3.
VCA can either be controlled by Trapezoid, or by external CV.
VCA has "Initial Gain" potentiometer added.
Attack / Hold (On) / Decay envelope triggered by Gate signal or front panel Button
Self cycling mode with Attack / On / Decay / Off parameters.
Voltage controlled Decay time.
VCA can be used seperately from Trapezoid.
2 SSB Modulators (Frequency Shifters)
Two complete Frequency Shifters ("Single Sideband Modulators"), sharing their Modulation Oscillator - Amount of frequency shift equals frequency of Osc. 1 Stereo Frequency Shifting
Upshift->Filter function->Downshift will turn a fixed filter bank etc. into a VCF. Output signal will be in the same frequency range as input signal (again), but filter courve will be virtually shifted down.
Barberpole Phasing
BBD Delay
Long delay (Echo) or short delay (Flanger, Chorus, Physical Modelling)
Various Filter functions (tracking, non-tracking)
1V/Oct control for tuned comb filters and Physical Modelling
Echo, Flanger, Chorus, Vibrator, Physical Modelling
Phaser Various Phaser modes. Stereo output.
Phasing, Vibrato, Crossover (HP/LP)
External FX Send / Return Loop for external effect.
Voltage controlled Wet / Dry Mix
For Reverb, Filter Banks, etc.
Wave Folder
The famous Serge Wave Multiplier, adapted to VCS3-like input and output levels
Dynamic creation of harmonics
Mangling modulation signals
Envelope Follower
Inspired by Eventide Instant Flanger
Features "Bounce" function and variable Release time

Inv / Amp
Amplifier, inverting or non-inverting
high Gain or low Gain
Built-in opto-electronic Limiter





Front Panel design
Construction Drawing
Photo of Front Panel back side
Sub D connector with wires to be mounted on back of front panel

Trapezoid Generator / VCA board
Trapezoid / VCA Schematics

Joystick
Joystick Schematics

Wave Folder / Inv Amp board
Schematics of Wave Folder and Inv Amp

Oscillator 3 / Noise / Filter board
Schematics Part 1: Oscillator 3 (incl. Random with variable probability)
Schematics Part 2: Noise and +12V/-9V regulator for Noise and LPF
Schematics Part 3: Filter
Photo of Osc 3 / Noise / Filter board

Oscillator 2
Waveforms Pulse, Ramp, Triangle and Sine, with variable waveshape and waveshape modulation from Osc. 3
Osc. 2 Schematics
Osc. 2 Photo

SSB Modulators (Frequency Shifters) and  Oscillator 1
Block diagram of signal path
QVCO Core schematics
QVCO Core board photo
Modulator schematics
NEW: Compander Board
I had built a compander system around the Philips NE572 chips, but whatever I tried in terms of operating level and time constants, that thing did sound horrible. (That's the reason I hadn't published that part in the first place.)
Now I've replaced it with an OTA-based compander, similar to the one I had used in the FS-1 Frequency Shifter. Which, in turn, has been inspired by the compander inside the Roland Vocoder Plus.
It has an almost infinite compression ratio (more ALC that compressor), and it's feed-forward sidechain allows 2-pole filtering after the level detector for a very smooth operation. Frequency Shifters are quite demanding for compressors, as the output signal can be so much different from the input signal, and there is none of the cancellation of distortion products many traditional companding systems rely on. Here, in the Matrix FX, the expander's control current is directly copied from the compressor's - and not derived from the expander's input signal. The only drawback of this method is that the CV is "a little too fast" compared with the signal, which is slighly delayed in the dome filter. I practise, this wasn't a problem, however.
Compander circuit for SSB Modulator 1
Compander for Modulator 2, and CV distribution amplifiers for output volume modulation and wet/dry mix of several modules
Two dome filters for two channel operation. The circuit is basically the same as FS-1 Dome filter.
PCB layout of Dome Filter board
Component side of Dome Filter board
Photo of Dome Filter Boards
Filter Transformation Demo (7MB .wav file)
  • A drone sound from the OB-8 is sent to a first Frequency Shifter, which shifts the audio signal up by a variable amount of Hz.
  • This upshifted signal is sent thru a filter bank (MAM vocoder, with a rather thin-sounding setting with all even channels up and all odd channels down.
  • This filtered signal is shifted back down with a second Frequency Shifter, by the same amount of Hz that the original signal was shifted up.
  • First you hear the very thin sounding unshifted signal - just the OB-8 drone processd by the vocoder filter bank.
  • Then we start to modulate the Frequency Shifter with the Joystick.
  • Please note that there is no filter modulation! The filter is fixed all the time (static formant filter) - it's the signal that is modulated!
To do: Build new Dome Filters that go to 32kHz, for whole-range upshift -> filtering -> downshift operations.
(As it is now, if there are ultrasonic parts of the spectrup left after upshift + filtering, they won't be processed in perfect quadrature in the down-shifter's dome filter.)

BBD Delay
  • 1V/Oct Tracking.
  • Half speed switch.
  • Long/Short delay (MN3005/TDA1022 chips).
  • A variety of filters: 15kHz like Dynacord TAM-19, 5kHz like Amdek Delay Machine, Tracking Filter, or unfiltered.
  • Opto-electronic limiter for controlled feedback buildup.
  • Voltage controlled Wet/Dry mix.
Block Diagram
Schematics of BBD clock circuit (inspired by Aries and Dynacord circuits; special thanks to Mike Irwin for helpful comments on Aries part.) Note the exponential -1V/Oct input, and 1/x law modulation sensitivity for the modulation matrix.
Schematics of BBD core circuit - TDA1022 (as in Dynacord TAM-19) for short delays; MN3005 (as in Amdek Delay Machine) for long delays; DG413 for switching.
Schematics of Anti-Aliasing and Restauration Filter:
Filter CV
Filter Signal Path.
10-pole-VCF (5 pre BBD, 5 past BBD) allows the following modes:

* 3.18kHz with optional "Bump" in the 2 ...3  kHz Range (Amdek Delay Machine)
* 15kHz (Dynacord TAM-19)
* Tracking (follows the BBD clock rate)
* "Open" (ca. 400kHz; hopefully emulates certain filter-less Electro Harmonix devices)
SE571 Compander
VTL5C4 Limiter
DG413 Switching
Select 5 LM13600's for identical gm (+/-3%) from Iabc = 1uA to 1mA
Circuit for selecting LM13600's
(Do not use LM13700 or NE5517!)
Sketches for making a "Pseudo-PCB layout" on Veroboard (BBD clock, and tracking filter)
- in case you ever wondered how to get from schematics to veroboard.

Phaser Board
Some thoughts about phaser control law
Modes
No phaser model on this planet sounds like any other. The underlying principle is always the same: all pass filtering, mixing of the filtered and dry signal to create notches, and some amount of feedback to create peaks. But that's about it, what all phasers have in common. The all pass filters (APFs) themselves can be implemented with different circuits. The number of stages differs. And there are countless ways to apply feedback and to create a second output for stereo effects.
I have chosen opamp-based all pass fiters with Vactrols to control the pole (and zero) frequencies. There are 2 chains of 6 APFs which can be used in series or parallel. A preset switch with 6 positions controls an array of 24 electronic switches for different configurations. Some of these configurations or topologies are inspired by classic phaser circuits, such as Moog, Schulte or Mutron. The APF technology used is different in these, though.

Mode 1
Actually not a phaser. In Mode 1, the LPF and HPF outputs of a state variable filter are routet to output 1 and 2, respectively. Think of this as a 12dB/Oct crossover with VC separation frequency and variable resonance.

Mode 2
Modelled after the Moog 12 Stage phaser, with only 4 stages in use. (4 stages phasing, and 4 stages feedback.) Feedback polarity is positive, resulting a very fat sound. Feedback is AC coupled to avoid amplification of offset voltages. Output 2 is the "Auxiliary" output of the Moog, created by subtraction of Output 1 signal and the dry signal. Compared to the original, which used OTA-based APFs, there is a bit of CV slewing.

Mode 3
Also modelled after the Moog 12 Stage phaser, but using all 12 stages in series. (12 stages phasing, and 12 stages feedback.) See Mode 2 for details.

Mode 4
Modelled after the Schulte Compact A Phasing. 8 stages of APF. Feedback taken from stage 2. Negative feedback for pronounced resonant sound. This unique compination creates 4 noches, but only one resonant peak on the main output (Output 1). Output 2, just as in the original, is the signal after the feedback summing , before any all pass filtering, creating only a peak and no notches. Slewing is less than on the original, as the Compact a uses incandescent lamps instead of LEDs.

Mode 5
Modelled after the Mutron Bi-Phase. 2 x 6 Stages of APF with opposite polarity modulation for wide stereo effect. Positive feedback for fat sound. Slewing should be similar to the original (LED->LDR technology).

Mode 6
"Thin Stereo". Just like Mode 5, but with negative feedback, and mixing of inverted APF signal with dry signal.

There's a table in the schematics PDF file that shows in detail what's going on in the various modes.
Phaser Board Schematics
Please note that these are my raw design notes, from which I've developed and built the phaser board. This is no clean drawing. For instance, some circuit parts are shown twice, in different context, on different pages.
Oh, and I used a NOS SSM2024 quad VCA chip for 2-channel wet-dry mix. I've used this chip only because I ran out of space on my circuit board. I recommend using a pair of CA3280s, or similar linear controlled current-output VCA chips, instead. Ground-reference the control current inputs with four pnp transistors with their bases grounded, in that case. (Collector goes to 3280 Iabc inputs, emitters act like SSM2024 inputs.)

audio clip 1 - playing around with phaser, BBD delay and wave folder
This is about 11 minutes long (10.6 MByte of mp3) playing around with the Matrix FX. Not music - just a sound demo. Apart from the Matrix FX, there is only some Farfisa Organ (VIP500), which is processed with the Matrix' BBD Delay and Wave Folder, and some reverberation (Quantec and SRS-56). No edits - just playing the knobs.

audio clip 2 - "infinite phaser"
A short clip (427 kByte of mp3) demonstrating the use of a Frequency Shifter as an "infinitely" rising phaser. White noise is fed into the SSB Modulator 1 (= Frequency Shifter), with the shift amount (= OSC 1 frequency) way into the sub-audio (LFO) range. Frequency shifter output is mixed with the direct signal. Frequency shifter output is also fed ito the phaser (12 pole Moog mode, only the wet signal used, i.e. acting as an all pass filter!), and then fed back to the Frequency shifter input. A certain anount of Loop gain added with the Inv/Amp to get enough resonance. (Can't use the phaser's own resonance here!).

NEW: audio clip 3 - "phaser and frequency shifter"
Six minutes (5.7 MByte of mp3) of playing around with the Matrix FX and Farfisa VIP500 organ.
The Farfisa's upper manual sound goes thru the phaser (set to Compact A mode, and modulated by the envelope follower with a large amount of "Bounce".), and then into two frequeny shifters (downshift on left stereo channel, upshift on right stereo channel).
The Farfisa's lower manual sound goes thru the diode ladder filter (slightly modulated by a self-cycling Trapezoid generator), and then into the phaser (see above). Also, the lower manual feeds the BBD delay, which is mixed to the center of the stereo output after the frequency shifter.
The whole sound is treated with reverb (Quantec, with a large cathedral setting).
I've deliberately used a rather thin organ, and not a synthesizer, as input for the Matrix FX, to demonstrate its sound modifying capabilities.
 Please read the FAQ.

Back to JH. homepage