JH SYNTHI CLONE
I've built a Clone of the EMS Synthi A.
Well, at least something similar.
The VCS3 / Synthi A is a true phenomenon. You can still
buy new ones from the factory, used ones are highly sought after, people
usually complain about high prices on the used market - and building one
yourself will not save you any money. This last one was surprising, but
it also was clear to me after a bit of research for the necessary components.
Patch matrix, large front panel, vernier dials, joystick, reverb tank ...
It was clear from the start: if I wanted to save money by building it myself,
I'd better forget it!
So the main reason for building a clone was that I could
make additions / modifications right from the start and build an instrument
that is specially tailored to my needs. I didn't want to change too much,
though, in order not to loose the character of the original instrument.
I've made the following changes:
A slightly larger Matrix. You can get a stock 20 x 20 Matrix
from Ghielmetti cheaper that a custom ordered 16 x 16 one.
A second Envelope Generator. I chose the Korg MS-20's HADSR
design for this, slightly modified to work from +12V / -9V supplies.
A Sample & Hold and Glide circuit. It has 3 modes
(1) Ordinary S&H with internal clock rate.
(2) Portamento / Glide (linear mode; actually a slew
(3) Glide & Hold. Clock source in this case is Pulse
wave from Osc. 3.
The idea is to have only one potentiometer to control
clock rate and slew rate.
An Envelope Follower and a Signal Inverter. I designed a
new Meter circuit for these functions. When the Meter is set to "Control
Voltage", it will give an inverted output. (You could emulate this function
one the original with external patching, btw.) But when the switch is set
to "Signal Strength", the output will be a rectified and lowpass filtered
signal for use as Envelope Follower. There is a switch to trigger the trapezoid
and HADSR from the meter channel, too.
A Precision Control path for the VCOs. Remember the original
Synthi A goes out of tune when you change the number of modulation sources
in one patch? Graham Hinton shows one way to fix this problem (on his EMS
web page): Buffering the matrix and get precision summing instead of weighted
summing of the modulation inputs. I'll try a different approach: Leave
the matrix as it is, and introduce a second, independent control path for
keyboard control voltage. The VCOs will have a 3-way switch for Keyboard
Tracking / Free Running Audio / Free running LFO.
Thanks for all the input from Synthi users who gave recommendations
how to make best use of the larger matrix. The final result is shown in
the picture below. Special thanks to Graham Hinton for his collection
of modifications for original Synthi's. Especially the unique Sync function
is very nice !
The Interface to the outside world is changed a little, also.
Gate and 1v/Oct inputs replace the multipin connector, the mains connector
is an IEC type, and the headphone output doesl monitor stereo left and
right (not output ch. 1 and 2).
Schematics and Circuit Descriptions
Due to respect of EMS, who still are in business and build Synthi A's,
I will not include original or redranwn EMS schematics here.
What you will find are my own additions and modifications.
Sample & Hold / Glide
An additonal Envelope Generator
The Trapezoid Generator contributes to the Synthi A's uniqueness, and
I wouldn't replace it, but sometimes a "normal" envelope could be useful
as well. I chose the Korg MS-10 / MS-20 style Hold-Attack-Decay-Sustain-Release
circuit as an additional EG. Component values are changed to match the
+12V / -9V supply. The 1 MegOhm potentiometers are a compromise. I've increased
the integration capacitor to preserve the maximum time, but this also increases
the minimum time. This is surely not optimal, and I may change it some
time. (several MOSFETs in parallel ?!)
Using standard components for the VC Reverb
I used what I found in a drawer - similar transistors should work as
well - no need to get the exact parts. The reverb tank in my Synthi Clone
is a standard Accutronics one (3 springs, long version), so it was no surprise
that the tank's output level was different from the original one, and the
resistor values for the wet / dry mix had to be changend, too. That's why
there are more trimpots than in an original Synthi.
BTW, I love the design of this voltage controlled wet / dry mixer.
If I would have designed it, I would probably have ended with two OTAs
and several opamps. But look at this: Two transistors for the audio path,
one acting as VC resistor, and a 4th for CV mixing. Not exactly precision
stuff, but it does the job, and does it well enough.
The "Precision Path" for 1V/Oct CV to the VCOs (might also be interesting
for owners of an original Synthi)
For more information, please contact
All drawings copyright J. Haible (C)1998