Rick's Formant page 

Welcome to Rick's Formant page 

If you're interested in some experiences with home-building a synthesizer,
read on.  

I have built/am still building a modular analog synthesizer. It is
the Elektor Formant designed by C. Chapman (an Englishman?), which
was published in 1977/78 in the European electronics magazine Elektor.
In those days I was still a student, and didn't have the time or money
for the project. So I shelved it, till a later date when I would have
both time and money at the same moment. This rare combination of factors
occurred in October '92, triggering the project from hibernation.  

The Formant is a conventional Moog-style synthesizer, voltage controlled
with a 1V/octave characteristic. It consists of separate modules, which
I mounted in a rack. Each module is a printed circuit board with a
front.  

The Formant was published in a series of articles in Elektor, and later
as a book with a compilation of the articles. A year later there was
a book with additions and extensions by M. Aigner, an Austrian.  

The first book describes the keyboard and interface, power supply,
VCO, VCF, LFO (consists of 3 LFOs), Noise, ADSR, VCA, COM (output module)
, RFM (Resonance Filter Module) and a 24dB VCF. Book 2 contains things
like Ring Modulator, Envelope Follower, Mixer, Phase Shifter, ADSR
Controller, VC-LFO's, Sample and Hold, and a few more. I'll briefly
describe the modules here, and how building the Formant went.  

---------------------------------------------------------------------------

The Modules 

Keyboard interface   
The keyboard is made with double switches, one switching a resistance
ladder, the other the gate signal. 

   inputs: frequency modulation 
   controls: portamento, octaves coarse and fine 

VCO  

The VCO generates sine, triangle, square, sawtooth and spaced sawtooth
waves. The pulse width of the square is adjustable, and even modulatable
(PWM).  

   inputs: frequency modulation, external control voltage, pulse width modulation.
A LED indicates the pulse width. 
   dials for octaves coarse and fine, output level, pulse width, pulse width
modulation and frequency modulation. 
   Switches to select each waveform (you can have sine and square at the
same time, the signals are simply added together). 
   External control voltage/keyboard voltage selector. 

VCF   
The VCF is a 12dB/octave voltage controlled filter. You can configure the
filter as Low Pass, High Pass, Band Pass or as a notch filter. Quite
versatile really. 

   inputs: Timbre Modulation, external control voltage, external signal.

   dials for timbre modulation, external signal level, envelope, octaves,
Q (resonance) and output level. 
   Switches to select HP, BP, LP and Notch filter function. 

ADSR   
The ADSR generates the envelope signals for the VCA and the VCF. Normally
you will have two ADSRs in a system, one for the VCA, and the other
for the VCF(s). 

   dials for Attack, Decay, Sustain and Release. 
   switch to select AD or ADSR envelope 
   LED indicator showing the envelope 

VCA   
The VCA has the following controls: 

   inputs for amplitude modulation and external signal. 
   dials for AM depth, ES level and Gain. 
   Switch for ENV/Off (Off means no ENV control). 
   LED to view the input signal level. 

LFO   
The LFO module consists of 3 LFOs. LFO 1 and 2 are identical, generating
square, triangle and sawtooth (/|/|) waves. LFO 3 is different, in
that it generates triangle and two sawtooths, one being the inverse
of the other. (/|/| and |\|\). Each LFO has a dial for the frequency,
and a LED that shows the triangular wave amplitude.  

Noise   
The noise module generates white noise, coloured noise and a random
voltage. The 'speed' of the random voltage is adjustable with a dial and
indicated with a LED.  

COM   
The Control and Output Module outputs the Formant signal, there are dials
for tone control (Bass, Mid, Treble) and Output level. You can connect
a headphone or external amp to this module. (All outputs of the Formant
you can connect a headphone to.)  

24dB VCF   
Is basically like the VCF, but 24db/oct. 

   inputs: Timbre Modulation (TM), ECV, External Signal (ES). 
   dials for TM, ES level, Envelope (ENV), Octaves, Q (resonance) and output
level. 
   Switch to select HP/LP, a switch to select VCO's/ES. 
   A dB/Oct selector: 6, 12, 18, 24 dB/oct. 

RFM   
The Resonance Filter Module is used to mimic existing instruments more
naturally, or just as some filter. Another name for an RFM is a 'parametric
equalizer'. This is not a voltage controlled filter. The Formant RFM
consists of 3 filters in parallel. Each filter has three dials for
f0, Q and A.  

Basic Formant  

A basic Formant consists of 3 VCOs, 1 VCF, 2 ADSRs, 1 LFO, 1 Noise
and a COM. So far I have built 6 VCOs, 2 VCFs, 2 24dB VCFs, 4 ADSRs,
2 VCAs, 2 LFO, 1 Noise, 1 COM, 1 RFM, 1 Sample&Hold, 1 Ringmodulator,
4 ADSR Controllers and 2 Mixers.  

Getting the parts  

One of the first difficulties when you're going to build a thing that was
designed 15 years ago is getting the components. Most of the parts
used in the Formant are just common opamps, there are no exotic custom
chips used, like the CEM chips.  

There are some (three) key components, that are getting pretty hard
to get these days. The following turned out to be 'difficult':  

   The exponential generator that converts the linear control voltages to
an exponential voltage in the VCO was made with a uA726 double transistor
from Fairchild. The uA726 is an ideal part, it contains both a precise
double transistor and an oven which keeps the temperature constant.
Thus temperature stability is guaranteed. It never was a commonly
used component anyway, and now it is getting rare. I have been able
to get a few still, thanks Georg :) You could use a modern double transistor
though, like a MAT02. Modern transistors are far more stable, and it
might be you can use one straight away. I haven't tried though. If
I would not have gotten hold of the uA726 I would have used the MAT02,
and built a temperature controlled heater myself (using a powertransistor
as a heater and a temperature sensor to control it).  

   The exponential generator in the other voltage controlled modules (VCF
and VCA) were built with another double transistor, a PNP transistor
array in an ic, the CA3084. I have been able to get 10 of those still.
You could probably use a MAT03 as a replacement though.  

   In the VCOs BC107C and BC177C transistors were used. Very simple and
common transistors you'd say. Well, they aren't. I have not been able
to locate a single BC107C. The A and B types are available though,
but it is explicitly stated that a C type is absolutely crucial. BC108C
or BC109C won't do, as they have too low a voltage rating. I used BC547C
and BC557C instead. Those are very common transistors these days. No
idea why there are no BC107Cs available any more.  The BC's are used
in a comparator, converting a sawtooth to square and in a sawtooth to
triangle converter. 

All other components are still easy to get. I ordered opamps from a mail
order company, and the things they didn't have from others. It is vital
to keep a list of companies you already phoned and what you asked for,
to avoid the embarrassment of asking someone the same thing again
a few days later. It happened to me only once. Things I first wanted
to have a look at I went to local electronics shops for.  

Upgrading to 'modern' standards  

Not a great lot seems to have happened in analog electronics in the
past 15 years. Nearly all parts common then still are common today,
with the exception of the few mentioned above. I have been able though
to upgrade the design very easily by just using some modern equivalents:

   741 I have used TL071C opamps instead of the old 741. If you buy
a hundred they're same price, or lower, than the bl**dy old 741. The TL071C
is a FET opamp with very low noise specs. Ideally suited for a synthesizer
or any other low-noise application, like preamps.  

   FETs In many places FET source followers are used as buffers. In
those days FET opamps were not widely available, and very expensive.
The disadvantage of using a FET source follower is that you have to
select suitable FETs and select a resistor with each FET. I used the metal
can version of the LF356H FET opamp instead, which, with some bending
of its legs, will fit beautifully in all places FET source followers
are used. No selection necessary, and no R either. I'd have liked to
use the metal can version of the TL071 or the LF411, but those aren't
readily available in a metal can package. The LF356 draws quite some
current and gets rather hot, the TL071 and LF411 should remain cooler.

   Rs Metal film resistors are almost as cheap as carbon film ones these
days. I simply used metal film resistors just everywhere. The advantage
of metal film Rs is that they are low-noise and more stable. In some
places 1% resistors are used, but they were as easy to get as the normal
(5%) ones. 

Printed circuit boards  

For all modules Elektor published PCB layouts. Back then you could also
buy the prints from them. Elektor only has prints a limited period
of time, but I found a company that still has the films, and can produce
the PCBs for you. All modules, except the VCO, are eurocards. The VCO
is a double eurocard. Price of the VCO board is Dfl 61,-, which is ca
US$ 30,- if I count two Dfl's to the dollar. The eurocards cost ca US$
12,- each. The PCB layout is also in the book, so you could etch and
drill the prints yourself, but I thought that far too much hassle, and
environmentally unfriendly.  

The prints from book 2 were never published in Elektor, but they are printed
in book 2 of course. I had the company mentioned above just take them
from the book.  

Keyboard and housing  

Building a synthesizer is not simply soldering parts onto printed circuit
boards. The total effort involved is more than *twenty* times the effort
of the plain soldering. Most of this effort and frustration is invested
in mechanical things like the keyboard and the housing.  

The keyboard is just a plain electronic-organ keyboard. There is a
double switch for each key, and those switches have to close simultaneously.
Such switches are made by a company called Kimber- Allen in England.
The mechanical interface between key and switch is a *lot* of work.
The original idea was that under each key there is a T-shaped piece
of plastic that will close the two switches at the same time. I have
not been able to get a keyboard with those T-shaped 'pushers'. My keyboard
just had a piece of plastic, with two hooks in it, as used in electronic
organs. I had to find my own solution, so for each switch I drilled
two tiny holes in the piece of plastic attached to the key. I have
a 5 octave keyboard, with 61 keys, so I had a marvellous weekend.  

The housing of the Formant turned out to be an awful lot of work too,
and very costly as well. The Formant consists of eurocards and double eurocards.
All pictures in the book get the idea across that everything fits
neatly in a 19" rack, six modules in a row. The truth is not so simple.
The pictures in the book show a 19" rack, divided in 6HE and 3HE. The
modules with the bigger fronts, like VCO and VCF are in the lower part,
the smaller modules in the upper 3HE part. And here things go wrong.
The VCO is a double eurocard, but the VCF is a single eurocard. A double
eurocard does *not* fit in a 6HE 19" rack, nor does a single eurocard.
As you see, this is a complete mess, you cannot put VCO and VCF prints
in rails in a standard 19" 6HE rack. I ended up with extra alu profiles,
and mounting both eurocards and double eurocards on a piece of aluminium
as to fit in a 6HE rack.  Add to this that my pcb manufacturer did not
cut the epoxy material really very precisely the 10 centimeter width
of a eurocard (3HE), but a few millimeters extra, that I had to grind
off using sandpaper. Of course I only found out after having already
soldered all components in place. I have called the good man some very
bad names.  

The fronts of the modules have given me headaches. Elektor has given a
design for them, too. You can copy those on 0.5 mm thick adhesive photosensitive
aluminium, as made by 3M. My pcb-making company could also do that for
me. What fun, as the fronts turned out to be some 5 millimeters larger
than 6HE. Also, the markers for the drill-holes on the fronts do not
match the spacing of a 19" rack at all. So, I ended up with fronts with
screws in the corners, and the marks very visible. Furthermore the
result was quite fuzzy, the lines and figures being far from sharp.
I guess during the years the films got scratched, and so my fronts also
looked rather scratched. Another things is that I use knobs for the
dials with a plexiglass clear disk. In the original design the figures
are then obscured by the plexiglass disc, making them unreadable. You
can see them, but as the clear plexiglass disc is curved its more like
looking through a lens. Awful.  

Thus I was not pleased with the fronts at all, and designed new ones myself
on a Macintosh. My idea was to print them on film with a 600 dpi printer
and use the same 3M alu front stuff. Alas, the printing would cost me
US$ 10 per A4 (two big fronts or 4 small ones). Plus US$ 15 per A4 sheet
3M alu foil. I thought that a bit steep. And then having to process
it all... nah. I printed my designs onto overhead sheets, which I simply
glue onto a piece of aluminium with transparent 2-component glue. (Image
flipped, with the toner on the inside of course.) Looks great!  

Oh, I haven't mentioned yet that you need to drill a *lot* of holes
in the alu fronts for potentiometers, connectors, switches, LEDs and the
lot, have I? Well, you have to drill a lot of holes in the alu fronts
for potentiometers, connectors, switches, LEDs and the lot. About a
hundred holes in total. And you think drilling holes in 1.5 mm aluminium
is easy? Very funny. It is not. The holes for switches and 3.5mm connectors
are still easy, but the bigger (ca. 10mm) holes for potmeters are a
pain. You cannot simply drill those, for you will end up with a triangular
hole. Instead, you have to use a conical drilling tool that I don't
know the english word for now, but I'd call it a 'widener' :-)  

Next fun thing is, you have to connect the prints and potmeters and
switches and all with pieces of wire. Many hundreds in total, each
of which I neatly soldered and protected with the insulating shrinking
stuff the English word for which eludes me right now. It is all a lot of
work. Especially as not all directions from the book are clear. I ended
up with quite a few potmeters connected in reverse, so you'd turn up
something and it would in fact go down. Very confusing if you don't
know exactly what you're doing. You'll learn though.  

Calibrating procedures  

An analog synthesizer has to be adjusted and calibrated. And again,
this is a lot of work. Plus, you need to know what you are doing.
Now I'm an electro engineer, but I work as a systems programmer so I don't
have much experience in the field of practical electronics, which I
was never very good at at all in the first place. (I still remember
nightmares of the electronics lab where we had to make a simple single
transistor work as an emitter follower or something as like that, and
nothing seemed to work. Least of all the oscilloscope, that had about
as many dials and trigger options as my synthesizer has now, which would
show nothing but an oscillating emitter follower. Aaaaargh!) Well anyway,
not all directions and procedures are very clear, and you have to fill
in a lot yourself. Sometimes directions are open for more than one
interpretation even. You definitely need an oscilloscope, frequency counter
and a digital volt meter for the final adjustment of the VCOs.  

Cost  

*important note*  

If you think that building a synthesizer yourself is a cheap way of getting
a synthesizer, here is what you must do: FORGET IT!   

The Formant has cost me a *lot* of money. I could have bought, for
example, a Waldorf microWave plus an Atari, or maybe even a Macintosh LC.
But, no regrets. There's no such thing as twiddling your own knobs
(no pun intended here) and knowing the ins and outs of your synthesizer.

Most expensive single items are the 19" racks. The rest is not very
expensive per se, only you need a lot of things. A potmeter and a knob
aren't very deer, but if you need a hundred it all adds up to quite
an amount. Same goes for the switches (also about 50), and 3.5 mm connectors.
Also you need tools. I bought an electric drill, drills (amazing!),
saws, what not.  

In total, very roughly speaking, the Formant has cost me about US$
2500,-. (Including tools and parts already bought for future use.)  

Time  

I had time off from work last year for 5 weeks in a row in the autumn,
time I had to take off or lose it. Being rather fed up with our management
after work on our Visualization Center (computer animations of scientific data,
video equipment and the lot) I decided to take it :-) It cost me a
full week to get more or less all the most important parts. I spent
much of the remaining 4 weeks fulltime putting it all together. Planning
is vital here, nothing is more frustrating than not having bought just
one silly tiny resistor. So, if you have to build a Formant just in
the evenings and weekend, be prepared to work on it some time.  

Playing it  

Well, after spending so much time and money I found out what I knew
before. I can play with it, but not really play it very well yet. I've
had music lessons as a kid, and even learned how to play a recorder.
The wooden type, not the tape type. I never learned to play a keyboard.
Only a few weeks ago I found out how to play scales. Not that it matters
much, I'll interface it to midi and a computer one day and be my own
Kraftwerk. Or I could practise and learn how to play, really. I'm definitely
having fun with it anyway.  

I already have been able to reproduce some sounds Tomita used on Pictures
at an Exhibition, I know how to do the Popcorn sound, and I have found
how to make the thin high-pitched sound in the background of Kraftwerk's
Spacelab, and how to do the deng-deng-deng-deng-... of old Radioactivity.

Who can build a Formant?  

Building a machine like this is not something to undertake lightly.
It costs a lot of money, it takes a lot of time, and requires a lot
of mechanical skills and not least some insight in electronics. An analog
synthesizer has to be adjusted and calibrated. And again, this is a
lot of work. Plus, you need to know what you are doing. Well anyway,
not all directions and procedures are very clear, and you have to
fill in a lot yourself. Sometimes directions are open for more than
one interpretation even. You definitely need an oscilloscope and a
digital volt meter for the final adjustment of the VCOs.  

I have been lucky that I haven't made many errors while soldering the boards,
but I have made some. Those can be a big puzzle, and you *will* have
to solve them. (Note the error in the ringmodulator print, where pin
2 and 3 of the ringmodulator ic aren't connected.) 

   If you have never built anything electronic before: don't do it.

   If you don't have the money: don't do it. 
   If you don't have *plenty* of time: don't do it. 
   If you don't have the mechanical tools: get them first. 
   If you have neighbours who don't like synthesizer sounds: get rid
of them first, or buy a pair of headphones. 

Future  

Well, there still are new modules to build and play with, like the
extensions from book 2 (ring modulator etc). I want a sequencer, the
simple analog type with just a few potmeter dials for pitch and duration.
I will design and build it myself, its rather an easy thing really.

---------------------------------------------------------------------------

Book 2  

---------------------------------------------------------------------------

The past future has now begun. I have now built some modules from book
2: sample&hold, ringmodulator, ADSR controller and a ring modulator.

Ring modulator  

The ring modulator module contains a ring modulator and an envelope
follower, which is also a sensitive input to connect a microphone. 

   inputs: A and B and External Signal. 
   output: the ring modulated signals and envelope. 

Book 2 contains a eurocard format ring modulator print. The same circuit
was published in book 1, but not as a eurocard. After soldering all
the parts and assembling the module... it didn't work. After a few hours
of happily (ahem!) bug-searching it appeared pin 2 and 3 of the ringmodulator
ic erroneously weren't connected on the print. A little piece of wire
and a blob of solder fixed this.  

Sample&Hold  

The Sample&Hold is fun! Simply connect a few slow LFO signals and control
a VCO with the result, and it plays silly little melodies. Or use the
(slow) random voltage from the Noise module to control it. The S&H has
two inputs (Sample and Trigger) and one output. A dial sets the trigger
level. There is a C at the sample input. To be able to use slow LFO
signals you should omit this C and put in a wire instead. No need for
AC coupling. This is a bug, in my opinion.  

Mixer  

The mixer is a 6-input mixer. One of the inputs has a high sensitivity,
to connect a microphone or other external signal to the Formant. The
inputs are AC coupled with a rather big capacitor (680nF). You better
leave this C out, i.e. mount a piece of wire instead, else slow LFO
signals don't work. It works fine without anyway. Rather nice to construct
complex control signals from the LFOs.  

ADSR Controller 

Normally the gate signal directly drives the ADSR module(s). If you want
to hook up a sequencer or an LFO instead there now is a way to do it:
with the ADSR Controller. The ADSR Controller lets you select an external
gate signal. Also it allows keyboard repeat, i.e. the external gate
signal is only fed through if a key is pressed. Another feature is
Delay. A dial for Delay lets you delay a gate signal from 0 to 5 seconds.

There is a bug in the circuit. Keyboard repeat doesn't work, unless
you change R21 from 10k to 22k.  


Rick Jansen, rick@sara.nl
