Assorted random project ideas

Before we begin, as general background, if you need pick a project that
requires matched transistor, check this out:

Bill and Will’s Synth
Moog Transistor Matching Tester Construction
. If you’re building
a VCO, I can supply you with monolithic matched pairs. If you’re building a
VCA or a VCF, you can probably get away with some rough hand matching.
If you need matched diodes, check out
thread on electro-music

OK, here’s some random ideas. These are circuits that have caught my attention
for one reason or another and that I would enjoy exploring more.
Remember you shouldn’t
consider yourself restricted to picking something on this list; if you
do have your own idea for a project you’d like to pursue, looking at the
things on this list will give you a sense of relative scope. Or, you may
not already have a project idea, and may not be interested in anything on
this list per se, but reading what’s here may inspire a new project idea.

Don’t be scared if you see strange part numbers. A good portion of these
parts are long out of production. I can suggest appropriate substitutions.

  • The Rhodes Chroma VCO uses a unique “charge pump” design using an
    Exar 4151. Here is the
    service manual
    ; in particular, the oscillator circuit is described

    and the vco schematic is given in the left part of

    (notice each Chroma voice board has two VCO’s;
    you’d want to build just one). The pitch input is labeled as “PITCH” A or B,
    and the sawtooth output is found at pin 3 of the 4151.
    are some notes about one person’s experience building it.
    (For a
    one-person project, I’d suggest just buffering the sawtooth output and not
    worrying about the rest of the waveshaping circuitry, and I’d suggest
    the one in the upper-left corner that is missing the sync circuitry;
    for a two-person
    project, I’d suggest doing in the one in the lower-left corner, which
    includes the sync circuitry, and also building up the sawtooth-to-pulse
    waveshaper. In either case, I’d leave out the 4011-based
    ring modulator, and just bring the various outputs to the front panel
    instead of using the 4052 switch.)

  • Ray Wilson designed a nice
    4-pole OTA-based lowpass VCF.
    It looks
    vaguely based on the Polyfusion lowpass VCF. Polyfusion also made
    a highpass VCF, which I’ll send out a PDF of.
    You could build a highpass version of Ray’s circuit,
    using inspiration from the original Polyfusion highpass VCF circuit.
    (I think this would be interesting since there are very very few 4-pole
    highpass VCFs).

  • I’m curious to see if a third-order-with-feedback filter would be
    as interesting as a fourth-order-with-feedback filter. You could do
    one of the Ray Wilson/Polyfusion/type OTA-C designs mentioned in
    the previous bullet point with voltage controlled resonance if with
    just two LM13700s. If this sounds interesting, we can get together
    and figure out what the maximum allowable feedback it. (There is such
    a design out there already, the Ian Fritz Threeler, although I don’t
    think the schematic is available unless you buy a kit, so I just
    mention it to note that a proof of concept is out there.

  • Many years ago,
    I ran across

    obscure voltage controlled filter patent by Hammond
    years I had no clue what actual product this might correspond to. I recently
    discovered that it was probably the
    . I’m really curious how this sort of filters sounds.
    It’s a strange circuit; it constructs a lowpass filter by putting
    a highpass filter in a feedback loop.
    You could build
    Figure 1, using the values given in the table in the text of the patent.
    You could just use TL08x or TL07x op amps, and J201s or MPF102s for the FETs.
    R151 and all the circuitry to the right of R151 just looks like an
    amplifier, so I think you could leave all that out; you could replace R151
    a 1K protection resistor and then put the output jack right there, where
    C35 currently is. The only thing that would need added might be
    a buffer amp for the control signal input, and we might need to
    experiment with some gain on this buffer to give a useful control voltage
    range. (This would be a case where you wouldn’t worry about trying to
    get a 1 volt/octave control characteristic.)

  • Here’s the state variable filter of the EML-101;

    here is a redrawn schematic of it
    . It’s a strange circuit that uses
    differential BJT pairs to feed “differential integrators” that use op amps
    and two capacitors.
    To clone
    this, you’d need to completely replace the circuitry creating the currents
    for the transistor pairs of the integrators, since the uA 726 is long out
    of production (and I think the circuit they use is probably unnecessarily
    complex.) You could replace it with any number of the more modern exponential
    current sinks
    we’ve looked at in class or that you find on the web.

  • The Roland Jupiter-6 has
    a filter consisting of two cascaded state variable filters, as you can see
    this schematic. Each filter has an electronic
    switch that lets the musician choose between the highpass and lowpass outputs,
    so you can have a 4-pole-highpass filter, a 4-pole-lowpass filter, or a
    bandpass filter consisting of a combination of the 2-pole and 4-pole filters.
    You could replace the electronic switches with ordinary mechanical switches.
    Notice that the voltage controlled resonance is the same for each SVF.
    (A good one-person
    project would be to do one of these filters – maybe bring out all three
    outputs. A good two-person project would be to do both filters, wired
    Jupiter 6 style, with the highpass feeding the lowpass.)

  • There’s
    thread on electro-music
    building a Korg MS-20 sytle VCO, which includes some schematics of a few
    versions of a modularized adaptation.
    It has a sawtooth core, but it’s a “thyristor” design that is
    different than the sawtooth core we covered in class. (The sawtooth core
    would be a good one-person project; a full-featured oscillator that also
    included the

    waveshapers from the MS-20
    would be a good two-person project.)

  • The highpass filter part of the Buchla 191, which has four
    highpass stages,
    uses BJTs configured as diodes
    as variable resistors.

    Here’s a rough sketch of my suggested version of this idea.
    The idea
    is that changing CV* changes the DC operating point of the diode which changes
    the effective resistance, and AC signals will then see a capacitor through
    an effective resistance to “AC ground.”


  • The

    generates harmonics from a fundamental triangle wave
    through a series
    of strange waveshaping circuits. These circuits are quite strange;
    they create deadbands,
    not the way we saw in class, but through op amp circuits with diode bridges in
    the feedback loop. They are quite strange. Let me know if you’d like to take
    a look at some of them and try getting them to work.
    (This could be a one-person
    or a two-person project, depending on how many circuits are tackled.)

  • I ran across information about the
    1005 Utility VCO
    Jim Patchell’s website.
    I can’t seem to find any information about this company “Tau Systems” – my
    google fu is failing me. (Juergen Haible cloned the “Tau Pipe” phasor
    a while back, so most google hits are realted to that).
    Anyway, looking at the
    you’ll see it’s a triangle core like the Buchla 259 core we discussed in class,
    except a CA3080 is used instead of the “roll your own” 4-transistor OTA
    Buchla uses. The HA2500 op ap is acting as the comparator, and the
    FET and BJT at the output of the 3080 are acting as a buffer. I think
    the weird configuration of Q5 and Q6 is acting as a voltage clamp. It
    appears to have been intended to be some sort of
    module for a bigger circuit
    . (This
    would be a good one-person project.)

  • I have some H11Fx series optocouplers. How about a
    Buchla Lowpass Gate
    uses the optocouplers (which have light-dependent FETs that act as
    resistors) instead of Vactrols (which have light-dependent resistors)?
    Maybe the
    optocouplers would react faster to control signals than the vactrols,
    and give it a different sound. To start off, I’d use regular switches
    (connecting wires on the breadboard, really) in place of the CMOS

  • The Korg Delta has a rather interesting VCF; you can find it in
    the upper right corner (once you flip it sideways)
    of “Section 4: Circuit Diagram” in the

    service manual
    . It’s like a 4 pole cascade, using LM13600s (we would use
    LM13700s), but the feedback loops are
    weird, making it look kind of Sallen-Key-ish.
    I’m not sure what’s going on in this circuit but I bet it sounds
    interesting. Anyway, it looks like the
    audio input is buffered by IC-6 (or IC-8?
    I am having trouble reading it), and immediately hits a LP/BP decision switch.
    The the audio output is the output of
    the last stage. The 15K resistors R122, R127, R132, and R137 just make sure
    current is flowing through the Darlington buffers. The main control voltage
    is in the upper right, labeled “ext fc,” along with the “Cut freq” pot – you
    could probably drop the “Joystick” input. If this looks fun, I will meet with
    you and hammer out the details and go over the smeared and confusing parts
    of the schematic. You should also look at

    EFM’s version of this filter.
    (This is a somewhat big circuit, but probably
    still doable by one person.)

  • The Roland
    Jupiter-4 filter consists of a 1-pole highpass followed by a typical
    4-pole-lowpass-with-feedback. You can find it in the
    service manual;
    it’s on the top of one of the page
    that says “module
    board,” about 40 pages in. The LPF part is quite similar to the polyfusion
    lowpass VCF we looked at. You’d probably want to simply the various bits of
    control current generation circuitry. If this one sounds interesting, ping me
    and we will brainstorm. (This is a big circuit; probably would make a
    good two-person project).

  • We’ve discussed how some manufacturers used diodes (or diode-connected
    transistors) instead of BJTs as way to “get around” the Moog ladder VCF patent.
    One such company was Roland.
    You could try the
    Roland 100 filter, the
    SH3 filter
    (which actually has five stages instead of the usual four), or the
    SH1000 filter
    (to make sense of “Pack #2” you will need to see how it fits in to the
    big picture, as seen on the second page of the
    full SH1000 schematics). Or, you could try the
    4-pole lowpass
    VCF from the SH-5.

  • OK, I just
    had a crazy idea… We’ve looked at four-pole-LPF-with-feedback
    cascades using OTAs in place of resistors. We’ll also briefly talked about
    vactrols, which are light dependent resistors combined with a light
    emitting diode.
    How about putting a single LED in the center of four LDRs, placed
    equally spaced around the LED? (Some very old guitar pedals, such as the

    , used this idea). Those LDRs could then replace the resistors,
    and you could control all four stages at once just by changing the current
    through a single LED. It would be a sort of make-your-own super-vactrol. You’d
    probably want to cover up the
    LED/LDR setup somehow so external light wouldn’t effect it.

  • Buchla 291 Dual Bandpass VCF. To modernize it, I’d use
    TL08x or TL07x op amps all over. Also, see the dual FET pairs with the
    680 ohm or 150 ohm resistors; those are just buffers. To modernize it
    I’d make
    those amp amps set as noninverting buffers. (I’m not 100% sure what the
    topology is here; it looks sort of like a state variable filter with a

  • The so-called “biquad filter”
    (see Figure 5 of this
    ) is a variation of the state-variable idea. Notice it consists of
    a single-pole filter and an integrator; the op amp in the middle is just an
    inverter. I’m not aware of any synth
    designs using this topology; is should be pretty
    easy to make this voltage controlled via OTAs. The single-pole part could
    be done easily using the single-pole OTA we’ve seen in lecture, and we’ve
    seen many voltage-controled OTA integrator designs. I can help with the

  • Sorting through my hard drive, I found this

    transistor ladder that uses PNPs instead of NPNs
    I can’t recall where I found it!

  • The Korg MS-50 had a vactrol-based “MVCA” – see the last page of
    set of schematics

  • Guitar effects, like wah pedals and distortion circuits,
    are often a great source of ideas; sometimes you can
    take such an effect and
    replace variable resistors with vactrols or JFETs (used as a voltage

  • The very last slide of my “ladder filter” slide set showed a diagram
    from the Moog patent that shows his idea for creating a four-pole
    highpass filter; it uses the same overal idea of changing the
    resistance of BJTs, but is quite different in the details. Although everyone
    and their cousin has created variations of the Moog ladder lowpass filter,
    there aren’t many implementations. Examples include
    the Moog 904B
    and this
    (simpler looking) implementation by “EFM”.

    I ran across
    thread on electro-music

    Fernando: I have an schematic
    from Tom Gamble (EFM) describing a 4th order Moog HPF.
    But I can’t recall if it was based on the original circuit or was a guess.
    edit: it was a re-interpretation.

    yusynth: Yes it was and if the basic idea was fine the
    schematic contains some
    errors in some resistor values…
    I think the reinterpretation by Osamu Hoshuyama is more sensible:

    francois: I agree with you that Osamu Hoshuyama’s
    interpretation is more sensible, and clearer also.

  • Find the
    Memorymoog schematics, and find the circuitry for
    the VCF and VCA, which in the Memorymoog are hooked together.
    You could clone this combined circuit by putting your input right to the left
    of the 0.22 microfarad cap (C26, I think it is). Note the “40 mv p-p” notation
    on the diagram; you’d need to put some circuitry to buffer the input and
    divide your 10 V p-p input to 40 mV p-p. Use a LM13700 instead of the two
    3080s. For the control current generator for the ladder, you could simplify
    the control input circuitry. You’d also need to add a buffer for the
    audio output. The emphasis, filter cutoff, and VCA control
    in the Memorymoog are generated by
    a microprocessor. It makes sense to just buffer your VCA control input, but
    for the emphasis and filter cutoff, I’d add some circuitry to allow you to
    mix a control voltage with a voltage set by a knob.

  • Here are the schematics
    for the
    Moog Rogue.
    Check out
    the lower right corner of page 7; you’ll find the VCA/VCF, which is ripe for
    Most of the advice I give for the Memorymoog applies here. I’d leave
    out the master volume pot, and maybe change R145 (or is it R143?) to 20K to
    get a +/- 5 V output instead of a +/- 1 V output. If you use an LM13700
    of a 3080A, you’ll have a second OTA handy, so you could replace the pot in
    the feedback path with a VCA to give voltage controlled resonance.

  • The
    Buchla 227 System Interface

    has lots of VCAs made using VTL5C3/2 type
    vactrols. The schematics can be found
    Look at the Board 4 (Program Buffer) and Board 5 (Monitor Buffer) schematics;
    on each of those schematics sheets, you will see four copies of this kind of VCA
    circuit. I think it would be interesting to try building just one of these.
    On both Board 4 and Board 5, the circuit that generates the control current
    for the vactrol LEDs is on the bottom part of the schematic; in fact, it looks
    like these are the same. The VCA circuits themselves are also similar;
    they use a 6.8K resistor to connect the middle connected point of the vactrol
    resistors to ground. (This must give some shaping to the control that Buchla
    liked.) Both versions use a 12K resistor in the feedback loop an output op amp,
    although some other details around them differ (you could probably leave these
    details out at first).
    The input op amps (which form inverting amps) on Board 5
    have 20K resistors in the feedback loop and 30K resistors
    at the input on Board 5. When trying to make a stand-alone circuit, I’d use
    a 100K resistor on the input, and then try using 100K on the feedback loop.
    I’d tweak these values as needed to get “unity gain” out of the VCA
    When applying maximum control voltage to the CV input. I understand Buchla
    schematics are hard to read and my explanation may not make much sense; if
    you’d like to try this one, I can meet with you to describe what I’m thinking
    in more detail.

  • The Yamaha GX-1 (or equivalently the MOTM-485)
    uses a Sallen-Key filter using
    diode rings as the voltage controlled element. How about a state-variable
    or four-pole-with-feedback filter using these diode rings? (If you’re
    interested, I can get you the salient sections of the MOTM-485 schematic.)
    (Warning: a people tried this a few semesters ago. The state-variable filter
    never got working – the student had a lot of trouble with generating the
    current sink and source; the four-pole got simplified to a one-pole,
    which I decided was pleanty interesting, and I saw it sometimes working
    on the breadboard, but the final in-solder didn’t work. This is a really
    interesting circuit though so it would be cool for someone to look at it.)
    Here is the Yamaha patent on the diode ring idea; I think just getting
    one pole filter on the first page working would be interesting.

  • The timbre
    generator in the Buchla 259
    . Your signal input would come in where you see
    the sine wave squigglies; there’s two of them. I have some CA3130s (which are
    out of production) that you can use if you want to try this, and can order you
    a 2N3565. Look at the lower left where it has the symmetry control; I think
    your symmetry CV woulg go in that lower left input, and you’d have the pot
    between 0 and +15 setting the main symmetry setting. You would ignore the CMOS
    switch above it. The trick to this one is I’m really not sure what’s going
    on with the timbre CV. If you’re interested in pursuing this, let me know
    and I will poke at it more.

  • The Sallen-Key filters I know of in real synths are all lowpass
    designs. The
    SSM2040 datasheet shows schematics
    for true highpass a bandpass Sallen-Keys, which would be fun to try building.
    You could use LM13700s. Note that I think the op amp in the schematics may not
    be strictly necessary; I think it’s there to compensate for the fact that you
    don’t have access to the positive inputs on the OTAs on the SSM2040.

  • On
    the SSM2040 datasheet,
    there’s something called a Cauer filter. I’m not sure what that
    is, but it could be fun!


  • The Yamaha CS-80 filter
    consists of two cascaded state variable filters,
    highpass and one lowpass. You can find them on
    , in the right part of the diagram, labeled as VCF high pass
    and VCF low pass. They are known for sounding particularly smooth and
    creamy, perhaps because of a couple of extra parts that Yamaha threw into
    the 2nd-order feedback loop.
    They look quite compact ince they are using custom Yamaha
    IG00156 chips. Here is a
    high-level block diagram of
    the CS-80 VCF
    , and here’s
    Juergen Haible’s
    (The SPICE schematic looks a bit weird
    but don’t let that scare you – the triangle {gm1} symbol combined with the
    boxes with the little current sources in them appear to be his way of
    specifying an OTA, and the boxes with the little voltage sources in them
    appear to be regular op amps.)
    You can also find a similar filter structure in
    the CS-50 (see
    schematic here) and the
    (see schematic
    ). I haven’t looked at these closely enough to see if these filter
    circuits are identical, and if not, what differences there are.
    I’ve found descriptions of the IG00156
    and here (see page 19).

    I think it would be interesting to try to replicate this circuit. It looks
    to me like the IG00156 basically had some OTA and buffers and an expo,
    and we could guess what is where, and build it using LM13700s and a
    voltage-to-current expo converter cribbed from any number of sources.
    this sounds interesting, let me know and I will meet with you and we can
    try reverse engineering the circuit together; I have a pretty good idea
    of how to do this.
    (The CS-80 has two SVFs, with the highpass output of
    the first going to the lowpass output of the second. A good one-person
    project would be to do one of these filters – maybe bring out all three
    outputs. A good two-person project would be to do both filters, wired
    CS-80 style, with the highpass feeding the lowpass.)

  • I came across a
    video demonstrating the
    Farfisa Syntorchestra, which sounds fantastic
    . There seem to be two
    VCF circuits in the
    (you’ll have to rotate them clockwise by 90 degrees
    in your PDF viewer for them to make sense),
    one in the “Poli” section (upper right corner
    of page 6) and one in the “Mono” section (upper right corner of page 8).
    They’re quite strange. It looks like they’re using
    JFETs as variable resistors; this is usually easy to do in a predicable way
    if one side of the “resistor” is grounded, but here they’re “floating,”
    which is likely to induce all sorts of odd behavior. The Poli one seems to
    be an cascade of two RC stages, but without buffering between them, so
    the poles of the filter will be spread out a bit, and I don’t seem to see
    any feedback. The Mono version seems to have a feedback loop from the
    output back to the input, but that feedback loop itself has some
    unbuffered RC stages in it, and there’s a swtich that you can use to switch
    this feedback loop in and out. I’d love for someone to rig this up and hear how
    it sounds! (This would probably make a good one person project.)