A New DIY Synthesizer Project

A New Synthesizer Project
First Posted 11-23-2002
Updated 9-18-2004

Modular Synthesizer

Modular Sequencer
Logic Module
Memory Module
Sequencer Module

Updated Schematics
Updated 3-28-2004

You will find a directory of schematics here.

Modular Sequencer
5-9-2004

This has been an ongoing project now for several years. I am finally starting to get everything designed. Originally, this was going to be a single 8U high panel, but, it has sort of expanded into three panels now. Such is the way of Feature Creep. I am probably going to be the only one who will enjoy using this thing, because in order to make it you, you must patch it. There are no default settings. Plus, in order to patch it, you do have to understand at least the basics of logic design. The system consists of a whole bunch of logic primitives that can be connected together with bannana plugs.

    One of the primitives, the Adder/Subtractor, is not really so primitive. It is more like a four bit ALU. It has three inputs that control the function of the outputs. Currently, it performs the following 8 functions:
    C=A+B
    C=A-B
    C=A
    C=A xor B
    C=A and B
    C=A or B
    C= 0000
    C = 1111

    The rest of the logic functions in the modules consist of AND, OR, XOR gates, registers, JK flip flops, decoders, shift registers.

    In the Logic module there are 8 4 wide, 4 input, AND-OR logic arrays. (See Below). These have programable (with a switch) logic inversion of each input. This module should come in handy.

    There are also a couple of 8 bit DACS as well as one 8 bit ADC.

    In the Memory module, there is a 16x8 pattern generator (a ROM that uses toggle switches as the memory element), as well as a 2048 x 8 ram module that can be loaded via RS-232.

    Since some of the same PC boards are used behind multiple panels, here is a list of the current schematics and PC board layouts.

    Pot Selector
    Adder Subtractor/Register
    Dual 4 input AND/4 wide Or
    Misc Logic Board I

    Pictures
    More to come....

Memory Module

I am calling this module a memory module. A picture of the module can be found here. Again, just a drawing of the front panel. Hopefully, I will start construction soon. The main feature you will note on the panel is the Pattern Generator. This is a 16x8 array of toggle switches. The ABCD inputs select one row of switches, and their state can be read out on D[0..7]. The second most prominent features are the RAM modules. These are 2048 x 8 bit ram modules that can be loaded via an RS232 port. I was debating if I should use MIDI to do this, but in the end, decided on RS232....I might change my mind before I start constructing, who knows. There are also a pair of DACs that can be used to provide control voltages to other analog synth modules.

The module is rounded out by what will hopefully be some useful random logic blocks. There are four Adders (really more like a mini ALU) that can perform ADD, SUB, A->OUT,XOR,AND,OR,0->OUT,1->OUT functions, and there are four registers, which in combination with the Adders can be used to make counters and what not. There are a pair of 2 to 4 decoders, and a pair of And/Or/Xor gates.

Logic Module

   A document of the module can be found here, it includes schematic as well as the front panel. As of today, nothing has been fabricated and I am probably going to change a few things before I start building. Hopefully

    Most of the panel space is occupied by the 4 input, 4 wide AND-OR modules. There are 8 of them. Each of the 16 inputs has a toggle switch associated with it. This toggle switch determines if the input is inverted or non-inverted. While these logic units are not very deep, when compared with a macro cell in say a 22V10, they should prove to be quite flexible.

    The next submodule are the adders. These are 4 bit Adder/Subtractors. There is an input that controls whether they add or subtract. There are also Carry In (CI) and Carry Out (CO) jacks so that the adders can be cascaded. Adder/Subtractors are used primarily to implement counters, using the Register Modules.

    The Register submodule is a 4 bit D type flip-flop register. It has a clock input (CLK) that is positive edge triggered and a reset input (RST) which is active high and resets the register to 0000.

    There are a pair of JK flip flops that can be used to extend the operation of sequencers. JK flip flops are the most versitile. There is a switch on the K input to make that input either Active High or Acitive Low. If it is Active low, you can connect the J and K inputs together and make a D flip flop. If you set the switch to active high, you get a T flip flop when you connect the J and K inputs together.

    Another module is an 8 bit A/D converter. It has a voltage input, that can range from 0 to 5 volts, or there abouts. There is a clock input ON the panel for the A/D converter, but, that will be removed..(I found no way to do this nicely). There is a signal for starting the convert, and another to indicate when it is done.

    And another module is a shift register module. These can be useful for generating pseudo random sequences.

    Update 7-20-2004

    Got the front panel yesterday, and I spent the day putting the part in it...I was off because I had a minor surgical proceedure....and I was supposed to be taking it easy....

    Anyway, here is the photo.....

    Update 8-15-2004

    Started to wire this panel up this weekend. Got the And-Or matrix wired, took about 12 hours so far this weekend.

    Anyway, here are the photos.

Sequencer Module

      This is what most of you will actually recognize as a sequencer, since, this module has the pot knobs on it. A picture of the front panel can be found here.

    This module consists of a selection of sub modules that can be patched together to create sequencers. Most of the panel space is occupied by the Pot Multiplexors. The inputs A, B, C are used to select one of the 8 pots. The IN jack can be used to add another control voltage to the default 5 volts that is built in. So if you apply 5 volts to the IN jack, that will be added to the 5 volts that is already there, making for 10 volts. In this particular case, the knobs will adjust the voltage on the OUT jack from 0 to +10 volts, rather than the default 0 to +5.

       One thing should be noted. The sections that contain the pots only have logic inputs that select the pot. To make it work like a traditional sequencer, you can use all of the logic primitives that are also on the panel (Adder/subtractor, 4 bit registers, XOR, JK flip flops, decoders) to create counters and logic sequencers. I did this because it allows me to have ultimate flexibility in how the system is configured. Although, this will make for a mass of patch cords...but, I like chaos.

       The pot sections have an output and an input. If you connect a voltage into the input jack, then the output will be that voltage attenuated by the selected pot.

    The decoder module is handy to select the various pot multiplexors. There are 3 in 8 out and 2 in 4 out. These decoders are active high. Only one output will go high for each input combination.

       There is also a midi interface that will provide for START/STOP (same jack), and RESET functions, as well as RAW midi clock and midi clock divided by 7 different constants.

Modular Analog Synthesizer

9-18-2004

A new Chassis

    Well, here it is almost a year later, and I finally got around to making the chassis for the Modular Synthesizer. I was going to use a Bud 17x14x3 inch box, but, it was too small in two dimensions by about 1/8 of an inch. So, I got my rear in gear, and purchased some sheet metal to use with the tools I purchased last year when I was up at the PNW SDIY meet (purchased a Grizzly Shear and Corner Notcher).

    So, here is a picture of the Parts.

    Still not really done at this time...more updates to follow...I hope.

12-16-2003

Latest Updates

    Well, I am still pretty happy with this synthesizer. I worked on the firmware a bit more. For those who may be following this, the uController on the Midi->CV converter is a AT90S8515. I will be changing this to a MEGA8515 in January of 2004...I hope. The MEGA part is capable of 16MHz and it has a multiply instruction.

    The VCO's sure do seem to be about as rock solid as you can get. While I have not actually checked to see what their scale factor is, all three oscilators are tracking in their tuning. These are the first oscilators I have ever had do this. In fact, I am suspecting that they are even more stable than the CEM3340s that I used to use. I will have to fire up a synth that used these parts to find out if that is indeed true or not.

    I am also hoping to get a Real Time Operating system running on the CPU. This is probably not going to be easy. I could use Salvo, but, I am too cheap to pay even that low price for a Real Time Operating System. I am planning to port the RTOS I wrote for the 68000 over to the AVR.

11-30-2003

    Well, I have pretty much got this synth completed, in that I can use it to play a tune. I met pretty much all of my goals. The Recuring Labor on this synth came in at about 40 hours, which was my goal. It did take a bit more time to actually build this first one, because there were a few boo-boos on some of the cards. All of them were minor, but it still takes some time. I still have to finish the code for the Midi->CV converter yet. But that is about 80% complete.

    I am very pleased with the way this project turned out. It is clean, and it sound, in my opinion at least, real good. I still have to build a box to go around it all. In order to do this, I need to set up the sheet metal tools I purchased during the summer.

    So to start off with, here is a 15 second sound sample I made from the synth....

    MP3 Sound Sample

    And here are some pictures of the assembled unit.

    Rear of panel (PC boards and wiring)....

    Closeup of Rear

    Front of panel

    This is all of the point to point wiring

    Side view of rear

9-2-2003

Here it is, September of 2003, almost a whole year since I started this project. One of the goals was to design a synthesizer I could build in a reasonable amount of time. So far, it looks like I am going to meet that. I have about 20 hours of work into the build of this project, and a good amount of the work is finished. You can see the photos below for the details. I do have to build a jig yet to solder the panel pots to the boards, but, that is just a detail.

In the photo of the boards, you can see pictures of the Midi->CV converter, 3 VCOs, 3 Multimode VCFs, 2 ADSRs, 2 VCAs, 3 Diode wave shappers, 3 CV processors (gain/offset), 1 noise generator (32 bit), 2 LFOs, and 2 4in/1out mixers.

Picture of PC Boards

Picture of Front Panel

Picture of "Killer Bees"

One of the problems I have been experiencing with my latest synthesizer projects is the amount of time it takes to do the front panel wiring. There are several reasons for this. The biggest one is I just can't see as well as I did about 20 years ago, and that really hampers things. So of course the big question in, what can be done to build a modular synthesizer module without having to do a lot of tedious wiring. Well, mounting the jacks and the pots on the printed wiring board is of course one answer. I discovered that this was a lot easier said than done. The jacks were fairly easy. I am using the Switchcraft SC112 phone jacks. Pots, on the other hand, are a big problem. A problem that is only partially solved, at this time. Currently, I am using the Vishay Model 248 pots. These are a bit pricey. But, the feel good. There is sort of a problem in mounting them to the board, in that the ceter of the pot does not line up anywhere close to the center line of the Jacks. I plan to solve this by making a jig to solder the pots in the board with. I would first mount the Jacks, the mount the board in the jig using the jacks to support the board, and then solder the pots.

Also, this will be the first Synth project that will use a panel made by Schaeffer. It will cost about $400 to get this panel made (I use a 14" x 19" panel), but they look great, and besides, it takes me somewhere between 10 to 20 hours to make a panel by hand. And these days, finding 20 hours of spare time is getting more difficult. The rest of the chassis will be a 14" x 17" x 3" aluminum Bud chassis box. I put pem nuts in the lip and then screw the front panel to the box. Looks pretty clean. The first module I did this with was the prototype of the Mikado. The only problem I have not solved is where to put the power supply...

    3 VCOs
    3 VCFs
    2 ADSRs
    2 4in1out Mixers
    2 Simple CV processors (Gain, Offset)
    3 Complex CV proecssors (Gain, Offset, 2 break points)
    1 Noise Generator
    1 Midi-CV Interface
    1 LFO
    1 Balanced Modulator
    1 Glide

The PC boards are 2.75 inches by however long they have to be to accomodate the number of jacks and pots the module needs. The Midi->CV converter actually got a couple of jacks added to it because I had to stretch the PC board out an extra two inches to accomodate the components.

This VCO uses the temperature compensating circuit that several others (especially Jurgen Haible) and myself were able to come up with. The core of the VCO is pretty much a standard sawtooth VCO, often refered to as an ASM-1 type VCO. Actually, this VCO was originally concocted by Terry Michaels and was published in Electronotes. The VCO also features a fairly sophisticated Triangle to Sine converter using a CA3280. Also, the VCO has an onboard driver for powering the main Frequency Adjustment control. It is intended that a multi turn pot be used here. The onboard driver has a gain control so that it is posible to set the VCO up so that 1 turn of the pot equals one octave exactly. There is also an auxiliary connector on the VCO so that it would be posible to have a master pitch adjustment knob as well as a master PW knob.

This is a fairly standard Voltage Controlled State Variable Filter (Multimode). It features a temperature compensation circuit similar to the VCO. Limiting is accomplished using a pair of LEDs. The LEDs can poke through the front panel so that you can get an idea of when the filter is going into limit. This filter also features voltage controlled Q control as well.

This circuit is by far the simplest of them all. It has 1 input and 1 output. You plug the output of a module into the input, and you can vary the gain, from -1 to +1, as well as the offset, from -10 volts to + 10 volts. Very handy for modifing the output of an ADSR.

Similar to the simple processor, except, you can set two break points where you want the gain of the module to change. In this way, you can shape an incoming signal if many ways. This module was primarily intended to be used with the filters to create phonemes.

The noise generator I chose for this projects is a 32 bit maximum lenth shift register type. It has an internal clock, but, this can be over ridden by using the clock input to supply it with another source. This is handy for generating random sequences of notes. Also, there is a reset input that will set the shift register back to its initial state so that the sequences can always be the same if you wish. It would have been cool if it were posible to set the "seed" that it reset back to, but, that would have been too complicated.

This modules uses an Atmel AT90S8515 micro controller (AVR). The board has an ISP connector on it so in therory, if you wanted to do some hacking, you can program this thing yourself. Other features of the board include 8 control voltages. Four come out on jacks, the other four are on an aux connector on the board. The standard firmware also support Midi Clock, RESET, and START/STOP.

Please Note, as of 11-23-02, I have not checked this schematic for acuracy. The Midi connectors could be backwards.

This is the quantizer out of the Mikado. I am not sure if I am going to include this in the synthesizer of not. Because of the way I designed the Noise generator, I probably don't need this module.