This project is more of a way for me to learn more about electronics than it is a project to design a synth. That said, I very much hope a synth comes out the end of it.
Having been inspired by instruments such as the PPG Wave and Prophet VS as well as more traditional analogue synths, I wanted to see what the possibilities were for developing something similar using modern parts. The ultimate aim is a a digitally-controlled, all-analogue-signal-path polysynth (see the uPoly project) but this project will culminate in a standalone monosynth that represents one voice in the larger project.
I don't expect to break with mould with the synth's capabilities but I do want to make something that I'd be happy to play and to buy - and I want to buid a platform for future projects.
As far as learning about analogue electronics goes I suggest the following.
Get out your Horowitz and Hill 'Art of Electronics' and read chapter one, 'foundations'. If that goes well have a scan of chapter two, 'transistors'. This book goes to the right depth for what you (and I) need. Notwithstanding how well it went it's time to get the hardware out. Starting at a point where you felt it wasn't easy to absorb what was written, build something, ie connect some components together and measure things. For doing this I favour drawing pins [thumb tacks] and floorboard offcuts, it's closer to the original 'bread board', allows you to spread out the components like the theoretical schematic, and the soldered joints are quick, easy and reliable. A good place to start is with a single bipolar transistor, preferably one for which there is a lot of data, characteristic curves etc. (e.g. BC108). First try and measure the characteristics, get a feel for what they mean, plot a few curves, see how they compare with the given data. Next build simple amplifier circuits— one transistor and a handful of components — start with a common emitter. Explore different biasing arrangements, figure out how to measure gain, input and output impedance, frequency response and compare these results with the theoretical values for this circuit — as explained in H&H. Then try different single transistor configurations — common collector , common base — all the time thinking how these simple circuits can be used to make filters, oscillators, VCA, power amps, and how these circuits could form a part of your project. https://flic.kr/p/rxYrMx is an example from when I was trying this out with a differential amplifier. You're going to need some test equipment for this, one or more multimeters, A DC power supply, An oscilloscope - probably the most useful tool, nothing fancy required, you only need to measure low frequencies at this stage. Dual channel at least though, you'll want to see the input and output at the same time to compare amplitude, phase and distortion in real time. Remember that a 'scope is also a good high impedance DC voltmeter and even a single channel one can be used to measure phase difference using Lissajous figures. A signal source — basic audio signal generator — nice if it had sine, triangle and square wave o/p. If you don't want to buy it's not too difficult to build one from say an ICL8038, or perhaps one of the chips you intend to use in the project. As a bit of a challenge see if you can build the analogue portion of your block diagram — from oscillators to output — using a single discrete device for each block, Your choice of device, I'd favour valves, and, though there are some interesting, even exotic discrete semiconductors out there, it could all be done with bipolar transistors. Single transistor oscillators*, filters, VCAs (variable mu valves are still used in high end studio compressors on sale now), power amps are all possible and if nothing else will show why a few more (but not thousands) of active devices might be beneficial. Digitally controlling the result would be a challenge, probably nerdier than you'd want to go.
*single device negative resistance oscillators are interesting, eg the dynatron using secondary emission effects in a tetrode and the transitron using a pentode valve, which I've tried but not at audio frequencies, and semiconductor tunnel diodes — rare now and no longer cheap though there are ways round this, viz. the 'lambda' diode. Being in effect two terminal devices the tank circuit determining frequency is easily variable. Single device relaxation oscillators producing sawtooth waves are possible, gas discharge tubes eg neon lamps, gas thermionic devices eg thyratron (surprisingly plentiful, from big power ones to sub miniature ones used in early radar proximity fuses) and semiconductor devices eg unijunction transistors and thyristors.
All suggestions are welcome! I'm gonna post some more about the oscillator methodology on Monday when I get some hard visio time in again, as there are a couple more methods I'm looking at, including the one I'm likely to use. I'll also post up whereI am in hardware land.
As far as learning about analogue electronics goes I suggest the following.
Get out your Horowitz and Hill 'Art of Electronics' and read chapter one, 'foundations'. If that goes well have a scan of chapter two, 'transistors'. This book goes to the right depth for what you (and I) need.
Notwithstanding how well it went it's time to get the hardware out. Starting at a point where you felt it wasn't easy to absorb what was written, build something, ie connect some components together and measure things. For doing this I favour drawing pins [thumb tacks] and floorboard offcuts, it's closer to the original 'bread board', allows you to spread out the components like the theoretical schematic, and the soldered joints are quick, easy and reliable.
A good place to start is with a single bipolar transistor, preferably one for which there is a lot of data, characteristic curves etc. (e.g. BC108). First try and measure the characteristics, get a feel for what they mean, plot a few curves, see how they compare with the given data. Next build simple amplifier circuits— one transistor and a handful of components — start with a common emitter. Explore different biasing arrangements, figure out how to measure gain, input and output impedance, frequency response and compare these results with the theoretical values for this circuit — as explained in H&H. Then try different single transistor configurations — common collector , common base — all the time thinking how these simple circuits can be used to make filters, oscillators, VCA, power amps, and how these circuits could form a part of your project.
https://flic.kr/p/rxYrMx is an example from when I was trying this out with a differential amplifier.
You're going to need some test equipment for this,
one or more multimeters,
A DC power supply,
An oscilloscope - probably the most useful tool, nothing fancy required, you only need to measure low frequencies at this stage. Dual channel at least though, you'll want to see the input and output at the same time to compare amplitude, phase and distortion in real time. Remember that a 'scope is also a good high impedance DC voltmeter and even a single channel one can be used to measure phase difference using Lissajous figures.
A signal source — basic audio signal generator — nice if it had sine, triangle and square wave o/p. If you don't want to buy it's not too difficult to build one from say an ICL8038, or perhaps one of the chips you intend to use in the project.
As a bit of a challenge see if you can build the analogue portion of your block diagram — from oscillators to output — using a single discrete device for each block, Your choice of device, I'd favour valves, and, though there are some interesting, even exotic discrete semiconductors out there, it could all be done with bipolar transistors. Single transistor oscillators*, filters, VCAs (variable mu valves are still used in high end studio compressors on sale now), power amps are all possible and if nothing else will show why a few more (but not thousands) of active devices might be beneficial. Digitally controlling the result would be a challenge, probably nerdier than you'd want to go.
*single device negative resistance oscillators are interesting, eg the dynatron using secondary emission effects in a tetrode and the transitron using a pentode valve, which I've tried but not at audio frequencies, and semiconductor tunnel diodes — rare now and no longer cheap though there are ways round this, viz. the 'lambda' diode. Being in effect two terminal devices the tank circuit determining frequency is easily variable.
Single device relaxation oscillators producing sawtooth waves are possible, gas discharge tubes eg neon lamps, gas thermionic devices eg thyratron (surprisingly plentiful, from big power ones to sub miniature ones used in early radar proximity fuses) and semiconductor devices eg unijunction transistors and thyristors.