I'm working on an audio project, and I have two frequency synthesis needs to address.

1. I'm looking for full polyphony, so I need an oscillator per note (for a limited range of notes). (Square waves, which I can then use to obtain other wave forms.)

2. I'm planning ahead to future prototypes (and the first is barely even on the drawing board!) and I'd like to be able to generate the sine waves for the second through sixth harmonics as well as the fundamental, each at the same amplitude as the fundamental.

Requirement 1 I can solve with lots of oscillators, but it struck me that frequency multiplication or division might be more efficient and/or straightforward. The problem is that it's very much non-integer multiplication/division (let alone as simple as divide-by-two) and I can't just retune the single oscillator because of the polyphony requirement.

Requirement 2 I could solve with a sawtooth and a series of band-pass filters, but I'm not sure I can get sharp enough bands, and I'm not sure I can get the amplitudes equal. So, I was thinking of using a multiplier or divider on a square wave (at the fundamental or sixth harmonic frequency, respectively) to generate square waves the the right frequencies and then using a low-pass filter to extract only the fundamental of each wave.

I was rummaging on the Mouser web site a few days ago, and I found a programmable frequency divider. I think it was by TI or Fairchild. It took a clock input (I think it might have had an internal oscillator too) and came in a range of sizes (7, 9, etc., I think), which different numbers of output pins, each programmable for a different frequency ratio. It had a very big datasheet covering programming, etc., and described the internals as a series of PLL dividers.

It seemed ideally suited to what I'm doing, but I can't for the life of me remember what it was, and I don't seem to be able to find it on the Mouser site now. Any clues? Could be that the part's insanely expensive too, of course

Any other suggestions, either for suitable parts, or alternative strategies?

 

The classic electronic organ used 12 squarewave oscillators, for the top octave. They then used 12 flip/flop dividers to get the next lower octave, followed by 12 more for the next, and so on. In older organs these would have been discrete circuits, and not chips - even valves in really old ones!.

The original 12 oscillators would be free running, and because of this they weren't 100% 'spot on', this produced a tone that people liked.

An enterprising component company bought out a crystal controlled chip for it as well, but this wasn't as well received, as all frequencies were then totally accurate, and not so musical.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

Wow, that's obvious! Wonder why I didn't think of it - thanks

In the application I'm designing, I think I'll actually only need six oscillators. I can get the first, second, and fourth harmonics by divide-by-two from one oscillator, third and sixth from another, and fifth from a third. That make 18 oscillators. I think I can cope with that.

 

By "free running oscillator" do you mean something like the cross-connected transistor type astable multivibrator? I hadn't given much thought to deliberate fuzzy tuning....

 

I've never dealt much with electronic music producing circuits, mostly only audio amp circuits, but it makes sense about the high octave oscillators driving dividers for the lower octaves. However at some point is there low pass filtering required to remove the harmonics from a specific lower octave divider to produce a pure note ?

Lefty

__________________
Measurement changes behavior

 

Musical instruments don't produce pure notes, you're probably aware that organs have various 'instruments' - in an electronic organ these are done by switching different filters in to circuit. The filtering is likely to be more complex than the generation.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

Well, my plan is to have the square wave oscillator also produce triangle, sawtooth, and sine waves (the latter through low-pass filtering). Any of those outputs are interesting, but the sine wave is a bit dull musically. But it's much easier (I think) to do the frequency division on square waves, since you can use flip-flops, etc.

The idea behind generating the first through sixth harmonics is to then replicate the functionality of "tone wheel" organs (like the Hammond) which mix the harmonics in different amounts to create different waveforms. Each of those will need to be sine waves, so low pass filters come into play there.

 

That's how it's been done historically!. There are various organ building groups around on the net (or at least used to be), I would suggest looking for those?.

I suspect you already know that it's a VERY complicated process.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk

 

If it were easy, it'd be no fun