[sdiy] Top Octave Modular

[Elain Klopke]

Thank you all for your amazing information :)

On Mon, Apr 3, 2017 at 6:24 PM, <rsdio at audiobanshee.com> wrote:

> On Apr 2, 2017, at 6:04 AM, Elain Klopke <functionofform at gmail.com> wrote:
> > So, more stupid questions.... In having 8 octaves of notes to switch to
> 8 busses, each with their own switch, that's 832 switches to control. At
> first I thought that daisy chaining 74HC595s would work, and it probably
> would, but there has to be a simpler way of controlling all those switches.
> >
> > CPLD, FPGA, Raspberry Pi?
> >
> > I'm getting pretty comfortable with the analog bits of the design
> (except maybe being a little fuzzy on  how to turn the 0-5V output to
> something centered around 0V that other modules would be expecting), but
> the programming bits of this are still over my head.
>
> I suppose that you may already be aware that the vintage Hammond keyboards
> have 8 switches per key? I still haven't bothered to trace the complete
> signal flow diagram, but the Hammond organ maintains several audio busses
> for the various oscillators and drawbars, and then still needs 8 switches
> per key to bring in the right set of harmonics in the selected proportions
> to get the final sound.
>
> Of course, you don't necessarily need to be limited to 8 mechanical
> switches when a single switch can control any number of analog CMOS gates.
>
> As Tom has pointed out, if you stick to square waves then you have a lot
> of options for digital combinatorial gates that might work to simplify your
> circuit. If you have sawtooth (or even sine) waveforms, instead, then it
> still might be possible to simplify the switching.
>
> The question of CPLD versus FPGA versus R-Pi is best left as a decision
> after you've designed the logic. I recently designed a complex interface
> between two parts and found that the cheapest CPLD was still more expensive
> than two or three discrete logic chips. In my case, I used the 74LS32
> 2-input OR, 74LS27 3-input NOR, and a weird NL7SZ19 1-to-2 address decoder
> chip. Vintage logic chip designs like those are so cheap that a CPLD is
> expensive in comparison. Another huge disadvantage of CPLD or FPGA is that
> the chips do nothing when soldered on the board - you have to program them
> first, and that can be a pain for manufacturing. In contrast, OR gates and
> NOR gates do what they're designed to do without any extra programming
> steps. Just design the PCB, assemble the chips, and apply power. However,
> it depends upon the complexity of your logic. At a certain level, a CPLD or
> even an FPGA makes sense because we no longer build giant PCBs. If you want
> to avoid programming, that's another vote for discrete logic.
>
> By the way, most of those 74LS* parts that I mentioned have equivalents in
> the CMOS 4000 series, and those run on various voltages instead of being
> fixed to just 5V or 3.3V. Depending upon your mix of analog and digital
> signals, you might want CMOS logic chips.
>
> Brian Willoughby
> Sound Consulting
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20170403/dcb2ac63/attachment.htm>