DIY Top Octave Generator

richardc64 · Posted: Tue Oct 07, 2008 11:48 am Post subject:

If existing CPLDs aren't dense enough for all 12 top octave notes, how about a set of, say, 3 chips, each covering 4 notes? Setting aside the top-octave obstacle, how about binary counters denser than the standard CMOS line, such as two independent 6-bit counters in a package?

And to whomever is bold (or crazed,) enough to attempt any of these schemes, please choose a package that doesn't need a $40 adapter to make it fit a .1"x.1" grid perfboard.

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Whoever does the project should probably have PCBs made since you won't find a DIP CPLD.

I would be shocked if it didn't fit in the smallest CPLD available.

Uncle Krunkus · Posted: Tue Oct 07, 2008 2:02 pm Post subject:

I'd much prefer to build it out of a dozen CMOS chips! Laughing

Why is that automatically not as good as a 50240?
It'd be a lot more fun. Shocked

blue hell · Posted: Tue Oct 07, 2008 3:59 pm Post subject:

It's just as good and even better in some aspects to build it from CMOS. It's a bit odd to do it that way in these times, but you'll learn good stuff from doing it, and soldering is a pretty relaxing thing to do.

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Too big, too power hungry, too much margin for error, too much soldering.

I understand analog synths, but CMOS (used purely digitally) over programmable logic? Razz

blue hell · Posted: Tue Oct 07, 2008 4:30 pm Post subject:

And where soldering is relaxing, programming fucks you up Shocked

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Ah, now I see. But I don't understand. Wink

blue hell · Posted: Tue Oct 07, 2008 4:52 pm Post subject:

It is ok to not understand, trust me, I have experience in that.

Of course it's madness to build the thing with CMOS, but well, people just did it 20 years ago, so why not now, as an educational exercise ... as long as it's not driven by digi-fear Wink

BOB-SNARE · Joined: Sep 26, 2008 Posts: 30 Location: Australia

There's definitely some logic optimisation, that can be done to avoid 12 separate 9-bit counters. An interval of a 5th up is 1.5x the frequency.
e.g. for a 9-bit counter (@1MHz) a D is 426 cycles, and an A is 284 cycles
426 = 1.5*284

I'm just downloading the Xilinx webpack, and I can test out a simple design and see if it fits in a cheap 5V XC9572 CPLD. These come in PLCC packages (for those that fear surface mount).

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Yeah, you should be able to do it without any programming at all now that I think about it. Just schematic capture.

Uncle Krunkus · Posted: Tue Oct 07, 2008 10:41 pm Post subject:

Rykhaard · Joined: Sep 02, 2007 Posts: 1290 Location: Canada

Coriolis · Joined: Apr 11, 2005 Posts: 616 Location: Stilling, Denmark

Just to flog that ol' dead horse again:

IT'S DIY, RIGHT? Wink

No doubt, from a commercial, parts-count, man-hours-to-solder, board-size, etc, perspective - CMOS isn't feasible.

No doubt, from a limited-time-for-hobby, choosing-between-soldering-or-programming-and-I-already-spend-my-workday-on-a-pc, kind of perspective - learning to program just isn't feasible... Razz

How about the uC guys do a version of this, and the CMOS-guys do theirs?
Could be fun!
Besides, the whole sdiy community seems to get terribly teary-eyed over the demise of the TOG's about ten times a year, so why not make it happen?

Of course I'll be watching from the side, being neither a programmer or cmos whiz... Embarassed

Erm..carry on!

C

Uncle Krunkus · Posted: Wed Oct 08, 2008 12:22 am Post subject:

Great attitude Coriolis! Very Happy

We could all gain from what can be learned here.
Believe me, I'd love to get into programming chips again. If it didn't cost me half of my spare time, half of my spare cash, and half of my spare brain cells, I'd be there in a flash! Laughing

BTW One of the secret (not so secret now Embarassed

) reasons for me wanting to do it with CMOS is that I'd like access to all those spare bits floating around! Very Happy

Dirty R/2R networks take up even more board space, and cost nothing! Yeah!!

frijitz · Posted: Wed Oct 08, 2008 6:58 am Post subject:

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Uncle Krunkus · Posted: Wed Oct 08, 2008 3:04 pm Post subject:

I'm quite interested in the 40103, and I just need to get a few to start some experiments. I'm thinking of using those 8bits for the smallest divisor, and then using the carry out or precounting to add a bit or two as they are needed. That way, I would be squeezing the most functionality possible out of each chip.

BOB-SNARE · Joined: Sep 26, 2008 Posts: 30 Location: Australia

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

I didn't realize how few flip flops are in the smallest CPLD. I may have to eat my words about putting a top octave generator in such a small part. Embarassed

But I still think it can be done. Smile

BOB-SNARE · Joined: Sep 26, 2008 Posts: 30 Location: Australia

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

What are you using a table for?

And are you using VHDL, Verilog, or schematic capture?

BOB-SNARE · Joined: Sep 26, 2008 Posts: 30 Location: Australia

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

Cool.

9 flip flops for the main counter, and 13 flip flops for the outputs. 10 flip flops left.

Excuse me while I brush up on some fancy combinational logic...or switch to FPGA. Very Happy

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

I didn't know much about top octave synths before looking into them a little today, but it seems like other dividers are needed to generate 5 octaves of each of the 12 or 13 notes. These should probably be included in a CPLD/FPGA solution, right?

That way I could get away with using something bigger than the smallest CPLD Wink

urbanscallywag · Joined: Nov 30, 2007 Posts: 317 Location: sometimes

I think the entire TOG + 13-5 octave dividers would fit into a $13 chip. It would require a small PCB to go along with it. Add in power regulation etc and this is getting expensive fast. Crying or Very sad

Sorry, I appear to have spoken before I knew enough about the problem earlier.