my first dual-SID design

[thaber]

I finished the first design of my dual-sid synth and was hoping to get some comments or ideas on it.

This isn't exactly midibox sid stuff since it uses an AVR chip, but I believe it is related enough and

I could definitely use the insights of hardcore SID users.

I put the design on my webpage: http://research.edm.uhasselt.be/thaber/junk/sid.pdf

Some issues that remain:

• Is 16K memory enough?
  
• The feedback control R12 should probably be digitally controllable, but I don't have a digipot that is
  able to do this yet. I have some but they allow only 0-5V on the inputs, so I would have to change
  the DC offset to 2.5V.
  
• Is 16Mhz enough to drive two SIDs (all control surface things are off-loaded to another device/microcontroller)
  

[TK.]

Is 16K memory enough?

depends on what are you planning to do with the SID. So long your synth only transfers the sound parameters to the SIDs, 16k is more than enough. Once you are starting to add features like LFOs/Envelopes/modulation, wavetables, arpeggiators, external CVs, you propably need much more.

It also depends on your programming skills - if you know how to compress code and optimize the memory usage, 16k will be fine. But if you never completed such a project before, and want to develop under relaxed conditions, use the microcontroller with the biggest memory you can find.

My first MBSID started with a 16k PIC16F874, in the meantime I'm using a 96k PIC18F4685 - and the firmware will propably allocate this memory completely.

Is 16Mhz enough to drive two SIDs (all control surface things are off-loaded to another device/microcontroller)

You really need to be more specific when asking such questions.

Prophet64 runs on a 1 MHz 6502, and handles the graphical user interface as well as the two SIDs in parallel.

Best Regards, Thorsten.

[cloudstrife]

Hi,

the design and layout are simple to do, compared to software, and when you really want to make a 2 SID design, do it controlling each SID of its own with some adress-logic, this makes it easier expandable.

I had similar plans. My idea was a full featured 8voice synth. One SID makes out one oscillator. Externally I planned to add some CV'ed vactrol filters (oldskool, but easier to handle than a few thousand opamps behind each of the 8 voices.) External VCA for each voice with S/H registers. Every module is connected to a bus system with analogue data (for the S/H) and digital...

This makes it really large, but thats the way even Korg or Kawai built their synths in the 80s.

[qoute]

    * Is 16K memory enough?

    * The feedback control R12 should probably be digitally controllable, but I don't have a digipot that is

      able to do this yet. I have some but they allow only 0-5V on the inputs, so I would have to change

      the DC offset to 2.5V.

    * Is 16Mhz enough to drive two SIDs (all control surface things are off-loaded to another device/microcontroller)

[thaber]

I created my AVR DualSID over the weekend: the hardware is done and working, the software is mostly finished.

In the attachment, you can see what it looks like. Nothing fancy, this is really just a test project before I start designing and building a complete synth. The CAN controller and transceiver are currently left out, but they will allow communication with other synths or a control surface in the future.

I made a few changes to the first design, mostly simplifying things. The opamps were removed and replaced with a simple emitter follower like in the commodore. Feedback was also left out, since I can easily control it externally.

I noticed a significant amplitude difference between the output of the two sids (one R3 and one original it seems).

Is this normal? The original also gets a lot hotter. Both are functioning correctly though.

Greets,

Tom

[TK.]

Yes, this is normal. In the first years Commodore optimized the chip production process, with the effect, that the SIDs - especially the filters - are sounding different. Also the leackage varied (high leackage -> hot chip).

Only the newer 6582s and 8580s are sounding relatively equal, therefore (and also because of the better filters of course) this should be the prefered choice for stereo synths

Here two samples for comparison:

Two 6581 from calendar week 40/84 and 25/84 which sound very different:

http://www.ucapps.de/mp3/midibox_sid/mbsidv2_filter_6581.mp3

Note that the filter characteristic is not linear, therefore a simple linear-based calibration doesn't really help here.

Two 8580 (or to be correct: 6582) from the same batch (but my two 8580 sound equal to 6582)

http://www.ucapps.de/mp3/midibox_sid/mbsidv2_filter_8580.mp3

Note also the silence at the end (nearly no background noise compared to 6581)

Best Regards, Thorsten.