[sdiy] Help, I'm Desperate! (Charge Injection with DG408)

[rsdio at audiobanshee.com]

I do remember this thread, and I thank you for sharing your success. It’s way more enjoyable when a thread ends with a good report, rather than trailing off into mystery and frustration.

Brian


On Dec 31, 2018, at 12:24 AM, David G Dixon <dixon at mail.ubc.ca> wrote:
> Hi All,
> 
> Remember this thread?  Well, I have some progress to report...
> 
> I took Roman's advice and used SPST analog switches instead of multiplexers
> for my scanner.  I just panelized the prototype board, and it all works
> perfectly!  The switching is completely noise-free, as expected.  Also, the
> board worked the first time with no layout corrections needed, which is a
> bit of a surprise given that this is a 4" x 6" one-sided PCB which is
> completely stuffed (12 ICs and a pile of resistors and diodes).
> 
> This was not a terribly straightforward redesign.  First of all, the CV
> circuit was originally designed to convert a -5V to +5V ramp to four
> truncated triangles from 5V to 0V (5-0-5-0-5-0-5-0-5).  This was perfect for
> 9 channels, which my original multiplexing scanner did.  However, with the
> SPST scanner, each channel requires its own switch, and the switches come
> four to a chip (DG444), so this means only 8 channels.  Hence, I want my
> triangles to only go 5-0-5-0-5-0-5-0 (no final 5).  This means that I want
> to convert the ramp to 3.5 truncated triangles instead of 4.  This took some
> serious thinking, but was ultimately pretty easy to achieve, with only a
> couple of additional opamps.  I designed it so that if the incoming CV
> signal is lower than -5V, the scanner will output Channel 1, and if it is
> higher than +5V, the scanner will output Channel 8.
> 
> As for the switching circuit, it required 6 comparators (LM339 + LM393), 8
> XNOR gates (2 4077), and 8 SPST switches (2 DG444).  I used hysteresis on
> the comparators, and wired things up using Roman's "Fat Dot" concept.
> Actually, the channels fall on two CV busses -- A and B (A is on when the
> 2164 CV is 0V and B is on when the 2164 CV is 5V).  Channels 1357 are on the
> B bus (5V), and channels 2468 are on the A bus (0V).  Rather than thinking
> of the XNORs as providing fat dots, if one simply sees channels 1357 as one
> circuit, and 2468 as another circuit, with the XNORs alternating between
> them, the one has two 4-channel logic circuits with regular skinny dots.
> Either way, it works.  Also, to clarify Roman's schematic, the top and
> bottom comparators are not necessary.  The top one can be replaced with
> Ground, and the bottom one with 5V.  Hence, only 6 comparators are needed.
> 
> Also, I opted for "negative logic" in this circuit, and thus used XNOR gates
> (4077) and DG444.  However, it would work just as well with "positive
> logic", which would use XOR gates (4070) and DG445.  These are
> pin-compatible with the others, so either could be used on the same board.
> 
> Finally, the "client" had originally wanted LEDs to indicate the channels.
> I had deemed this impossible with my original circuit.  However, it is very
> straightforward with this new circuit.  The XNOR outputs provide "ground"
> for the LEDs, and the truncated triangle CV signal drives the LEDs.  Since
> only one XNOR output from either the 1357 or the 2468 side is on (at ground)
> at any given time, only one LED from each bus will shine.  The "inactive"
> LEDs have 5V instead of ground at their negative end, and therefore cannot
> conduct.  The triangles cause these two active LEDs to fade one to the other
> in exactly the same fashion as the audio signals themselves.  Really, this
> works better than I could have hoped, and it was very easy to do.  I just
> had to invert and level shift the truncated triangle CV signal so that one
> goes 5-0-5-0-5-0-5-0 while the other goes 0-5-0-5-0-5-0-5, and use these two
> different signals to drive the 1357 and 2468 LEDs, respectively.  If
> "positive logic" were used instead (with 4070 and DG445), then one would
> simply have to swap the CV signals around and wire up the LEDs the other
> way.
> 
> I'm going to make some kind of video to show how this works, and I'll put up
> a link here for it.  It will probably be tomorrow.  I also plan to create a
> big long thread on Muff Wiggler explaining the entire circuit in detail with
> schematics and stuff, as a sort of tutorial on how I approach design.  I
> think that might be pretty useful for other DIYers.
> 
> Thanks again for all your help, especially Roman!!
> 
> Cheers,
> Dave
>