PolyModular Standards

[Haible Juergen]

	>From:	Harry Bissell [SMTP:harrybissell at netscape.net]
	>
	>Hi. I just thought I'd piss off everyone on the list, starting with
JH.

(;->)  (;->)  No problem ...

	>The first thing to do would be to define a module STANDARD
(remember that
	>fiasco...)

Hmm, yes, you're right. I was aware of it as I typed my mail to start this
thread.
I'm breaking two of my own vows, not to work on a standard, and not to
participate
in pcb stuff. Oh well. But things look different when you start to get
emotionally
involved in something again - as it is the case with a PolyModular (PM) for
me.
So yes, we have to find standards of some kind.

	>We might (optional) want to agree on some kind of form factor. If
the pots are
	>pcb mounted, this is vital. 

Probably not that important here. It *is* the best way for a monosynth, no
doubt, but as you only distribute DC from a front panel pot in the PM, there
is no such need. Rather the contrary.

	>I'd like to suggest that we mount the DIN jacks on
	>the PCB (there are great inexpensive PCB mount jacks) 

Mechanical stability is most important. Are there pcb mount jacks that are
fixed
to the panel nevertheless ? I'd even say mounting these to the pcb is not
*that*
important, but it can't hurt if we do. Mechanical forces must be applied
to the front panel, and not to the PCB, in any case.

	>I think there might be fewer VCA functions than you think. There
are lots of
	>good voltage controlled (name of function here)... like VC-ASDR...
They would
	>only need standard pots... only the output level needs a VCA.

You hit the point. VCAs at  the *outputs*. Let's use the VCS3 concept here
(in an
adapted form). One output level pot (5 VCAs) for each module.
Inputs are just connected
in parallel (summing node concept), and take what ever level they get. Make
these
inputs as "soft limiting" as possible, so it's completely up to the user to
drive them
for 0.1% or 10% distortion.

	>Personal preference (no need to do the merits post again, I read
them all) for
	>V/oct. If we really use a modular format, maybe I'll make a
	>pentaexpoconvertermodule. V/oct was always the de-facto standard,
and V/hz the
	>orphan cousin...

Hm, what I want is to do a "best of both worlds". Yes, you can use your
standard
expo VCO. But use the linear FM input (reference current ...) for the
Keyboard
voltage (KOV), and you get rid of many drift problems. I mean the KOV that
is
routed over a bus (and not over patch chords) would be the only V/Hz signal
in
the whole system. And you still can choose *how* to make these V/Hz KOV's
in the first place. My idea was a special DAC, but I like your idea of a 
penta-expo-converter even better ! And here is the place (the *only* place
IMO) where I'd like to see Ed's suggestion of a multiplexed system. I
wouldn't like to see
MUX and clock stuff inside modulation and feedback loops, but  for the mere
KOV it's ok, and probably the best solution. I'm fascinated by this design
in the
Korg Polysix: One rather cheap expo converter is enclosed in a MUX/DEMUX
system, and one of the channels is a reference channel that will adjust for
temperature drift with a servo system, the variable resistor not being a
tempco resistor, 
but a Vactrol.
But that's just an idea, and you might come up with a better one (Vactrols
aren't considered "easy" components either !). But the bottom line of it all
is that:
If you choose to do V/Oct  to  V/Hz conversion at all, then do it at *one
point*
for all voices, and do it just immediately after the keyboard (and after a
possible 
sequencer / transpose input).

	>I like the idea of six voices... 

The same for me, but what connector would you use ? 
Using cheap standard cables is one of the cornerstones of the whole concept
...
To tell the truth, I would even go for 4 voices if there wouldn't be a cheap
5-pin connector. I rarely see the 4-voice polyphony of my CS-50 as a
limitation.

	>Maybe someday, someone will make a guitar to
	>CV converter that really works... like in REAL TIME (not 30mS into
the
	>past...). 

You're getting in conflict with nature's laws, not just technical skill,
here ...

	>The DIY-PCB group might want to play... There are some good panel
layout ideas
	>there...

what was the adress to subscribe again ? (got me)   8-)

	>Juergen H: maybe you could think a bit about what some signal
STANDARDS (oh
	>yeah I forgot I said I'd not use the "S" word again) might be. I
mean
	>suggested (oops thats "S" too...) recommended?? Anyway... voltage
levels for
	>Audio signal and CV. I know the fun of a modular is to plug
anything into
	>anything... but a two transistor VCA is not going to handle 10V
signals
	>without being unhappy. I'm kind of wide open on this one... I like
10V
	>signals, but some folk like 5V, etc. 

Here we go. I suggest a standard of  100uA @ 0.1% THD as a rule of thumb.

What, "uA" ?? Yes. That's the way to go if we have output attenuators rather
than input attenuators, isn't it ? And these DIN connectors have a long
history of high impedance output  /  low impedance input connections, too.
Maybe we cannot realize *perfect* virtual GND summing nodes as inputs
(you might need some small series resistors to avoid HF problems), but
this should show the direction. OTAs for the VCA functions of module outputs
anyway, so why even add voltage buffers there ?
Disclaimer: OTAs does not necessarily mean 3080's, and "no buffer" might
still allow a finite impedance / load resistor in parallel (just as a
"voltage"
system ususally has 1k series resistors at the output ...) But I want to go
current design, not voltage design, here !

	>If we had signal standards, then we might form design teams.... We
have a wide
	>rande of talent in this group, but I know I'd feel more comfortable
with ENV
	>GENs and LFOs than 5 or 10 VCO that must all track...

I can offer to provide the circuits for the first few basic modules (VCO,
VCF, VCA),
which would in a way *set* a kind of electrical standard. Then it would be
easy to
add other modules, and more sophisticated VCOs and VCFs, just using the
same input / output structures. What I'm asking is that someone makes the
layouts for these first modules after I have tested them in simulation, and
sends
me a set of pcbs. After that, we will know if the concept is working. If
not, so be it. 
If yes, we're off the ground and I everybody can bring in his talents for a
variety of modules.
I wouldn't dare to do such a thing normally, but with this list, I see a
chance.
There is some risk to do it all in vain for me, and for the one(s) who
volunteer
to make the first PCBs, but after that, it could be a really great thing.

	>I just read a "get a life... get a micro" post. I don't think they
are a
	>necessity but I could be convinced... The problem is the lack of
support a lot
	>of us have for micros... 

I already sent a reply to Barry and the list, but I will say it more
pointetly again:
If there is someone who does the keyboard scanning and/or Midi stuff with
a uC, this is the best that could happen on that part. Personally I have
no clue here, so I cannot even decide who should do it. But this should be
done by someone who is familiar with uCs in a way that he can do it
"blindly".
I'm sure we have several people who qualify for this on the list. (But if
you consider
it a "challenge", you should probably *not* do it.)

	>Maybe someone wants to head a mirco interface team
	>and work with the analog folk on some standards for handshaking.
Maybe a
	>system that just reads the analog voltages and can reproduce them
at will. But
	>(just being playful here) I dont want to build a modular DX7
either. The
	>analog system has infinite resolution... 7 bits won't get it. The
keyboard is
	>a good place for a micro....

Exactly my thoughts. Could not have said it better. 

	>Well I bet I've pissed of everybody enough by now. When the dust
settles we
	>should have a real good picture of what this machine might be...
Did I say
	>"when"?  I meant "IF".   ;-) Harry (V/oct) Bissell

You will get your V/Oct inputs. They are an (almost) free side effect of the
VCO
scheme I have in mind (Very MS-20-influenced). Just make 5 resistors tempco
rather than normal types (and live with more drift, of course (;->) ).
I hope that *I* will get "my" V/Hz KOV from the uC guy (or from the 
"pentaexpo" guy ) as easily (;->).

JH.