From analogue-request@magnus.acs.ohio-state.edu Fri Apr 23 16:53:04 1993
Received: by quark.magnus.acs.ohio-state.edu (5.65/3.910213)
	id AA12190; Fri, 23 Apr 93 16:52:24 -0400
Errors-To: analogue-request@magnus.acs.ohio-state.edu
Sender: analogue-request@magnus.acs.ohio-state.edu
Received: from att-out.att.com by quark.magnus.acs.ohio-state.edu (5.65/3.910213)
	id AA12185; Fri, 23 Apr 93 16:52:22 -0400
From: wbf@aluxpo.att.com
Received: from alux1.cnet.att.com by aluxpo (4.1/DCS-aluxpo-040293)
	id AA07155; Fri, 23 Apr 93 15:42:58 EDT
Received: by alux1.cnet.att.com (4.1/DCS-aluxpo_client-022293)
	id AA28057; Fri, 23 Apr 93 15:42:59 EDT
Date: Fri, 23 Apr 93 15:42:59 EDT
Original-From: aluxpo!wbf (William Fox)
Message-Id: <9304231942.AA28057@alux1.cnet.att.com>
To: analogue@magnus.acs.ohio-state.edu
Subject: Re: Programming Tips (fwd)
Status: OR

| mike metlay * atomic city * box 81175 pgh pa 15217-0675 * metlay@netcom.com
|
| >It wasn't until years after both of those events that I read that by
| >making the modulator the SAME frequency as the carrier, you
| >generate the usefull result of HARMONIC overtones.  <LIGHT BULB TURNS
| >ON>  So when the two frequencies don't match, the overtones are
| >INHAMONIC like in bells!
|
| Not just the same, but any integer multiple. 

I knew that but I incompletely stated the situation in an attempt for
clarity of (my own) thought.  I guess I blew it in the outside world!

| >               ...  FM synthesis on an analog synthesizer; what a
| >concept!
|
| Bill, this may have just made your "what do I buy next" question a lot
| simpler. The Xpander does this straight out of the box, and a lot of
| well-known Xpander users like ex-KEYBOARD staffer and recording artist
| Ted Greenwald believe that it's the last great unmined motherlode of
| neat new sounds for the beastie. (My vote goes to the Lag Processor.)

Vas iss loss Lag Processor?  (Please pardon my phonetic pig-German!)
And with the wealth of DX/TX patches floating around, Ted Greenwald
could spend years porting patches over to the Xpander, even being
selective about which ones to port!

BTW, Mike, thanks for answering my post on the VS/Xpansions lists and
for keeping my needs in mind while answering this programming tips
thread.  I may as well make it official and post that I'm looking to buy
an Xpander (at a "reasonable" cost).  Anybody out there know of one for
sale?  Analogics hasn't returned my call (I only called yesterday) and
Media Sonics is out of the office until Tuesday.  Rogue, as Mike pointed
out to me, is WAY overpriced since I don't HAVE to have one IMMEDIATELY.

Bill Fox

From analogue-request@magnus.acs.ohio-state.edu Fri Apr 23 22:19:57 1993
Received: by quark.magnus.acs.ohio-state.edu (5.65/3.910213)
	id AA22825; Fri, 23 Apr 93 22:17:43 -0400
Errors-To: analogue-request@magnus.acs.ohio-state.edu
Sender: analogue-request@magnus.acs.ohio-state.edu
Received: from acid-rain.lucid.com by quark.magnus.acs.ohio-state.edu (5.65/3.910213)
	id AA22820; Fri, 23 Apr 93 22:17:39 -0400
Received: by acid-rain.lucid.com (4.1/SMI-4.1)
	id AA13343; Fri, 23 Apr 93 19:17:36 PDT
Date: Fri, 23 Apr 93 19:17:36 PDT
From: till@lucid.com (Don Tillman)
Message-Id: <9304240217.AA13343@acid-rain.lucid.com>
To: analogue@magnus.acs.ohio-state.edu
In-Reply-To: wbf@aluxpo.att.com's message of Fri, 23 Apr 93 09:33:42 EDT <9304231333.AA19875@alux1.cnet.att.com>
Subject: Programming tips...
Reply-To: till@lucid.com
Status: OR

   From: wbf@aluxpo.att.com
   Date: Fri, 23 Apr 93 09:33:42 EDT

   When I took electronic music at Ohio State, like every novice, I ran a
   VCO's output directly to another VCO's FM input.  I "discovered" that if
   the first VCO's frequency was subaudio, playing the role of an LFO, I
   got this cool sound.  But if I ran the frequency up into the audio range
   where you start to hear sidebands instead of the frequency sweeping
   effect, it sound horrible and was "totally useless."

   When Yamaha came to Columbus, Ohio in the personage of Bo Tomlyn (sp?)
   to push the DX7, I asked the "simple" question, "What IS fm synthesis?"
   After a few brush offs, and through my persistence, he finally told me
   to go read Chowning.  Well, we were so pissed at each other that even if
   I'd have been able to understand what name he said, I was incapable, due
   to my state of mind, to remember who he said, anyway.  Only in hind
   sight do I know what he said.

Well yeah, I can understand; "What is FM sythesis?" ain't a simple
question.  (But what's the *real* reason you two got so pissed at each
other?)

   It wasn't until years after both of those events that I read that by
   making the modulator the SAME frequency as the carrier, you
   generate the usefull result of HARMONIC overtones.  <LIGHT BULB TURNS
   ON>  So when the two frequencies don't match, the overtones are
   INHAMONIC like in bells!

   Using my TX81Z to do this, I now wish I could take what I know now back
   to the OSU emusic lab and patch a VCO through a VCA (and maybe even a
   VCF!) to feed the FM input of a VCO->VCF->VCA.  (Both VCOs set at
   sinewave.)

	   VCO1->VCA1------>VCO2->VCF->VCA2

   So for me, will somebody please patch this on their modular?  How
   easy is it to match the frequencies of both VCOs?  Since analog's fatness
   comes from, in part, the fact that the VCOs will drift, how does it
   affect this patch?  FM synthesis on an analog synthesizer; what a
   concept!

It sounds pretty crappy on my ARPs Oddyssey and 2600.  One issue is that FM
synthesis requires a VCO with linear control for modulation and exponential
control for tuning.  Additionally it's nice if the VCO can handle negative
control voltages (and negative frequencies).  A breeze for digital, I don't
know of any analog synths with VCOs that qualify.  On my ARPs, the
exponential VCOs pitch up something fierce when they're FM modulated in
this way.  If you compensate for this, it can sound kind of interesting,
but nothing like a DX7.

Drifting isn't a serious issue by itself.  Heck the DX7's detuning is an
important feature.  But on my Arps, if I attempt to compensate for the
pitch-up, things drift all over.


I know this is off the subject of an Analog mailing list, but the DX7's
implementation is worth discussing for a little bit.  Back in the early
eighties when this puppy came out, Yamaha wanted to make a digital
synthesizer (heck, everybody did) but implementing a digital filter was
very difficult.  Y'see a digital 2-pole low-pass filter requires something
like two multiplies and two adds 40,000 times a second, times the number of
voices you want to do.  So for an 8-voice machine you'd need to do a
16-by-16 multiply in something like 1.5 uSec.  Back in those days, a fast
multiply was *expensive*, a few hundred dollars for the chip or the
equivalent circuit.  And that's just the VCF multiply operation, you also
need VCAs.

Nowadays modern DSP chips will do a 24-by-24 multiply, with an accumulate,
simultaneously updating a couple of address registers, in a single 150 nSec
clock cycle, and all for $25 or so.  ("You kids these days have it far too
easy!")

The DX7 is an incredible hack in that they use FM modulation instead of
filtering to do interesting things to waveforms, and can implement it with
zero real time multiplies (a "real time multiply" is one that has to be
performed during each sampling period) so the whole thing can do 16-voices
for dirt cheap.  The only real time operations are adds, stores, and ROM
lookups.  (Even the VCAs are implemented without real time multiplies!)
Pretty damned impressive as engineering feats go.

   From: wbf@aluxpo.att.com
   Date: Fri, 23 Apr 93 10:34:17 EDT

   >                                                  ... the cutoff of
   > VCFs, and maybe even the VCA (for a little ring-mod action). These make
   > the tone much more complex. You only have to put in a little to get some
   > action going; don't overdo it. ...

   You're not just whistling Dixie!  Subtlety or else FM (digital or
   analog) goes astray quite quickly.  That's why my newbie experiments led
   me to believe for so long that subaudio modulation was the only useful FM.
   Maybe that's a contributing factor as to why so many people have been
   afraid to program their own FM synths.  (Ignoring the lack of sliders
   and knobs, of course!)  Too bad I had nobody to correct my course way
   back when.

FM modulation was developed by John Chowning and the folks at Stanford's
CCRMA.  Check out his book, and the Howard Massey DX-7 books for details on
how it works; the math involves Bessel polynomials and can get real hairy.

Programming an FM synth is the electronic music equivalent to solving a
Rubic's cube; it's very nonintuitive.  On top of that is Yamaha's complete
ineptitude in user-interface design and general software writing (the
envelope generators are almost too goofy to use at all!), the thing is
*really* difficult to program.  So most folks just use those Godawful
factory patches.  Yech.

  -- Don

