[net.music.synth] Midi Interfaces cause Mac to "blow up"?

jer@peora.UUCP (J. Eric Roskos) (02/16/86)

Until a couple of weeks ago, for about a month I had been working on
writing a MIDI sequencer for my IBM PC.  I built a serial-to-MIDI interface
out of a Motorola 6801, wrote the basic sequencer, and was starting to
work on the Hard Problems like how to insert notes of shorter duration
into the middle of a sequence without messing up the timing, when I started
reading reviews in here about the Deluxe Music Construction Set.  Since
I had a Macintosh that was basically sitting around collecting dust except
when I needed to draw pictures for a presentation at work, I decided to
quit working on my program and buy the DMCS.

So I ordered the program, and it came much sooner than I expected, and I
started working on the MIDI interface for the Macintosh.  That's when the
problems started...

Maybe because the clock frequency on the UART for the Macintosh is so close
to an integer multiple of the MIDI baud rate, and because on the IBM PC
with it's ancient UART you can divide the UART clock rate by any 16-bit
integer, and maybe too because I still believe, during weak moments, all
the great stuff they said about the Macintosh back when it first came out,
I thought, "He'll just set the divisor in the UART to give the MIDI baud
rate, and all I have to do is make the current loop interface the MIDI
standard requires, and it'll all work fine."

No... nothing that simple... the current loop interface worked fine (when
tested with a terminal emulator program); it even lit one of these
microscopic LEDs Jameco sells that run on 5 milliamps (same as the
optoisolator in the MIDI standard interface).  But not when it was running
DMCS... then, it just sat there...

After a lot of thinking, the probable truth finally occurred to me... one
of the pins on the Macintosh serial port (pin 7) is an external clock
input, and they must be setting the UART to the external clock mode,  so
those MIDI interfaces must all have a clock on them that clocks the serial
port.

After a few abortive attempts to clock the port with a 1MHz oscillator I
had handy, I decided "shucks, this isn't worth all that trouble when they
already have them built for $75 or so."  So I just called my local music
stores, and... more problems...

Most stores had answers ranging from "We only sell guitars" to "Gee, uh,
I don't think we have any instruments you can connect to a computer" to
"We don't carry any synthesizers by `Middy'".  Finally, I found one who
knew what I was talking about, at least.  "What kind of software do you
have?" he asked.  He had never heard of it either.  But... he did know
about the two MIDI interfaces mentioned in the manual.  "Don't carry
the bloody things," he said.  "We used to, but the blighters kept blowing
up people's logic boards; one bloke said he was going to sue me."  (He
talks that way because he's from some foreign country over there near
Japan, you know.)

Is this true?  What MIDI interfaces do you folks out there use?  Have you
had any "blighters" "blow up"?  I mean, it's certainly believable, given
the fact that the machines seem to burn out their flyback transformers
spontaneously if you leave them on for any period of time, but I wonder
if maybe *that*, not the MIDI interfaces, was the cause of the problem...
On the other hand, it could be because they are probably drawing their
power from the +5 or +12 volt pins on the serial port to run that oscillator
and whatever else they use to drive the interface.

Better yet, has anyone built one of their own?  It seems easy enough to
do, since for DMCS you only need the output side of the interface; the
only questions being "what clock frequency do they expect," and "what does
the clock signal to the serial interface look like?" (I tried connecting
the output from one of these TTL-compatible hybrid oscillators to the
clock input, since there's only one input pin (not two), and since the
output voltage with respect to ground on the serial port output was about
5 volts, but that didn't work.)
-- 
UUCP: Ofc:  jer@peora.UUCP  Home: jer@jerpc.CCUR.UUCP  CCUR DNS: peora, pesnta
  US Mail:  MS 795; CONCURRENT Computer Corp. SDC; (A Perkin-Elmer Company)
	    2486 Sand Lake Road, Orlando, FL 32809-7642