Re: [AH] DIN Sync Splitter and DIN Sync timing gotchas

From Robin Whittle
Sent Thu, Nov 27th 2003, 00:32

It really doesn't matter how you make the electrical connections.  The
Run/Stop (pin 1) and Clock (pin 3) are robust signals - 5 to 15 volts -
with no special fussiness with capacitance, shielding etc.   You could
probably run them a kilometre or more on unshielded cables and it would
be fine.

Beware the pin numbering of 5 pin DIN connectors.  At one end is pin 1,
and at the other pin 3.  In the middle is pin 2.  That was for the 3 pin
arrangement.  A 5 pin DIN plug has the same numbering for those pins,
but adds pins 4 and 5 in between.

Some Roland gear receives signals on these these pins - for instance the
TB-303 and TR-808 receives Tap on pin 4 and an some undocumented
function on pin 5 - the circuit diagram has a blank for the label of
this pin.  But these are not necessary for starting, stopping and
running in sync.  The TR-606 is the same, but it also sends Tap on pin 4
- an approximately 5 volt signal (6 volts via a diode).  There's no
resistor in this circuit, so if pin 4 was shorted to ground, then
pressing the Tap button would cause excessive current to flow through
the diode.


You could get some existing leads - including any MIDI lead which has
wires for all 5 pins - as someone suggested, and cut off the connectors
at one end, or cut them in half, and solder all the common wires
together at one point.  That would be an octopus or star structure.  The
only thing to beware of is that some old audio leads for connecting
stereos had the pin 1 and 4 of one end connected to the pin 3 and 5 of
the other - to automatically connect the two output channels of one
device to the inputs of the other.

My suggestion of a daisy chain is to use three different coloured pieces
of hookup wire into connector A, with the other end of those wires going
to connector B, together with a second set of three wires which goes to
connector C etc.

The Run/Stop signal goes high at the start of the song and low when the
song is not playing.  The Clock signal is a square wave, 24 pulses per
quarter note.  It can be running when the Run/Stop is low too - just to
make the slave device's lights blink.  However, ideally, the Clock
signal should be brought low before the Run/Stop rises at the start of a
song, which can lead to arbitrarily short Clock signals, because there's
no telling when the song will be manually started.

Unfortunately there's no technical specification on the timing of Roland
(AKA DIN) Synch.  As far as I know, there's no spec on the minimum width
of the clock pulse, or even the voltages which must be used.

For instance, its a bad idea to send the first Clock pulse (its positive
edge) too soon after the rising edge or the Run/Stop - because
particular slave devices, such as a TB-303 or TR-808 won't have got
their act together after the Run/Stop activation before the Clock pulse
arrives, so they will miss that, and actually start their sequence on
the second Clock pulse.   This is something which should be born in mind
by all people who create MIDI to Sync converters.  How long is "too
soon"?  There's no spec - but its best to try it on a TB-303 or similar
and then add quite a large margin for Justin (just in case).  This means
writing software in the device to count up potentially several MIDI
clock codes which might arrive faster than they can be sent out to the
Roland Synch device - and then catching up with them by sending Clock
pulses at some reasonable (whatever that means) rate, involving some
limit on how short the pulse can be, and how short the delay until the
next one.


  http://www.firstpr.com.au/rwi/dfish/

  Devil Fish mods for the TB-303