Analogue-sequencer

I   just got my hands on a nice   memorymoog + with a dbm mod.
A nice   synth indeed   in honor of   Moog even though the late great Bob 
Moog   wasnt   the designer . .   that filter   *sigh* sounds nice.
Now   i am looking forward to the drive to Pomona   and snatching a P3.

what ive noticed about   the Memrymoog. .   the midi   timing is   pretty 
sloppy.
is   this due   to the z80   cpu?
even the internal arpeggiator   is very loose when   using the sync input ,   
sync'd   w/ the 909 rimshot. It appears to be way behind   & stumbly.   hmm   
old   midi?
 
Its   a bit like that on my oberheim xpander when i turn knobs   while 
running a sequence via   an mpc 60.. it s timing stumbles as well .   im used to   
the rock solid timing of the tb 303 sequencer & my arp seqs . 

Im hoping    to have better results   with the p3   & my midi gear as far as 
timing goes.

I have   a kenton pro 2 which seems to have lots of   aux   cv outs   to   
shake hands with my other analog goodies. minimoog, korg mono-poly ,arp odyssey 
/   arp 2600.   yep im a braggart.

a really happy one.



[Non-text portions of this message have been removed]

> what ive noticed about   the Memrymoog. .   the midi   timing 
> is   pretty 
> sloppy.
> is   this due   to the z80   cpu?

It's not the CPUs fault.
I measured the latency on my JP8 (also a Z80) with Groove MIDI interface,
against my 909 triggering a rimshot synced to the P3.
The latency on the JP8, including MIDI latency, was between 4 and 10 ms,
averaging about 6ms.
This sort of delay is imperceptible unless you are using very, very short
transient sounds, and listening very carefully.
Even then, it's more in the comb filtering range than audible delay.

But absolutely rock solid timing didn't seem to have been a top priority
back in the day, and usually suffered at the hands of poor software design.
I've got an old magazine article somewhere, where the author set up a timing
test for various synths to measure the latency between the arrival of a MIDI
note on, and the appearance of audio at the output.
The worst cases were like 30 to 40ms IIRC. Which is    very     noticeable
indeed.

A shame really, since all those MM hardware envelope generators should be
capable of super snappy sounds.
I have to avoid synths like that, since they tend to make me want to 'fix'
them.

Best regards,
Colin Fraser
Sequentix Music Systems Ltd
http://www.sequentix.com

"Colin f" <colin@c...> wrote:

I feel like an old timer, rocking in his rocking chair saying this...

Back in the day, when I was a session musician in LA, we carried a
ruler, scaled in milliseconds (15ips and 30ips).  When tracking a
basic track with several parts, we would record it onto 24 track
analog so that we could measure the delay while recording, and the
parts that were delayed could be recorded again with an offset to
compensate, lining up the bass and drum tracks.

gary

> and the
> parts that were delayed could be recorded again with an offset to
> compensate,

Recorded again? I thought the usual trick was to thread the tape
backwards and bounce the track through a delay line...

	-- N.

-- 

  nick rothwell -- composition, systems, performance -- http://www.cassiel.com

> The latency on the JP8, including MIDI latency, was between 4 and 10
> ms, averaging about 6ms.

This is considerably better than the official SCi midi interface on the
Prophet 5. This suffers from around 12mS to 15mS and it is noticeable to
the discerning ear.

But this again is not midi at fault but the low priority that the
Prophet's firmware puts on midi messages. The poor old CPU has to do
everything including update all the LEDs... meaning that if you throw a
lot of midi or note data at it, all the LEDs go dim.

Of course, I am waiting for the Sequentix Prophet 5 complete OS re-write.
I'd buy one ;-)

Tony

www.oakleysound.com

> > The latency on the JP8, including MIDI latency, was between 4 and 10
> > ms, averaging about 6ms.
> 
> This is considerably better than the official SCi midi 
> interface on the
> Prophet 5. This suffers from around 12mS to 15mS and it is 
> noticeable to
> the discerning ear.

Both figures suggest voice assignment is being done in the main program
loop, rather than in the MIDI reception interrupt routing.
If I were re-writing the OS, voice assigment would be the highest priority
task, and would certainly be done within 1ms or so of the message arriving.
Maybe the JP8 and P5 are similar enough that a re-write for one would adapt
easily to the other.
I tend to think I'd use a replacement CPU, rather than a re-write, both for
reasons of performance, and to prevent it being a waste of time
commercially.

Best regards,
Colin Fraser
Sequentix Music Systems Ltd
http://www.sequentix.com

--- In analogue-sequencer@yahoogroups.com, Nick Rothwell <nick@c...>
wrote:
> > and the
> > parts that were delayed could be recorded again with an offset to
> > compensate,
> 
> Recorded again? I thought the usual trick was to thread the tape
> backwards and bounce the track through a delay line...
> 
> 	-- N.

Back in the day, the typical situation was using two machine lockup,
so we usually would just bounce from the master machine to the slave
with the offset.

gary