Sequential circuits samplers

<<... but I still wonder>>

have a quick look at the guide in the files section. 

you can, on a preset-by-preset basis, change the way the voices are allocated 
within the sampler. more scientifically, the sampler d>a is directed in turn 
at the 8 analogue "synth" channels (these things are like p600's without the 
oscillators) in the same way as a polysynth uses it's voices... only better. 

so you can build a preset in each "half" of the machine and ringfence 4 
voices each for them; a very basic multitimbral function. 

or (as you are hearing) the voices can be allocated on a seemingly random 
basis that sometimes ends up being L-R-L-R because you're playing a certain way 
(non-legato). 

the thinking is that if you are only using one instrument (i.e. not 
multitimbral) you'll probably be hooked up in mono and not get the dancing outputs. 
four voices each are hardwired to the stereo left and right outputs, unless a) 
there's a mono plug in the right output or b) I've remembered all this wrong.

(gets book out)
it's in "dynamic allocation". on or off. saved w/ the preset.
(v3 and above, there's a "fixed output" option. this is intended for use in 
conjunction with the "plus" 8-output card which is a bunch of op-amps and jack 
sockets you could weld in, and locked the voices one per socket. useful for 
drums or.... something.)

if it's off, you get the left map through the left voices and 4-voice 
polyphony (& the right...). so if you're only playing one map (i.e. not multitimbral, 
split or merging two maps) you should switch it on and hook-up in mono or at 
least pan the two outputs together a bit on y'r mixer; this can be quite good 
with a bit of reverb. 

dynamic allocation steals from the first note played when you get to 8 notes 
(or less if there's a stack on). 
(5-17 in the full manual).

duncan/r.m.i.