AW: Re: Analog Sequencer Poll

[Haible_Juergen#Tel2743]

> I have in front of me here an ARP Sequencer manual. On the ARP sequencers,
> they have this funky random mode, where it picks out which notes to play 
at
> random... how would ya implement this ? (without using a CPU!!!). I had 
the
> idea of using a high speed oscillator, and selecting voltages off it
> periodically... would this work ?

You don't need CPUs for everything (;->) (;->)

Seriously: As others have pointed out, the ARP sequencer does something
like this indeed. But I think there is one thing that wasn't mentioned so
far: The ARP also has a "masking" circuit that will hide the random
counting from the output. I. e., there is a little delay between the 
internal clock
and the gate at the output, so you won't hear the random counting.

But even without the masking, it's not that much audible, if you do the
random counting very fast.
I had a patch on my modular that did random steps on a "normal" sequencer.
The sequencer clock was gated by an AND gate. One input of the AND
gate was my random source (see below), the other input was a very short
AD envelope which was triggered by the clock. So with every rising edge of
the clock, there was a small time window in which the random source could
count thru the sequencer.
For the random source, I tried white noise at first, but it was difficult to 
set
the right level of the noise. Then I took a VCO (can be a very simple one),
and modulated its frequency with the white noise. The VCO was set to very 
high
frequencies (AD-time was about one or 2 ms, and you want a few random 
triggers
within this time window!), and its pulse output worked great for the AND 
gate.

Ok, this was with my own modular system, so I don't know if it is 
reproducable
on other gear, but my sequencer is really simple (4017 ...), so I think 
Audio
rate capabillity is the only requirement.

Another point: For a future sequencer design, a "skip" capabillity would be
important to me. Now there is a similar problem with "skip" as with random:
you either need some additional high-speed-clock to quickly jump to the next 

stage (and you'll need N high-speed-clock cycles to skip N stages!),
or you need quite some overhead logic to detect stages that shall be skipped
*before* the actual clock pulse. This second solution would probably
need a discrete-flipflop (vs. binary counter) design. Things become more
complicated if you want a skip function in up *and* down mode as well.

JH.