New project: Sequencer

[jhaible]

Something I carried around with me for a long time.
A big analogue sequencer to match my JH-3 system.
Finally started doing experiments now.
I want to have a random mode and skip function 
(like the ARP seq), and I want up/down modes like 
the Korg SQ-10.
Several Gate busses a la ARP, too.
Not yet sure about number of steps and rows.

The core of my Seq will be the CD4516 up /down counter.
What does "Reset" mean, when the counter counts backwards?
I think about using the4516's Preset function to jump to the
last stage in that case.

Skip and Random:
I have simulated things like that on my simple 4017-based 
sequencer that is part of the JH-3 system. The sequencer clock
does not trigger the counter directly, but opens a gate and lets
a HF VCO clock the counter. Works, but needs a lot of patching,
and my new Seq should also have a masking function to hide
these fast clocks at the beginning of a note.
I looked into the ARP schemos, and I was surprised how clever
their solution was: While the GATE is on, they do all the Random
calculation, and the last step number is stored in a latch. In the short
time when the capacitor of the seq clock VCO is discharged, they
do the Skip operation. Problem: How do you know which steps to skip
before the output is updated ? I'd almost used a separate MUX to
control the Skip bus, with the MUX connected directly to the counter,
while the main MUX for the potentiometers would be latched. But then
I found how ARP have done it: The used a *transparent* latch !
So, the "random" counting is masked by the latch, but the short
"skip" counting takes place while the latch is transparent. Brilliant.
I threw my first veroboard away, and use the ARP method now.

Something I did not like was ARP's way to implement the random
clock, which is actually pseudo random, and makes use of an 
(undocumented?) ability to slightly voltage control the time of a CMOS
monoflop. I'll use a 4046 instead: The zener voltage of the 4046 is
amplified 
by a single 741 (feedback 1megohm output to inv input, 10uF directly
from inv input to GND, the slew rate of the 741 does the rest ...), and
the 741 controls the VCO of the 4046. Works great - have just prototyped
it.

Quantizer:
I want to build a quantizer where I can select which chromatic tones I want

to include and which I won't, with a set of 12 individual switches. 
I guess I'll do this with a normal binary counter and resistor network
(main
DAC), and a second divide-by-12 counter plus 4067 MUX and a separate
resistor network for the semitones (semitone DAC). The two DAC outputs
would be exactly scaled to each other with trimpots, and then added.
Then I can put a switch into each 4067 output line, so I can remove evey
unwanted semitone. I couls also make each of the semitone DAC's
resistors slightly variable with a resistor + trimpot combination, so I can
tune each semitone (for all octaves in common), if I don't want equal
tempered
scales. What do you think about this ??

Back to the bench. If you have suggestions for features you wouldn't like
to miss in a sequencer, please tell me during the next days !

JH.

PS.: One last thing that has impressed me in the ARP sequencer: How they
built
        FET opamps from standard opamps plus a 4007 CMOS chip. Sure, they
have
        done this because thy didn't have these cheap LF356 or similar
chips (?),
        but one thing is still impressive today: The common mode range of
these 
        "4007-opamps" goes up to the positive supply voltage, which is a
great plus
        if you build a saw vco and discharge the cap to the positive rail.
No auxiliary
        voltage required ...