[sdiy] All things Lowpass gate

[Peter Grenader]

OK, I might as well make a self-propelled string out of this.

The issue of LPG bleed through has been nothing short of an obsession with
me.  I've tried a myriad of things, even contemplated adding a rather
complicated noise gate after this circuit to kill it.  I've set one up using
a comparator driving an amplitude VCA in a patch and it works, but it's a
major pain in the butt to find one noise gate circuit which works with both
fast and slow attack signals through the LPG.

Garry Chang has a software plug-in that does the trick, but it can only work
with recorded passages as it looks into the future 10ms before the sound
occurs and makes all necessary corrections before hand.

Some things you can try is dicking around with the .001 (i think Don has it
has a 920 picofarad or something like that) going into the buffer. Adding
about 200 pf there will attenuate the bleed, but it comes at a cost of
increasing the filtering which in turn kills a bit of your high end.  The
bottom line, if you look at the databook you'll notice vactrols don't shut
all the way down and there lies your problem.  So to attenuate the bleed
you've got to try other things and  without going into the details of how I
finally got it to what I consider  manageable level (.3mv), I'll give you a
hint  -  180 degrees and feedback.  there you go.

Another 'cute' personality trait of the LPG you'll notice is that without
presorting them beforehand, no two vactrol have an identical curve and this
effects the amount of  'ring' in combo mode.  Some ring more than others. If
you look at the datasheet you'll realize Perkin Elmer is well aware of this
as their dark resistance spec is listed as something like 'off resistance
after ten seconds'.  This time delay compensates for the variance between
the devices which P. E. explained to me if the nature of photo resistors.

If you want to standardize the curve across multiple LPGs, I recommend you
purchase more vactrols than you need and sort them before hand.  At 5 bucks
a pop, this is a tall order however.  The fact that they are diffewrent form
one another doesn't bother me - I actually embrace that part of its
personality.

Another thing - Don Buchla's  LEDdriver circuit, his sort of quasi expo via
a zener trick is on the near side of crappy and you'll lose a bunch of the
energy in long attacks. Take note that  the Buichla EGs are linear. but
instead of a nice linear slope on long attacks you'll get more of a
'nothing nothing nothing - there There THERE' sort of thing which isn't
necessarily bad, but it doesn't track the control voltage it's getting.  The
292 rev B's led driver is slightly better in this regard.  You can however
improve the response using other methods of turning the vactrol led on.

hope this helps,

P-