[sdiy] Saw core JFET mystery

[David G. Dixon]

Hey Team (especially Ian Fritz):

I soldered up my latest board revision for the 2164 Expo VCO tonight, with
the proper JFET at last: PN4391.  After much simulation, I decided that I
could greatly simplify the circuit around the ramp integrator and reset
comparator, and eliminate the high-frequency trim circuit in favour of a
Franco compensation resistor (just like everybody else).  So far so good.
However, tonight, when I got it all hooked up, it seemed that the reset
comparator RC time constant was too large.  I only used 10k and 22pF, or
220ns.  However, this seemed to be way too long.  I took the resistor off
and rigged up a 20k trimmer instead so I could play around with R...

Long story short, using the scope I found the best resistor value to be
about 1k.  Hence, the reset RC is only 22ns!  I'm thinking that this can't
be right, but the scope confirms that the ramp is resetting all the way, and
the bottom corner of the derived triangle looks very nice over the whole
audio range, with just the tiniest inaudible reset glitch (this is the real
indicator of resetting goodness).  I hard-wired a 1k resistor in, and it all
works swimmingly.

So, why can this RC be so much smaller than everyone else uses?  Is the
PN4391 that much different than the 2N4391?  They should be virtually the
same.  My integrator cap is only 1nF, so it will drain pretty fast, but
still...?

Now, bear in mind that I have no idea how well this beast tracks yet.  I'll
have to look into that tomorrow, and I probably still have some optimizing
to do on sizing the Franco resistor, but I'm expecting it to track very
well; it certainly resets fast enough!