From: Winfield Hill
Subject: Re: circuit wanted-SQUARE PULSE GENERATOR
Date: 27 Nov 2002 08:45:10 -0800
Organization: Rowland Institute
References: <1038144646.265208@athprx02> <email@example.com> <firstname.lastname@example.org> <3DE4D81E.email@example.com>
X-Newsreader: Direct Read News 4.11
Fred Bloggs wrote...
> You still have the comparator tpd's which approach 1us or more at
> small overdrives like 5mV at threshold crossings.
That's a small effect, in the 0.05 to 0.1% level, which is roughly
what we're talking about for a 10-turn pot, etc. BTW, using what
we learned recently from Jim, even that small non-linearity can be
largely removed with one resistor in series with the capacitor. :)
> Your circuit is basically open-loop, and, although clever, will not
> compete with a 4046 wherein the operator closes the loop with a
> stable output frequency display.
Hey, I'm not claiming the world for this simple circuit, competing
with a crystal osc + PLL + dividers, a uP DDS, etc., but neither is
it a piece of #### as you would seem to be implying.
> The tolerance on the threshold resistor divider chain is something
> like 25% and your calibration pot can only get at one of two nodes.
You're overly pessimistic, Fred.
The worst-case 7555 divider-ratio tolerance is under 5%, the typical
value is probably closer to 1%. However, it really doesn't matter
so much. The CAL pot may control only the upper threshold, but the
lower threshold merely sets a charging dV as a fraction of the total.
The final result is a CAL operation that accounts for ALL the various
parameters affecting frequency... capacitor value, resistor ratios
(whether inside the chip or not), etc. We also know that after the
CAL the resistors track well with temperature, etc., from the chip's
typical 50ppm/C drift quoted on the data-sheet's cover. So while an
improvement would be gained if the 7555 were replaced with discrete
IC equivalents, the ciruit can serve well for a simple 0.1% bench-use