[sdiy] Sequ... (mistake!)

[jh.]

>I've made some experiments today with these staircase generators, the first
>finding was  that one needs to do some bias correction to prevent
integration
>errors. (The input bias current makes the cap rise slowly, that gives an
>oscillation even without input pulses.)

My circuit (http://home.t-online.de/home/jhaible/hjvcdbyn.gif)
has such correction. The offset trimmer must be set such that the
integrator goes slightly downward without input pulses, and
it is clamped to approx. -0.6V with a diode. No problems
here - it just starts its regular action from -0.6V rather than
0V. But I'm also using BiFet opamps (there *must* be some
advantage over the cheap 3900 version, after all!).

>The circuit that I've used is the staircase generator on page 11 of the
>LM3900
>datasheet. This makes two steps for one complete cycle of the clock input.
>(The diode between the inputs does that.) I notice that your circuit (and I
>assume the original) does that as well, it may be another hint.

The core of my circuit makes two steps from one input pulse. The Flipflop
at the output puts it back to (1 pulse / N) per input pulse. The Flipflop
is needed to get a reasonable pulse width at the output.

The strange looking output stage is to get a 5V level from 15V supply
without using a (slow) opamp or a current consuming class A circuit,
and it also provides a low impedance switch trigger output (the extra
transistor).

JH.