[sdiy] Frequecy Divisor Question (TTL pulse stretcher needed)

[harrybissell]

I'd try setting a D flipflop, and reset from the Q output, delayed through a couple
of inverters. This uses the propogation delay as the timing element.

OTOH I did NOT do the math to see if this timing is really practical, that is
left to the student :^P

<bfg>

H^) harry

Michael Buchstaller wrote:

> Hallo friends,
>
> my Journey with the Oscillator continues...
> After my fruitless efforts with the 74LS193, i built Ken Stones
> 74LS193 based design, which worked from the 1st try.
>
> Now, i want to use a fast clock to gain higher accuracy.
> So i decided to use 74F193 counters, which are (according to
> the data sheet) good for up to 133 MHz.
>
> I cannot directly measure here, because my 20 MHz scope
> does not trigger such fast signals. But when i use one of the counters
> free-running, i can see the outputs work as expected (QD: 6.25 MHz,
> QC: 12.5 MHz, QB: 25 MHz (barely visible), QD: no trace, but will be 50 MHz)
> So, the counter chips do work.
>
> The carry signal from the chip is only 1 clock cycle long - that gives a pulse
> of 5 ns width (assuming a 50% duty cycle from the oscillator).
> It popagates properly over my board; even the last counter gets it and reacts
> correctly.
>
> But there is the monostable that is fed by the last counter´s carry bit. It is used to stretch
> the pulse duration, so all counters will be reliably reloaded with their respective values.
> If i insert a 25 MHz or slower oscillator, everything works fine. 33 MHz and faster does not
> reload the counters. No output pulse from the 74LS123 monostable. I tried several brands,
> and some of the ´123s seem to work with 33 MHz.
> So, i assume the pulse is too short for the monostable to fire. 20 ns with the 25 MHz oscillator
> does work, but if i can get the signal to stretch to 20 ns, i could eliminate the 123 completely and
> feed the LOAD input of the counters directly with that, correct ?
> I tried to simply use the carry output for that (directly and inverted), but this does not work.
>
> So, how do i stretch the 5 ns pulse to (say) 20 ns ?
>
> -Michael Buchstaller