[Lowfer] LF Xtal on eBay

[mikea]

On Thu, Apr 23, 2009 at 12:41:46PM -0400, Bill Ashlock wrote:
> 
> Hi Peter,
> 
>  
> 
> I usually figure a .1% 'pull factor' is all one can hope for using
> a series capacitor or inductor. That Xtal, therefore, would fall
> short of 185.3 by quite a bit. I keep preaching a 5.185 MHz 60 cent
> microprocessor Xtal divided by 28 works very well but I guess coming
> up with a divide by 28 counter is a stickler.

Divide by 28 is just divide by (4*7). Divide by 4 is trivial, and divide
by 7 is only a tiny bit less so. I can think of three good ways to do it
right off the top of my head, and I don't doubt the real EE-for-pay
folks on the list can come up with others. 

The commercial divide-by-7 counters I've seen have used a 3-bit counter
(discrete flipflops) with the "1" outputs ANDed together to form a
RESET input to all three bits. That way, as soon as all bits go to "1",
all three flipflops are jammed to "0". If you do this with the output
of the divide-by-4, then the divide-by-7 counter only has to work at
5.185/4 MHz (1296.25 KHz) clock rate, which is pretty slow, and all
you have to worry about is getting the flipflops reset quickly enough:
there will, inescapably, be some time during which the counter is in an
all-"1" state, before the RESET pulse jams the flopflops to all="0", but
the duration depends on the speed of the logic you choose. You could,
alternatively, detect an all-"0" state and jam the flipflop state to
"001", but that's a bit harder.

Or you could use a 3-bit Linear Feedback Shift Register; that gives you
a 7-bit-long sequence automatically: just jam it to all-"1" to start
and provide an output when all three flipflops are "1". 

Or use a 7-bit-long circular shift register. 

I like the LFSR myself: it's easy, fast, very efficient in terms of 
hardware, and not used as much as it could be for applications like
this. 

-- 
Mike Andrews, W5EGO
mikea at mikea.ath.cx
Tired old sysadmin