From: email@example.com (Bill Sloman)
Subject: Re: clock divider
Date: 24 Oct 2002 15:09:11 -0700
NNTP-Posting-Date: 24 Oct 2002 22:09:11 GMT
John Larkin wrote in message news:...
> On Tue, 22 Oct 2002 18:08:17 GMT, "Jamie Morken"
> >I am building a circuit that requires a 120MHz clock and a 24MHz clock and
> >they also need to be synchronous. Is there a "divide by 5" clock divider
> >chip I could use to get the synchronous 24MHz clock from the 120Mhz clock?
> >Jamie Morken
> a 74AC160 or 162 decade counter would probably work. I've used the AC
> and ACT parts at 125 MHz; they're hot and noisy, but they work.
Individual 74AC160 or 162 decade counters might work. The Fairchild
74AC161 which you can actually buy from Farnell can't make 120MHz, or
anything like it, on the basis of worst case design.
You'd have to hook up the terminal carry output (TC) to the Parallel
load Enable input (notPE) and the worst case propagation delay from
clock to TC is about 11nsec - rather more than the 8.33nsec available
You've also got to deal with the 7.5nsec set-up time on the notPE
input. Since you'd have to invert TC before you used it to drive
notPE, you'd probably use a separate D-type bistable to re-synchronise
TC to the clock and use the notQ output to drive notPE.
IIRR these parts all fall over at around 50MHz, at least if you are
doing worst case design. You might be able to get a TTL-based
shift-register-based divider to work, but it is pretty much a waste of
time - as you suggest, a fast PLD is a much better solution, if you
must stick with TTL levels.
When faced with a comparable problem a few years ago, I redesigned an
existing 74F board with surface mount parts, and - in the extra space
I'd made available - I threw in an ECLinPS quad latch and a couple of
discrete TTL-ECL levels shifters and a quad ECL-to-TTL driver, as well
as a switch to allow us to select the right phase of the 200MHz ECL
differential clock to pick up a stable level out of the TTL.
It got rid of the systematic sub-nanosecond jitter that our client had
been bitching about, and we didn't have to replace any more of the
selected 74F parts that had had the nasty habit of drifting out of the
timing window as they got older.
Bill Sloman, Nijmegen