[time-nuts] 100 MHz decade divider advice needed

[Gerhard Hoffmann]

Am 30.11.19 um 00:50 schrieb Perry Sandeen via time-nuts:
> Yo Bubba Dudes!,
> I'm considering attempting  making a 10 MHz frequency difference meter sort of based on the Ticor 527.
> The design calls for using 100 MHz OCXO's and using the PLL difference to steer the EFC and then mix it with a PLL 90 MHz locked to the reference 10 MHz.  Then that 10MHz difference is fed into the next stage PLL and the process repeated in the following stages.
> Hy problem is this.  The only generally available 100MHz decade divider I've found is the 74LS196N.
> The problem arises that on the TI website it's advertised as going to 100 MHz BUT when one reads the data sheet it says that it only works to 85 MHz or so.
> I found on ebay the MC12080 100 MHz to 1.1GHz chip that can be configured to divide by 10.
> Now the ON units carried by the big name distributers are about $6.  The Chicom no-brand-name units on ebay are less that $2 each.
> I'd really like to use the LS296 chip as it is a dip type and *genuine* ones sell for less than $2 on ebay.
> Since I need one OCXO for each decade stage plus one for the 90 MHz reference, at $50 each used on ebay this starts to get very expensive quickly.
> I did find some obscure 100 MHz decade counters on some design circuit on the net but I'd never heard of them and expect they will be unobtainum.  The same seems to be true for the 74S and 74AC series.

74ls, 74s and mostly 74AC are all obscure now. 5V is mega-out, 3.3V is

still standard but the trend goes to even less. 74LVC is mainstream.

Not all chips of the old 74 series have been ported to 74lvc because nobody

designs system around MSI function blocks any more. There are just the

essentials: octal bus drivers, gates, flipflops (mostly only 1 gate/ff 
per pop in a tiny

SO-8 or sot-23-5). There is still the 74LVC163 counter. I think there is 
no LVC191

whose precursor I have used a lot. These SSI/MSI chips only save as glue 
logic to make

LSI ICs to fit together or to connect them to IO.  Heavy functions that 
do not

justify a full custom chip go to FPGAs or at least to CPLDs.

I personally like the Xilinx Coolrunner-2 CPLDs as garbage collectors, 
but they

are getting old also.


The 74LVC163 + inverter should work for you, up to 200 MHz at least.

But remember: with the speed come problems, too. Your output may be

only 10 MHz, but rise and fall times are easily as fast as 500 ps.

Good rf engineering is absolutely required.


The lower-left quarter of the picture should work for you.


There are more fast families, like Fairchild NC7SZ04P5X  (04 = one 
inverter in sc70-5)

or TI SN74LVC1G04DCKR (mostly the same)

or ON Semiconductor NL17SZ74USG  (D-type FF), Fairchild NV7SV74K8x

or NXP 74LVC163PW ( counter )

as search strings for your google-foo, as typed from the anti-static bags

in my drawer. If you find one of the family, the brothers are easy.


local bed time now,   Gerhard  :-)




-------------- next part --------------
A non-text attachment was scrubbed...
Name: Bildschirmfoto vom 2019-11-30 01-20-59.png
Type: image/png
Size: 86274 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20191130/5359eccb/attachment.png>