10Ghz Prescaler

[Jørgen Brandt-Larsen]

Dose anybody have a suggestion how to make a "divide by 1000" prescaler with
a max input frequency of 10 to 12 Ghz. You can get divide by 2,4, and 8
chips for 10Ghz but how to get form 2,4,or 8 to 1000 is the problem.

73’
OZ1ETE
Jørgen

 

[anthony wooldridge]

Dose anybody have a suggestion how to make a "divide by 1000" prescaler with
a max input frequency of 10 to 12 Ghz. You can get divide by 2,4, and 8
chips for 10Ghz but how to get form 2,4,or 8 to 1000 is the problem.

73'
OZ1ETE
Jørgen

Am I stating the obvious in dividing by 8 then all you need is divide by 125
at 1.5GHz
does that make the problem any easier?
Regards
Anthony

 

[Adam S.]

I'm not at all familiar with high speed logic, but could you
first divide by 8 then feed through three cascaded divide by
5 counters ? Maybe connect up three MC10E451 6-bit D
registers into a divide by 125. Will this work ?

 

[George R. Gonzalez]

Dose anybody have a suggestion how to make a "divide by 1000" prescaler with
a max input frequency of 10 to 12 Ghz. You can get divide by 2,4, and 8
chips for 10Ghz but how to get form 2,4,or 8 to 1000 is the problem.

I guess it depends on what the final use for this is.

If you can get by with divide by 1024, the answer is pretty obvious.

If you need a closer answer, you could do some digital or analog math on the
result.
Maybe a good application for those little-used rate multipliers.

If you need an EXACT divide by 1000, you could divide by two,
that gets you down to 6GHz max, then maybe you can cobble together a divide
by five,
maybe as a ring counter, or some other topology that doesnt depend on
ripple-carries.

 

[maxfoo]

Use the following combination;

/5 * /5 * /5 * /8
Hittite sells them for about 2 bucks a piece.
http://www.hittite.com/index.cfm?bo..._div_det&catid=7&sort=function&source=leftnav

 

[Winfield Hill]

George R. Gonzalez wrote...

If you need an EXACT divide by 1000, you could divide by two,
that gets you down to 6GHz max, then maybe you can cobble
together a divide by five, maybe as a ring counter, or some
other topology that doesnt depend on ripple-carries.

He could divide by 8 with a Zarlink ZL40810, then by 5 with
a 1.25GHz ring counter or pulse-swallowing divider, then by
a final 25 with more conventional stuff rated at 250MHz.

Or perhaps the whole final divide by 125 could be handled
by On Semi's MC100EP016A programmable synchronous counter,
although there might be little safety margin.

Thanks,
- Win

 

[Andrew Paule]

You're now in the NBSG ballpark (although the parts are expensive) - 10G
is not easy - been there, still there, you are going to have to talk
SiGe and be very familiar with layout and simulation - traces,
dielectric, and pad shape/launch matter here big time. The standard ECL
parts (some quoted earlier) die at 1-3GHz, divide using the NBSG parts
(use a flop - the 53) and then use the programmable divider.

Andrew

 

[Volker Winterscheid]

Hi !

There is one readily available
See at: http://www.ame-engineering.de/

12GHz prescaler.

He is also a radio amateur and made a nice lecture about
YIG Oscillator drivers on the weinheim ukw-Tagung this year.

73!
Volker

 

[maxfoo]

?? the parts Isuggested from Hittite work from DC-18GHz
10GHz is childs play...been doing it for 25 years...