div-3 divider with 50 percent duty cycle?

[bill_sun]

hi guys,do u have any suggestion to me?
operating freqency:3GHZ;differential signal;50% duty cycle;
SCL is impossible to get this.
Any advice will be welcome.

 

[jecyhale]

Do you mean you need the frequcny of 1GHz with 50% duty cycle?

can you double the operating frequency to 6GHz and divided by 3 and then divided by 2.
That will get 50% duty cycle.

Or you can use the Low pass filter to filte spectrem higher than 1GHz.

 

[Mazz]

It is a possible approach, but not very power efficient!
It will take for sure an higher current to go up and down in freq.
Other (low power) solutions?
Mazz

 

[bill_sun]

The background of this question is that we need a differential clock ranging from 100Mhz to 750Mhz with low jiiter(rms jitter is less than 1ps).The ducy cycle is 50%.The ring VCO is easy to cover this range,but the jiitter is too high.So the LC VCO is selected for this application.As you know,the large freq range is a big challenge,but I can try.I think the VCO can oscillate at 2Ghz~3Ghz and then we get the clock by dividing the freq to a low value.
From the data file from FAB,I find the Q of inductors is a bit higher in 2~3Ghz than in other freq range.

In this way,I can get what I want:
VCO freq: 2GHZ---3GHZ;
1/2 freq: 1GHZ-----1.5GHZ
(1/2)*(1/2) freq:0.5GHZ----0.75GHZ
(1/2)*(1/3) freq:0.333GHZ----0.5GHZ
(1/2)*(1/4) freq:0.25GHZ----0.375GHZ
100MHZ~250MHZ is not a problem.

As I know, the div-by-3 and div-by-5 ckts in SCL cannot output a signal with 50% duty cycle.
Is there a better choice for me?
Thanks a lot.

 

[linkfox]

hi guys,do u have any suggestion to me?
operating freqency:3GHZ;differential signal;50% duty cycle;
SCL is impossible to get this.
Any advice will be welcome.

see the provided link

 

[svicent]

Divide by 3 with 50% duty cycle (Not valid for 3GHz)

 

[Mazz]

I think analysis of dividers is OK. But are you taking into account the difficulties to make a 2-3GHz LC VCO?
Most probably you have to divide the whole band in sub-bands, assure that you will be able to cover it considering process spread.
Mazz

 

[bill_sun]

Thx for yr help.yes,it's really a challenge for me.I would design a wideband lcvco with a calibration module to adjust it.