Injection locked frequency dividers

Question

I am beginning to design a monolithic frequency synthesizer (around 2.4GHz) in which I need the maximum of spectral purity. I was looking at injection locked frequency dividers (ILFDs) for use as a prescaler as an alternative to flip-flop based prescalers. Are there disadvantages to using ILFDs in terms of spectral purity?

As a sub-question, there are two main types of ILFDs--LC based and ring oscillator based. Ring oscillators have a reputation of being jittery as opposed to LC oscillators. However, within the context of an ILFD, does this jittery-ness come into play? In other words, WRT spectral purity, would a ring-based ILFD be sub-par?

Answer 1

You did not specify the frequency accuracy required and phase noise, but LC clocks will have a Q of 100 or so and poor capture range and phase noise, and a cheap 50MHz Xtal or whatever oscillator PLL VCO prescaler fractional N divider will have a Q of 10,000 with low phase noise and 50 ppm frequency error. typ. but phase noise is multiply by divider ratio up to loop bandwidth. If you need widetemp range, then thermal VCXO can be done with varicap and lookup table with known xtal profile. I've developed 1 PPM synth for Tx @ 1GHz using a 25 cent Xtal, quick thermal 30 second test, and test jig for varicap. for circuit -40~+70'C range. 1PPM VCXO using part of HC05 code for VCXO compensation table

Answer 2

1. Why design your own; there are many companies that make synthesizers?
2. If designing, use an LC-based one, it will have better spectral purity since they have fewer digital second-order effects.