Easier way to realize a 2FSK LO in Xband

[lewisFeng]

Dears,
My current solution is Generate it at DDS and then modulate on L-band, then up-convert to X-band. This solution need a lot of components include a dds, two PLL, two Mixer and lots of amplifiers and filters and attenuators.
Because the 2FSK modulation is two simple, I'm wonder if there is some simpler method to go. Maybe switch between two normal X-band oscillator.
PS, the jitter should be less than 20 ns, the switch time should be less than 20 ns, the spur suppression should be greater than 60dBc, the phase noise should below -80dBc/Hz@1KHz.

Thank you

 

[VEBill]

Back in the old days (1970s ), experimentors would sometimes (often?) directly apply the baseband modulating signal to the microwave oscillator. Perhaps the oscillator (likely a Gunn diode in those days) was slightly sensitive to applied voltage. Or they'd put somnething in the resonant oscillator chamber. They'd directly shift the frequency of the x-band oscillator.

This approach was also related to AFC techiques. With appropriate design, the hardware can do both at once.

The data rates used by those experiments were quite low.

 

[lewisFeng]

The AFC is more like a frequency fine tune method with switch time of several us. My LO has 240MHz adjust range with step of 1MHz. So if I can use MOS SPDT between two Fractional-N PLL with the speed of 20ns with isolation of 60dB, the circuit is greatly simplified. But during rise-fall time the isolation is gradually changed, the smearing effect is a big problem, the total setup time is very long. I haven't seen prior art using several cascade switches to conquer the trade off of rise-fall time to isolation.