I'm designing a receiver in which an ADF4351 acts as the main local oscillator. We want to implement a second oscillator to inject a known signal to the chain for self-calibration and self-test. We need to inject a signal of known relative phase with a frequency between 18 MHz and 4 GHz. We already know the ADF4351, we have it characterized and tested in other designs so we wanted to keep it as a candidate for this task, besides we've found no other synthesizers from ADI or other manufacturers that cover the whole range. We thought about switching between the two branches of RFOUTA with an SPDT switch, sending one (RFOUTA+) directly and the other (RFOUTA-) though a /2 divider. The problem is that we found no RF-specified dividers up for this task that aren't way too overkill and expensive (forr example HMC432 would work but it's a waste to use such a device to cover a range from 35 MHz to 70 MHz input range and 18 to 35 MHz ouput range). Would it be OK to use a simple digital D flip-flop with its input AC-coupled, terminated to 50 ohm and DC-biased to VDD/2 or something similar and AC-coupling the output and series-terminating it with a 50 ohm resistor?
Any other idea?