I was told by AD's technical support that REF CLK could not be a short-duty cycle pulse...though I've now read in these forums that a higher slew rate on REF CLK is better for phase noise performance, so I am now a little confused. A short, sharp pulse would have a high slew rate, so why could this not be used as a REF CLK signal? That is, why is there a requirement of a 45-55% duty cycle (as stated on the AD9912 datasheet, for example).
In my application, I am using a pulsed laser as the REF CLK source. I have an ultra-short-pulse pulsed laser on a fast photodiode. The laser pulses at about 80MHz; the pulses are so short that the electrical pulse from the photodiode is limited by the risetime of the photodiode (<1 ns). So in the time domain, I have short 80MHz pulses; in the frequency domain, I have an RF 'comb' with approx. equal components at 80MHz, 160 MHz, 240 MHz, etc. I wanted to use the 80 MHz pulse as the REF CLK. If I cannot do this, then I will now plan on bandpassing just one of the harmonics of the RF comb to obtain a sinusoidal signal that I can then use as a REF CLK, albeit with a lower slew rate. So for example, I'll filter out a comb line at ~1GHz and use it as REF CLK for an AD9912 eval board. Are there any problems with this?
Thanks in advance for any help from anyone! I realize this is possibly an unusal REF CLK source, plus I am a physical chemist, not an EE expert, so any help would be greatly appreciated!