I'm following this tutorial on si5351, with the goal to generate variable frequency range (eventually with two clocks in quadrature).
Anyway the tutorial is quite interesting, but it give some constraint as assumption I can't find in the datasheet.
Let's consider the formula they use to calculate the output frequency (to use same language here):

Where FMD is the feedback multisynth divider and OMD the Output multisynth divider. They state:
The VCO frequency can only be from 600 MHz to 900 MHz. Yes, over-clocking is possible, and so is under-clocking, to extend the output frequency range beyond 2,3 kHz to 200 MHz. But let’s leave this for now. The FMD ratio can be from 15 + 0/1 048 575 and to 90 + 0/1 048 575, i.e. a = 15, b = 0 and c = 1 048 575 to a = 90, b = 0 and c = 1 048 575. Please remember that c can never be zero (0), as this will violate fundamental math principles. The OMD ratio can be 4 and from 6 to 2048. Please remember that f can never be zero (0), as this will violate fundamental math principles. The R divider can only be 1, 2, 4, 8, 16, 32, 64 or 128. If the output frequency is above 150 MHz d is always 4. If the output frequency is below 500 kHz R should be used, i.e. higher than 1.
A part from the frequency constraint, and the 1 048 575 constant (due to 20 bit), what is the explanation to the constraints below? Where these limits are found?:
- The FMD ratio can be from 15 + 0/1 048 575 and to 90 + 0/1 048 575, i.e. a = 15, b = 0 and c = 1 048 575 to a = 90, b = 0 and c = 1 048 575.
- The OMD ratio can be 4 and from 6 to 2048
To clarify even more the question: the article does non mention how these constraints cames from, and, with my knowledge, I can't find mentions of such limits in offcial datasheet.