LMX2592: Phase noise performance between a 150MHz and 100MHz reference.

Hi Kevin,

As a general rule, decreasing reference frequency tends to mean decreasing the PFD frequency, which increases the N divider used and increases the reference noise multiplication at the output. So, usually a reduction in PFD frequency results in an increase in integrated noise within the loop bandwidth, and a reduction in loop bandwidth overall. Sometimes the reference frequency is independent of the PFD frequency, and there would be no change in this case: assuming you have an equally-good frequency-scaled 100MHz and 150MHz reference, but a PFD frequency of 50MHz, both cases would have identical phase noise performance.

Usually the 100MHz + OSCin doubler will outperform the 150MHz reference with no doubler, but there are restrictions on the minimum N divider permitted by LMX2592 which make it hard to say definitively that one is always better than the other. For example, the minimum N-divider restriction of 30 with 4th order modulator makes it impossible to use 200MHz PFD frequency below 6GHz VCO. Enabling the doubler in this case would just add to the 1/f noise, and the doubler output would need to be divided again to satisfy the minimum N-divide constraint.

I strongly encourage you to simulate your desired conditions in PLLatinum Sim, a tool we built specifically to make these comparisons possible for our customers. PLLatinum Sim will warn you graphically if you have prepared an invalid configuration (e.g. violate the minimum N-divider requirement for a certain modulator order), and can calculate the integrated phase noise for any desired configuration.

Regards,

Derek Payne