Good day,
We have been evaluating the performance of the LMX2594 (LMX2594EVM) and found that using the recommended setting for PFD_DLY_SEL (given in table 2 of the datasheet), when the MASH order is set to 3 gives a measured phase noise that is worse than using a smaller value for this setting.
To provide some context, our measurement setup used is as follows:
We don’t have a phase noise analyser at hand so instead we measure the spectrum of the signal using a spectrum analyser. The close-in (< 100 kHz) noise is dominated by the instrument.
The LMX2594 settings used are shown below and set using the TICS Pro software:
Our frequency range of interest in between 10.5-10.6GHz, and we set our DEN value to 33554432, with the MASH order at 3. For example, according to table 2 of the datasheet we should set PFD_DLY_SEL equal to 4 for an output frequency of 10.5025 GHz, with N = 52 . We are aware that using a large fractional setting for the denominator will result in worse phase noise performance, but we didn’t expect this to be worse when using the recommended setting for PFD_DLY_SEL. We consistently found that using a value of 3 (and the MASH order of 3) that we had a lower noise floor, see the attached measurement. This measurement was made using a spectrum analyser with a resolution bandwidth of 100Hz, hence the power level. We not concerned with the close-in “hump”, we suspect it’s the response of our measurement setup/instrument. The noise floor is about 6-7dB lower using a value of 3, even-though the TICS PRO software gives a warning for this setting.
There is almost no information in the datasheet as to what exactly the PFD_DLY_SEL setting is. Could you provide us with a more verbose description of this? Secondly, can you see any reason why we should not use this setting and stick the the recommended value?
