Synthesizing clean 40MHz from EVAL-ADF4351EB1Z

Hello, we have been working with an EVAL-ADF4351EB1Z for a couple days now, and we have modified it to accept our 10MHz reference clock at 50 Ω.  Termination in our test setup is 50 Ω throughout.  The ADF4351 is locking and producing a reasonably clean 640MHz output sine wave, using the AD SW to program the registers over USB.  We are powering the synthesizer off a clean power supply, and we've made a few changes to the settings, such as turning on the Ref Doubler, MTLD, INT-N mode, more sensitive lock, etc.  Low output phase noise is important to us here.  We are using only RFoutA+, so RFoutA- is 50 Ω terminated.  Register settings are (R0-R5): 0x00400000, 0x08008011, 0x02004FC2, 0x000004B3, 0x00EA043C, 0x00580005.  So all appears well.

The issue comes in when we lower the output frequency.  As we do, the waveform becomes distorted, and I'm guessing this is due to the shunt inductors on this board.  Ultimately, our target is 40MHz, but even at 320MHz, we already see distortion, as shown below:  [blue: 640MHz, red: 320MHz, yellow, 160MHz].  At 40MHz, the output is horrific.  We're dropping by halves to keep all settings the same and only Div changes: INT 128, FRAC 0, PFD 20.



UG-435 states "The board contains 7.5 nH shunt inductors on the RF output stages, which is optimum for a frequency range of 1 GHz to 4.4 GHz. Lower frequencies than this require larger output stage inductors, or a 50 Ω resistor to Vdd, which provides lower output power, but wider frequency bandwidth than the inductors."  In the ADF4351 data sheet, there is Table 8 which lists LC Balun component values for what appears to be a slightly different design than what is implemented on the evaluation board.  The inductor values go up exponentially as frequency drops, but there is no entry near 40MHz, nor any information about how to calculate this value.  I notice that the other values in the table change drastically as well, and there is not always an obvious 1:1 correlation with the components on the evaluation board.

We're wondering what the best way to proceed is.  Extrapolating shunt inductor values from the table gives a value of around 1.1uH, but would swapping the shunt inductors (L2 and L3 on the evaluation board schematic) for this value be sufficient?  Or would we also need to change the values of the related capacitors as shown in Table 8?  If we decide to go the 50 Ω route, do we remove these inductors entirely?  Where would be a good place for the 50 Ω resistor?  Output power is somewhat of a concern as we're ultimately feeding into a AD9361 setup, which we already have working well, but with a different reference which has slightly too much phase noise for a recent new application.

We're under a bit of time pressure here, so we'd appreciate some guidance from more knowledgeable folks before we go ripping into this 0402-sized section of the board.  We have studied the ADF4351 "Cheat Sheet" and it was very helpful in other areas, but not in this regard.  Thanks in advance.