The second harmonic specification on the ADF4350 is -20 dBc. Is this measured with a single-ended or differential output? If it is measured differential, I would have expected even-order harmonic cancellation to kick in and reduce this level.
All RF specifications are measured with each RF output pin terminated in 50 Ohms. The performance was measured through RFoutA+ to the spectrum analyser, with the unused RFoutA- driving a 50 Ohm resistor. Some even order cancellation is to be expected if the RF outputs are combined in a balun, or if driving a differential load.
Thanks. This is a useful piece of information. Even-order harmonics tend to mess with the timing of the edges of the fundamental. This is bad news for IQ Modulators and IQ Demodulators that use a 2XLO where any duty cycle degradation in the LO will degrade the quadrature accuracy. So if you can connect a device like the ADF4350 differentially to an IQ Mod or IQ Dmod (with 2XLO), you will get cancellation of the even-order harmonics which in turn will remove the need to filter the LO.
I'm also having a similar problem. The ADF4350 supports generation of output frequencies from 137.5MHz to 4000MHz (using dividers of course). But it is not possible to put a harmonic filter at it's output, since the second harmonic falls within the output bandwidth (We are interested in an output frequency range: 50-2200MHz). So the second harmonic suppression is a critical parameter. We have purchased an evaluation board for the same (Part no: EVAL-ADF4350EB1Z, Rev B), but in that there is no provision for putting a balun at it's output. Do you have any results showing the improvement in harmonic performance while using differential outputs.
As Eamon points out if a 2x MOD/DEMOD is used then the ADF4350 harmonics will not impact the mod/demod performance in any way, as the square wave nature is beneficial to such a modulator architecture. If you choose one of these you should not see any performance degradation due to LO harmonics.
Such a modulator is the ADL5385.
Also, used with a Mini-circuits balun TC1-1-43 we get 20 dB of rejection of the 2nd harmonic at 2 GHz compared to single-ended.
"TC1-1-43A+" and "TC1-1-43+" has no middle point, and no applicable in differential one. IMHO.
Retrieving data ...