I am attempting to verify the linearity of the AD8302 evaluation board at 2.4GHz. I first tested the board as was specified in the datasheet: I used two RF sources, offset slightly in frequency (approximately 100 kHz). The output power of one source was verified to be approximately -30 dBm (using a spectrum analyzer) and was connected to port INPB. The output power of the second source, which was connected to INPA, was varied from approximately -10 dBm to -50 dBm (40 dB dynamic range). A digital multimeter was connected to the VMAG port and the voltage at each gain setting was recorded. This test was successful, and the linearity performance outlined in the data sheet (at 2200 MHz) was confirmed (including the predicted slope in V/dB and offset in V).
The application in which I am hoping to use the AD8302 is one in which I have a single RF source with two complimentary phase-locked outputs (the ADF4360 evaluation board, programmed to 2.4 GHz). I attempted to verify linearity again in the following manner: One output of the ADF4360 was fed to an attenuation block such that the output power was roughly -30 dBm, and this was connected to INPB. The other output was fed to a variable attenuator (which was confirmed to be linear by VNA measurements), the output of which was connected to INPA. VMAG was connected to a DMM. The attenuator was adjusted such that the power at INPA varied between -10 dBm and -50 dBm, and the output voltage was measured at specific points. In this case the plot of output voltage vs. gain did not meet linearity specifications; a cubic fit was required in order for the linearity specifications to be met.
Although the cubic fit does work, I really am looking for a linear response with error less than +/- 0.5 dB. Is it possible to achieve this with my setup?
Thank-you for your time,