I'm using the ADRF6750 with a 20MHz reference clock and an output frequency of 1030MHz. At 0dB attenuation, the output is within specifications and is stable UNLESS we run the output directly into a TriQuint 857216 SAW (BW=10MHz). With the SAW, the output amplitude becomes unstable. Adding 1dB of attenuation within the ADRF6750 brings back stability. The PLL lock detect remains stable in the "locked" state during these tests.
The SAW is supposed to be well matched to 50 ohms. Are there any known stability issues with the ADRF6750 output amplifier?
Is the RFOUT ac-coupled to the load? Can you share the details of the components on the path from the RFOUT to the SAW unit?
A 100pF cap couples the output of the ADRF6750 to the common terminal of a SKY13251 Single Pole Triple Throw RF switch, then, with the problematic path, another 100pF capacitor couples it to the SAW.
Here's the output of the SAW on a power meter with no attenuation on the ADRF6750:
And this is what it looks like with the ADRF6750 set to 1dB:
And here's the ADRF6750 set to 0dB with an external DC-blocked 1dB attenuator between it and the SAW (almost identical, but just slightly flatter):
Thank you for the plots, these are helpful in understanding the issue.
In order to isolate the problem would you mind repeating this experiment but driving the SAW filter using a signal generator with the same output power as the ADRF6750? If you could also include the RF switch in the path it may be useful.
It would take a little while to get the signal generator modulated in synchronization with the ADRF6750, so I just relied on the switch for isolation which gives only around 10dB. Below is the result. I varied the input signal in 1dB increments around this point, and the results show no anomalies.