I've reconfigured my AD9523 PCBZ Eval board by replacing the on board 122.88 MHz VCXO with a 100 MHz VCXO of the same part number (CVHD-950) with the goal of generating multiple-of-10MHz frequencies (20 and 40 MHz). I'm not using a reference signal for PLL1, so I assume it is in holdover mode and thus is being driven by the VCXO. When I observe the output spectrum I find that the frequency is off by about 30ppm (e.g. off by 600 Hz at 20MHz, and off by 1200 Hz at 40 MHz), which causes problems for the equipment that uses this signal. I assume the error comes from the VCXO, though I am somewhat surprised that it has such a large offset.
My question is two-fold:
1) Is there a way to effectively bypass the VCXO, allowing the outputs to be determined solely by a (tunable) reference input? This way I could test that the frequency offset is the cause of my secondary equipment problems.
2) Is there a way to correct this frequency error by providing a reference input?