I used the AD9910 to design a custom PCB, which acts as a DDS based frequency converter. The PCB is working just fine, wherefore I did a couple of phase noise measurements to characterize my PCBs. I have three identical PCBs, but as you can see in the plot below, I get three slightly different phase noise characteristics. The measurements are taken one after another within a short period of time, so the influence of temperature fluctuations are negligible. The measurement setup is also the same for every PCB, so the measuring device, the cables and the clock source are identical in every measurement. I also did more measurements distributed over a whole week, but the results that I see are always the same expect for some very minor noise fluctuations (not even close to the differences you can see in the plot below). So as you can see the repeated measurements of the same PCB lead to the "exact" same result (custom PCB 2), but the measurements of two different boards are leading to measurable differences. I also got more measurements of PCB 1 and 3, but the observation is always the same.
As a clock source, I use a 1 GHz MXO-PLD and the output of the AD9910 is 19.53125 MHz for every measurement.
So my question is: Is it possible that these differences in phase noise come from manufacturing fluctuations of the AD9910 (particularly of the A/D converter of the AD9910)? Did you ever experience those fluctuations within your own measurements?
Just for the record: on the custom PCB are two other major components in the REF_CLK and I_OUT signal path as you can see in the excerpt of the schematics below.
REF_CLK path to AD9910:
I_OUT to SMA connector:
If you need further information, feel free to ask. Thanks in advance.
The values for phase noise indicated on the specifications of the data sheet are typical values and not min or max because this parameter fluctuates from part to part. In our experience in testing…
The values for phase noise indicated on the specifications of the data sheet are typical values and not min or max because this parameter fluctuates from part to part. In our experience in testing SFDR of the AD9910 evaluation boards, we are getting values that are different for each board but these values are close to the typical value indicated on the data sheet.
thank you for your reply. You confirmed our own assumption, so the mystery seems to be solved.