Why I get two different states of BPSK output when I use AD9852 evaluation board?
Hi zh _allan,
What codes are you using for the BPSK? Random code or just a sequence of 1's and 0's.
just 1MHz square wave (2.5V LVTTL level)
In my understanding, there are times that, it so happens that, the transition is near phase continuous. While most of the time, it is phase coherent. As a result the spectrum varies. Maybe if you can make it such that it is only operating either of the two, then you can have a single state spectrum.
I see from your plots the carrier frequency is either 60MHz or 70MHz.
1) What system clock frequency are you applying to the DDS?
2) What level of PSK are you employing (e.g., pi/2, pi/4 or other)?
1.the system clock frequency is 280MHz.
2.the two phase levels of BPSK is 0 and pi.
I think the main difference between the two spectra stems from the fact that a 70MHz carrier happens to coincide with exactly 1/4 of the system clock frequency. In such a case, if we ignore the modulating signal, then the output of the DAC is a repitition of the same 4 phase points (0+theta, pi/2+theta, 3pi/2+theta, 2pi+theta -- where theta is an arbitrary initial phase offset). When modulating phase by pi, however, the DAC will always output one of these same set of 4 phase points.
When the carrier is 60MHz, however, the carrier frequency is no longer exactly 1/4 of the system clock frequency. Again, ignoring the modulating signal, the DAC puts out many, many different phase samples. Hence, when modulating phase by pi, the DAC can hit a wide range of different phase points (not just the same 4 as in the above case).
I think the absence of the carrier is an expected consequence of phase modulation by pi. However, in the first case with a carrier equal to 1/4 the system clock, I suspect the appearance of the carrier is due to a Nyquist image. I would expect that most any carrier that is not a power-of-two submultiple of the system clock frequency will yield a spectrum similar to the one in the "1.png" image of your first post.
After looking more closely at your plots, I realized I had misread them. I failed to notice that one plot used the CENTER/SPAN display and the other the START/STOP display. This lead to my incorrect assumption of two different carrier frequencies.
I assume (correct me if I am wrong) that both plots are for a carrier frequency of 70MHz.
I have had a chance to dig more deeply into the phase modulation spectrum (see attached simulation plot). Indeed, there should be no carrier present in the output spectrum. I cannot explain the presence of the carrier in your "2.png" plot.
Just out of curiosity, how are you doing the modulation (e.g., via the profile pins, SPI, other)?
Regarding my previous post, there is a possible explanation for the appearance of the carrier in your "1.png" plot...
If you had inadvertantly used a modulation phase jump of pi/2 instead of pi, the resulting spectrum comes very close to your "1.png" plot (see attached simulation plot).
Retrieving data ...