AD9959
Recommended for New Designs
The AD9959 consists of four direct digital synthesizer (DDS)
cores that provide independent frequency, phase, and amplitude
control on each channel. This...
Datasheet
AD9959 on Analog.com
Using the AD9959 to generate a 5.115MHz BPSK modulated carrier using the profile pins. Judicious selection of SYS_CLK ( 327.36MHz ) allows an integer relationship between modulation at 1.023MHz and the 5.115MHz. By adjusting timing delays this has achieved a close to zero crossing waveform from all 4 channels. However, close observation of the modulated signal shows intermittent non-zero crossing, Profile pin transitions meet the set-up times WRT to the SYNC_CLK but very close examination shows that a negative going transition of the profile pin can result in an early non-zero crossing by one SYNC_CLK cycle. Positive going always seems OK.
Other forum entries suggest the profile pins should be driven by an AND gate with one input being the SYNC_CLK. This clearly gurantees the setup time, at the cost of one SYNC_CLK delay, making the overall latency for the phase change SYS_CLK times 29 ( see data sheet) plus one SYNC_CLK.
Is this the correct solution?
John Newell
Jules.Nikko - Moved from Clock and Timing to Direct Digital Synthesis (DDS). Post date updated from Wednesday, June 12, 2024 11:26 AM UTC to Tuesday, June 18, 2024 8:01 AM UTC to reflect the move.
Hi jnewellhaysys ,
Apologies for overlooking this issue. Let me take a look at this and get back to you.
All the best,
Jules
Hi Jules,
Any upate on this? I have now determined that to get a truer zero crossing I am using one channel of the AD9959 to output a sine wave and a fast comparator (LTC6752) on the raw AVdd referenced differential outputs to give an exact (+-1ns) logic output. This can control the profile pin transitions to allow for the delay and latency. By small adjustments to the reference sine wave phase the phase modulation must occur close to the zero crossing. However, my original question was eliminating the SYNC_CLK uncertainty by using AND gates to drive the profile pins (only three now). The adjustment can only be as fine as the SYSTEM_CLK though, but hopefully good enough.
Regards
John