AD9858 chip will take how long to respond the order it receive and begin to generate the wave? If I already use FPGA chip Cyclone_EP1C6T144C6 and AD9858 generate a single_tone mode wave , in there I use 80MHZ clock to deliver control word 、frequency word and phase word.when I want a 0.3us width wave ，my solution is counter 0.3us time ，then set the control register CFR and CFR to logic 1，and updata this control word to DDS core .But there don't generate the wave I expect. When I cut the deliver clock down to 20MHZ ,the same set will be generate the expect wave .But observe the wave on oscilloscope its quality is not good ! why cut down clock can generate? How to do this type wave?And how to improve the wave's quality?
I presume that you are working with parallel communication since you are using an FPGA and in my assessment, the fastest possible update speed for parallel communication would be around 60 MHz. SYNCLK is SYSCLK/8 and it takes 2 SYNCLK pulses to latch the value from the IO buffers to the control registers. It seems that at 80MHz you're updating the I/O buffers at a rate faster than what the control registers could handle. Using CFR and CFR is suitable for your application since it generated a waveform at 20MHz. Are you adjusting the core clock frequency to the FPGA whenever you change the data speed? If that is the case, I think your timing for the 0.3 us width wave would be different since the value at the timer register would increment at a rate relative to the core clock frequency.
Thanks a lot for you explanation,
I already figure it out the issue about how fast and cut wave at the point when I want .BUT I have another issue about phase change ，use FPGA chipCyclone_EP1C6T144C6 and AD9858 generate a single_tone mode wave ,FPGA clock is 30MHZ and DDS clock is 240MHZ ,I want generate a 3us width continuous wave who is contain the same frequency and 3 part different start phase， each part is 1us width. MY solution is set three profiles of DDS , and select profile in turns, then DDS generate the wave i need ,but between two different phase part there has about 100ns blank space .HOW to make the wave change phase but the wave is continuous without the blank space?
I just want to clarify. When you say blank space, do you mean that the signal settles within the "zero crossing"(if it is ac coupled) of the waveform?
I have take some photos about my questions.
This photo show the two part of same frequency but different start phase ,by set different profile of DDS to generate(FPGA_CLOCK=30MHZ,DDS_CLOCK=240MHZ). But I think the transitional period is long ,you can see it almost 100ns width.In this condition how to shorten the width? Make the wave like this photo
Another question is if I want genertate a wave which contain 8 part,each part's wave is 0.33us width and continuous but has different start phase. The output wave is like this photo .
You can see that each part wave's quality is not good,amplitude is not consistent ,look like uneven.IS there some solution to avoid uneven?
I believe that AD9858 doesn't have the amplitude scaling factor capability, thus I'm suspecting that you're using a different module to vary the amplitude of the signal. From what I observed in the first figure, considering that the transition occurred at the center of the oscilloscope, the amplitude envelope is not in sync with the phase transition of your signal. When the transition occurred, the phase of the signal changed "instantaneously" but not simultaneously with the amplitude envelope. I think it has something to do on how you modulated the amplitude of the signal. Did you mix your signal to a modulating signal or you used an amplifier wherein the gain is controlled to match the timing of your transition? Some of our DDS have features for this kind of application, an example of which is the AD9910. It has an amplitude scaling factor fitted for amplitude modulation, wherein you can control the amplitude of the signal the same way on how you control the FTW. This capability would allow you to properly time the phase,frequency, and amplitude transition of your signal.