Post Go back to editing

AD9837: ignored commands

Hi all,

I'm using AD9837 as an OOK (on-off keying) modulator on a prototype board.

The DDS is driven by an Altera Cyclone IV FPGA providing MCLK (15.6MHz) ,SCLK (4MHz),FSYNC and SDATA.

On the output side we're driving a non-inverting op-amp amplifier, which is used  as a pre-amplifier.

Sometimes, it seems that the DDS ignores the command I send from the FPGA.

off keying modulator, I've initialized the DDS using the following command sequence, then I switch FSEL whenever I've a transition between 0 and 1. This is my init. sequence:

0x0100

0x7ac1

0x8000

0x1100

0x409D

0x8000

0x0000

For switching FSEL, I use 0x0000 (f0) or 0x0800(f1).

In the attached image tek00109, you can see that our command (0x0800) respects all timing requirements.

On the upper slot of the image, you can also see that after this command (marked with the gray cursor) the output of the DDS still generates a 152kHz sine wave (Yellow trace).

On the other hand, in image tek00112 you can see that the same command forces the DDS to stop the output wave (see gray cursor on the upper slot).

I've verified all timing requirements in the datasheet, I've tried to reduce the SPI frequency and I also tried to give many times the same command for each transition in the output data flow, but I didn't solve the problem.

The only thing I noticed is that the error is more frequent at the maximum SPI frequency, even though the timing requirements are always respected.

I assume the DDS is the only responsible for this problem.

Please, could You provide a feedback about a possible cause of this behaviour?

Thanks,

Fabrizio Lucini

attachments.zip
Parents
  • Hi Miguel,

    I can see the same problem in both transitions (0->1 and 1->0), event though it's more likely to see the error when I switch to freq1.

    I'll check again the power supply and the grounding, but I must say the PCB follows the rules described in datasheet. In my first check, the 3.3V power supply seemed to be good.

    In attachment, you can also find the portion of our schematic in which the DDS is used: EXP_SPI lines are directly connected to the Cyclone IV FPGA (3.3V LVCMOS mode), while CARRIER is directly connected to an AD8672 op.amp. used as a non-inverting amplifier.

    As you can see, AGND is decoupled w.r.t. to digital GND using a ferrite bead. and also 3.3V is decoupled in the same way.

    Could You see something strange in the attached schematic?

    Regards,

    Fabrizio

Reply
  • Hi Miguel,

    I can see the same problem in both transitions (0->1 and 1->0), event though it's more likely to see the error when I switch to freq1.

    I'll check again the power supply and the grounding, but I must say the PCB follows the rules described in datasheet. In my first check, the 3.3V power supply seemed to be good.

    In attachment, you can also find the portion of our schematic in which the DDS is used: EXP_SPI lines are directly connected to the Cyclone IV FPGA (3.3V LVCMOS mode), while CARRIER is directly connected to an AD8672 op.amp. used as a non-inverting amplifier.

    As you can see, AGND is decoupled w.r.t. to digital GND using a ferrite bead. and also 3.3V is decoupled in the same way.

    Could You see something strange in the attached schematic?

    Regards,

    Fabrizio

Children
No Data