Forum Discussion

Samo's avatar
Samo
Icon for New Contributor rankNew Contributor
1 year ago

Agilex 5 reconfigurable PLL - emif

Trying to reconfig a PLL with the EMIF Calibration IP debug via system console Clocking and PLL User Guide: Agilex™ 5 FPGAs and SoCs enabled dynamic config within IOPLL and included EMIT Calib...
Fpga Devkit
LenzHyd's avatar
LenzHyd
Icon for New Contributor rankNew Contributor
3 months ago

Hello Sir or Madam,

I have a similar problem. I am able to perform multiple phase shifts (more than seven), but some of them do not work properly.
To be more precise, most shifts behave as expected (roughly four out of five, although this is not entirely consistent), but occasionally the clock signal is shifted by much more than the intended 100 ps.

I would like to build a tapped delay line, and inconsistent phase shifts make this difficult.

I look forward to your reply.

Kind regards,
Lenz

  • AqidAyman_Altera's avatar
    AqidAyman_Altera
    Icon for Regular Contributor rankRegular Contributor
    3 months ago

    Hi Lorenz,

     

    For your issue, are you using the HSIO or HVIO I/O PLL block type?

     

    Thanks.

    • LenzHyd's avatar
      LenzHyd
      Icon for New Contributor rankNew Contributor
      3 months ago

      Hi,

      we are using the calbus and EMIF calibration IP to reconfigure the PLL. Thus according to the Agilex 5 Clocking and PLL user guide, this shoud result in a HSIO I/O PLL block type. 

      Thank you for your help.

      Kind regards,
      Lenz

      • AqidAyman_Altera's avatar
        AqidAyman_Altera
        Icon for Regular Contributor rankRegular Contributor
        3 months ago

        Hi Lenz,

         

        The minimum supported phase shift step is fixed by the VCO frequency (e.g., at 1 GHz VCO, the step is 125 ps). If the requested phase shift (e.g., 100 ps) does not align with this step size, the PLL cannot accurately deliver it.
         

        When multiple shifts are performed, quantization and rounding errors can accumulate, leading to some shifts being much larger than intended (i.e., the “occasional large jump” that you observed).
         

        Regards,

        Aqid