Forum Discussion

aucamera1's avatar
aucamera1
Icon for New Contributor rankNew Contributor
5 months ago

GTS SDI II IP Core...what is the tx_vid_clkout frequency

I'm using an Agelix 5 (the Altera development board) with Quartus Pro 25.3.1 w/patch 1.02.

i'm trying to set-up the GTS SDI II IP Core (version 2.3.0)

on the Main screen, the only configurable values are:

  • Video Standard = HD-SDI
  • Direction = Transmitter
  • Insert payload ID = off
  • SDI_II wrapper = Both BASE and PHY

everything else is grayed out.

I have connected the tx_pll_refclk to the 148.5MHz input.   I do see the tx_pll_locked signal behave as expected and it does lock.

the problem is when I look at the tx_vid_clkout signal its 58.3MHz.  I expected it to be 74.25MHz.

I'm somewhat confused as to whether tx_pll_refclk should be 148.5MHz or 74.25MHz, but if the 148.5 is wrong then I would have expected tx_vid_clkout to be twice, not a somewhat random value of 58.3MHz.

when I reconfigured the IP core for 3G-SDI, the tx_vid_clkout frequency doubled to 116MHz.  The doubling would be what I expected, but still a wrong frequency.

i'm not sure what i'm setting wrong and why i'm getting a clkout of 58.3MHz

23 Replies

  • xcrouzy's avatar
    xcrouzy
    Icon for New Contributor rankNew Contributor

    Hi aucamera1​ 

    Thank you very much for your very quick answer.

    I checked in the new version of Quartus 26.1 and there's no change.

    Maybe my problem is not exactly the same as you at the end. I have my own electronic board with two VCXO to provide a 100 MHz input clock (for GTS System PLL) and a 148.5 MHz input clock (for TX PHY PLL). For now I faced two scenarios :

    1. When I use pins REFCLK_GTS_CH1P/N pins for 148.5MHz reference clock and pins REF_GTS_RX_P/N pins for 100 MHz reference clock, my system behaves as you described in your ticket : the PHY TX PLL locks but the frequency is not correct.
    2. When I use pins  REFCLK_GTS_CH1P/N pins for 100 MHz reference clock and pins REF_GTS_RX_P/N pins for 148.5 MHz reference clock (swapping pin locations in QSF file and swapping my VCXO configurations), the PHY TX PLL sometimes locks (and TX CLOCK OUT is 74.25 MHz as excepted) but most of time it does not lock.

    Unfortunately I cannot provide a 297 MHz clock because my VCXO are limited to 250 MHz.

    I did create my SDI sub-system based on exemple design generated by Altera (in mode "Both Base and PHY") and the example design (signal names and clock constraints in SDC file) seems to expect a 148.5 MHz reference clock for TX PHY. But it does not mean it will work with this frequency, I cannot easily test this example design on my electronic board.

    I am going to continue my investigation keeping in mind all your remarks. I will try to use option "Base only" option with an external GTS PMA/FEC Direct PHY IP. When I generate an example design in this mode, indeed, I can configure the reference clock frequency for TX PHY PLL (which is configured to 148.5 Mhz by default, both for 25.3.1 and 26.1 versions of Quartus).

    Thanks again for you very fast reply.

    Regards

    Xavier 

  • xcrouzy's avatar
    xcrouzy
    Icon for New Contributor rankNew Contributor

    Hi aucamera1​ ,

    I found your topic about TX PLL lock issue with GTS SDI II IP.

    I'm writing to ask if you've solved your problem, as I'm currently experiencing the exact same issue (same version of Quartus, same version of IP) : the lock signal is asserted, the tx_ready signal is asserted but the TX clock out frequency is not 74.25 MHz as expected.

    Thanks in advance for your answer.

    Best Regards

    Xavier

    • aucamera1's avatar
      aucamera1
      Icon for New Contributor rankNew Contributor

      Hi Xavier,

      the answer is buried in the chain above, but its a bit confusing to follow (i'll go back and mark the solution post to see if that helps).

      basically, the GTS SDI II IP, when "Wrapper Options >> SDI_II wrapper" is set to "both Base and PHY", then the txpll_refclk input clock must be 297MHz.  If you set the wrapper to Base only, then you can configure that clock but you have to use the GTS PMA/FEC Direct PHY IP as well.

      I can't remember exactly at the moment, but there were issues with me getting that 297MHz clock to the pin when using the "both".

      IMHO.....I think they set that "both" wrapper up using an eval board that had a 297MHz on a daughter card and left it in there.  It doesn't make sense to require that high of a frequency clock to be produced simply because of a setting variable gets hardcoded when they combine the two IPs.  Hopefully they fix this in a future revision 

       

  • aucamera1's avatar
    aucamera1
    Icon for New Contributor rankNew Contributor

    I'm checking the frequency via a clock measuring logic circuit I have.  I have several clocks I'm doing this with.  

    I'm using the Agilex 5 FPGA E-Series 065B Premium Development Kit

    • Wincent_Altera's avatar
      Wincent_Altera
      Icon for Regular Contributor rankRegular Contributor

      Hi aucamera1 ,

      Your configuration and 148.5 MHz refclk are likely fine. The ~58.3 MHz (HD) and ~116 MHz (3G) clocks indicate the SDI II IP selected a wider internal word width (effectively ~25 bits), 
      which is expected behavior when the “Both BASE and PHY” wrapper auto-manages transceiver constraints on Agilex 5. 
      If you need tx_vid_clkout = 74.25 MHz, regenerate the IP so you can select 20/40-bit widths, or synthesize 74.25 MHz externally and re-time.
      Your numbers line up with the SDI II IP auto-selecting a wider internal parallel word width than the “classic” 20-bit/40-bit path. In the SDI II transmitter, the user video clock (tx_vid_clkout) is essentially:

      • tx_vid_clkout ≈ serial line rate / internal TX parallel word width

      If the IP uses:

      • 20-bit width (HD) → 1.485/20 = 74.25 MHz
      • 25-bit width (HD) → 1.485/25 = 59.4 MHz
      • 40-bit width (3G) → 2.97/40 = 74.25 MHz
      • 25-bit width (3G) → 2.97/25 = 118.8 MHz
      • 50-bit width (3G) → 2.97/50 = 59.4 MHz

      Your measurements:

      • ~58.3 MHz in HD mode is close to the ~59.4 MHz expected for a 25-bit path (allowing for measurement tolerance and fractional-rate variants).

      Regards,
      Wincent

      • aucamera1's avatar
        aucamera1
        Icon for New Contributor rankNew Contributor

        Hi Wincent_Altera​ ,

        Sorry, but i'm a bit confused by your answer.  There is no 25-bit HD-SDI setting.  The bit width is set by the Video Standard (except for SD-SDI which can be selectable 10 or 20-bits, but not sure its even supported for Agelix 5).  So if you're saying 25-bits is being selected, how do I force it back to 20?

        thanks,   -rob

         

  • Wincent_Altera's avatar
    Wincent_Altera
    Icon for Regular Contributor rankRegular Contributor

    Hi aucamera1 ,


    Just want to confirm, how you check the frequency ?
    Through analyzer or oscilloscope ? I might need to try and see if I can replicate your issue in my place.
    Which devkit you are using ? Modular or Premium devkit ?

    Regards,

    Wincent

    • aucamera1's avatar
      aucamera1
      Icon for New Contributor rankNew Contributor

      I just went back and pulled the tx_vid_clkout onto a user IO pin and verified it is 58MHz with an o'scope.