• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 2 answers
  • Subscribers 30 subscribers
  • Views 462 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AWR2243: Wilkinson Divider for Cascading Radar

Manwoo Kim
Prodigy 170 points
Part Number: AWR2243
Other Parts Discussed in Thread: AWR2188,

Tool/software:

Hi, I have some question about wilkinson divider.

When I review the PCB file about cascade radar (with AWR2243 and AWR2188), the length for LO divdier is quite worng.

I think lambda/4 is about 3.7475mm (because LO frequency is 20GHz), but wilkinson divider's length is quite different.

Could you please explain how to design the wilkison divder? (calculation value, designed value, etc)

Thanks,

  • 0
    TI__Expert 5341 points

    Hi Manwoo,

    I'm the designer of the AWR2188 board so I'll comment on that specifically, but the Wilkinson splitter concept remains the same on the AWR2243. You're right that λ/4 in air at 20GHz is about 3.75mm. However, on the board, we are using Isola 370HR material which has a Dk (dielectric constant ) of 4.24. That shifts the quarter wavelength to approximately to 1.83mm ~ 72mils. I rechecked the layout and the actual λ/4 arms of the divider each measure about 72.9 mil, which aligns closely with the theoretical length. The trace width of these arms is 4.75 mil, which is designed to yield a characteristic impedance of ~70Ω as required for a 2 way wilkinson in a 50Ω system. So, while the surrounding routing may appear longer, the critical λ/4 sections are correctly designed to match the 20GHz frequency and material stackup. I'd recommend simulating your design structure on HFSS or other tools using the actual stackup to confirm the behavior rather than relying only on λ = C/F in air. 

    Regards,

    Aydin 

  • 0 in reply to Aydin Kiasat
    Prodigy 170 points

    Hi, Aydin!

    Thanks for your reply.

    I have more questions about layout of divider.

    1. When we are using 370HR, quarter wavelength is almost 1/2 shorter than theoretical length. Is it general?

    2. When we do simulation, should we use Dk for 20Ghz?

    3. Around Divider pattern, solder mask should be opened? Is there some recommend guide?

      (Near BGA ball should be closed? Is there recommend length or width?)

    Thanks,

  • +1 in reply to Manwoo Kim
    TI__Expert 5341 points

    Hi Manwoo, 

    1. When we are using 370HR, quarter wavelength is almost 1/2 shorter than theoretical length. Is it general?

    yes, that's expected. The theoretical λ/4 in air is longer, but in a PCB dielectric, the guided wavelength is shorter because the wave travels slower in the material. This is not specific to 370HR, it applies to any PCB substrate. The quarter wavelength reduces roughly by a factor of 1/√DK.

    2. When we do simulation, should we use Dk for 20Ghz?

    yes, always use the frequency dependent Dk. Although some datasheets only specify the range up to a certain frequency, but you can always estimate based on the best fit line or for accuracy reach out to them for the exact Dk value at your frequency of interest. 

    3. Around Divider pattern, solder mask should be opened? Is there some recommend guide?

      (Near BGA ball should be closed? Is there recommend length or width?)

    It is generally recommended to remove solder mask and have exposed copper to improve impedance control especially around the wilkinson. Near sensitive areas like the BGA, we recommend following the reference design as is. 

    Regards,

    Aydin

  • 0 in reply to Aydin Kiasat
    Prodigy 170 points

    Hi, Aydin!

    Thanks for your kind answer!

    I will design the wilkinson divider following your guide!

    Thanks,

  • 0 in reply to Manwoo Kim
    TI__Expert 5341 points

    Hi Manwoo,

    Sure, I’ll close this thread for now and feel free to reach out if any further questions in the future.

    Regards,

    Aydin