Forum Discussion

Altera_Forum's avatar
Altera_Forum
Icon for Honored Contributor rankHonored Contributor
13 years ago

PDN Tool Zh = Zs + Zv

Hello All!

The PDN Tool shows the impedance between the decoupling network and BGA on the graph as Zh = Zvia + Zspreading, but if I leave the plane and BGA vias unchanged, the Zh value changes if I select another power rail. e.g. changing from VCC to VCCIO, Zh goes up. Can anyone explain what's happening here?

I imagine that Zh is also including the package internal Z value for the different rails, but a clarification on this would be useful.

If that is the case, then what power rail should be used in the PDN tool when sharing several rails on one supply. The one with the lowest or highest Zh?

Thanks, Ken

2 Replies

  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Hi Ken,

    See Page 1-22 of the PDN tool user guide here: http://www.altera.com/literature/ug/ug_pdn_gen_device.pdf

    In the PDN guide for combining rails your supposed to choose the one with the highest Feffective. Feffecitve is the frequency above which PCB decoupling caps cease to help, due to the inductance between them and the die, due to bga vias spreading, etc. So it makes sense that in my PDN tool for VCC I see that when Zh becomes significant (is this the 3dB point?), that is where Feffecitve is, so there must be some relationship between the two, but in general I don't think you have to worry about Zh. You're also supposed to combine the current for all rails, and design for the lowest ripple voltage.

    Steve
  • Altera_Forum's avatar
    Altera_Forum
    Icon for Honored Contributor rankHonored Contributor

    Hi Steve,

    Thanks for the reply, and the reference in the UG. I am working with an Arria V and will combine, for example, VCCL_GXB, VCCR_GXB and VCCT_GXB as they are all 1.15V. They all have F-eff at 70MHz, but different values of Zh:

    VCCL_GXB Zh = 85mOhm

    VCCR_GXB Zh = 30mOhm

    VCCT_GXB Zh = 20mOhm

    It makes a huge difference to the decoupling results depending on which one is chosen in the PDN tool, so in the end I have used the value for VCCR_GXB as this is the supply drawing the most current in my design.

    It would be nice if Altera gave a bit more explaination about what the Zh values really are. They're obviously not just Zv + Zs.

    Ken