[Arm-netbook] HDMI High-Frequency Layout: Impedance

[Richard Wilbur]

On Fri, Aug 4, 2017 at 4:16 AM, Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:
> On Fri, Aug 4, 2017 at 8:51 AM, mike.valk at gmail.com <mike.valk at gmail.com> wrote:
>> 2017-08-04 9:19 GMT+02:00 Luke Kenneth Casson Leighton <lkcl at lkcl.net>:
>>> On Thu, Aug 3, 2017 at 11:53 PM, Richard Wilbur
>>> <richard.wilbur at gmail.com> wrote:
>>>
>>>> 3.  Instead of using two separate anti-pads on signal vias, combine
>>>> them into an oval shared antipad (on every layer) to reduce parasitic
>>>> capacitance.
>>>
>>>  oo.  never heard of this practice.   never heard of "anti-pads"
>>> either!  so sorry, if you put some references i missed them.
>>>
[...]
>> Basically a barrier between a via and a passing layer. Something
>> default to prevent shorting.
>
>  oh, ok - just a hole where you'd expect one to be :)  didn't know its
> name was "anti-pad".

Sorry for not explaining very well and no references.  Thank you Mike
for adding both to the discussion.

>> With differentals have them combined so there is nothing in between
>> them in a horizontal line/z axis as well.
>
>  turns out that the anti-pads from PADS are big enough to create a figure-8.

That's cool.  The reason for an oval (or ellipse) shape instead of a
figure-8, if you can muster an oval or ellipse, is that the sharp
points greatly concentrate electric fields--leading to
capacitance--which is pretty much the opposite effect of a smooth
curve.

I'm glad to hear that successful HDMI layouts have been achieved using
5mil width, 5mil spacing for differential pairs.  The reason I
recommended 5.1mil width, 5mil spacing is because TI's geometry
recommendation includes the condition width > spacing.  Sounds like
that is not a necessary condition for a working board.