Computer repair shop

Drew Van Zandt drew.vanzandt at gmail.com
Mon May 5 20:51:55 EDT 2008


Briefly, as I am bound for the club to dance until I can't:

* Continuous ground plane, or nearly continuous, is highly desirable for
prevention of crosstalk and reduction of radiated emissions.
* For high-speed purposes, ground plane and power plane are nearly
indistinguishable re: above.
* Every chip on the surface would put a hole in the power or ground plane if
things were run there.
* signals between nearby chips are optimally run on the surface because then
you don't need vias.  To descend a layer and then return would require 2
vias, which would block routing resources on at least two layers at each
via.  This adds up fast.
* Putting the power and ground planes right next to each other (e.g. layers
2 and 3 when 1 and 4 are the two sides) gets you some free extremely low ESR
capacitance.  More true in boards with lots of layers where layers are
packed closer together.
* Thermal/soldering issues: power and ground are usually big, fat, wide
pieces of copper, i.e. heat sinks.  You can ruin your board yield /
soldering reliability by getting these big heat sinks too close to component
pins.
* Assuming blind vias are being used, you can supply quality power even to
mid-field pins in a BGA - a surface power plane would be too perforated with
vias to be terribly useful.
* Large continuous copper areas on the surface can make the boards more
likely to warp.  There are ways around this, you'll notice often they
crosshatch large areas like this with empty space in many cases.
* EMI is least if the signals are all next to a plane.
* crosstalk is least if there is a plane between two high-speed signals.
* As someone else said, it's awfully nice to be able to get to the important
signals for debug.
* All generalizations are false.
* I've left out two or three things that I thought of and then forgot; music
and pretty girls call.

--DTVZ

On Mon, May 5, 2008 at 7:27 PM, <VirginSnow at vfemail.net> wrote:

> > Date: Mon, 5 May 2008 14:00:29 -0400
> > From: "Drew Van Zandt" <drew.vanzandt at gmail.com>
> > Cc: gnhlug-discuss at mail.gnhlug.org
>
> >
> > Power *should* be on the internal layers.  That's where it belongs, for
> > several engineering and manufacturability reasons.  Call ideas
> dim-witted
>
> Care to enlighten us?  I guess PCB design is on-topic enough for this
> list. (/me ducks).  The only reason I can think of to run power on
> internal layers is that if the designer's loose screw shorts out
> something on the surface of the board, the sparks will only be 1/2 the
> size.  You sound like you've designed you share of PCBs.  What other
> reasons are there?
>
> For the record, my list of "dim-witted" ideas also includes things
> like soft power, spam filters, and iPhones.
> _______________________________________________
> gnhlug-discuss mailing list
> gnhlug-discuss at mail.gnhlug.org
> http://mail.gnhlug.org/mailman/listinfo/gnhlug-discuss/
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.gnhlug.org/mailman/private/gnhlug-discuss/attachments/20080505/77d85db0/attachment.html 


More information about the gnhlug-discuss mailing list