to put the via in the BGA pad is feasible but it must be a micro via
(otherwise there are paste problems with bigger vias in pad)
these are about 2 mils in dia
the ones i have investigated are laser fab'd and as such only go to the
layer directly below where you put a blind or buried via to escape
this can open up routing channels and save layers at the expense of the
higher tech features

given all the special fab stuff we elected to go with the staggered
conventional via approach
the BGAs were fine pitch (1mm) and close to 700 pads
these particular ones did not have the power all clumped in the center
and the center was fully populated with pads
so it was truly swiss cheese

all dimensions following are in mils
we used via dia = 18, via hole = 10 (pretty aggressive)

routing density was such that we had to go to 4 mil track 4 mil gap to
get 2 through each channel
(it still took 16 layers)
BTW, it worked!

on another job requiring probing and access points as you described we
used something from Ironwood that had balls on the bottom and a forest
of pins sticking up on top
they make bds that plug on that that hold the real BGA and have many
fannout test points

you need to leave a LOT of low profile room around the BGA to allow for
the plug on adapters

Dennis Saputelli

Matt Polak wrote:
> At 11:09 AM 2/5/2002 +1100, you wrote:
> >Yes, but the power connections are made through the Vias are they not?
> >There are over 100 ground and 3.3V/2.5V balls that connect to internal
> >planes through Vias. At least for this particular BGA. Therefore, wouldn't
> >the thermal effects still be relevant in this case?
>          Actually, a related question I had that seems to tie in to this one...
>          I have been told by some engineers that they often like to put a
> 'via per pad' when working with BGAs, especially when doing prototypes.
> Part of the reason for this being you have all of the connection-points on
> the BGA part available as testpoints and/or solderpoints on the opposite
> side of the board for those lovely last-minute fixes on prototypes. (Also,
> several companies make direct BGA<->PGA snap-in adapters for prototyping
> that keep the EXACT BGA footprint extended into PGA, making it possible
> (from what I have been told by a sales rep) to open up your vias and simply
> drop their adapter in as a fine-pitch PGA thru-hole device. When it comes
> time to do a production run, you simply shrink the diameter of the vias and
> attach your BGA devices normally.)
>          I can obviously see that putting vias into the pads directly could
> potentially cause 'robbing' to occur if too large of a diameter was used,
> as well as the 'thermal stressing' due to cooling from power planes. (Colin
> mentioned 'plugging' vias under the BGA - is there a way to do this without
> using blind/buried? I'm unfamiliar with the term...) I have also been told
> by an assembly technician that a lot of times they prefer all BGA pads to
> have a via in them for thermal conduction from the other side, as it can
> potentially help the reflow process.
>          My question (for the BGA masters out there!) that may help to
> partially answer the question about thermal relief for the vias... Do you
> typically place your vias directly IN the pads, or do you create an array
> of vias offset from the actual landing pads? (I've seen the 'offset via'
> approach documented by several companies in various appnotes) In the direct
> via-in-pad approach I can see the need for the thermal-separation as was
> mentioned... But in the offset-grid approach, I suppose it wouldn't matter,
> since not much heat will theoretically get conducted across the trace from
> BGA pad to via to make an appreciable difference.
>          Any comments about this from those who have worked with BGA?
> Trying to get all of the facts, since some upcoming designs we are planning
> on working on need to be based on parts only available in BGA. :/
>          (P.S. THANK YOU again to all of the folks that replied to my
> question last week concerning autorouting vs manual routing! You all are an
> absolute goldmine of technical knowledge. Thanks for sharing!)
> -- Matt
> >At 06:10 PM 4/02/2002 -0500, you wrote:
> >>At 08:34 AM 2/5/2002 +1100, Colin Weber wrote:
> >>
> >>>I was talking with another designer, Phil Dutton, and he aslo
> >>>recommended the Power Connections Relieved.
> >>>"Make sure that your Power connections are relieved, or they will cool
> >>>down faster than the rest of the device, inducing stresses."
> >>
> >>That is "Power Connections." We've been talking about vias. Different animal.
> >>
> >>Abdulrahman Lomax
> >>Easthampton, Massachusetts USA
> >
> >
> >Regards,
> >
> >Colin Weber

___________________________________________________________________________            Integrated Controls, Inc.    
   tel: 415-647-0480                        2851 21st Street          
      fax: 415-647-3003                        San Francisco, CA 94110

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
* To leave this list visit:
* Contact the list manager:
* Forum Guidelines Rules:
* Browse or Search previous postings:
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to