I would reply that Dave Hoover's response to his customer was right on the mark.
This is a situation that we all too frequently encounter with large surface area
boards and large perimeter parts.  At its worst we have seen up to 12 components
removed and then replaced where spot underside preheating has been used leaving
a condition that for lack of a better term can only be described as "egg-crate."

Once the rework/repair operator has created this situation we enter into a
damage control scenerio.  The dimpling or egg crate effect can leave large
residual stresses in the board structure along with all the other various types
of heat related damage.  We have had some success with placing the board in an
oven and SLOWLY bringing the board up to temperature and then back down.  The
actual temperature and time must be determined (based on board material / size /
damage / etc.) on case-by-case basis.  By the time we have become involved
several rework and repair scenerios have already been attempted which severly
limits the available options.  Normally we will bring the board up to
temperature over a 2-3 hour period; hold for a minimum of 4-6 hours; turn off
the oven and allow to cool overnight.  This can only be described as a salvage
operation for a high value assembly.

--------------------------------------
  [log in to unmask]
  Corporate Training Administrator
  PACE Inc.
  9893 Brewers Court
  Laurel MD 20723-1990

  301.490.9860 x329 voice
  301.604.8782      fax
--------------------------------------

<!-- On 7/12/95 Dave Hoover Wrote -->

Old-Return-Path: <miso!hadco.com!dhoover>
Encoding: 1301 Text
Resent-Message-ID: <"TGtij3.0.lSF._s31m"@ipc>

Resent-From: [log in to unmask]
X-Mailing-List: <[log in to unmask]> archive/latest/686
X-Loop: [log in to unmask]
Precedence: list
Resent-Sender: [log in to unmask]
               I have a customer who is experiencing excessive warpage
          after reworking BGAs. The PCB is a .062" thk - 10 lyr made
          from 140'C Tg FR-4. He uses forced hot air under the BGA
          device that heats only the region/area where the device is
          located. I discussed with him the issues
          (stress/TCE/expansion/contraction) that he may be
          encountering by heating only specific areas. I also stressed
          issues associated with controlled heat rise and cooldown of
          PWAs. His next tests will be to heat the entire board
          slightly below component degradation temperatures then
          perform his BGA rework procedure. Once rendered he was going
          to allow the entire PWA to cooldown at a slow rate to
          minimize any potential stresses that could happen during
          rapid cooling.
               Since I'm a fabricator and don't completely understand
          the assembly/rework issues, I thought I would throw this up
          for suggestions.


          Any responses will be greatly appreciated.

          Thanks in advance.

          Dave Hoover
          Senior Process Engineer
          Hadco Corp/Tech Center Two-Watsonville, CA
          [log in to unmask]