Mitch,
From your post, I'm assuming you're at a OEM with your own products,
and I think the very best place to start is with your design engineers.
Try and make sure that someone is completely "up to snuff" when it comes
to the "latest requirements" for good board layout utilizing Fine Pitch
devices.
I don't know what kind of equipment you have, but I'll assume that you
have vision capability seeing how you're talking 20-mil...
Here's a couple of things I always see get missed on prototype boards we
get from our customers:
* Fiducials-Absolute necessity!!! Global marks for the whole board, AND
local marks for the fine pitch. If ya' got components on the bottom, even
if it IS only passives, put some there too! A lot of times I've seen that
the designer remembers about the fiducials on the top-side because that's
where the QFP's are, but then forget about the bottom-side. A resistor
that comes off in the wave because the epoxy dispensing was inaccurate
due to no fiducials will keep a board from working just as surely as an
open joint would on a QFP.
* Component Footprint and Pad Geometry- Screw-ups happen here more than
you might think it would. It happens for various reasons...the part
libraries that the design engineers utilize when laying out a board may
sometimes be out of date, or, I've seen boards that did have 20-mil
center to center pad layout, but the QFP's pitch was actually 19.7-mils..
it looks like placement is okay at first glance, but as you look closely
at a row of leads on one side of the part, the lead at one end will be
dead perfect, but as you look down the row, each lead gets further and
further off the pad until it's unacceptable at the end of the row.
* Mask- Use Liquid Photo Imagable (LPI) with "solder-dams" between the
Pads!!! Do anything, and I mean anything (promise your first born,
sacrifice a virgin to the solder god, etc.) to stop someone from spec'ing
dry film mask for a board with 20-mil fine pitch...IT WILL GUARANTEE YOU
SOLDER BRIDGES!
* Stencils- Personally, I use only single level stencils. Now I know that
there are those out there that will say; "Whut's wrong with stepped
stencils?" for one thing, the extra squeegee pressure needed to clear the
the step-down, may cause scavenging on the locations that have large
aperture openings, and maybe it's just me...I still wonder how they can
control the etching of the step-down so precisely? Within 1 or 2 mils?
Boy, that's really something if they can do that each and every single
time! For a board with 20-mil fine pitch on it, I usually spec' a 6-mil
single-level stencil, micro-mod the 20-mil pads to 10-mils wide and allow
growth on everything else. I then get the whole thing electro-polished..
works like a charm!...by the way, I don't know what kind of printer you
have, but a slow, controlled separation of the PCB from the stencil after
the print stroke (make your stroke speed nice and slow too) will give you
a beautiful 20-mil print...nice straight, tall sidewalls.
I agree with the things Jim Maguire said in his post below, but have a
few things to add to that. We've experienced the same problems with
co-planarity issues that he spoke of, however, in my experience the
problems with lead co-planarity come from all the people handling and
counting the darn parts (not being aware how easily they're damaged) than
from anything happening on the line...that's one of the reasons I can't
wait for BGA to really take off so we won't have to deal with those
flimsy leaded things anymore!!...I know, I know, that won't be for a
little while yet!
The other thing Jim brought-up was about the SMD solder joints becoming
liquidus during wave. I had made a post a few weeks ago looking for some
help with precisely that problem...I got a little help, but still haven't
got it nailed yet. Part of the problem I know is because the QFP's pads
have a via almost right at the very end of the pad, heat transfer from
the pot, thru the via, up to the pads will do it everytime! For the time
being, we're using a liquid mask to cover the via's before wave..it works
okay, but we need to do something better...just be glad you don't have to
deal with this "Board-from-Heck" that we've been dealing with! It's 14.5"
X 17", 16-layers, almost .125" thick, with 20-mil AND "J-leaded" parts on
BOTH sides, as well as having the PTH coming into the board from BOTH
sides too!!
Sorry this was so long, but I wanted to do a "brain dump" and try to give
as many of my "lesson's learned" as I could....
If you would like anymore information, email me back, or give me a call.
///// ///// ///// ///// ///// ///// Steve Gregory
| @ @ | @ @ | @ @ | @ @ | @ @ | @ @ Pragmatech Inc.
C _) C _) C _) C _) C _) C _) 101 Nicholson Lane
\ O \ O \ O \ O \ O \ O San Jose, California 95134
/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ Phone:(408) 943-1151
Fax:(408) 943-1461
email:[log in to unmask]
*************************************************************************
Our major problems were in the areas of controlling part placement and lead
misalignment to prevent opens during assembly. This was mostly equipment and
training related and essentially had to do with getting people used to the
fragility of these leads when placing the devices (we autoplaced but VERY SMALL
forces can move leads out of alignment (vertically or horizontally).
As we conquered this, we also had some problems on "mixed technology" PWAs with
fine pitch on top and also wave soldered (PTH) devices. Problem was that the
wave solder schedule was "too hot" and tended to cause localized reflow/voiding
under the fine pitch device leads. These joints would look OK but have very
little mechanical strength. We had to lower wave solder preheats/dwells to
overcome this. IBM has had similar issues (I believe this was reported in a
recent IPC Tech Review article.
Jim Maguire
Boeing Defense & Space Group
[log in to unmask]
______________________________ Reply Separator _________________________________
Subject: transition to fine pitch
Author: [log in to unmask] at esdigate
Date: 7/5/95 8:56 AM
Hello,
Our design requirements are driving us to finer pitch technologies. We
currently assemble 50mil pitch devices. My expectation is that our next
generation of products will be 20 to 25 mil pitch. I am interested in hearing
war stories of people who have been through this process. Specifically, what
was the number one problem encountered and how was it overcome?
Thanks
Mitch Austin
Itron
[log in to unmask]
|
