Vinit and the group -
I have seen this before -
Apologies up front for the long post -
I am assuming that we are talking about the last four photos at the
webpage stevezeva.homestead.com, which are labeled Bad Board, Bad Board
2, Bad Board 3, and Bad Board 4.
If so then here is my take as relates to the "raised vias" shown in Bad
Board 2 and Bad Board 4, and this may be the key to understanding the
related problems.
About 12 to 13 years ago, I came on board at Datatape., in Pasadena
California, as a Sr. Designer. Although my work background had been
primarily as a PCB Designer working my way up to become a Sr. Design
Engineer for TRW, at Datatape, I was hired into the Electrical
Engineering Department as a troubleshooter primarily because while not
degreed, I have a broad background in Electronics, and hence because of
by cross disciplines I could communicate not only to Engineering, but
also to folks in Design and Drafting , as well as those in
Manufacturing.
My first crisis was an overdue Military Contract VME Style Backplane for
16 Cards with 3 96 Pin Connectors each, that was 11 layers at .238"
thick due to impedance requirements for ECL Logic.
Yes, you read correctly, almost ¼ inch.
Well the first batch from the first PCB supplier in Monrovia California,
ended up with massive shorts of almost all of the holes, to almost all
of the supply lines and ground. To make a long story short, this was due
to the fab house using the same feeds and speeds and drill changes on
that board as they would have on a 1/8 inch thick board, which, needless
to say dulled the drills immediately and ended up snagging most if not
all of the internal pads and ripping them out, which in turn caused all
of the material in each hole to be ripped out also, leaving all (or
most) of the holes with a clean entry and exit (due to aluminum backing
sheets), but a large caveran inside each hole which exposed all of the
edges of the clearance holes in the 5 power and ground planes, and
plated everything into one massive short.
Well I finally convinced the fab shop as to exactly what they had done
wrong and then had to convince them that they really could make the
boards properly after all, and I had them contact their Excellon Rep
(for help with drilling feeds and speeds and drill life for a ¼ inch
board), and made some minor adjustments to hole sizes and allowed them
to remove unused pads from the internal layers. Then I just sat back and
waited for the next turn of the board. I new that we had failed when the
shop foreman failed to call me back after an ohmmeter check after final
etch, and when I called them I was told he had gone down the street
cursing and screaming to a local bar. This time the holes were
beautiful, but the layer to layer registration was off since they didn't
use pins and just threw the whole large stack up into the laminating
press with only a few "hot glue" spot bonds to hold everything together.
That's like an In 'n Out Double Double Hambuger with extra mayo, extra
cheese, extra tomato, and extra onions, when you try to mash it down to
get in into the wrapper, everything shifts!
This left me to find someone who could make the boards properly and
quickly. I went to 2 different shops with the same order, one shop in
Huntington Beach California for 3 boards 1 week @ $13K each, and the
other shop in Florida for 3 boards 1 week @ $1K each. Major difference
in price, but Florida was a backup anyway, which I had located after we
contracted with the other guy, and we needed the boards done right three
weeks ago by then, so that was a small price for insurance.
The supplier in Huntington Beach, who used an autoclave, delivered
perfect boards on time and went out and bought a new car off of that job
(or so he told me). However the supplier in Florida delivered 3 boards
that had all of the holes (small holes) and vias in the board raised and
looking exactly like the photos on the website, labeled "raised vias".
This is what I found:
The first thing that I noticed was that a surprising number of the holes
had some peculiar "black flakes" in them which you could see when
looking into the hole with a lupe. While I did not have any fancy "bore
scope" or other tool other than a lupe, to examine the hole with, I came
to the conclusion that these "flakes" were actually protruding from (or
stuck into) the wall of the PTH. It almost looked like there was a 2 mil
thick layer of graphite or carbon that was in the middle of the board
that hat chipped in some holes and left a flake protruding from the PTH
wall.
My initial conclusion was that the boards were bad and I wanted to
reject them, but I had to either find a reason to reject them or eat the
$3K, which by then was a small piece of change anyway.
I used an ohmmeter to check a few of the worst looking holes, but they
showed continuity. Nonetheless, I knew they were bad. I called the
Company who had done the Mil Continuity Certification, which was
different than the supplier, and he assured me that the boards were
checked and all passed, but that no, he did not remember any "raised
vias" as I had described them to him.
That description, by the way was essentially this: When you held the
board up to a light source so that you could see the light reflected off
of the surface of the board, you could see that board surface
surrounding virtually each and every via was raised above the normal
flat surface of the board, and that the groups of 96 holes at each
connector location were also raised. It looked like a little island
sticking up out of a calm sea, and while I never measured the height of
the rise, it was visually very noticeable, similar to those shown in
photo Bare Board 2.
Well, I didn't know what to do to prove that the boards were bad, since
I didn't want to pay for them I couldn't very well section one of the
boards and then try to send it back, and the test coupons had not been
properly masked so the top and bottom layers as well as all plating from
the holes had been etched away, and then they were separated from the
boards and so they did not exhibit any of the problems.
I finally got some aluminum foil, Reynolds Wrap if I remember, and made
a large pad that I could hook to one of the test leads of the ohmmeter,
and use as a massive "contact" to contact all of the bottom pads on the
board in an area that I wanted to check. I then set about checking the
continuity of each and every hole of each connector (over 4600 holes),
as well as all of the vias (several hundred additional holes as vias). I
checked for continuity from the pad (annular ring) on the top side of
the hole to the pad on the bottom side of the hole, of each and every
hole. I wasn't doing a net list check, all I was doing was checking each
hole for top to bottom continuity.
I was about to give up when I found my first hole with an "open", and
when I checked the hole with the lupe, I expected to see flakes, but saw
instead a beautiful hole, and couldn't even tell visually that there was
a break in the wall since it looked so smooth.
I ultimately found that that board had a total of 8 holes that exhibited
an open from top to bottom. 4 of the holes were unused electrically
speaking, but the remaining 4 were active holes that were supposed to go
to other places, of which all were connected to the proper place by at
least one of the pads (either top or bottom (but not both)), but at
least one of the holes did occur in the middle of a net connection which
did in fact cause the net to be broken in the middle. This was enough to
reject the board for failure to exhibit the same continuity as the other
2 boards that it was checked against for the Mil Certification.
I started asking questions of the Florida supplier and I finally came up
with what I believed to be the answer, and when I explained it to the
guys in Florida, they concurred with my evaluation, and although they
said they themselves would section the boards, they never got back to me
with any contradictory findings.
The supplier in Florida took every precaution and was very careful every
step of the way, and in fact had produced perfect boards (with the
exception of the Test Coupons), right up to the very last step, which
was the only uncontrolled step in their entire process.
After they had gone down the street to get the boards Mil Certified for
Continuity at another vendor, they gave them to their "touch up girl" to
do a very small amount of touch up of scratches and voids in the solder
mask before shipping. When she was finished with her touch up, she put
the boards into her uncontrolled oven which she only used to cure the
"touch up jobs".
Well, my take on it was that the oven was hotter than it should have
been, and that the boards expanded very rapidly, and also expanded by a
large amount, such that every single small hole in the board had its
plated thru wall stretched so far that many of the walls actually broke
apart, some with a clean break, and some with a jagged break, and some,
probably most, just stretched.
Next in the process, the boards were taken out of the oven and simply
allowed to cool by themselves in a rack. This is when the stretched and
broken hole and via walls now stood up in protest, and simply would not
return to their original dimensions. I believe that a major part of this
was due to the fact that when the holes with jagged breaks in their
walls tried to contract, the 2 halves of the barrel just mashed into
each other, which, among other things produced the little "black flakes"
that I saw in many of the holes, which were actually broken pieces of
the plated wall of the hole. The reason that most of them looked "black"
is that I was seeing the outer side of the hole wall.
As it turned out, the 8 discontinuities that I found in the one board
that I tested, all exhibited exceptionally clean walls in the hole, and
I concluded that these 8 holes broke so cleanly that when the board
material in adjacent holes around them failed to return to the original
dimension, these 8 hole walls simply did not re-engage their
counterparts, and hence showed up as an open to the ohmmeter.
While I am sure that there are some additional factors which lead to
this catastrophic damage to the board besides the massive thermal shock
and massive expansion of the board, I could only put my finger on one in
particular that I believe contributed to the problem. I believe that a
primary contributor was the "reduced reproducibility" ratio of the .238"
board thickness to the .039" dia hole used for the connectors, which I
am sure produced a very thin wall (barrel) in the middle of the hole,
and while I know that 039" is not a very small hole, it is comparable to
a .010" hole in a .060 board of today.
I believe that my experience as described above will account for the
"raised vias" in the pictures on the website, which I believe were
caused by thermal expansion of the PC Board at some point in the final
manufacturing process, where the holes and vias were "stretched" so far
out of shape (dimensionally) that they refused to return to their
original "pre-expansion dimension".
I would stress that I do not believe that the hole walls would have to
be stretched to the point of breakage as did happen in the case I
described above, so there may be no evidence of the "flakes" that I saw
in the holes of the above example. I believe that the holes could
stretch without breaking, especially if there was thicker plating than
experienced in my example above, and still refuse to return to their
original dimension, and therefore leave an otherwise perfect looking
wall behind with no signs of the thermal expansion other than being
raised.
The one thing that I saw as a big problem in the whole episode as
described above that has scared the heck out of me ever since, and which
is why I will not allow a test coupon to be separated prior to delivery
of the board to me, is the fact that most of this type of damage occurs
to the board after all of its primary inspection, you know, the
continuity check, the QA inspection, and the visual inspection, since
touch up is usually the last thing that happens before it gets sent out
the door, and usually it is not even looked at by the person who did the
touch up once the board dried, which explains how an otherwise good
supplier can ship such a #4(&!\%* product. (pardon my ascii).
I am very confident that this explains the "raised vias", and I am sure
that when this scenario is taken into account, it will go a long way
towards explaining the remaining problems with the boards.
JaMi Smith
Optical Crossing Inc.
-----Original Message-----
From: [log in to unmask] [mailto:[log in to unmask]]
Sent: Friday, March 29, 2002 8:19 AM
To: [log in to unmask]
Subject: Re: [TN] Blistering in PCBs
Don and Everybody!
The pic's are now up! Go to: http://www.stevezeva.homestead.com
I gotta say, those are some UGLY boards!
-Steve Gregory-
Hey Earl
Where can I see these photos? Don't see them on Steve's site.
Don Vischulis
-----Original Message-----
From: TechNet [mailto:[log in to unmask]]On Behalf Of Earl Moon
Sent: Friday, March 29, 2002 6:49 AM
To: [log in to unmask]
Subject: Re: [TN] Blistering in PCBs
Vinit,
Ok, here we go. Never seen nothin' like these before. Oh well, maybe
once
when I was a "young gun" process engineer at Xerox. That's another story
I'll not bore anyone with now.
Concerning the pics, it is obvious there is delam everywhere but maybe
the
edges. Vinitdelam3's discoloration also is delamination, not solder mask
related.
Someone else take shots at this and Vinit's supplier. This shop
shoundn't
exist unless they just lost it all on his boards ONE PING ONLY PLEASE.
A pleasure to serve Vinit. Don't often get to see how bad something can
be
besides a train wreck.
MoonMan
---------------------------------------------------------------------------------
Technet Mail List provided as a free service by IPC using LISTSERV 1.8d
To unsubscribe, send a message to [log in to unmask] with following text in
the BODY (NOT the subject field): SIGNOFF Technet
To temporarily halt delivery of Technet send e-mail to [log in to unmask]: SET Technet NOMAIL
To receive ONE mailing per day of all the posts: send e-mail to [log in to unmask]: SET Technet Digest
Search previous postings at: www.ipc.org > On-Line Resources & Databases > E-mail Archives
Please visit IPC web site http://www.ipc.org/html/forum.htm for additional
information, or contact Keach Sasamori at [log in to unmask] or 847-509-9700 ext.5315
---------------------------------------------------------------------------------
|
