Carl is correct that the scariest part of this problem is that it may not be evident for a while. Years ago we had a problem with circuit packs were failing in and when they were returned there was usually a blown chip capacitor near the edge of the board. It took quite a while to resolve this issue but eventually it was realized that the capacitors that were blown were perpendicular to the edge of the board and the boards usually failed during or just after a severe thunderstorm or rain storm (i.e., failed during a high humidity event). The cause was flexure cracks in the capactor ends. The cracks didn't change the capacitance value very much because they were only near the termination end. When they cracked they would not short the internal layers because they would "Crack Open". However when humidity got into the crack it would cause a short between internal layers and pop goes the cap.
When the cause was discovered we modified our design guide lines of how close capacitors could be placed near an edge. More importantly we instituted a new test before any new design could be placed in the field. The new test was really a modification of an old test. We would expose a new design to 85C/85%RH for 24 hours but rather than bake the boards after the temperature humidity test we would test the boards immediately after removal from the humidity testing and sure enough if there was a cracked capacitor it would pop when powered up because there was humidity in the crack causing a short.
George (Retired FMA Engineer)
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Carl Van Wormer
Sent: Wednesday, August 15, 2018 2:46 PM
To: [log in to unmask]
Subject: Re: [TN] EXTERNAL: Re: [TN] MLCC spacing/orientation to V-score or Perforated Tabs
The scariest part of this problem (caps cracked from depaneling forces) is that the problem may not be evident for a while. A production run of 100 boards passed all our outgoing tests, even though the 12V power input filter (ferrite bead and ceramic cap, both 0603) were near the board edge. The cap orientation was perpendicular to the board edge, with one end spaced 0.10" from the edge. We had 6 units come back as field failures within the first 3 months of shipping. The ferrite bead burned the board (and damaged the pads) after the cap shorted to ground in the customer's environment.
We did a quick re-design of the board, rotating the cap and moving it to about 0.23" from the board edge. After several hundred of the "fixed" units were shipped, we've seen no failures of the newer layout boards.
Later,
Carl
Carl B. Van Wormer, P.E., AE7GD
Senior Hardware Engineer
Cipher Engineering LLC
21195 NW Evergreen Pkwy Ste 209
Hillsboro, OR 97124-7167
503-617-7447x303
[log in to unmask] http://cipherengineering.com
-----Original Message-----
From: TechNet <[log in to unmask]> On Behalf Of Drew meyer
Sent: Wednesday, August 15, 2018 12:36 PM
To: [log in to unmask]
Subject: Re: [TN] EXTERNAL: Re: [TN] MLCC spacing/orientation to V-score or Perforated Tabs
Wayne,
We have used the pizza cutter style as well and it can induce some strain on the board, more than you might think possible. The depth of the V-score can vary and the wheels may not track exactly down the middle of the V-score. In one case the V-score depth varied enough that the wheel walked out of the groove and into the PCB. Scrap that panel!!!!
A well maintained router is the best method. But as you can see above, even here it is not applied to every product.
Drew
-----Original Message-----
From: TechNet [mailto:[log in to unmask]] On Behalf Of Wayne Thayer
Sent: Wednesday, August 15, 2018 10:20 AM
To: [log in to unmask]
Subject: Re: [TN] EXTERNAL: Re: [TN] MLCC spacing/orientation to V-score or Perforated Tabs
Thanks Scott.
I'll see what I can divine from CALCE.
I have a hard time imagining the "stress lines" for the V-score (aka "Pizza
Wheel") singulation process. With tabs, bending is clearly the way shock waves propagate, so you'd think aligning parts with the long access parallel to the edge would be the orientation for minimal differential stress. But the "Pizza Wheel" is trying to rip the board apart in a very different way. I'm suspecting the stress waves would be of similar shape to the waves around the bow of a canoe when you're moving through the water:
They start out as propagating in the same direction as the bending waves, but end up turning nearly 90 degrees as the wheel wedges the board apart.
Probably CALCE has modeled that.
Wayne Thayer
On Wed, Aug 15, 2018 at 7:47 AM Decker, Scott UTAS < [log in to unmask]> wrote:
> Wayne,
> We have requirements here also regarding the location of MLCC's
> near the edges and areas of stress, like mounting holes, mouse bites,
> scoring, etc. and the electrical engineers are urged to use flexy
> leaded type caps when possible. Like Steve mentioned, .200" from a lot
> of things mentioned is a good start along with orientation of the
> components along the stress lines, etc. We also have restrictions on
> the soldering of boards with MLCC's on them as far as temp rise rate
> to avoid thermal shock. This is also related to the size and soldering type used, wave or re-flow.
> Something that might help also is to check into The Center For
> Advanced Life Cycle Engineering which is a research center at the
> University of Maryland. They have a calculator that you can use to
> help predict cracking issues with the parts. I can't share exact
> numbers and other related information without congressional approval
> from people I don't even know, but I will say that the .200" number is
> pretty good and Steve said the same thing. These parts have really
> been a thorn in the side for designs with always having to remember which way, and how far, etc. but it is what it
> is... :-/ Good luck.
> Later...
>
> Scott Decker – Staff Engineer, PCB Design Services CID+ – Electronic
> Systems Center UTC AEROSPACE SYSTEMS
> 3445 S. 5th Street, Suite 170, Phoenix, AZ 85040 U.S.A.
> Tel: 602 308 5957 FAX: 602 243 2347
> KE7MWT AKA:PadMasterson
> [log in to unmask] www.utcaerospacesystems.com
|
