Hey Groovy,
It's good to hear about your #1. I've been waiting for some
mitigating news about number up. Just to elaborate... We saw a 12-up,
18Fx24W, .005 core, 3oz/3oz, shrink 40 mils [!!] in the 24 inch
(grain) direction - and both sides were ground planes. By contrast, a
one-up backplane with at least one solid ground plane won't shrink at
all, i.e., it's structurally reinforced by the copper, which of course
wants to grow with heat during lam, not shrink. Half a mil/inch is a
good starting point for a .005" 1sig/1sig, but in my experience, the
numbers become unpredictable by laminate manufacturer for .005 core
and below. Another small consideration: normally the warp/fill ratio
is around 1.5-1.8 to 1; but watch out for builds that use exclusively
2113 pre-preg. It is the only weave that has twice the thread count
in the fill direction as in the warp (all the others are balanced),
thus more resin volume "soaked" into the fibers in the fill; the net
result is roughly 1:1 shrinkage ratio.
More discussion is needed on this subject - any other hypotheses out
there?
Joe Felts
PC World, Toronto
______________________________ Reply Separator _________________________________
Subject: Re: FAB: Film compensation
Author: [log in to unmask] at INET
Date: 8/5/96 5:37 PM
Lou,
How important is compensation. Well.....That all depends
on the technology of the product you're building. If you have pad
sizes that are >.014" over the drilled hole size then it may not
be so critical. If your trying to get to drilled hole size +.006".
Well....Then it's vital.
I'll add two more variables to your list.
#1 (The most important one) Quantity up on a panel. I'm seeing
more and more smaller parts. (I think due to die shrink stuff.)
The multiple up plane layers <at some point> start behaving like
signal layers due to copper density.
#2 Core thickness. (Remember the same prepreg we use to relam is
what is used to make the cores. 1060 shrinks a heck of a lot more
than 7628.)
Your other comments/observations are correct. The other variables
all have valid influence/impact on dimensional movement within a
multilayer package. The object is "how do we accurately predict"
or "model" all these variables? I started with a scaling matrix of
5 and now have 10 and still haven't got it where I feel completely
comfortable. I think that number may double shortly <for me anyways>.
Good Luck!
(Always in search of registration.)
Groovy
______________________________ Reply Separator _________________________________
Subject: FAB: Film compensation
Author: [log in to unmask] at SMTPLINK-HADCO
Date: 8/5/96 3:38 PM
Here's a question for TechNet that may be vague and imprecise, but I
will ask it any way. It has to do with film compensation for inner layers.
Our plant compensates inner layer films according to simple rules: add
0.5 mils/inch and 0.3 mils/inch for ground planes in the grain and
non-grain directions, respectively; add 0.6 mils/inch and 0.4 mils/inch
for signal planes in the grain and non-grain directions.
The engineering
team believes that these rules, made 4 or more years ago, are too simple.
Some factors that, we believe, should affect compensation are
- layup pattern, location of ground and signal planes in the board stack
- weight of copper (could be many variations here)
- number of layers
- trace density
- orientation of traces with respect to grain
- prepreg styles used (many variations seem possible here)
- supplier of prepreg
- type of material (polyimide, epoxy formulation).
Anyone have any ideas of the relative importance of these factors?
Anything we have overlooked?
How important is it to be concerned about compensation?
Lou Hart Compunetics
***************************************************************************
* TechNet mail list is provided as a service by IPC using SmartList v3.05 *
***************************************************************************
* To unsubscribe from this list at any time, send a message to: *
* [log in to unmask] with <subject: unsubscribe> and no text. *
***************************************************************************
***************************************************************************
* TechNet mail list is provided as a service by IPC using SmartList v3.05 *
***************************************************************************
* To unsubscribe from this list at any time, send a message to: *
* [log in to unmask] with <subject: unsubscribe> and no text. *
***************************************************************************
***************************************************************************
* TechNet mail list is provided as a service by IPC using SmartList v3.05 *
***************************************************************************
* To unsubscribe from this list at any time, send a message to: *
* [log in to unmask] with <subject: unsubscribe> and no text. *
***************************************************************************
|
