First of all, none of what I have written was intended as a personal 
attack, you stupid idiots!

***JOKE!!!*** No one writing here is stupid, though sometimes some of us 
could use a little coffee.

At 05:41 PM 7/9/01 -0400, Michael Reagan wrote:
>Abdul and Brad,
>If your design is critical to   a particular thickness, then yes you must
>specify it. no doubt, however if it aint, specify the min acceptable and
>give your design house the latitude to build up.

Brad's point, and mine, is that one may not think the thickness critical, 
but thickness *will* vary the high-speed performance of the board (and a 
few other things which are not so likely to cause trouble). If the board is 
all low frequency analog, no problem. But if the board has even one or two 
digital nets with fast transitions, thickness variations will cause 
differing amounts of ringing, crosstalk, and radiated noise.

So a board that works from one vendor might not work with another. It would 
be a shame to discover that after one has tested a board with prototypes 
from one vendor and maybe a small production run, and then, for a large 
batch where the economics are forceful, has bought the boards elsewhere.

*I have seen this happen.*

Now, I have actually followed Mr. Reagan's described practice. The vast 
majority of my designs, over the years, have not specified many details of 
board fabrication but have instead relied upon the fabricator's standard 
production characteristics. So I have not followed the advice I am now 
giving. I should.

>    Funny we are all on this
>subject, I plan to submit a paper to PCD magazine, about board specification
>and fab notes later this month or early next month.

Therefore it is a great time to review the subject and to consider new ideas.

In another post, Mr. Reagan wrote:

>to clarify  IPC states"
>  Class C is a fully documented procurement package. Documentation is to the
>extent that the information is self sufficient and be sent to multiple
>vendors, with each producing the identical product. This documentation
>package requires that all the full manufacturing allowances are disclosed
>and documented.   This is an IPC spec and a very important one.   I write
>specifications all the time, I did aprox 80 designs last year ranging in
>various sizes and speeds to 2.7 Ghz. ( fast enough)  Any specification
>should be written with the minimum requirements without tying your vendors

IPC 2221 notes that some laminate or prepreg characteristics 
*shall* not be included on the master drawing. I do not specify the 
thickness of the material, but of the finished product. How the fabricator 
gets there is his business; but if the board is a multilayer board, the 
finished internal dimensions are normally a critical attribute. Normally, 
I've seen layer distances specified as nominal value without stated 
tolerances, unless the board is controlled impedance, in which case 
impedance may be specified, with tolerance, rather than dielectric gap. 
Again, the fabricator can figure out how to get there.

The advantage of specifying impedance is that it can be measured without 
slicing up the board.

So, unless it is critical, fully specifying dielectric thickness can be 
omitted, but the nominal values should *not* be omitted, because it would 
then be possible for a fabricator to wildly vary the thickness, making it 
more likely that a significant change will take place in board performance.

I don't have a copy of D-325, which is the PWB documentation standard, but 
I would read the certification guide's restatement of it on this point as 
requiring nominal dimensions on the master drawing for Class B 
documentation (5.1, p. 132). Class C documentation, essentially, is fully 
defined including tolerances, Class B is less formal. Class A is "usually 
used for internal use."

What is the difference, by the way, between "usually used for internal 
use," "usually for internal use," or "normally for internal use," the 
latter two being good writing? But for the IPC publications, this is not 
bad, at least it is reasonably clear!

Mike, why not submit your article to a few designers for review before it 
is finalized for publication? Many heads, almost always, think better than 
one; you'd still be the organiser and writer!

The best list for discussion of this topic is probably the Designer's 
Council list.... Most Protel people are electronics engineers less involved 
with formal specifications and more likely to do a Class A-documented 
design. There are exceptions, of course.

Abdulrahman Lomax
P.O. Box 690
El Verano, CA 95433

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
* To leave this list visit:
*                      - or email -
* mailto:[EMAIL PROTECTED]?body=leave%20proteledaforum
* Contact the list manager:
* Browse or Search previous postings:
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to