Roger,

As Bill has pointed out in a parallel response, most things need to be
"specified" or agreed to up front, before the work begins.

One thing that we usually need to remember to do when we are consulting or
contracting, is to point out that our "bids" or "quotes" are contingent on
no "change in scope" on the job.

I believe that you have a good case for saying that there was at least one
if not more major "changes in scope" of the job once you had an approved
parts placement and layout of the design to that placement, and that you are
due additional compensation for any changes after that point in time.
However, good luck in trying to get something out of them.

Maybe in this instance you would be better off to try and "push" your
"friend" into trying to follow up and get you a few additional "sheckels".
But it seems to look like you just lost out on this one.

Why does it always happen that we have to lay things out several different
times just to make everyone on a project happy. This brings up the point as
to whether or not we need to additionally specify at the time of bidding a
job just who is the "one responsible party or person" who has the power to
approve or make changes in the design.

I guess Bill does have a pretty good point in that if we follow the outlines
given in IPC 2221, and use a little common sense up front, that we can avoid
some of these problems.

Maybe it is time that we as designers and engineers here in the Designers
Council come up with a standard checklist of things that we need to clarify
and specify when we do make bids for the occasional odd consulting job, just
so that we can nip these problems in the bud, before they become problems.

Respecting the question of: "Should the specification for the Impedance to
the Board house be in stack data, or Impedance value?"

I am of the opinion that if you are going to specify impedance at all, then
you define a specific stackup which has a specific tolerance, and specify
that certain conductors must have a certain impedance. So I guess that my
answer to your question would be "both".

If you specify a certain impedance, you are going to have to pay for it, and
my opinion is that if you have to pay for it, it is the responsibility of
the board house to prove that they are giving it to you.

In other words, if you specify it, then the board house is going to charge
you for it and bill you for the NRE of having to consult a specialist to
double check your design, and insure that it meets the specific impedance
requirements.

In this case, you are paying the board house to specifically meet the
impedance requirements, and prove to you that they do, and if they and their
specialist do not like the stackup or traces widths, etc., it is their
problem to tell you and work out an alternative with you that does conform
to the end product that you are requesting and paying for. The simple fact
here is that the board house is not only going to charge you more for NRE on
a board with a "specified" impedance, but they are going to charge you more
per board also.

There is yet another option that you haven't mentioned, but which is my
favorite way to deal with the problem.

I personally prefer to dictate the exact material, and the exact stackup,
and the exact trace widths, and the overall tolerances on the stackup and
traces, and simply state that they must stay within tolerances, and never
mention the word "impedance".

Sometimes the board house will understand what's going on and ask whether or
not I am trying to make a controlled impedance board, and sometimes it will
go completely over their head and they will never know the difference, and
if they just make it to spec, it will come out OK.

I have actually gotten better results sometimes by not telling a board house
that it is a controlled impedance design, and just insisting that they
follow the fab drawing and keep within tolerance, as opposed to spending
more money and letting the board house try and build a "controlled impedance
design".

In either scenario, the one thing that you really do have to do, is to make
at least one good test coupon per panel, that you will get back from the
board house, so that you can specifically test the impedance of all of the
traces on all of the different layers by yourself, by using a network
analyzer (most companies today can at least find someone to do such a test,
even if they do  not have an engineer and network analyzer available in
house). You also want to plan to destructively test some of these coupons
yourself (by cross sectioning them, etc.), just to see if stackups and trace
widths are met.

This is not to say that the board house themselves cannot conduct such a
test, but you would be surprised at how much closer tolerances seem to be
held when there is separate "outside" testing involved.

I guess the thing that I am trying to say here is that when an engineer or
designer really understands controlled impedance and high speed design, that
he can sometimes get as good or even better a product, for less money, if he
just does everything right himself and controls the tolerances in the
fabrication processes correctly.

As a side note here, one important factor that is almost always overlooked
here, and one that should also be strictly controlled, is that of trace
thickness. Be aware that most board houses are going to use pattern plating
where they can get away with it, and that that will probably affect the
thickness of certain traces, and that this may in fact slightly affect the
impedance of certain traces. You can actually control and dictate some of
this yourself by the judicial use of some ground planes to insure "balanced
copper" (with appropriate setbacks and clearances where required).

Going back to your immediate scenario, I would say that you need to specify
both "stackup" and "impedance value" to the engineer at the company who the
design is for and let him specify it anyway he wants it on the fab drawing.
If he is going to somehow hold you "responsible" for the way it is
specified, I would specify the stackup, the impedance value, and require
testing for compliance to the impedance value, and let the company eat the
cost of insuring that they get what you specify.

JaMi

* * * * * * * * * *

----- Original Message -----
From: "Roger Gathright" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Monday, June 09, 2003 1:29 PM
Subject: [DC] When to request specifications? When not provided up front!


> Hi everyone....
>
> I was wondering at what point in the design cycle for a board layout
> is it appropriate to request specifications?
>
> AND:  Should the specification for the Impedance to the Board house
> be in stack data, or Impedance value?
>
> To clarify the reason for the question:
> A recent layout (done as a favor for a friend) with his firm resulted
> in  much rework due to several people making changes as work
> progressed.   It was a fixed Price and Should have been renegotiated
> based on the changes, I know!       I actually laid out the board
> seven times, with an additional five sessions of major rework.
>
> This board had a 14 pin equivalent density of 125 parts on 4 square
> inches.  Almost all surface mount, and was impedance controlled.
> After much indecision the placement was approved and I did the first
> of many layouts.  After the third engineer was assigned to "Shepard
> the board / circuit" into production reviewed it, and more major
> changes were made, and a faraday shield was added, with much
> rearrangement to connect it in place.
>
> I realize now, that my "Friend" was only trying to get off the hook.
> But the original questions remains!
>
> Oh yes, I am an Electronic Engineer, with experience in Emissions and
> virtually all phases of design and Fabrication and test.  I also lay
> out almost all my OWN boards, and occasionally (Rarely) for a
> customer.  For myself, I always have ALL requirements called out
> BEFORE starting any layout!
>
> Roger Gathright
> Sr. Project Engineer
> Astromation Company
>

---------------------------------------------------------------------------------
DesignerCouncil 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 DesignerCouncil.
To set a vacation stop for delivery of DesignerCouncil send: SET DesignerCouncil NOMAIL
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
---------------------------------------------------------------------------------