On Fri, 27 Apr 2001 20:07:30 -0700, Jim McGrath wrote:

>> I hope it isn't that simple - testers, particularly roving probe testers
>> only test continuity from each end of a 'net' + any stubs. For testing
>> non-continuity (shorts) they don't bother looking for shorts between nets
>> which don't get close to each other. You need more than a netlist and pad
>> locations to figure out net ends, stubs, and net proximity
>
>This is confusing me. If they dont get close to each other how is it a short?

Sorry,  do we talk about testers testing to make sure you have what you
want or to make sure you don't have what you don't want? 

Roving probe testers should not bother checking net to net shorts for nets
that are not physically close. If they tested every net against every other
the test time grows exponentially with the number of nets. 

>> Much needed?
>
>Yes it is Much needed.
>>
>> While Gerber is not an ideal format for generation of connectivity
>> information at least it is a standard and one that PCB houses are going to
>> get from all customers.

>That is a MAJOR  problem. This File Needs to come fron the database
>so it verifys the gerbers. I have been almost bitten and the D-356 saved me.

But it doesn't verify the gerbers until the PCB is made and on the tester -
too late. Also you are responsible for making sure the Gerbers and D-356
files are correct (apart from screw ups the board house makes plotting
which are maybe their problem).

Why should it be easier to make a correct bug free D-356 generator than a
correct bug-free Gerber generator?

>> If you choose to supply the board house with additional information then it
>> is your problem if it is wrong - and you are going to be less familiar than
>> the board house is with the conversion tool and less able to 'tune' it to
>> suit their tester
>
>Not if it is pulled directly from the database. The Gerbers can be wrong for
>many reasons.

And why can't D-356 files be wrong? Gerbers come from the database too. 

>> I think it would be more useful to be able to check a conectivity netlist
>> against a Protel design, the PCB house gives you the netlist generated from
>> Gerbers the way they think they plotted them. That would check the
>> connectivity netlist and Gerbers and check them before they have made a
>> batch of possibly bad PCBs.

>Yes this would work but... I have TOO much work as it is and I'm not getting
>paid to verify the Board houses work. I can Prove the IPC file does this and puts
>the responsiblity in the board houses hands where it belongs.

Are getting paid to produce D-356 files for the board house doing their
work for them? 

You can't *prove* anything. If the boards fail on test it could be the
D-356 file - your fault, or the gerbers which could be your or their fault,
or manufacturing problems which are likely their fault. Or the Gerbers and
D-356 could be similarly wrong which wont fail test at all and be your
fault. 

It is more likely that tools from the same vendor working with the same
database will screw up in the same way. I would not have much faith in a
conversion tool kludged from a Specctra export for example (which requires
Protel not to have screwed up the export also). 

A netlist from a third party tool analysing a gerber (particularly if it is
part of the CAM package the board house is using to plot) is a better cross
check. A netlist compare (fed back to the CAD package or external from an
additional netlist from the CAD package) will show problems before you make
and waste boards. 

I don't know much about D-356 format, but generating an efficient and
accurate netlist for a board tester (particularly roving probe testers) is
not trivial. The web reference in this thread mentioned something about
indicating 'via' (or something) pads which mean they are not the end of
nets or stubs and do not need to be continuity checked - that is not
trivial to work out. The logic required to decide on net proximity and the
lack of need for open testing between particular nets is more complicated. 

It is a lot simpler to check a test netlist matches a design database or
another netlist  than it is to generate a good test netlist in the first
place. 

That is why I suggest the board house ought to be doing the work from
Gerbers with a tool they understand and matches their plotting and testers.
If you want to feed a netlist in what ever format to or from the board
house for cross checking before PCBs are manufactured that is cool.

I know personally the guy at the board house I use who sets up the testers
and scans gerbers for test netlist generation - In the past he has moaned
to me about bugs and slowness of the generation software but he never asked
even is passing if I could supply a D-356 netlist. 

I can see the attraction for board houses asking - they do less work, if
the netlist is wrong it is your fault and assuming you are paying for test
jigs and/or tester time they don't give a damn how inefficient your
netlists are? 

I would go along with a D-356 export or DRC check for Protel but only for
use as a cross check not for actual board testing (well not till P2002SE
+SP?  anyway). 

Cheers, Terry.


* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
* To post a message: mailto:[EMAIL PROTECTED]
*
* To join or leave this list visit:
* http://www.techservinc.com/protelusers/subscrib.html
*                      - or email -
* mailto:[EMAIL PROTECTED]?body=leave%20proteledaforum
*
* Contact the list manager:
* mailto:[EMAIL PROTECTED]
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Reply via email to