[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: gEDA-user: RFC --- Land Pattern (Footprint) Names



> > > Also what about creating an alias file so things like 0803 can still
> > > be used and simply map over to an IPC name?
>
> > I meant to keep 0803, 1206, etc. as a legal component group name. This
> > is an error in the syntax specification (and an omission in the
> > component groups section) that I will correct. In the examples there
> > is a 0402 and a 2220 (with the optional manufacturer specifications).

> Still an aliasing method would allow all of the standard SO20 and
> things like that using the (semi)standards on
> (http://www.geda.seul.org/docs/current/symbols/node14.html) and
> convert them to the IPC names.

Aliasing is always possibly using symlinks. A configuration file could
also be used to generate aliases. A shell (or Perl) script could be used
to do a rename (or generate symlinks). Symlinks seem like the safest.

> > > If it generates/could generate the PCB lands, how hard would it be to
> > > integrate the perl script into gsch2pcb so that upon generating your
> > > pcb it would create the necessary land patterns and put them into a
> > > folder.
>
> > I am not sure that I understand your question. Generating the land
> > patterns requires a specification that defines pad size, hole size,
> > part dimension's etc none of which is contained in a schematic. Using
> > a set of rules you could generate *land pattern names*, from a
> > schematic, that could be merged into the schematic or passed to
> > gsch2pcb as a separate file.

> I meant if you had a footprint attribute on your symbol, normally
> gsch2pcb searches through the elements-dir parameters for it.  I was
> wondering about integrating this script in so that if you had an IPC
> footprint name then it could dynamically create the footprints you
> need and place them in the new pcb file.

It is not always possible to generate a footprint from the
specifications embedded in an IPC footprint name (or in the footprint
names that I am proposing). For example --- an example of a legal
footprint name (in my convention) would be
CON_USB_TYPEB__Keystone_924. There are no physical specifications in
that name. Also in IPC-7351 none of the connector names contain
physical specification data.


> > > Would this create major issues with updating pcb's if that
> > > script had a new revision of the script though?
>
> > Possibly. Using incorrect land patterns is a problem that is
> > independent of the method used to calculate them.

> I guess I was suggesting more if you had a footprint already place on
> a PCB and then the script changed the layout of that land it could
> break a PCB.  At same time such a change would represent either a
> problem in the original footprint or a bug right?

If the script changed your PCB then it could produce an error in the
PCB but I do not believe that gsch2pcb would change a footprint in
your PCB file. I believe if the name of a component's footprint
changes then gsch2pcb removes the original component from the PCB
layout and places the new component in the .pcb.new file.

(* jcl *)

On 7/8/05, James Cotton <peabody124@xxxxxxxxx> wrote:
> > > Also what about creating an alias file so things like 0803 can still
> > > be used and simply map over to an IPC name?
> >
> > I meant to keep 0803, 1206, etc. as a legal component group name. This
> > is an error in the syntax specification (and an omission in the
> > component groups section) that I will correct. In the examples there
> > is a 0402 and a 2220 (with the optional manufacturer specifications).
> 
> Still an aliasing method would allow all of the standard SO20 and
> things like that using the (semi)standards on
> (http://www.geda.seul.org/docs/current/symbols/node14.html) and
> convert them to the IPC names.
> 
> > > If it generates/could generate the PCB lands, how hard would it be to
> > > integrate the perl script into gsch2pcb so that upon generating your
> > > pcb it would create the necessary land patterns and put them into a
> > > folder.
> >
> > I am not sure that I understand your question. Generating the land
> > patterns requires a specification that defines pad size, hole size,
> > part dimension's etc none of which is contained in a schematic. Using
> > a set of rules you could generate *land pattern names*, from a
> > schematic, that could be merged into the schematic or passed to
> > gsch2pcb as a separate file.
> 
> I meant if you had a footprint attribute on your symbol, normally
> gsch2pcb searches through the elements-dir parameters for it.  I was
> wondering about integrating this script in so that if you had an IPC
> footprint name then it could dynamically create the footprints you
> need and place them in the new pcb file.
> 
> > > Would this create major issues with updating pcb's if that
> > > script had a new revision of the script though?
> >
> > Possibly. Using incorrect land patterns is a problem that is
> > independent of the method used to calculate them.
> 
> I guess I was suggesting more if you had a footprint already place on
> a PCB and then the script changed the layout of that land it could
> break a PCB.  At same time such a change would represent either a
> problem in the original footprint or a bug right?
> 
> > (* jcl *)
> >
> >
> > On 7/8/05, James Cotton <peabody124@xxxxxxxxx> wrote:
> > > That looks like really good work.  I wasn't sure exactly what the
> > > script does though.  Does it generate a PCB layout file from a name,
> > > or create the correct formated IPC strings?
> > >
> > > Also what about creating an alias file so things like 0803 can still
> > > be used and simply map over to an IPC name?
> > >
> > > If it generates/could generate the PCB lands, how hard would it be to
> > > integrate the perl script into gsch2pcb so that upon generating your
> > > pcb it would create the necessary land patterns and put them into a
> > > folder.  Would this create major issues with updating pcb's if that
> > > script had a new revision of the script though?
> > >
> > > Good work,
> > > James
> > >
> > >
> > >
> > > On 7/8/05, Xtian Xultz <xultz@xxxxxxxxxxxx> wrote:
> > > > Em Sex 08 Jul 2005 15:09, John Luciani escreveu:
> > > > >   I have placed a first draft of my land pattern naming convention at
> > > > > http://www.luciani.org
> > > > > The naming convention is based on IPC-7351.
> > > > >
> > > > > Please send questions, comments, observations either to the list or to
> > > > > (jluciani) *AT* gmail.com
> > > > > (as appropriate).
> > > > >
> > > > > (* jcl *)
> > > >
> > > > Absolutelly fabulous!!!!
> > > > I didnt know that IPC have a free document about it.
> > > > I have a doubt: would it be possible in gschem, when I draw a component (like
> > > > a resistor) to have multiple footprints associated to it, and when I place a
> > > > component and open the Atrib Editor window (I dont remeber the correct name
> > > > of this window because my gschem is in portuguese) to choose one of the
> > > > footprints of the component?
> > > > Or the best should be have one component symbol for every kind of footprint?
> > > > (thats because is hard to remeber the correct syntax for a simple resistor,
> > > > for example...)
> > > >
> > > >
> > >
> >
>