Guys -
I just re-read the thread from 31 Oct 2004, and I'm trying
to make some BGA footprints using the high resolution mode.
Even when I check out pcb from CVS, I don't see any footprints
that use [ 1/100 mil ] mode. When I try, I get parse errors
on the load. Is this really supposed to work? Am I doing
something wrong?
This works:
Element(0x00 "Xilinx SF363 BGA" "" "XILINX_SF363" 100 0 0 100 0x00)
(
Pad(0 31 0 31 13 "B1" 0x00)
Pad(0 63 0 63 13 "C1" 0x00)
etc.
This does not:
Element(0x00 "Xilinx SF363 BGA" "" "XILINX_SF363" 100 0 0 100 0x00)
(
Pad[0 3150 0 3150 1350 "B1" 0x00]
Pad[0 6299 0 6299 1350 "C1" 0x00]
etc.
The above snippets of element definitions were generated by
sh QueryLibrary.sh .:/home/ldoolitt/share/pcb pcblib XILINX_SF363 XILINX_SF363
XILINX_SF363
The error message in the second case is
1: ERROR parsing file 'pcblib'
line: 12
description: 'syntax error'
And, after correcting for blank lines I snipped above, line 12 is
Pad[0 3150 0 3150 1350 "B1" 0x00]
Trying the same trick for the QFN series suggested by Dan, the query
sh QueryLibrary.sh .:/home/ldoolitt/share/pcb pcblib geda_QFN48_7 QFN48_7
QFN48_7
yields
Element(0x00 "Square Quad Flat Nolead (QFN) package" "" "QFN48_7" -157 -217 0
100 0x00)
(
Pad(-142 -108 -125 -108 11 "1" 0x100)
Pad(-142 -88 -125 -88 11 "2" 0x100)
etc. (plus a lot of comments and white space). So while that footprint
uses 1/100 mil internally, it passes coordinates to PCB proper using
1 mil resolution. That doesn't sound like what Dan claimed last month.
- Larry
