Re: Has anyone produced a board using Kicad?


Stuart Brorson wrote:
I see you only *considered* patching PCB. Hmmmm . . .

Ought to do your homework first ... I've sent Harry patches in the
past.

Donating general ideas is very easy, but doing implementation is
difficult. However, implementation is what counts. As an
open-source guy, I'm sure you know this.

Actually, if the only form of bug reporting acceptable is fixes, it
will be a very long time before your project is complete and stable.
I've provided demonstratable boards to Harry, DJ, etc that demonstrate
the unreasonable slowness.

As for the unfairly maligned GTK port of PCB: It was done by popular

Actually, not unfairly maligned. A clean Gtk port would not have broken
the existing Xaw usage until everybody agreed it was a stable
replacement. One of the boards I had here last year, took well over a
minute to redraw under Gtk after a simple pan, that was under a second
with Xaw ... that's not just SLOW, that's unusable. Doesn't mater how
pretty its GUI looks.

* "A patch is worth a thousand posts." Put another way: ideas are
cheap, implementation is what counts.

When you stop listening to other peoples experience, the only other
choice is to make all the same mistakes yourself along the way, and
hope you actually learn the lessons too, and not do as many less
clueful folks do, repeat the same old mistakes for ever, because thats
the way it's always been done.

There are far more clueful people in the world willing to share
experience, if not treated like clueless right off the bat. There was
no reason to jump into this discussion complaining about not supporting
gschem and gEDA ... I don't like gschem, and don't use it. one more
*** drawing UI to have to learn. The whole eEDA cludge between
gschem and PCB is painful at best. The lack of a consistant UI was one
prime complaint to DJ.

What I was talking about, are technical reasons for doing it right as
part of PCB, which has nothing to do with gschem, or the turf you where
defending by mistakenly attacking me.

In don't think it's nearly as difficult as you were complaining about,
and I've already looked. Maybe it's just because I'm not easily scared
by complexity, and do a general itterative "Keep It Simple Stupid"
(KISS) approach to tackling difficult projects until I become much more
experienced with the code. As I noted to Harry several years back ...
all the pieces are there in pcb already ... just startout by treating
schematic symbols as footprints and keep dual footprint libraries
initially (schematic and pcb) and two wire tracks for the design at
first. Then clean up the internal interfaces slowly, to include a
reasonable formal architecture. Things like crossed wires not forming a
connection, unless explictly joined. Like linked references between
schematic symbols, pin function lists, and actual foot prints based on
industry standards and vendor data. Things like automatic cross
notation between instances of the netlist (traces/rats). Be able to
pull spice data for not only the design, but the implmentation traces
as well, tied to vendor part data. Maybe not all the first year, or
even the second or third.

You end up with ONE UI, one project file, and hopefully one consistant
parts library.

Years ago, you did a schematic, then laid out the board. These days
with FPGAs, I frequently do the PCB then the schematic as nearly all
the pins are assigned based on ease of layout, not predefined as you
have with comodity parts. It becomes very useful no days, to have both
the schematic and the pcb up at the same time, and draw both at the
same time, one net at a time.

Used Sony 21" monitors are $50/ea on ebay and dual/triple/quad head is
supported in both Windows and Linux. A dual processor 1Ghz system with
4GB of ram is under $500, which makes one heck of a CAD system under
linux. My desk has three SGI GDM-5011P's on it which take VGA in. Most
peoples Best Buy or Circuit City systems cost more than I paid for the
parts on eBay. With large glass tubes not being "cool" these days,
high res sony monitors are a MUST BUY for any hardware hobbiest doing
CAD while the supply lasts. Get several and use them till they die.
I'm 54, and find using large fonts makes web surfing easier on my eyes
when I'm not busy doing another design.

I'm an avid hardware hobbiest, and most of the really dense and fun
boards I've done are for personal research. I mostly get paid for doing
contract software work and networking stuff, with hardware projects a
secondary part of my living. I like hardware, and don't turn down the
contracts when I can get them. When two pieces of a 16"x22" six layer
SMOBC panel are $185 from certain suppliers, that will hold a half
dozen projects ... doing quality PCBs is both cheap and fun. I
frequently run a homebrew club from my home, and share panel runs,
making two peices of most projects $20-60. And about double that if we
need to do stencils for both sides. I will be mfg boards as a business
later this year, with dual smt pick and place lines and N2 relow ovens.
Mostly to produce my own research boards, plus low cost hobby and
student FPGA project boards from recycled parts that I have extra. The
lines were a couple grand off eBay, and picked up initially to build my
home FPGA super computer boards -- which is another fun project I've
been working on for a few years. Several thousand FPGAs, MIPS/PPC CPUs,
memory, water cooling and a lot of power :)

The boards sent to DJ and Harry to demonstrate the Gtk slowness are all
proof of concept designs from my own research projects, some of which
I've also sold a few of. I think DJ thinks they are "interesting" too.
So when Stuart was getting off that anyone that needs more than a toy
design that can be done with the crippled student version of various
demo products, I pretty much feel he doesn't have a clue what real
hardware geeks want to do in their spare time with $50 of recycled eBay
parts :)

I'll be going back to grad school soon, and need my "home computer" for
research :)

.

Loading