[Cad] [Open Manufacturing] Which CAD Kernel Should the FOSS and OSHW Communities Focus On?

[Joshua Newth]

My two cents, as a mechanical designer, electrical designer, and programmer:

>>Board-level circuits are *absolutely* designed via declarative languages
often

Like Matthew, I have only ever seen and done board design and layout by
hand in a GUI (altium, orcad, eagle - the list goes on). I think this is
because it is very hard to state a complete enough set of constraints to
fully define a board of any meaningful complexity and because board layout
is a high dimension, highly coupled problem. For highly coupled problems it
seems to me that the "state of the heart" is a high-bandwidth way of
experiencing the design (GUI) coupled with an awesome parallel processor
(your brain). And certainly the market has already spoken on this point.

That said: The rules processing for verifying the board meets manufacturing
constraints is wonderful, because rather than using rules to generate a
design it uses rules to assist the person laying out the board. Electrical
CAD packages are very good at this. As an appeal to your intuition,
consider: Verifying a solution to a NP-hard problem is still in P.

I wonder if this similar approach could be successfully applied to
solidmodeling programs, that is, skip the declarative stuff for design
creation, but use declared rules for verifying the part or assembly (fits,
clearance, mold flow thickness, t-stacks, etc).

I also think that scripting would be a poor replacement for a good GUI for
3D CAD, but I do think that a bit of scripting to automate certain types of
feature creation is a great idea. I think of linear/circular patterning and
transforms as a very, very simple version of scripting for solidmodel cad.
For an interesting commercial example of this approach, see Rhino +
Grasshopper.

Sorry I bailed on the last two meetups. I've read the papers though! Great
stuff so far. Looking forward to Sunday.

 -Josh

PS In my own work, Ive stumbled on a practical area for improvement in
current CAD tools: Designing things that are fabricated in 2D (lasercut
parts, stampings, packaging design) but being able to visualize and
manipulate in 3D at the same time. I've been using Onshape to design
packaging for my startup and my workflow is "design the package as discrete
panels, then lay them flat and move them around to get the 2D pattern".
Painful.









On Wed, Nov 11, 2015 at 12:24 PM, Matthew Keeter <matt.j.keeter at gmail.com>
wrote:

> Do you have any examples of declarative PCB design in industry?
>
> I’m curious about what software tools folks are using for this kind of
> design;
> in my experience, most hardware companies are using Altium or similar.
>
> -Matt
>
>
> On Nov 11, 2015, at 3:09 PM, Ilia Lebedev <ilebedev at mit.edu> wrote:
>
> Setting aside the quality of opencores, which is mainly an FPGA circuit
> repository, Board-level circuits are *absolutely* designed via declarative
> languages often in order to remain maintainable. There is a wealth of
> graphical user interfaces, and one-off boards make great use of them, but
> this is by no means a summary of the entire field of PCB design :|
>
> On Wed, Nov 11, 2015 at 3:06 PM Bryan Bishop <kanzure at gmail.com> wrote:
>
>> On Wed, Nov 11, 2015 at 12:43 PM, Nancy Ouyang <nancy.ouyang at gmail.com>
>> wrote:
>>
>>> Very few people seem to argue for text-based circuit-board design, and I
>>> only know of one person who designs circuits that way.
>>>
>>
>> There's millions of lines of source code written in Verilog/VHDL
>> available on github, or http://opencores.org/ is an interesting starting
>> point.
>>
>>
>> - Bryan
>> http://heybryan.org/
>> 1 512 203 0507
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.mit.edu/pipermail/cad-rg/attachments/20151112/4ce701a7/attachment.html