Mailing list archives : pcb-rnd

ID:714
From:ge...@igor2.repo.hu
Date:Thu, 1 Jun 2017 19:28:23 +0200 (CEST)
Subject:Re: [pcb-rnd] common part format
in-reply-to:713 from James Battat <jb...@gmail.com>
Hi James,
 
On Thu, 1 Jun 2017, James Battat wrote:
 
> Igor2, all,
>
> Just came across this project -- EE Concierge -- a scheme to specify a 
> component (symbol, footprint and 3D model) in a way that is reliably 
> exportable to any user's desired format.  These guys charge a fee for each 
> requested part (average of $34/part):
>  https://blog.upverter.com/2017/06/01/announcing-ee-concierge/
>
> From that page:
> "We store all of this parts data in a common format that we?ve been developing 
> over the years at Upverter. This common format is our second secret weapon. It 
> allows us to build intelligent export configuration and translation tools so 
> that the users of EE Concierge can configure their parts library to export in 
> exactly the same style as parts they make internally."
 
Thanks, I know about upverter and side-projects. My analysis on these:
 
1. Format: we have tEDAx for interchange; the difference is that tEDAx is 
designed for easy implenetation so I hope I can get it spread a bit in 
smallish open source EDA tools.
 
We could support upverter's format, we've already looked at it, but 
there's no active user demand behind it while we have a queue of other 
formats with active user support or even pending test cases.
 
 
2. "Tapping their database for footprints": this idea often comes up with 
open, free or semi-free databases. It sounds like a great idea frist, but 
if you think over the details, it's not as great as it first sounded. The 
problem is that different tools model the world differently. No matter how 
good your format is, no matter how good your library is, no matter how 
good anyone's import/export code is: at the end, if you import a footprint 
from "somehwere else" you will need to open the datasheet and manually 
check all relevant aspects. In a real complex case this includes:
 
  - mass production soldering consideration details, like how longer an smd 
pad exactly is or that you want round cap on one end and square cap on the 
other
 
  - tricky mask and paste patterns
 
  - sometimes even tricky silk pattern
 
At the end you will spend almost as much time on making sure the imported 
footprint won't mess with your design than if you just draw it from 
scratch, deailng with exactly the same considerations.
 
Of course this is somewhat relaxed on the low end, for hobbysts trying to 
build an arduino shiled with 1206 and to92 packages, 2 layers, hand 
soldered. But they are not our true target audience: for them, pcb-rnd and 
gEDA is just too complicated and they will probably land at KiCad or 
eagle.
 
I don't say this means we shouldn't import or reuse footprints. I only say 
that the value of tapping a big database is often not as high as it first 
seems.
 
 
3. web service for sharing resources: we already have edakrill.
 
4. converting between formats: we have that already. Inventing yet another 
format from which conversion to everything else is "possible" or "easy" is 
not required for this. If your starting format is feature-rich enough, 
exporting to others will work well. Whether this format is a specifically 
designed for that or just happens to be a good format originally used 
natively by some EDA tool - doesn't really matter. So the "we have a magic 
format SO we can convert to anything" is rather just PR in my opinion.
 
So I think we already have our version of this part of the service: 
pcb-rnd as a conversion tool and edakrill as a service that transparently 
does the conversion.
 
5. the "pay-for-service, get open source as result" model: those lucky 
bastards! Honestly, I would be happy if we could introduce this in pcb-rnd 
and side projects. Like someone wants a functionality that noone wants to 
implement or if it's on the TODO list but somewhere on the bottom and 
wouldn't happen this year or even next year. The user needs the feature so 
badly that he rather buys a developer's time to see it implemented ASAP.
 
I'd happily jump on such offers - I would just take unpaid holiday at my 
daytime job, or reduce my workhours to 4 days a week. I am pretty sure 
some professional users would be surprised to see how cheap such dev time 
is, and how efficiently it is spent - or in other words, how cheap they 
can get their feature.
 
(It's a funny world we live in; long story short, most of our users are on 
the "West side" of the world while for example I'm on the "East side". The 
economical gap, the gap in prices is huge. That $34, probably a meail at a 
moderate restaurant on that side of the world, ~9 hours of coding on this 
side, lol. But I am sure this could work with some of the other devs too, 
even without the financial gap, just because the given developer 
understands the given part of the code so that he can do the task in 20% 
time compared to anyone else - less time costs less).
 
Unfortunately I see no sign of this happening around here. Our user base 
is too small, even those who are not donating money, "only" their time. So 
I don't see we could introduce such a service for code or footprints any 
time soon.
 
 
Regards,
 
Igor2
 

Reply subtree:
714 Re: [pcb-rnd] common part format from ge...@igor2.repo.hu