Mailing list archives : pcb-rnd

ID:3108
From:ge...@igor2.repo.hu
Date:Fri, 31 May 2019 18:35:58 +0200 (CEST)
Subject:Re: [pcb-rnd] discussion: place order feature
in-reply-to:3100 from ka...@aspodata.se
replies: 3109 from Gabriel Paubert <pa...@iram.es> , 3110 from Peter Stuge <pe...@stuge.se>
 
 
On Fri, 31 May 2019, karl@aspodata.se wrote:
 
>Igor2:
>> a possible feature is "order the board". It would be a trivial next step 
>...
>> Question is wheter we want such a featurea, whether our users would use 
>> it. 
>
>I think it could be a selling point.
>
>Reacently I did a arm/can/test/labcard (#1) for a collegue.
>Local fab wanted about 500EUR for five cards, which my collegue thought 
 
Wow... Unless it's something exotic... I mean if it's the usual 4..6 layer 
board, there are real good fabs in China that can make them much cheaper, 
even shipping included. 
 
>was to expensive so he contacted some english website, which ended in 
>that they wanted a kicad project or a surcharge... and in the end there
>wasn't anything ordered.
>
>#1 small.* in http://aspodata.se/git/openhw/boards_arm_aspo/stm32f105_can/
>
>///
>
>I think for the tree fabs in Sweden, this might not be needed because
>they solve whatever you throw at them. But if you like, I could talk 
>with them and see whether they are interested.
 
Yes please! There are multiple levels on which we can cooperate with fabs, 
and if the fab is willing to invest a minimal amount of engineering hours 
on their side, I am willing to double that on pcb-rnd side.
 
These are the common things we can cooperate in (written in much greater 
detail than what you want to tell the fabs especially in the first round, 
but if you have a clear picture about our side, that may ease the 
communication with them):
 
1. most basic level: we have a cam export job, named after the fab, that 
has been tested with the fab. This doesn't require them to fab any 
physical board for us!
 
What they need to do:
 
- provide us a spec (written in English) about what they expect to get
 
- ask one of their engineers who normally handles incoming gerbers to look 
at our output and tell if it would work fine with their processes
 
- tell us the problems so we can reiterate if the above fails
 
What they get: 
 
- a CAM job named after their fab, which may drive some users to use that 
instead of some random fab that doesn't have a turn-key, fully tested CAM 
config available. (But you can tell them our userbase is small yet)
 
- if other EDA software figure we have this feature, that could motivate 
them to have it too
 
2. more advanceed: automated "get a quote" service
 
As described in the original mail.
 
What they need to do:
 
- provide a reasonable API which we (and potentially other PCB editors) 
can use for getting quotes directly from the fab using board data
 
- provide us some support while we are developing our client for that API
 
What they get:
 
- same as in 1., some potential customers
 
- if they did not have an API before, we can help testing it out (or if 
they want, we can even join in brainstorming/designing it, putting in our 
client-side point of view)
 
 
3. pro: place order
 
Built on top of 2.; requires a bit more API, some sort of user login 
management. It's important that I do not want any payement to be processed 
from pcb-rnd, so this process should get only as far as the order created, 
the options determined and uploaded, files generated and uploaded and an 
URL given to the user where he can log in to the fab's web site and 
actually start the fabbing and do the payement.
 
What they need to do:
 
- same as in 2., just a bit more of it (but probably if 2. is done, 3. is 
a small thing)
 
What they get:
 
- same as 2., with an even smoother workflow for the user; building on 
lazyness of users this could bring more orders
 
4. extra: support us on figuring how to export fancy features
 
At the moment we have full support for bbvias in the data model, in core 
and in the GUI, but we can not export them, simply because we don't know 
how exactly it should be done _in practice_. I mean we know it in theory, 
(please don't link 200 documents explaining it), but we don't have any 
user or fab going to test it for us and bear with the extra cost and 
consequences of potentially botched boards. 
 
A few hours spend by an experienced engineer at a fab can help us bug 
times on this.
 
What they need to do:
 
- same as in 1: we agree on testing some fancy feature, they explain how 
they expect to receive it
 
- same as in 1: we then generate the files and they look at them and tell 
if they look good (no physical board fabbing required)
 
- ... and then the reiterations if we didn't get it right for the first
 
 
What they get:
 
- acknowledgement: I'd name the fab in our communication of the feature 
and thank them for their support
 
- bbvia is a relatively simple thing; some fabs offer more tricky stuff, 
like filled/plugged vias, cavities and other non-uniform board shapes, 
etc. Knowing how some other EDA tools model things, I think most of them 
wouldn't be able to represent these features properly and users would be 
forced to do workarounds and ugly hacks. I can imagine this keeps back 
some users, especially users of free/open or at least small CADs from 
doing boards with those features. In pcb-rnd our data model is more 
capable and more extensible. So we could potentially provide the tool for 
the small budget user to finally do that one special board that needs some 
of these special features. Which may turn into more orders for the fab.
 
- or if not more orders now, at least more advertisement of those special 
features - I mean we'd probably advertise anything unusual we can do and 
got tested by a fab
 
 
 
 
5. Now what I don't plan to offer in general:
 
- exclusivity: I do not want to put ourself in a situation where we can 
support only one fab, and I do not want to make fabs feel like we are 
trying to compare them - but if we do any feature that has to do anything 
with specific fabs, we do that feature in a generic way that potentially 
allows any other fab to join in at any time
 
- being a billboard: I am glad if we have good connection to fabs, but I 
do not want to have useless dummy features shoved into the code just to be 
able to put a label with the fab's name on it. In practice this means our 
users need to get something real, something useful from any cooperation we 
do.
 
- promising anything specific about how many users there would be, how 
many orders they would get, etc
 
- promising anything specific about if a feature will end up implemented 
and released or that it won't get removed later
 
 
Best regards,
 
Igor2
 

Reply subtree:
3108 Re: [pcb-rnd] discussion: place order feature from ge...@igor2.repo.hu
  3109 Re: [pcb-rnd] discussion: place order feature from Gabriel Paubert <pa...@iram.es>
  3110 Re: [pcb-rnd] discussion: place order feature from Peter Stuge <pe...@stuge.se>
    3111 Re: [pcb-rnd] discussion: place order feature from ge...@igor2.repo.hu
      3113 Re: [pcb-rnd] discussion: place order feature from Peter Stuge <pe...@stuge.se>
        3114 Peter, very last warning (was: bbvia, Re: [pcb-rnd] discussion: from ge...@igor2.repo.hu