Good news from Sao Paulo

[Werner Almesberger]

Good news ! Prof. Marcelo Knoerich Zuffo of the University of Sao
Paulo (USP) has offered gta02-core the use of his SMT facility. As if
this wasn't generous enough, all he asks for in return are a few of
our prototypes for his team to play with.

Here's a picture I took:
http://people.openmoko.org/werner/smt.jpg

Just kidding :-) This is just a manual rework station. Here's the real
thing:
http://people.openmoko.org/werner/smt-line-usp.jpg

There are three pick-and-place units with a total of about 200 feeders.
The long white thing at the end is a 12 zone lead-free nitrogen oven.

I'll upload some more pictures later today, when I'm back home.

Regarding PCBs, there are no PCB houses that are good at making this
kind of board in small quantities in Brazil, so the production team at
USP recommended that we look for a PCB maker elsewhere. If anyone has
a recommendation, please post it.

As a reminder, our PCB design parameters are 6 layers, 1 mm board
thickness, 5 mil track width and clearance, and ENIG coating. (USP had
some nasty experiences with cheaper coatings, and we wouldn't want to
lose half our boards to SMT problems.) 1 mm for six layers is a thin
board, which may drive the price up. 5 mil is also more demanding than
the 6 mil many mainstream PCB houses offer.

I also asked about whether we could use micro-vias under pads and
the production team promised me a response in a few days. Oh, and I
forgot to ask for a recommendation for a place that makes stencils.
I'll do so by mail.

We also owe a lot of gratitude to Jon "Maddog" Hall who brought me
in touch with Prof. Zuffo and who convinced him to support our
project.

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Evgeniy Ginzburg]

On Thu, Sep 17, 2009 at 11:20 AM, Werner Almesberger
<[hidden email]> wrote:
> Good news ! Prof. Marcelo Knoerich Zuffo of the University of Sao
> Paulo (USP) has offered gta02-core the use of his SMT facility. As if
> this wasn't generous enough, all he asks for in return are a few of
> our prototypes for his team to play with.
Fantastic news!
Werner you and all the guys in gta02-core are doing very fascinating job.
And of course special thanks to Prof. Marcelo Knoerich Zuffo for
generous offer and Jon "Maddog" Hall for all he do.
[snip loooong quote]

--
So long, and thanks for all the fish.

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[AlvieBoy]

Werner Almesberger wrote:
> Good news ! Prof. Marcelo Knoerich Zuffo of the University of Sao
> Paulo (USP) has offered gta02-core the use of his SMT facility. As if
> this wasn't generous enough, all he asks for in return are a few of
> our prototypes for his team to play with.

Indeed good news! Good work Werner.

> Regarding PCBs, there are no PCB houses that are good at making this
> kind of board in small quantities in Brazil, so the production team at
> USP recommended that we look for a PCB maker elsewhere. If anyone has
> a recommendation, please post it.

We have that one here in pt, not sure if Ruben requested a quote already.

> We also owe a lot of gratitude to Jon "Maddog" Hall who brought me
> in touch with Prof. Zuffo and who convinced him to support our
> project.

Kudos to Jon and Prof. Zuffo :)

Álvaro

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Dave Ball]

Werner Almesberger wrote:
> Good news ! Prof. Marcelo Knoerich Zuffo of the University of Sao
> Paulo (USP) has offered gta02-core the use of his SMT facility. As if
> this wasn't generous enough, all he asks for in return are a few of
> our prototypes for his team to play with.
>  

This is fantastic news!  Thanks to you and Jon for sorting out the
details on this, and of course huge thanks to the good Prof. Zuffo for
inviting us in  :-)


Dave

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Jon 'maddog' Hall]

Werner,

It did not take a lot of convincing.  I think that USP and its
collaboration on GTA02-core will form the start of what I hope is a new
breath of life for the Open Phone project.

md


_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Werner Almesberger]

?lvaro Lopes wrote:
> Indeed good news! Good work Werner.

Thanks :) All I had to do was nod my head and say "thanks" :)

> We have that one here in pt, not sure if Ruben requested a quote already.

It would be great to have a specific quote. I've been given a number
of references to PCB houses that have online quotes, but they all tend
to lack some of the features we need.

I still need to find out if we can use microvias and if we can use
microvias directly under pads/balls. In the latter case, we could
probably get away with 6 mil traces.

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Rene Harder]

Werner Almesberger wrote:
> Good news ! Prof. Marcelo Knoerich Zuffo of the University of Sao
> Paulo (USP) has offered gta02-core the use of his SMT facility. As if
> this wasn't generous enough, all he asks for in return are a few of
> our prototypes for his team to play with.
>  

Great news!
Many thanks to Prof. Zuffo of USP for his offer to utilize his SMT
facility and of course Jon "Maddog" for making all this possible.

> Regarding PCBs, there are no PCB houses that are good at making this
> kind of board in small quantities in Brazil, so the production team at
> USP recommended that we look for a PCB maker elsewhere. If anyone has
> a recommendation, please post it.
>
>  

So far, I've had good experience with Advanced Circuits [1] . They are
fast, reliable and can almost do everything (just a matter of the price
;-) ), also there exists a nice online quoting tool with design file
check . However the capability of this online quoting tool is limited
and If you need specific spec you are better off to directly contact a
sales person. You can find a summery of their capabilities here [2].

[1] http://www.4pcb.com/
[2]
http://www.4pcb.com/media/Capabilities%20August%2009%20-%20PDF%20version.pdf

> As a reminder, our PCB design parameters are 6 layers, 1 mm board
> thickness, 5 mil track width and clearance, and ENIG coating. (USP had
> some nasty experiences with cheaper coatings, and we wouldn't want to
> lose half our boards to SMT problems.) 1 mm for six layers is a thin
> board, which may drive the price up. 5 mil is also more demanding than
> the 6 mil many mainstream PCB houses offer.
>  

I think to get realistic and comparable quotes It would be good to
summarize "all" specs including board array size, panel frame width etc.
(requirements of SMT fab).


Rene

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Werner Almesberger]

Rene Harder wrote:
> [1] http://www.4pcb.com/
> [2]
> http://www.4pcb.com/media/Capabilities%20August%2009%20-%20PDF%20version.pdf

Thanks ! When I run the following specs (ballpark figures):

  Board 3 in x 4 in
  Array 15 in x 6 in, 4 boards (wild guess)
  FR4, 40 mil thick, ENIG, 1 oz copper, 5 mil traces,
  solder mask and silkscreen both sides, default colors,
  smallest hole 8 mil, SMT pitch 15 mil, 100 pads on top,
  1000 bottom, no extra features.

Then I get a per-board price, for 20 units:

        1 week 2 weeks 3 weeks 4 weeks
        93.82 71.82 66.32 60.81 (all USD)

plus Tooling USD 530
        Testing USD 620

So for the 2 weeks option, that would be a per-board price of about
USD 130 plus shipping, import, etc. They don't have online quotes
for stencils, but if I take the price from wedirekt, then a stencil
for the whole array should cost about EUR 160, again plus
international shipping. (If we use a two up, two down configuration
for the array, we may need only one stencil.)

That's pretty good. My estimate was in the EUR 200 range, while this
is only a bit less than EUR 150, all costs considered.

No microvias, though. Sniff.

Note that the fastest option (at USD129.58 per board) would be just
2 days. That's quite an improvement over the 2 weeks it always took
us to get those GTA01/GTA02/HXD8 boards. This is good to know for
future "real" development, because it means that we won't have to
idle endlessly waiting for PCBs as long as our specs are of this
kind.

> I think to get realistic and comparable quotes It would be good to
> summarize "all" specs including board array size, panel frame width etc.
> (requirements of SMT fab).

Yes, I still need to get these.

Thanks a lot !

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Rene Harder]

Werner Almesberger wrote:

> Rene Harder wrote:
>  
>> [1] http://www.4pcb.com/
>> [2]
>> http://www.4pcb.com/media/Capabilities%20August%2009%20-%20PDF%20version.pdf
>>    
>
> Thanks ! When I run the following specs (ballpark figures):
>
>   Board 3 in x 4 in
>   Array 15 in x 6 in, 4 boards (wild guess)
>   FR4, 40 mil thick, ENIG, 1 oz copper, 5 mil traces,
>   solder mask and silkscreen both sides, default colors,
>   smallest hole 8 mil, SMT pitch 15 mil, 100 pads on top,
>   1000 bottom, no extra features.
>
> Then I get a per-board price, for 20 units:
>
> 1 week 2 weeks 3 weeks 4 weeks
> 93.82 71.82 66.32 60.81 (all USD)
>
> plus Tooling USD 530
> Testing USD 620
>
> So for the 2 weeks option, that would be a per-board price of about
> USD 130 plus shipping, import, etc. They don't have online quotes
> for stencils, but if I take the price from wedirekt, then a stencil
> for the whole array should cost about EUR 160, again plus
> international shipping. (If we use a two up, two down configuration
> for the array, we may need only one stencil.)
>
> That's pretty good. My estimate was in the EUR 200 range, while this
> is only a bit less than EUR 150, all costs considered.
>
> No microvias, though. Sniff.
>  

If you want I can try to get a quote with micro- and/or filled vias, so
we can place the vias directly in the solder land.


Rene

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Werner Almesberger]

Rene Harder wrote:
> If you want I can try to get a quote with micro- and/or filled vias, so
> we can place the vias directly in the solder land.

It would be good to have an indication of what effect this has on
the price and on the PCB production time.

If it's small, we can consider using them (provided the SMT production
team at USP is happy with the idea). If it's big, we'd rather try to
avoid them.

Thanks,
- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Rene Harder]

Werner Almesberger wrote:
>
> It would be good to have an indication of what effect this has on
> the price and on the PCB production time.
>
>  

I requested a quote from Advanced Circuits for the following specs
http://rene.saweb.de/gta02-core/PCB-specification.pdf. (One including
microvias, the other one with buried and blind vias only)

It was quite a surprise for me that both configurations end up at the
same price.

Release Quantity Unit Price Delivery

        20 $244.286 2-Week
        20 $221.429 4-Week

Tooling Cost:    $278.57
Quantity:        20   E.Test Cost:    $278.57



So the final price per board would be around 273 USD (2 weeks turn-time).



Rene



_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core

[Werner Almesberger]

Rene Harder wrote:
> I requested a quote from Advanced Circuits for the following specs
> http://rene.saweb.de/gta02-core/PCB-specification.pdf. (One including
> microvias, the other one with buried and blind vias only)
>
> It was quite a surprise for me that both configurations end up at the
> same price.

Perfect, thanks ! So it seems that we have all the parameters for
the layout then ?

Oh, one more perhaps: can microvias be stacked ?

- Werner

_______________________________________________
gta02-core mailing list
[hidden email]
https://lists.openmoko.org/mailman/listinfo/gta02-core