[Open-graphics] Re: inexpensive project vga card batch?

[Michael Meeuwisse]

On 8 Mar 2008, at 22:43, John Griessen wrote:

> It would have to change.   It's not worth starting until there is a  
> batch size of twenty.

I'm not a magician. If people are unwilling to be early adopters,  
then no game. Sorry :)

> But you could help with a layout in pcb free-open-tool and data  
> format, and great postscript and gerber output.
> What is your layout done in?

It's done in Cadsoft Eagle, it's all explained on frontpage of the  
website. Feel free to grab the files and play around with it. That's  
the whole point of the GPL 3 license. Don't expect me to actively  
help porting it to PCB, I'm busy enough as it is with working on the  
HDL.

> So...  what's the total BOM cost?  Take that and add about $5 if  
> batch size is 50  add maybe $10 if
> batch is only 20 boards.

Depends on where you're going to get the components. I guesstimate  
it's roughly $125 per card.

>> Also, I went for the €150,- pricepoint for much the same reason as  
>> you're saying that you can boot-strap your operations
> I'm talking about using the profit from this to pay for devel time  
> on others when I say bootstrapping.
> This product is a very low volume kind....to bootstrap to some  
> higher kinds I am working on.

Sorry, 'others'?

> Sure.   Best thing is to find 20 30 units buyers.  They'll need to  
> pay in advance.

I'm not stopping you :)

>  The existing layout you have could accelerate things.  Even a  
> postscript of it that I can make a light grey
> image of and put in background while I redo layout in geda pcb.
> Part of my selling strategy for open-hardware-eval boards
> is they are a "better" eval board, TAPR open hardware license and  
> geda pcb and gschem docs are ready to
> add onto, not just start from zero and enter all
> the footprints and schematic symbols for a long while before even  
> starting to add new design details.
>
> having geda gschem and pcb docs for the boards makes them a speedup  
> tool for developers, hobbyists.

Images of the top and bottom layer are already available, from the  
front page no less. If you feel like porting the entire design to PCB  
it'd be fantastic. I can't guarantee in any way however, that you'd  
make a buck out of it.

> Thanks,
>
> John Griessen

> PS the same thing could be done for OGD1, except the layout is so  
> much, probably needs 6 weeks
> and a prototype test and bugfix run of 3 boards, I can't see
> starting without running a spreadsheet on how many prepaid buyers  
> there are and if parts cold be bought
> in >100 lots for most parts.  I don't know if the TT folks are OK with
> me being the hardware supplier since they propose to be the  
> supplier at present.   The OGD1 probably needs to be 4 layer.  It  
> has lots of chips, lots of connectors...it's harder, more of a  
> hurdle to be a product.

At this point I'm getting the impression you don't know how much work  
is involved in designing a card like this. FYI, the OGD1 is a 10  
layer design, and it took me a few months to design the 2 layer pVGA  
card. Since Eagle is completely unable to check for signal integrity  
and such I don't even know how stable the design is. pVGA 'only' runs  
a few signals at ~150 MHz, the rest goes slower. OGD1 has signals  
going more than double that. And AFAIK the latest revision of OGD1's  
circuit layout isn't even released, so you'll have to start from  
scratch.

3 prototype cards of OGD1 will cost you about the same as the 25 pVGA  
cards I might build if there's enough interest.

Cheers,

Michael
www.projectvga.org