Ugh, finance make my head asplode. ;) But, you are absolutely right. The cost of the final product is the cost of producing all the parts divided by the number of operational parts. Semiconductor firms work really hard at keeping yield high, especially those that have their own fabs and so bear the cost of screw ups in the fab process.
It is quite interesting to hear about this daily at work, and to now be involved directly in the process. I would recommend never selling an engineering board to the general populace, even specifying no warranty. There are some good uses for them though. For example once we have taped out the ASIC and have a run of production boards, say a few are found with errors in some part of the 3D core. The folks working on the VGA BIOS and drivers would be prime candadites for a reduced price card that they can use for development. And then they are those of us (and it's nice to not be alone) who want "trophies" of our work. Patrick M --- Timothy Miller <[EMAIL PROTECTED]> wrote: > The first thing you do with a bad board is try to figure out what's > wrong. Often, you get things like poor solder joints for pins on a > QFP, etc. Most of the time, you can just fix it and ship it, although > any fixed board would undergo more extensive testing before being > sent. If the problem isn't an easy fix, then you start hacking at it > to determine what's wrong. The instant you do that, it's toast as far > as being a sellable product. There are some other sorts of "minor" > problems that also toast the board, like a problem with a solder ball > for the BGA. If we can't figure out what's wrong, then we try to > salvage parts. Any board that's been hacked at or any part that's > been salvaged is usually not fit to be sold, so we turn those into > "engineering" boards that we use internally or give away for free. > Warranty returns will undergo similar processes. > > BTW, now that we're on the topic of fall-out... as engineers, we think > about boards costing $X in parts and selling for $Y, making $(Y-X) in > profit. But the big picture is that we bring in $(n*Y), but we spend > $(X*(n+b)), where b is the number of boards that fail testing and > can't be repaired. This affects Y in a non-trivial way. Furthermore, > while there's the idea of selling hacked/salvaged boards for a > discount, that's only worth-while if there's no warranty, but having > no warranty is always bad news, especially when what you're selling > has a higher probability of breaking. > > On 4/28/05, Patrick McNamara <[EMAIL PROTECTED]> wrote: > > Yeah, I know. :) Same things happen for boards, though at a much lower > rate. > > Well, one would hope it's at a much lower rate. > > > > I'd love to be able to do the same with a development board too, but we'll > have > > to see if a) somebody manages to kill a debug board or b) if my wallet will > > survive it. > > > > Patrick M > > > > --- Timothy Miller <[EMAIL PROTECTED]> wrote: > > > There's a certain fall-out rate when chips are produced. We could > > > have the fab house send us the bad ones separately and we could sell > > > them for a few bucks apiece. :) > > > > > > On 4/28/05, Patrick McNamara <[EMAIL PROTECTED]> wrote: > > > > So, I'll go on record as wanting to get a development card (maybe two). > > > I'll > > > > also want to get a production ASIC card (probably two), as well as a > bare > > > ASIC. > > > > > > > > Because someone is going to ask what I want a bare ASIC for, I share > ahead > > > of > > > > time. I'm surrounded by professional chip designers on a daily basis. > > > When we > > > > finally succeed in getting a shipping product, it will be something all > of > > > us > > > > should be proud of. That bare ASIC is going to be embedded in Lexan, > > > framed, > > > > and hung on my cube wall. :) I'm considering doing the same thing > with > > > one of > > > > the production cards. > > > > > > > > Patrick M > > > > _______________________________________________ > > > > Open-graphics mailing list > > > > [email protected] > > > > http://lists.duskglow.com/mailman/listinfo/open-graphics > > > > List service provided by Duskglow Consulting, LLC (www.duskglow.com) > > > > > > > > > > _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
