Hi Jonathan,
A few more ideas how to keep costs down...
We found out that producing 50 PCBs instead of 20 or 100 instead of 50 had
a relatively small incremental cost, maybe 25% or even less. The price
does not fall signifivantly after that (exceptin very big quantities).
In my opinion it doesn't make sense to order less than 100, except for the
prototype batch.
It also helps if all the cheap passive components are on one side of the
board and the expensive ones on the other side. This way you can produce
the boards to "almost complete" and add the expensive parts when you need
them. In this case gold-plating is a must, the tin oxídizes in a few
months. And keeping the number of different components down helps a lot.
At least the cellphone industry uses a lot of high-speed low-accuracy
mounting machines and a few low-speed high-accuracy ones for the critical
components. You are doing a small batch so this is not necessary, but
having worked one week beside a convection oven in hot summer days I can
say that having only the high-accuracy machine can be a bit tedious.
For low-volume production (ten to twenty) you could use etched copper
stencils but laser-cut stainless steel is better if you aim for production
quantities, as you should.
Best regards,
Jouko
...and my apologies for those that are not interested in HW production
issues...
"Life is pretty simple: You do some stuff. Most fails. Some works. You do
more of what works. If it works big, others quickly copy it. Then you do
something else. The trick is to do something else."
On Mon, 15 Sep 2008, Jonathan Weintroub wrote:
Thanks to everyone for their inputs on the budgetary costs of the new CASPER
hardware.
I certainly understand that the its best to spread the fixed costs, NREs etc
over a larger production run. What has worked in the past is to batch
different groups' requirements. i expect there'll be future opportunity to
do so. For now I think I have the rough data I need to propose for funds to
get our hands on some new, bigger, faster, CASPER toys.
Jonathan
On Sep 14, 2008, at 4:46 AM, Jouko Ritakari wrote:
Hi Jonathan,
I don't have exact numbers, but went through the math some six years ago
when we produced the VSIB boards.
A batch of ten boards is a seriously bad idea. In our case the startup cost
was about half of the total cost, we produced one hundred boards.
Most significant was the cost of the PCBs, producing a small number costs a
lot more. Work for buying components is the same for a small or large
batch, automated assemly line costs maybe 50 to 100 dollars per hour, you
can expect two hours setup time and ten minutes for every board.
And if you order a lot of boards it's a good idea to have them gold-plated
or stored in protective atmosphere (nitrogen I think), otherwise they spoil
in six months.
Cheers,
Jouko
"Life is pretty simple: You do some stuff. Most fails. Some works. You do
more of what works. If it works big, others quickly copy it. Then you do
something else. The trick is to do something else."
On Sat, 13 Sep 2008, Jonathan Weintroub wrote:
Can someone please advise rough numbers (which one might use in a funding
proposal), for the cost to fabricate, assemble and test Roach PCBs?
Please assume donated FPGAs, or break out their cost). This application
does not not use large volumes, so assume a run of say 10 boards.
Also, is there a roadmap for compatible ADC boards? In particular we seem
to recall discussion at the workshop of an ADC, with two input channels
*each* capable of sampling at a 2 GSa s^-1 rate, which connects to a
single ZDOC. Is this called iADC2? Has this been built yet? Is there a
rough estimate of cost to fabricate, assemble and test?
Thanks for any guidance the list can offer.
Regards, Jonathan
