SVT Production Table

The purpose of this table is to count how many units do we have to buy of the expensive and/or critical (mainly because obsolescent) parts for the coming production. The preliminary step is to count how many units have to be bult of each board. The table is work in progress, YOUR opinion is wellcome and much needed. The idea is that here we list the maximum number of boards that we will reasonably build under normal conditions, even if we will not assembly all of them right away, but want neverthless stock critical items so that more boards may be assembled and needed. The rational is that board assemby and testing is quite expensive, likely more than the bare parts cost.
In addition to that I allow 20% extra for critical components for damage during assembly, repairs, use for tests, prototypes, or just in case we have some minor accident and need to build a bit more boards then foreseen. (In case of major disaster it will be cheaper to re-build from scratch the needed functionality with the technology of the time, then to store obsolete parts). The definition of critical components is somehow obscure, it mostly mean stuff that is very hard or unconvenient to buy in small number at later time (i.e. nobody sells ONE fifo chip, but there is no problem in buing one ALS32 whenever needed). For really obsolescent components (EPLD, FPGA's etc.) that are going out of production soon, I put a wider safety margin. For AMchip we are more limited by budget, then by safety... too bad !
The guidelines I followed to count the boards are:
  1. one "sample" baord to be kept wherever each board production is made (the factory).
  2. one complete 1/12th of the system (phi slice) in the test stand in B0, to be put there asap and to be left there "forever" as test setup in B0 to diagnose/test problems on-line.
  3. as many as are needed in the trigger (!!!)
  4. as many boards of each kind as are needed for a minimal functioning test stand for other boards production/repair. Such a test stand will e.g. only need one HF (instead of 3) one only AMB etc. I assume that 2 such test stands will eventually be needed, one in Chicago and one in Italy. In this test stands one needs 2 mergers to feed data in and out of the system, such mergers may be only partially assembled since not all their channels will be required, I have not taken into account this component saving yet.
  5. about 15% spares. As these are somehow interchangeable with what is e.g. in the test stands, these are the boards that we will likely not build right away, but just store parts for.
I think the numbers in this previous exercise to be quite conservative, i.e. we may need to buy less, but definitely no more. I hope with everybody's help we will soon define the number we really need.
Stefano Belforte ( stefano.belforte@pi.infn.it)