atari email archive

a collection of messages sent at Atari from 1983 to 1992.

I.C. Fab Costs

(1 / 1)


Just F.Y.I.

I talked to Jim Wallin and got the following information regarding
CUSTOM I.C. WAFER FABRICATION:

  - DATE: 		7/23/85
  - VENDOR: 		NCR
  - LEADTIME: 		3 WEEKS; TYPICALLY 6 WEEKS
  - MINIMUM ORDER: 	25 WAFERS ( this is considered a production order)
  - TYPICAL YIELD: 	22-23 WAFERS (at wafer level or die level?)
  - COST PER WAFER:	$400.00


------------------------------------------------------------------------
For Reference, Don Paauw says there are 1000 die on a wafer for both
SLAPSTIC AND SLAGS. 

The cost per die is:

		($400.00)/1000=$.40 per die, untested, uninspected


Our minimum order quantity is: 
		
		(25 wafers)*(1000 die)= 25000 die untested, uninspected


In the case of SLAPSTIC which has (8) versions on one wafer, we get

		(25000 die)/(8 versions)= 3125 die of each version, untested
--------------------------------------------------------------------------
These two devices are relatively small. Hence, yields have typically
been in the 70-80% range. SLAGS had lower yields (50%) initially due to
poor processing by the vendor. When I say that the yield is 70-80% I mean
that 70-80% of the packaged parts received by Atari passed ELECTRICAL test
and inspection. Typically, the vendor performs one to three optical inspections
to screen out mal-formed die, reject die attach, and reject wire bonding.  
At this time I do not think Atari is paying for the creation of test programs
to electrically test die at the wafer level (wafer probe..."Sentry Testing").

In summary, actual yield from fabricated die to functional part, will be 
less than 70-80% since die fallouts at the wafer fab and packaging vendors
are not included. I have no data on yields in those two areas at this time.
Message 1 of 1

Jul 23, 1985