Hactar wrote:

In article ,
Michael A. Terrell wrote:

How many products have you see with less than 100 components? The
last design I worked on had over 5,000 SMD components.

Hey, I didn't come up with the "100" number. You did, 21 lines up. 5000
independent failures (each of which has a probability of 1% of happening)
would mean a net failure rate of 1-(1-1%)^5000 = about 1 part in 10^22
less than certainty. I suspect typical failure rates (per component)
are several orders of magnitude better.

I said in that example of one failure in 5000 parts, but there are
other failure modes in electronics manufacturing. BTDT, wore out a box
full of soldering iron tips.


You've never met Mr. Murphy have you? Motorola had well over 100%
failure rate at the end of the production line on their TV sets when
they sold their consumer electronics division to Matsu****a. The
components they were using were well over .1% failure rate, yet they had
multiple defects in most sets, and very few that worked in final test.

Well I guess they count a board as failing double if there are two
separate faults. Most humans would put things into two categories, PASS
and FAIL.


It is two failures, if it takes two repairs to make it work. You
troubleshoot to find the first problem and have to stop, till that
repair is made. For instance, you are missing a rail from the power
supply, and the CPU is bad. HITH do you find the second failure, until
the supply is repaired? If it takes ten attempts to repair a $8,000
circuit board, then it has ten separate failures. A bad solder joint is
a failure, but multiple bad joints may, or may not be multiple failures.


Electronics manufacturing is not simple Pass / Fail, except on $1
throwaway Chinese crap.



--
My sig file can beat up your sig file!