11 Jun
2006
11 Jun
'06
3:03 a.m.
William Donzelli wrote:
I've not noticed that. I use cost solely as the criteria for
when to specify a socketed part on a board of mine. But, that
is OVERALL cost (which is burdened by in-warranty repairs, etc.)
I socket anything that interfaces to the "field" as well as
anything that would be too costly to program *reliably* in-circuit
(adding ISP adds recurring costs, too, so that's a tradeoff).
Anything that is likely to "disappear" during early production
or that is export controlled, etc.
And, anything that can add value by being socketed E.g., in
some markets, it's easier to get a set of DIPs through
customs than it is to try to get a whole *board* through.
So, "firmware upgrades" are easier to accomplish on-the-sly
than they would be with other schemes (relying on FLASH and
available machines -- "PC's" -- isn't always appropriate
for some markets).
I think it
depends on the quality of the components
and construction.
No, it generally does not, unless really cheap, out of spec sockets are
used. The humans and pick-n-place robots still make mistakes - twice as
many when you have twice as many insertions.