On Aug 14, 9:09 am, wrote:
On Aug 14, 9:30 pm, Jim Wilkins wrote:
I believe that is an unintended consequence of Surface Mount
Technology, specifically thermal stress cracking
Agreed - its a known problem, if designed PROPERLY it can be avoided -
as you say, heart pacemakers dont fail after 3 years (with monotonous
regularity)
Pacemakers operate at a constant temperature until they are no longer
needed...
The gear in the ambulance certainly doesn't, I've examined the
internal data logs of field returns. It's real obvious when the thing
was left out overnight in Toronto or all day in Phoenix. (They took it
well, BTW)
Critical equipment is tested in burn-in chambers. I built them for the
auto and electronics industries back in the 70's. GM parts were cycled
rapidly between -25C and +125C while operating. They had CO2 plumbing
all over the factory for their temperature-shock chambers.
Condensation makes high volume cold testing for commercial-grade
products difficult.
And they keep on making the same old design "errors" after, what, 50
years of solid sate - cheap flexi bendable PCB's, low grade
electrolytics (with known failure curves) heat generating components
with inadequate heatsinking, even to the point of making the pads they
connect to too small...
They don't need to - they know, from engineering knowledge, how long
it will last. Automated production machinery wil take care of the
tolerances involved and make a "perfect" product...
There isn't any one THEY behind this. What you describe is a
compromise between conflicting requirements. The PCB designer picks a
library of component decals that may fit someone's idea of a solution
to his particular problems rather than industry 'standards'. Whoever
bought that library probably hadn't test-driven it. Very likely the
pads were sized for packing density and manufacturability on a certain
brand of pick-and-place machine. People solve their own problems
first. For prototypes I generally extend the pads half a millimeter in
one direction so I can hand-solder a replacement without overheating
the chip.
Thermal calculations are handed off to an ME who is too busy to care.
They are really quite hard because of all the uncertainty, especially
flatness and surface finish. If the heatsink is adequate someone will
object to the cost and size. You can design for a Belleville washer
properly torqued, then see a flat washer overtightened with an air
tool later on the production line. Dilbert isn't fiction.
There aren't enough capable and willing engineers to complete the
boring end of the design process. They move or are pushed on to the
next project while co-ops or lab techs manage the details. From what
I've seen, the board designer at a contract house almost never
understands the circuit well enough to get the subtle details right
and they are under too much pressure to finish it yesterday even if
they did.
Funny that - oh, and BTW - when they can charge you more than the
cost of the appliance for a spare part, they can say they have
honoured there spares supply obligation......like, say, the $10
fusible resistor....and as for custom VLSI ic's - pick a number, any
number, bang a few zeros in there, and add exorbitant shipping and
handling to it...
www.digikey.com
I try to buy either older, repairable expensive equipment or cheap,
disposable new stuff, like a $30 DVD player. That's difficult with
cars but my welding machines all have large transformers and fairly
simple schematics. My machine tools don't even have DROs.
Custom VLSI chips are probably a good part of why electronics are
disposable. They are a real pain even at the factory. Most of the
board's logic is hidden, you don't have the internal schematic or
source code and the programmer quit years ago. I've designed a couple
myself and a year later couldn't easily reconstruct how they worked.
What's the point of troubleshooting a bad $4 chip anyway? The repair
paperwork time costs more than shipping a new board.
I am slowly becoming the parts maker for the old washing
machine and lawn mower.
Good On Ya - thats what RCM should be, and is, all about - to develop
the skills to do this stuff ourselves...its "not economical" but we do
it for the sheer joy of beating the system.
I know how I do and don't use the thing and make a simple functional
replacement rather than a full copy of a molded part. It isn't as
pretty and may need greasing or tightening or replacement more often
than the original. For example, I build up car starter contacts with
brazing rod. The brass might last only 40,000 miles but it's easy to
do over. The washing machine now has a plug to add transmission oil so
I don't have to pull the agitator, and a slinger cup on the bottom
where the oil leaks out, to keep it off the belt. The reel lawn mower
sits on tires from cheap replacement wheels turned down to fit its
hubs (messy!). The ball bearings from those wheels are part of my
sawmill, the metal hubs are baseplates for a hoist.
Maybe its an age thing, but it rankles with me that perfectly good
gear has to be thrown out because a 10c part is No Longer available -
thats why I got into metalworking, so I could make the 10c part....
(its turned into another bloody obsession, which I dont need, but what
the hell...)
Someone else's loss is our gain. Yesterday I acquired a free pool pump
motor that doesn't always start. I showed him where to whack it to
shift the armature but he wants it to run on the timer.
Andrew VK3BFA.
Jim Wilkins