On Wed, 15 Dec 2010 22:16:34 -0700, D Yuniskis
wrote:

My 30+ year *design* experience has been very different.

To be fair, I haven't designed anything really complicated for many
years. Between about 1973 and 1983, I played RF designer for various
companies. After that, a mixture of repair, speculation, and
consulting. Most of my horror stories and abominations are from "rent
a project engineer" style short term consulting in the last 10 years
or so.

I see the problem today as one of folks cutting and pasting
"app notes" together -- letting the component manufacturers
do their engineering for them.

Yep. I just love "reference design" clones that don't work any better
than what the applications engineer threw together. Some of them look
like test fixtures or evaluation boards, not finished products. With
RF, layout is actually more important than component selection.
However, we're not talking about RF here. It's ordinary mundane
capacitors that are the current problem.

Much to my amazement, I opened a 2 year old Lenovo desktop and found
polymer capacitors around the CPU. Nice. That desktop is going to
last a long time.
http://www.capacitorlab.com/capacitor-types-polymer/
The whole mess would go away if manufactories would begin using
polymer caps instead of electrolytics in temperature critical
locations. Too bad they cost a few pennies more, some values are not
available, and availability tricky.

The unfortunate ugly truth is that many of these "typical
applications" are *not* designed well. And, the folks
who may have done the designs aren't available to answer
questions about their designs (and shortcomings thereof).

Ok... confession time. When I was doing RF, I would steal as much as
possible. Motorola had a nifty line of RF application notes, with
designs by Helge Granberg. Most were close, but it still took a while
to shake out the problems. If I had sufficient experience with the
components and applications, I probably could have done better
starting from scratch. However, everyone was pioneering back then.
Any manner of head start was beneficial. Incidentally, I didn't do
any better copying designs from competitors radios. Most of those
barely worked.

Reverse example. Just before I was hired, the company had a problem.
The standard IF frequency of 10.7MHz was too low to be usable for the
rather wide marine band (156 to 163MHz). 21.4MHz was the next
available frequency, but none of the commodity integrated IF chips
would work reliably at 21.4 at the time. Something in between was
needed. After many meetings, the high command got tired of the
technical discussion and conjured a frequency out of thin air. It as
16.9MHz which was his daughters birthday. Orders for crystal filters
and offset oscillators were duly ordered. They worked, but the choice
of frequency was horrible, resulting in major birdies (spurious junk)
on top of common channels. Some tweaking helped, but it never really
went away.

Meanwhile, the filter sales droid leaked our choice of IF frequency to
all his other customers, also known as the competition. Since
everyone else had the same problem, they presumed we knew what we were
doing and immediately adopted 16.9Mhz as their new IF frequency. I'm
sure they weren't amused with the results.

A "finished" design is usually outsourced to someone/somewhere
with the emphasis on low cost. The vendor *always* assures the
customer that they will produce a quality product.

The vendor always lies. Most of the data sheets are science fiction.
Production parts never resemble the engineering samples. Welcome to
life in hell.

Long ago, Fairchild Semi did just about everything they could to ruin
my project. In retaliation, I systematically designed their parts out
of the product line. I kept records and kept the lying sales droid up
to date on how much in sales I cost his employer. I don't recall the
total, but it was something like $2 million in the first year. One
would expect that someone would notice or care, but nothing ever
happened except for a visit by a well dressed incompetent that
delivered numerous promises and then disappeared.

But, product lifecycles being as short as they are, design defects
never get fed back into the pipeline -- a "new" product has already
replaced the old; a new vendor doing the fab, etc.

Yep. That's also a major problem. Nobody wants to fix an obsolete
product when its replacement is in the pipeline. In some industries
(i.e. hard disk drive), sometime 3 generations of replacement products
are in the pipeline at the same time.

The end user is the root of the problem, IMO. They neither demand
quality nor are willing to pay for that quality.

Blame the victim? Much as I hate it, I agree with your assessment. I
suspect that simple economics is at the root of the problem. If you
can buy 4 junk contrivances for the price of one quality product, that
becomes a real temptation. With the economy in bad shape, going cheap
versus going quality is no longer an option.

With many devices,
they often don't even know how their device is *supposed* to
operate so don't feel emboldened to complain when it *doesn't*
work.

History will some day remember Microsoft Windoze as having trained the
population to accept mediocrity as normal. There are many things in
life where we simply have never seen quality. I never liked wine
because I always seem to end up drinking junk. My favored method of
cooking salmon produced dry rubbery fish, which I assumed was normal,
until I had some that was properly cooked. However, I got my start
with various Unix mutations, so I never got suckered into thinking
Windoze, MacOS, or OS/X were anything more than a bad imitation. If
you could give the GUM (great unwashed masses) a taste of quality, I'm
moderately sure they will demand better products.

Or, they are all too willing to use the product's failure as an
EXCUSE to buy something new -- "reward" themself. I've seen this
"rationale" all too often.

Yeah, I have the problem myself. I have a 1989 vintage SCO Unix
3.2v4.2 server running in my office that refuses to die. I keep
wishing that something would fail, but it just keeps running. I had 9
months of uptime before I had to move things around and start over. I
keep wanting to replace it with a shiny new Linux box, but only after
it fails. My previous vehicle went for about 290,000 miles.

New products often introduce new features. Why settle for last
year's model when *this* year's model has blah?

Yech. Because the new features are often useless or cause side
effects, such as speed and uptime problems.

"Service" is usually not available, locally (for the majority
of clueless consumers). So, now they have to deal with packaging
the device up for shipment ("Gee, I sure wish I had the forethought
to save the original packing materials!"), paying for shipping
(with "tracking"), living without the device (or its replacement)
for the duration of the repair, an unknown repair time *and* cost.

Sure, but some things are getting smaller and companies are becoming
more efficient at dealing with failures. Many companies have a flat
rate exchange program. That's not going to work for high ticket HDTV
monsters, but it does work for smaller devices. For example,
HP/Compaq ships out an empty padded shipping container for warranty
repairs.

Most "repairs" are board level swaps. And the prices of those
boards are ridiculous.

I'm still doing board level repairs, but I lose money on all of them.
The time and effort necessary to just get access to the PCB is often a
major problem. Products are just not designed to be repairable any
more. I just ripped apart a Roland D-5 junk keyboard synthesizer.
Wires and cables on all 4 sides of the sole PCB. No slack or service
loops anywhere, which means I can't run the board while inside the
unit (without building custom extension cables). So how much more
would it have cost to add a few inches of cable length? More than
Roland was willing to spend.

You would think that high end products would be more repairable, but
apparently not. I tore into an Icom IC-7000 radio. Street price
$1300. The mounting screws would either bind or strip the threads in
the soft aluminum casting. Many of the major components were buried
under layers of unrelated parts and pieces that had to tediously
removed. The driver hybrid was screwed to the casting with no access
holes through the PCB. A large number of leads had to be unsoldered
in order to gain access to the mounting screws. Basically, it's a
nice radio with absolutely no consideration for service and repair.

[E.g. we have avoided purchasing an on-demand hot water heater
out of the *realization* that any failures will be "repaired"
by a PLUMBER. Makes about as much sense as having a carpenter
do your dental work!]

Out of curiosity, if not a plumber, whom would expect to hire to
repair your on-demand water heater? Some of my friends are getting
into solar contracting and are doing everything from wind to solar
electric, including some really complicating plumbing for the water
heater. It's not really that difficult if one is willing to learn.
While I don't do well at plumbing, it's not unusual for me to do a
service call to repair a computah, and end up working on the printers,
network hardware, cell phones, PDA's, backup drives, UPS, game
machines, wireless, and all other manner of marginally related
hardware. While the losers specialize, those with a clue are picking
up adjacent technologies.

Exactly. The same is true with most consumer kit. Businesses are
even worse -- replacing their IT kit every 2-3 years *just* so
they can buy the latest buggy version of windows... (does that
secretary REALLY need a dual quad core 3GHz machine to type up
business correspondence??)

No. But they need it to run MS Office 2010, which requires Windoze 7,
which requires major horsepower. To a hardware geek (like me), the
whole thing seems backwards, but to the business owner, his
requirements are applications driven. He needs to run a short list of
major apps, they require a particular OS, and that requires a hardware
upgrade.

However, you are correct about one item. One of the major reasons to
upgrade is to obtain the latest release of the operating system in the
vain hope that it will have less bugs than its predecessor. Miracles
and bug free operating systems are possible, but unlikely. Still,
hope springs eternal.

One of my servers has triple redundant power supplies. I pulled
two sets-of-three power supplies from similar servers that were
headed to the scrap pile. If I had to buy one today, I suspect it
would be hundreds of dollars -- *if* I could find one!

I do quite well searching for obsolete components and parts on eBay
and various web sites. I don't recall buying much directly from
vendors in many years.

Ditto. Though only where practical. E.g., I often treat color
printers as disposable -- when my supply of toner/ink runs out,
the printer makes a graceful exit (I refuse to spend $200-300
for a set of cartridges!)

Sigh. I buy the refill kits and replacement toner refill protection
reset chip on eBay for my various ancient color laser printers (HP
2500 and 2600). I can usually refill a cartridge 3 times before it's
no longer with the effort. Looking on eBay, clone cartridges are
about $30/ea while toner bottles are about $24/ea. That's quite a bit
cheaper than new cartridges.

Get rid of it. If I haven't had a pressing enough need to fix something
that is broken, it goes out. Life's too short! :

Yeah, I know. I just can't get myself to throw away (recycle) the
junk. However, I'm doing it. Most of the antique machines are gone.
I'm keeping bits and pieces that might be useful, just in case. The
rest is sloooooowly going to the recycler. Sniff.

Whew....

--
Jeff Liebermann
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558