Re: 2 Fried G210's in a single day :-(

Hi,

I've been following this thread with attention, so before replying I
decided to research my "deep-fried gecko" database to see what light
it would shed. I looked up the results for the last 12 months, April
2001 to April 2002.

In that time period 6,813 geckos shipped out and 129 came back fried.
That is an unadjusted failure rate of about 1.9%, which is not too
good until it is analyzed further.

Every returned drive is tested and the cause of the failure is
entered into a database. This discards about 110 drives because
typically 1/2 of the drives I see coming back have no discoverable
problem. Of the 129 "fried" ones, every effort is made to determine
the precipitating event that led to failure.

Of the 129 that were actually fried, 69 were determined to be user
induced failures. Some were truly spectacular, such as the 2 that had
115VAC and the one that had 230VAC applied to the power supply inputs.

As probably everyone knows by now, we have a "one time stuff happens"
deal. This means a drive is replaced one time, no charge, no matter
the reason. All that is required is a full confession as to what
happened along with the dead drive. This helps me to see what the
common mistakes are.

Of the remaining 60 drives, it was not possible to determine with
certainty what precipitated the failure with 38 of them. Either they
were too badly damaged to do a forensic examination or the failure
was of such a nature that it could not conclusively proved to be
caused by the user.

That leaves 22 drives where the fault was entirely ours. Here there
were logic ICs that got bored doing Boolean algebra, capacitors that
wanted to be short circuits and resistors that felt infinite ohms
should be their existential goal. Throw in a cracked PCB trace that
only reveals itself with board flexure, a solder-paste joint that
failed to wet an IC pin and cold wave-soldered joint because
someone's guacamole got on the board. That pretty much runs the gamut
for these 22 failures.

I then compared these failure rates against our OEM accounts versus
individual end-users. Our mix is about 70% OEM versus 30%
individuals. OEM by our definition is someone that uses more than 100
drives a year and incorporates them into a product they manufacture.
Some numbers shifted dramatically here.

User induced or undeterminable failures were 19* drives out of 4,680
OEM units shipped, or 0.4%. Non-user induced failures (our fault)
were 12 units, for a failure rate of 0.25%. This is a much more
tolerable number.

Mariss


*The user-induced failure of 19 drives was skewed by one OEM customer
that had a bad (shorted turn) motor in their incoming inspection test
stand. The receiving inspector went thru 8 drives before someone
thought there might be a problem here. They sent the motor to me as a
trophy after things got straightened out. To my knowledge, it has
killed more geckos than any other motor I know of.

--- In CAD_CAM_EDM_DRO@y..., beer@s... wrote:
> On 7 May, CAD_CAM_EDM_DRO@y... wrote:
>
> > All the problems I had appeared to be due to false contacts on
motor
> > leads (two times) or a spike on the motor connection when
plugging
> > the fuse into its socket (the caps were still loaded ... ).
>
> > I'm a product designer and in my job I know an important rule:
every
> > product should be designed with a reasonable fault tolerance.
>
>
> There are limits to what constitutes "reasonable". Certainly,
asking
> the Gecko to survive a roaring fire would not be considered
reasonable
> by most people. Asking the Gecko to survive being run over by a
> bulldozer would also not meet most people's reasonable test.
>
> Well, the electrical nastiness you've imposed ( most likely as a
result
> of arcing ) would very likely be placed in the same class as the
above
> examples by many electronics designers.
>
> It may seem like a small thing to you, but to others it's a really
big
> thing.
>
> For myself, being basically poor, I'd rather have the most stripped
down
> ( and therefore lowest cost ) unit possible and then simply be VERY
> careful.
>
> When I feel the need for insurance, I buy a completed finished
product
> with a warranty.
>
> Alan
>
>
> --
>
> Alan Rothenbush | The Spartans do not ask the number
of the
> Academic Computing Services | enemy, only where they are.
> Simon Fraser University |
> Burnaby, B.C., Canada | Agix of
Sparta