at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or is it
just a convenient label ?


--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

On Tue, 16 Oct 2007 08:28:43 +0100, "N Cook"
wrote:
at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or is it
just a convenient label ?

Hi, with an electronmicroscope you can check the die. If it was
destroyed due to ESD, you can see this. A "before and after" foto can
be found at: http://www.jrwhipple.com/sup_notequal.html

Also read what they say about AMD.

Without advanced equipment, it is very difficult to tell the cause.

Pieter

Pieter wrote in message
...
On Tue, 16 Oct 2007 08:28:43 +0100, "N Cook"
wrote:
at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or is
it
just a convenient label ?

Hi, with an electronmicroscope you can check the die. If it was
destroyed due to ESD, you can see this. A "before and after" foto can
be found at: http://www.jrwhipple.com/sup_notequal.html

Also read what they say about AMD.

Without advanced equipment, it is very difficult to tell the cause.

Pieter

What I fail to see is how the most extreme conditions in the original HV
discharge can fail to produce any immediate defect but produce a latent
defect that only becomes a full defect, at some later date, in normal
service conditions of voltages and currents.

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

"N Cook" wrote in message
...
Pieter wrote in message
...
On Tue, 16 Oct 2007 08:28:43 +0100, "N Cook"
wrote:
at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or
is
it
just a convenient label ?

Hi, with an electronmicroscope you can check the die. If it was
destroyed due to ESD, you can see this. A "before and after" foto can
be found at: http://www.jrwhipple.com/sup_notequal.html

Also read what they say about AMD.

Without advanced equipment, it is very difficult to tell the cause.

Pieter

What I fail to see is how the most extreme conditions in the original HV
discharge can fail to produce any immediate defect but produce a latent
defect that only becomes a full defect, at some later date, in normal
service conditions of voltages and currents.

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/


I would guess from the photos that I've seen of this damage, that only part
of an internal structure is blown away such that the manufacturing tests
don't pick it up as being out of spec in any way. Imagine if a particular
transistor was designed into the chip to be part of say a clock oscillator,
and one of the connections to it gets partly blown away due to an ESD on an
external pin. The connection might now be so thin that under the constant
stress of the oscillator current flowing in it, it might heat up more than
it was designed to, until eventually, it fractures and voila ! no clock and
the chip completely dies. As far as proof of this, I think it has been shown
pretty conclusively by s.e.m. photos, as Pieter says, that it is a very real
effect. I seem to remember that some years ago a plane crash, or maybe one
that was shot down accidentally or something along those lines, was shown in
the subsequent investigation, to have suffered some kind of navigation
equipment anomaly, that was caused by an ESD damaged chip. I could be
telling utter lies there, but I'm sure that something like this is in the
back of my mind amongst the cobwebs that seem to settle there a little more
every year ...

Arfa

In article ,
"N Cook" wrote:

at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or is it
just a convenient label ?


--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

Personally, I'm extremely skeptical of the ESD goblin, particularly in
latency. I think the horror stories have been greatly exaggerated, at a
minimum. But, I have no scientific data on which to base my skepticism.

In article ,
"N Cook" wrote:

at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or is it
just a convenient label ?


--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/




Yes, it really does happen. The leakage current on the input may be
specified as 1 nanoamp or some other value relevent to the geometry or
materials involved. An electrostatic dischange that doesn't distroy the
device may cause the leakage current to jump in the hundreds of
microamps. This current can be easily suppled in most cases by the
driving device. However, in such small structures, a phenomenon called
electromigration can take place. The many microamperes of leakage can
cause metal migration to take place along the leakage path. Eventually
it leads to a sufficiently low resistance that the driving source cannot
supply the current to change the logic state. Then you have a failure.
It may take days or years depending on many factors.

Al

On Oct 16, 8:04 am, Al wrote:
In article ,
"N Cook" wrote:

at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause of
defect , ie not stress cracking of die, internal bond failures etc. Or is it
just a convenient label ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

Yes, it really does happen. The leakage current on the input may be
specified as 1 nanoamp or some other value relevent to the geometry or
materials involved. An electrostatic dischange that doesn't distroy the
device may cause the leakage current to jump in the hundreds of
microamps. This current can be easily suppled in most cases by the
driving device. However, in such small structures, a phenomenon called
electromigration can take place. The many microamperes of leakage can
cause metal migration to take place along the leakage path. Eventually
it leads to a sufficiently low resistance that the driving source cannot
supply the current to change the logic state. Then you have a failure.
It may take days or years depending on many factors.

Al

After 44 years at Bell Labs, and head of the Bell Labs EMC Committee
for many years, there are ESD effects at the atom level that lead to
failures in later times, due to disruption of the normal atomic
crystalline (SP?) structure. I wish it wasn't so, but it is.

H. R. (Bob) Hofmann

hr(bob) wrote in message
ups.com...
On Oct 16, 8:04 am, Al wrote:
In article ,
"N Cook" wrote:

at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause
of
defect , ie not stress cracking of die, internal bond failures etc. Or
is it
just a convenient label ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

Yes, it really does happen. The leakage current on the input may be
specified as 1 nanoamp or some other value relevent to the geometry or
materials involved. An electrostatic dischange that doesn't distroy the
device may cause the leakage current to jump in the hundreds of
microamps. This current can be easily suppled in most cases by the
driving device. However, in such small structures, a phenomenon called
electromigration can take place. The many microamperes of leakage can
cause metal migration to take place along the leakage path. Eventually
it leads to a sufficiently low resistance that the driving source cannot
supply the current to change the logic state. Then you have a failure.
It may take days or years depending on many factors.

Al

After 44 years at Bell Labs, and head of the Bell Labs EMC Committee
for many years, there are ESD effects at the atom level that lead to
failures in later times, due to disruption of the normal atomic
crystalline (SP?) structure. I wish it wasn't so, but it is.

H. R. (Bob) Hofmann


Any idea of failure ratio, due to ESD damage, causing immediate failure and
proportion due to delayed latent failures ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

In article ,
"N Cook" wrote:

hr(bob) wrote in message
ups.com...
On Oct 16, 8:04 am, Al wrote:
In article ,
"N Cook" wrote:

at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as cause
of
defect , ie not stress cracking of die, internal bond failures etc. Or
is it
just a convenient label ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

Yes, it really does happen. The leakage current on the input may be
specified as 1 nanoamp or some other value relevent to the geometry or
materials involved. An electrostatic dischange that doesn't distroy the
device may cause the leakage current to jump in the hundreds of
microamps. This current can be easily suppled in most cases by the
driving device. However, in such small structures, a phenomenon called
electromigration can take place. The many microamperes of leakage can
cause metal migration to take place along the leakage path. Eventually
it leads to a sufficiently low resistance that the driving source cannot
supply the current to change the logic state. Then you have a failure.
It may take days or years depending on many factors.

Al

After 44 years at Bell Labs, and head of the Bell Labs EMC Committee
for many years, there are ESD effects at the atom level that lead to
failures in later times, due to disruption of the normal atomic
crystalline (SP?) structure. I wish it wasn't so, but it is.

H. R. (Bob) Hofmann


Any idea of failure ratio, due to ESD damage, causing immediate failure and
proportion due to delayed latent failures ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/


If you Google latent esd you will find about 75000 hits. It sorta hard
to pick out the relevent ones. Good luck!

Al

When I worked at a computer company,
if ever a PC came in for Lightning Damage, and the Insurance company
INSISTED only replacing the damaged parts (instead of replacing the whole
machine), we put in a disclaimer saying that Latent damage resulting from
that strike may cause other components to fail later. And Guess What,
They usually always failed within 3 to 6 months.

P

"Al" wrote in message
...
In article ,
"N Cook" wrote:

hr(bob) wrote in message
ups.com...
On Oct 16, 8:04 am, Al wrote:
In article ,
"N Cook" wrote:

at or before pcb assembly.

Does anyone have proven evidence of this and definitely this as
cause
of
defect , ie not stress cracking of die, internal bond failures etc.
Or
is it
just a convenient label ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/

Yes, it really does happen. The leakage current on the input may be
specified as 1 nanoamp or some other value relevent to the geometry
or
materials involved. An electrostatic dischange that doesn't distroy
the
device may cause the leakage current to jump in the hundreds of
microamps. This current can be easily suppled in most cases by the
driving device. However, in such small structures, a phenomenon
called
electromigration can take place. The many microamperes of leakage can
cause metal migration to take place along the leakage path.
Eventually
it leads to a sufficiently low resistance that the driving source
cannot
supply the current to change the logic state. Then you have a
failure.
It may take days or years depending on many factors.

Al

After 44 years at Bell Labs, and head of the Bell Labs EMC Committee
for many years, there are ESD effects at the atom level that lead to
failures in later times, due to disruption of the normal atomic
crystalline (SP?) structure. I wish it wasn't so, but it is.

H. R. (Bob) Hofmann


Any idea of failure ratio, due to ESD damage, causing immediate failure
and
proportion due to delayed latent failures ?

--
Diverse Devices, Southampton, England
electronic hints and repair briefs , schematics/manuals list on
http://home.graffiti.net/diverse:graffiti.net/


If you Google latent esd you will find about 75000 hits. It sorta hard
to pick out the relevent ones. Good luck!

Al

"Peter K" wrote in message
...
When I worked at a computer company,
if ever a PC came in for Lightning Damage, and the Insurance company
INSISTED only replacing the damaged parts (instead of replacing the whole
machine), we put in a disclaimer saying that Latent damage resulting from
that strike may cause other components to fail later. And Guess What, They
usually always failed within 3 to 6 months.


Al



Well how do you go about repairing a pc thats been hit by lightning??? Just
replace everything except the case?

"Michael Kennedy" wrote in
:


"Peter K" wrote in message
...
When I worked at a computer company,
if ever a PC came in for Lightning Damage, and the Insurance company
INSISTED only replacing the damaged parts (instead of replacing the
whole machine), we put in a disclaimer saying that Latent damage
resulting from that strike may cause other components to fail later.
And Guess What, They usually always failed within 3 to 6 months.


Al



Well how do you go about repairing a pc thats been hit by lightning???
Just replace everything except the case?




Surely depends on how direct the hit was. A direct one would induce
currents that would blow the case apart violently from magnetic repulsion.

As for damaged parts, I guess the answer is to zap the lot afterwards just
to keep the insurance company honest.

On Tue, 16 Oct 2007 23:16:47 +0200, the renowned "Peter K"
wrote:

When I worked at a computer company,
if ever a PC came in for Lightning Damage, and the Insurance company
INSISTED only replacing the damaged parts (instead of replacing the whole
machine), we put in a disclaimer saying that Latent damage resulting from
that strike may cause other components to fail later. And Guess What,
They usually always failed within 3 to 6 months.

P

A great way to scam a new HDD or monitor?


Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com

Insurance would only pay to replace the items that showed physical damage
(normally modems, PSU's and monitors.

"Michael Kennedy" wrote in message
. ..

"Peter K" wrote in message
...
When I worked at a computer company,
if ever a PC came in for Lightning Damage, and the Insurance company
INSISTED only replacing the damaged parts (instead of replacing the whole
machine), we put in a disclaimer saying that Latent damage resulting from
that strike may cause other components to fail later. And Guess What,
They usually always failed within 3 to 6 months.


Al



Well how do you go about repairing a pc thats been hit by lightning???
Just replace everything except the case?

I managed to get myself several modems and monitors that were easy fixes
(normally opto on modem, HOT on monitor, sometimes even fuses on monitors).
Just cos the insurance didn't want them after the claim.

P

"Spehro Pefhany" wrote in message
...
On Tue, 16 Oct 2007 23:16:47 +0200, the renowned "Peter K"
wrote:

When I worked at a computer company,
if ever a PC came in for Lightning Damage, and the Insurance company
INSISTED only replacing the damaged parts (instead of replacing the whole
machine), we put in a disclaimer saying that Latent damage resulting from
that strike may cause other components to fail later. And Guess What,
They usually always failed within 3 to 6 months.

P

A great way to scam a new HDD or monitor?


Best regards,
Spehro Pefhany
--
"it's the network..." "The Journey is the reward"
Info for manufacturers:
http://www.trexon.com
Embedded software/hardware/analog Info for designers:
http://www.speff.com

"Peter K" wrote in
:

Insurance would only pay to replace the items that showed physical
damage (normally modems, PSU's and monitors.


This is why I never pay for insurance.

And I DO mean never, apart from a brief period a few years back paying for
life and injury insurance. The whole point is meant to be trust, sharing of
the load in cases of loss, but the entire industry is founded on profit and
distrust.

If a computer worked, then after a nearby lightning strike it does not, and
can easily be shown as not able to pass a POST or show any other seign of
life, the insurance company must either accept that they are liable if it
was insured against lightning strikes, or accept that people are not wrong
to refuse to accept their'services' and therefore not pay for them.

I'd rather put my money steadily into small assets than can be easily
reconverted to cash fairly quickly at a decent rate. I could lose them, or
my life. So long as I did not lose both, either I, or someone else close
enough to me, will benefit in some way. The worst case is very serious
injury combined with loss of all assets. That's very unlikely, and even
people who get everything 'right' can't immunise themselves against that if
it happens.

The whole problem with insurance comes from a situation where people want
to believe they somehow cannot lose. That, coupled with the fact thet they
are ALWAYS losing the moment they start paying premiums, is absurdity. It's
bad enough that we are sometimes forced to do this by law if we want to
drive or own a house, but there's no point in encouraging it.

If someone knows electronics well enough to handle their computers and
other tech stuff, where is the sense in trusting crucial decisions about
that stuff to people who don't, and whos primary motive is profit, or at
least retention of what they now see as their assets?

It's all very well saying that the small amounts from many people cover the
high cost of unlikely events, but what possible use is that if they won't
pay out when the **** hits the fan?


/rant.

Peter K wrote:
Insurance would only pay to replace the items that showed physical damage
(normally modems, PSU's and monitors.


Maybe. But that is not my experience from the early '80's,
for whatever that is worth.

We had a major thunderstorm that took out the phone company
and cable TV company equipment at their offices, and wiped out
most of the electronics in our neighborhood. I had an Apple II
at the time, which was destroyed, but with no visible physical
damage. A bunch of other stuff was killed in my house, too,
including the cable box and a cordless phone, both of which
had to be replaced. I fixed the rest, but the Apple was totaled.
The insurance company required a repair estimate from a local
repair place. They said it was not repairable, so the insurance
paid for a replacement PC. I suppose they recognized that there
was no way they could deny the claim, with all the damage in the
neighborhood and at the phone company and cable TV offices.

Ed

On Oct 16, 5:02 am, "N Cook" wrote:
What I fail to see is how the most extreme conditions in the original HV
discharge can fail to produce any immediate defect but produce a latent
defect that only becomes a full defect, at some later date, in normal
service conditions of voltages and currents.

Besides the other causes mentioned in this thread, it's also possible
(though
I don't know how commonplace) that the initial strike destroys the
protection
diodes (shunts to power and ground) but not the IO transistor itself.
So the
circuit functions normally but has no more ESD protection; a
subsequent
(unnoticed) static event at lower voltage then takes out the
transistor.

TM

On Oct 17, 11:21 am, Lostgallifreyan wrote:
The whole problem with insurance comes from a situation where
people want to believe they somehow cannot lose.

Gallagher on life insurance:
You're betting that you're going to die - you lose either way!

TM

wrote in
ups.com:

On Oct 17, 11:21 am, Lostgallifreyan wrote:
The whole problem with insurance comes from a situation where
people want to believe they somehow cannot lose.

Gallagher on life insurance:
You're betting that you're going to die - you lose either way!

TM



Nice. Another way of looking at it, the house always wins. Ok, it might not
be exactly the same as a casino but the similarities are shocking. First
duty is to preserve the house. Second duty is to honour the winnings of the
ordinary punter. Third duty is to try to minimise the unusual demands
especially those which conflict with the first law. Sounds almost as
simple as Asimov's Laws of Robotics, and it is. Anyone who has really been
hit hard by injury knows that the insurance people never pay enough to
cover, the promise is always vastly outpaced by the reality. Most people
with really severe needs are more likely to be funded directly by the heath
services as the results of the unusual procedures directly benefit them as
well as the patient. Agreements like that usually begin after the insurance
payouts have long since failed to cover.

It won't be easy in many cases to work out what is the best way to do what
insurance promises to do, but with the growing problems with insurance
(another instance made national news on the BBC this evening), people will
start to keep their money more carefully instead of spending high and
payting what's left to other people for what amount to promises that can't
be kept.

Also read what they say about AMD.

Their presentation read very well up to that point. Evidently they've been
content to ignore the large number of other supporting components made by
AMD, including Ethernet controllers, ROMs and other devices. I would believe
that a lot of the failures they see are due to cheap motherboards, failed
fans and bad/underrated power supplies, but those are just as available and
common for Intel CPUs.

About the only thing I can think of that AMD processors fail dismally at is
running without a heatsink. They smoke, blow up, or explode while most Intel
P4 and later CPUs simply downclock or lock up.

Anyone who has run a multitude of microprocessors side by side will note
that reliability is just about identical in a properly functioning system.

William