Flip-Chip Package Substrate Solder Issue

[Emile]

My Xilinx distributor shipped an affected lot to us and we are now in
a line-down situation. We've been told that lead-time is 10 weeks.
I'm looking for 50 XC2V6000-4BF957C from an unaffected lot. Thanks.

 

[Austin Lesea]

Emile,

Just so you have the facts, we have 100 2V6000's from the affected lot,
that are at 12 upsets in 44,496 device hours of operation. It takes
from 6 to 40 flips to cause a functional fail (on average over many
customer designs). Assuming 10 flips on average, that makes the mean
time to fail 4.2 years (approx). Could be half that on average, to more
than four times that, on average.

Austin

Emile wrote:

My Xilinx distributor shipped an affected lot to us and we are now in
a line-down situation. We've been told that lead-time is 10 weeks.
I'm looking for 50 XC2V6000-4BF957C from an unaffected lot. Thanks.

 

[Martin Euredjian]

"Austin Lesea" wrote:

mean time to fail 4.2 years (approx).

So, if the product in question is not operating 24/7 (powered on/off on a
daily basis or when it gets used) the failure in question is as good as
non-existent?


--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian

To send private email:
0_0_0_0_@pacbell.net
where
"0_0_0_0_" = "martineu"

 

[Austin Lesea]

Martin,

Excellent point. If the unit is reconfigured (doesn't even have to be
powered off) it does not change the probability, but it will prevent
upsets from acumulating bit flips (lessening probability of functional
fail), as well as lessening the time of the outage, or the probability
that an outage is even noticed.

Austin



Martin Euredjian wrote:

"Austin Lesea" wrote:


mean time to fail 4.2 years (approx).


So, if the product in question is not operating 24/7 (powered on/off on a
daily basis or when it gets used) the failure in question is as good as
non-existent?

 

[Emile]

Austin Lesea <austin@xilinx.com> wrote in message news:<bvbamr$eet1@cliff.xsj.xilinx.com>...

Martin,

Excellent point. If the unit is reconfigured (doesn't even have to be
powered off) it does not change the probability, but it will prevent
upsets from acumulating bit flips (lessening probability of functional
fail), as well as lessening the time of the outage, or the probability
that an outage is even noticed.

Austin



Martin Euredjian wrote:
"Austin Lesea" wrote:


mean time to fail 4.2 years (approx).


So, if the product in question is not operating 24/7 (powered on/off on a
daily basis or when it gets used) the failure in question is as good as
non-existent?

Thanks for your help Guys,

With your help, our franchise supplier has agreed to accept the
devices in return for credit. We were also fortunate enough to find
an independant supplier with unaffected Lot Numbers. Thanks again.

-Emile