Hi Russell, I work on a lot of stuff with similar
fets and "dimple" mounted. Never seen this fault.
Normally you can crush the heck out of a TO220
pack on the plastic and get many years use.

I work on similar stuff from almost new to many years
old, never seen this.

Maybe try a different brand fet??
-Roman


Russell McMahon wrote:
>
> SUMMARY
>
> Apparent FET failure due to mechanical stress from dimple on mounting clamp
>
> Has anyone else any experience of this ???
>
> +++++++++++++++++++++++++++++++++
>
> DETAIL:
>
> I have experienced an apparent FET failure mode caused by mechanical stress
> that I have not encountered before.
> Has anyone else experienced this or similar.
>
> I have a design with two TO220 package MOSFETS which are mounted on a common
> heatsink. The customer specified the mounting hardware and supplied samples
> for prototype assembly. Silicon rubber thermally conductive but electrically
> isolating pads are placed between each FET and the heatsink and the two FETS
> are held against the  heatsink by a clamping bar which has a single tension
> screw midway between the FETs.
>                        SS
>       CCCCCCSCCCCCCCC
>           \/            S                   \/
>        FFF         S              FFF                        Side View
>   HHHHHHHHSHHHHHHHHHHHHH
>   HHHHHHHHSHHHHHHHHHHHHH
>                      NSN
>
> HHHH = heatsink
> CCCC= clamping bar
> F = FETs
> S = screw
> N = nut
>
> \/ = dimple in mounting clamp bar
>
> HHHHHHHHHHHHHHHHHH
> HHHFFFHHHHHFFFHHHHH   Top View
> HHCCCCCCCCCCCHHHH
> HHCCCCCSCCCCHHHHH
> HHHFFFHHHHHFFFHHHHH
>          |  |  |               |  |  |
>
> There is a "dimple" in the clamp bar where it rests on each FET so that it
> makes point contact to the FET body.
> The dimple is punched in the clamp bar and while not sharp it concentrates
> the force in a small area.
>
> During testing, one PCB exhibited a Drain-Source short in one FET as
> measured by meter in circuit and demonstrated by circuit (non) performance.
> On removing the FET it was found to be non-short. Curious as to cause I
> re-inserted the same FET and after assembling the circuit the short circuit
> reappeared. On dismantly the short again vanished. Using a new FET cured the
> problem.
>
> It APPEARS that the localised clamping pressure is causing internal
> shorting, possibly due to mechanical flexure.
>
> This was originally  not a problem so some degree of "creep" with time and
> or temperature may have occurred.
>
> A friend simultaneously reported similar problems with a power IGBT actually
> splitting across the mechanical contact point when under heavy electrical
> stress. The clamp bars were supplied by the same customer (different
> project). He also reports similar occurrences of failure in a USA produced
> product some years ago where  a vast number of failures occurred in the
> field some time after production. .
>
> I would be interested to hear from anyone with similar experiences.
>
>       Russell McMahon
> _____________________________
>
> --
> http://www.piclist.com hint: The PICList is archived three different
> ways.  See http://www.piclist.com/#archives for details.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.