Re: SSTC Reliability

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Jim Lux" <jimlux-at-earthlink-dot-net> 


----- Original Message -----
From: "Tesla list" <tesla-at-pupman-dot-com>
To: <tesla-at-pupman-dot-com>
Sent: Monday, November 08, 2004 5:26 PM
Subject: Re: SSTC Reliability


 > Original poster: Terry Fritz <teslalist-at-twfpowerelectronics-dot-com>
 >
 > Hi Steve,
 >
 > You carefully design it for 100% and then run it at 25% ;-))

Indeed... but that's only part of the story.

 >
 > It would be as reliable as any IGBT generator.  Probably not a good super
 > long term machine since electrolytic caps and all would die just from
 > age.  Getting parts 50 years out would be a problem let alone finding
 > anyone who knew how to fix it.  Trace bits of chlorine and such in the ICs
 > would kill it in about 100 years...  Try to find parts they use in space
 > craft.  I have a 486-50 Intel processor just like is in the Hubble ;-)
To
 > this day, the Space Shuttles main engines use magnetic core memory from
the
 > 60's.  It survived the Challenger explosion...

Not necessarily Class S parts... Turns out that plastic parts (Plastic
Encapsulated Microcircuits, PEMs in the space rel business) might last
longer, depending on a lot of things.  Especially if there are temperature
cycles (as in a museum installation that turns on and off 10 times an hour).
For space they also worry about radiation tolerance, which is probably not a
big deal for terrestrial applications.

 >
 > No problem for say 20 years out since the glass coating industry and all
 > uses pretty darn reliable AC generators now.  Under sea com equipment is
 > "burned in" for 5 years before installation...
 >
 > You would have to build it like those super high end audio amplifiers
whose
 > parts simply don't degrade with time.  MiIitary/Space hermetic parts...


Or, in a more modern vein, something that has a design that doesn't depend
on device parameters.  Self adjusting, if you will.  Part of it is doing a
good worst-case analysis.

 >
 > Document everything so those that have to make "modern" replacements, 100
 > years from now, will know what to do. Be sure the "paper/ink" those
 > documents are written on will last that long too!!
 >
 > Today's parts are just made too cheap and short term so you have to get
 > pretty extreme about finding super long term reliable parts.  This is one
 > vast advantage of the old spark gap systems ;-)

Yep.. pretty darn hard to have a microscale defect in a tungsten electrode.

 >
 > Find an antique electronics store and note what fails, and what does
 > not.  Older IC's from the 70's were much tougher, but not as capable.
Most
 > modern plastics will fall apart after 100 years.  And all this is even
 > before you turn it on...

Not all 60s and 70s ICs were tougher.  Things like tin plating, thermal
expansion coefficients matching, etc.  Of course, the performance was so
grim, that you HAD to design with wide tolerances.  But, I'd sure rather
have modern diodes in a series string than even 80's vintage.


 >
 > I note we had tube based line carrier amps that were working perfectly
fine
 > after 50 years with zero care.  But they were "built" for it!!  Those
tubes
 > were running at about 10% ;-))  Modern version had three pound transistor
 > heat sinks dissipating two watts...  The line power industry is pretty
good
 > at making equipment that lasts forever...
 >
 > Heatsink grease turns to dust after 20 years...  Just don't use anything
"wet".

Chotherm is your friend.


 >
 > Making machines that are designed to last and operate for say 500 years is
 > really hard...  But a fascinating area of study!  I am making an
autonomous
 > robot that "lives" in the wilderness indefinitely.  Batteries are the
 > limiter there along with temperature, humidity, animal attacks....

You might want to look at JPL's bible for design (JPL document D-17868)
which we call "design principles", but has some other formalized
name."Design, Verification/Validation and Operations Principles for Flight
Systems"
http://acquisition.jpl.nasa.gov/rfp/OVWM_TWTA/exhibit1/D-17868.pdf  might
work.
 >
 > Cheers,
 >
 >          Terry
 >
 > >
 > >My question is, what is the likelyhood of being able to produce high
powered
 > >SSTCs (e.g. 10 KVA or more) with museum quality reliability?   That is,
 > >coils that one can trust to run several times a day, day after day, year
 > >after year, without exploded IGBT bricks and other failures?  Can
reliable
 > >"overload" monitoring, anti-shoot-through, shutdown and restart
protective
 > >circuitry be developed?  We know the challenges - highly variable
secondary
 > >loads and reflected impedance changes, severe RFI and EMI for the driver
and
 > >protection circuits, etc.  Quite an engineering challenge.