[sdiy] Do Electros Die Slowly?

[harrybissell]

Tantalums were fairly new then, and people had not become used
to the drawbacks.

We used some 35V tantalums in a 24VDC circuit and saw quite a few feild
failures  (and in our business they do not ship to you, they ship your
@ss
to them... anytime day or night :^)

We switched to 50V and the failures stopped.

I like to derate capacitors at 50% voltage if possible. Obviously that
will not work in a guitar amp.

Exceptions are the metallized foil caps I've udes lately.  I spec'd a
1000V
part thinking that the 850V part would be 'marginal'... but the
ratings on these go like
1000V - forever
1100V - three hours per day
1200V - one hour per day
1300V - fifteen minutes per day
1400V - fivi minutes per day

At this rate (with 1200V semiconductors) its moot point.  The 850V parts
were probably more than enough.  When these caps do punch through, a
segmented
foil area disconnects the bad area and the rest of the cap continues to
work...
end of life is 5% loss of initial capacitance (/boom)

Tants and electrolytics are another story, I'd derate of safety,
especially
considering the failure mode <boom>

H^) harry

Bob Weigel wrote:
> 
> Hehe.  I have no idea Harry.  All I know is, they put them in that way
> and given the amount of heat they were putting off combined with the
> series 100 ohm resistor, they most certainly would serve to help protect
> the local circuitry against OV.
> 
> :-) I doubt that's why it was put there in that value though.  I'm not
> sure you caught the humor in that remark. They probably just used 1V
> over there because they didn't have parts of higher value...even though
> in the power supply I think they used 10uf 35V didn't they?  It's been a
> while.  I have no idea why they did it.  but they did it.  I'm always
> prone to speculate the closest thing I can think of even if it's crazy.
> It's a humor style.  Forgive me.  -Bob
> 
> Harry Bissell Jr wrote:
> 
> >"Tantalum as intentional fuse" is a really good
> >idea...
> >as the basis for a novel or urban legend.  :^P
> >
> >I'm sure that no Tantalum manufacturer specifies
> >(specified) the voltage their caps would "blow"
> >at... or how long they would take to "blow" or
> >how much current they would handle if they did... and
> >guarantee that they would remain shorted.
> >
> >In most cases, the damage to other components would
> >probably be done by then.
> >
> >Usually engineers do not rely on unspecified component
> >parameters.  Next weeks part can/will be different.
> >
> >If the tantalum goes and is saves your @ss... you are
> >lucky, nothing more.
> >
> >Much more likely is the difference in PRICE of a 16V
> >tantalum and the next size up, the 25V tantalum.
> >This is a massive difference, especially as the
> >capacitor size (uF) gets larger.  Check out a Mouser
> >or Digi-key catalog.
> >
> >Now... they had a regulated supply, so they threw the
> >dice and said that 16V is good enough for a 15V
> >regulated supply.  I doubt they anticipated, or
> >planned
> >for failure.
> >
> >People who planned for failure, even then... used a
> >zener diode, an SCR and a fuse to crowbar the supply
> >voltage if it gets too high.
> >
> >Don't make the mistake (always) of choosing the same
> >voltage part if it was marginal in the application.
> >No reason not to sub the 25V part if it is available
> >and it fits in the board.
> >
> >H^) harry
> >
> >--- Bob Weigel <sounddoctorin at imt.net> wrote:
> >
> >
> >
> >>Yeah it was kind of funny.  The customer said "Hey I
> >>tried even swapping
> >>in this other board I had out of an omni 1 since
> >>they looked the same.
> >>It heated up in exactly the same place! "
> >>
> >>I promptly mailed him a cheap meter I had here which
> >>I precalibrated.
> >>He reported 14.9V on the supply to the 100ohm
> >>resistor that isolates
> >>that section.  That's about right.  They pair up 2
> >>22uF tantalums in
> >>paralell to get 44uF there.  At least one of them
> >>failed on both
> >>boards.  I was thinking possibly we had an
> >>overvoltage situation but it
> >>appears the power supply was fine. So he just
> >>soldered in the 47uF cap
> >>that I sent him and hopefully all is well. Haven't
> >>heard back. --Bob
> >>
> >>megaohm wrote:
> >>
> >>
> >>
> >>>I never knew that, Bob. Thanks for the heads-up.
> >>>
> >>>
> >>This section has a
> >>
> >>
> >>>fuse, but... if the 16V electros worked/work for
> >>>
> >>>
> >>28 years (it was made
> >>
> >>
> >>>in'78), maybe I'll stick with what is specified.
> >>>
> >>>peng
> >>>
> >>>
> >>>On 4/3/06, *Bob Weigel* <sounddoctorin at imt.net
> >>><mailto:sounddoctorin at imt.net>> wrote:
> >>>
> >>>    I just advised a guy today on an arp omni
> >>>
> >>>
> >>where they DO use that exact
> >>
> >>
> >>>    setup.  I think the tantalum caps are designed
> >>>
> >>>
> >>to be a kind of
> >>
> >>
> >>>    fuse....so that they soak down the power in
> >>>
> >>>
> >>that section so
> >>
> >>
> >>>    nothing else
> >>>    gets fried if the PS regulator fails.  -Bob
> >>>
> >>>    megaohm wrote:
> >>>
> >>>    >
> >>>    >
> >>>    > On 4/3/06, *Loscha* < loscha at gmail.com
> >>>
> >>>
> >><mailto:loscha at gmail.com>
> >>
> >>
> >>>    <mailto:loscha at gmail.com
> >>>
> >>>
> >><mailto:loscha at gmail.com>>> wrote:
> >>
> >>
> >>>    >
> >>>    >     running 16v caps at 15v is much too
> >>>
> >>>
> >>close for comfort
> >>
> >>
> >>>    >
> >>>    >
> >>>    > I agree. This is what is called for on the
> >>>
> >>>
> >>schematic and what was
> >>
> >>
> >>>    > actually used in the tape echo. If I replace
> >>>
> >>>
> >>them, it will be
> >>
> >>
> >>>    with 25V
> >>>    > or 35V. The pcb is tightly spaced but caps
> >>>
> >>>
> >>now are generally smaller
> >>
> >>
> >>>    > so it should be no problem fitting them.
> >>>    >   Still, I think this has little to do with
> >>>
> >>>
> >>the problem I'm
> >>
> >>
> >>>    facing (of
> >>>    > course I could be wrong about that).
> >>>    >
> >>>    > peng
> >>>    >
> >>>    >
> >>>    >
> >>>    >
> >>>    >
> >>>
> >>>
> >>>
> >>>
> >
> >
> >
> >