[Prev][Next][Index][Thread]

Re: Latest MMC Calculations...

Generally, you multiply life multipliers... So, say you had a part spec'd
at 1000 hours, and if you have a factor of 2.1 for reduced voltage, and an
factor of .2 for pulse rate, the overall life would be 2.1*.2 = 1000*.42 or
420 hours....

This presumes the factors don't interact...

Now, if you are dealing with probabilities of failure, it gets a bit more
complex....
Say you have 2 parts with certain probabilities of failure in a given time
span:

Pfailure (total) = PfailureA + PfailureB - PfailureA*PfailureB

If the Pfailures are very small, then the product is very very small and
can be ignored..  You can turn MTBF into failure probability, however, it
is a bit tricky... Say a year has 2000 hours (heh,heh),, If the MTBF is
1000 hours, the probability of failure is NOT 2.. It isn't even 1 (some of
the parts won't fail, just the mean time to failure is half a year)...

Of course, if the MTBF is 1,000,000 hrs, then you can probably get away
with assuming Pfailure (in one hour) is 1E-6

The Reference Data for Radio Engineers (aka the ITT book) has a chapter on
this stuff that has plug and chug equations, as do most other handbooks. If
you are a glutton for punishment, there is always MIL-HDBK-217

----------
> From: Tesla List <tesla-at-pupman-dot-com>
> To: tesla-at-pupman-dot-com
> Subject: Re: Latest MMC Calculations...
> Date: Thursday, July 15, 1999 6:28 PM
> 
> Original Poster: Terry Fritz <twftesla-at-uswest-dot-net>
> 
> Hi Reinhard,
> 
> 	I think I'll agree with you.  If we have three lifetimes that are all
900
> hours, My equation would give only 300 hours total.  That is not right.
> You assumption would give 900 hours total which is much better.  If we
knew
> the distribution of the failure functions we would see that it is really
a
> little less than 900 hours but that detail is meaningless (it is like
> figuring all this out to 30 digits of accuracy...).  I think your taking
> the lowest number as the life is best.
> 
> Cheers,
> 
> 	Terry
> 
> 
> At 08:06 PM 7/15/99 +0200, you wrote:
> >Hi Terry, all,
> >
> >I donīt know if I can agree with this. Personally, I think
> >one needs to calculate the lifetime for each situation (L1,
> >L2, L3) and simply take the lowest number as THE life
> >factor. I know this is somewhat pessimistic, but at least
> >you wonīt be disapointed. For an analogy lets have a
> >look at a normal light switch. Letīs say the life of the
> >plastic handle (before it gets brittle and breaks) is 10
> >years, the life of the pivot (before it wears out so far
> >that it wobbles all around) is 5 years and the life of the
> >contacts themselves is around 2 years (Of course these
> >are just imaginary numbers). This means the effective
> >lifetime of the switch is 2 years and not 1/(1/5+1/10+
> >1/2) years. In our cases, I donīt think Life3 (DvDT)
> >will ever be a real consideration, simply because (if you
> >near or exceed dv/dt) you will also decrease Life1
> >(the temp factor) by a much higher rate. Your method
> >of seeing them as "parallel resistors" would be a sort
> >of statistic value (Iīm not a believer in statistics).
> >However, as not every coiler will be using the exact
> >same building technique or caps (etc), I donīt think
> >this will give more realistic values, then if you simply
> >just view the lowest of the 3 lifetime values.
> >
> >We shouldnīt forget that we are not using the caps in
> >a 24 hr/day mode. If this were the case, then I would
> >tend to agree that the different lifetime factors affect
> >one another (like the temperature factor aggrevating
> >the ionization factor and perhaps dv/dt being lowered
> >due to the lowering of the ionization factor).
> >
> >Coiler greets from germany,
> >Reinhard
> >
> >
> >
> >
> >> Original Poster: Terry Fritz <twftesla-at-uswest-dot-net>
> >> Hi Stefan,
> >> I too was wondering how to combine the lifetimes.  I was thinking of
> >> combining them like parrallel resitors.  Suppose you have three life
> >>times of 10000, 1000, and 5000 hours.  1/(1/10000 + 1/1000 +
> >>1/5000) = 769 hours.
> >>  Seems reasonable to me...
> >
> >
> >> At 04:39 PM 7/14/99 +0200, you wrote:
> >> >Hi Terry, all,
> >> >
> >> >the "15th_power-law" is a very interesting find. I looked over
> >> >the formulas, you posted at
> >> ><http://www.peakpeak-dot-com/~terryf/tesla/misc/MMCCalc1.jpg>.
> >> >
> >> >You wrote:
> >> >> These equations have the life estimates for corona, temperature,
> >> >> and Dv/Dt
> >> >
> >> >and you calculate the lifetimes L1, L2, L3.
> >> >
> >> >But don't you think that all those three major factors play
> >> >a role TOGETHER in decreasing the lifetime? In my opinion, the
> >> >lifetime reduction factors should be multiplied as each one is
> >> >decreasing the lifetime at the same time.
> >SNIP
> >
> >
> >
> >
>