Calculation of FIT or MTBF

[Alex1000]

Hello all,

from my customer I got the requirement for my system:

5 ppm at lifetime of 10.000 hours. How can I calculate the FIT and the MTBF out of this?

What is wrong in this calculation?

MTBF= 10.000h * 5 *10 exp -6 = 2 exp9 h
Fit= 1/MTBF = 0.00024


thanx
ALex

 

[IRstuff]

you're mixing apples and soda pop. It's completely unclear from your posting what the 5 ppm refers to.

If the "lifetime" is 10 hrs, then that's the MTBF as well, is it not?

TTFN

 

[Alex1000]

HI,

The requirement is that during lifetime of 10.000 h 5ppm of the ECU's are 'allowed' to go bad..

Is it now better defined?

Regards
Alex

 

[qassure]

There are various means to find out the MTBF.
Some time we conduct two types of test to find out the MTBF,. One is time terminated test and other is failure terminated test. The above problem can taken under time terminated test.
No. of test samples - 1 Million
Hrs tested - 10000 Hrs
Failures - 5
Failure time (t1, t2, t3, t4, & t5)
i.e first sample failed at t1, second sample failed at t2 etc.
MTBF = [10000 ( 1000000-5)+ (t1+t2+t3+t4+t5)]/1Million

Note: We normally use 10 to 50 samples

 

[IRstuff]

Accurate failure rate is a data-driven metric. When do 50% of them actually fail?

Barring that, your specification, 5 ppm at 10 hrs, basically implies a 5/(10^6*10) or 0.5 failures per million hrs, which is an MTBF of 2,000,000 hrs

TTFN

 

[Alex1000]

Hello all,

Thanx for the replies...
@IRstuff: your using a linear distribution...isn't it? Should we not use a gaus-distribution?

Here is the possibility which came into my mind
(At the weekend after 5 beers.. )

barring this picture would be a gaus-curve...
The mid of the curve will then represent the MTBF.


|
| _
| /|\
| / | \
|------/ | \-------------------
|---------|----------------------------

MTBF

No I have to calculate the area under the curve and set it equal to my 5 ppm..

The formula would be something like:

Integral f(10.000 h) = 5 ppm

What do you think to this?

 

[IRstuff]

You can try, but you'll find that it's impossible to do that integral without both MTBF and the standard deviation.

Moreover, the concept of MTBF is a linear function, constant-number-of-failures per unit time

TTFN