UPD70108C Ver la hoja de datos (PDF) - NEC => Renesas Technology

Número de pieza

componentes Descripción

Fabricante

UPD70108C

QUARTERLY MICROPROCESSOR/MICROCOMPUTER RELIABILITY REPORT

NEC => Renesas Technology 

NEC Electronics Inc.

Roseville Manufacturing

Failure Rate Prediction

This report contains reliability test results of all microprocessor

devices assembled in NEC Roseville that were subjected to

routine Monitor Reliability Testing (MRT). It also contains

failure rate predictions for these devices, calculated using the

Arrhenius method shown below.

This report will be updated in September 1998.

When predicting the failure rate at a certain temperature from

accelerated life test data, various values of activation energy,

corresponding to failure mechanisms, should be considered.

This procedure is done whenever exact causes of failures are

known by performing failure analysis. In some cases, however,

an average activation energy is assumed in order to accomplish

a quick first-order approximation. NEC assumes an average

activation energy of 0.7 eV for CMOS-4 and lesser

technologies and 0.45 eV for CMOS-5 and greater

technologies for such approximations. These average values

have been assessed from extensive reliability test results and

yield a conservative failure rate.

Since life testing at NEC is performed under high-temperature

ambient conditions, the Arrhenius relationship is used to

normalize failure rate predictions at a system operation

temperature of 55°C. The Arrhenius model includes the effects

of temperature and activation energy of the failure mechanisms.

This model assumes that the degradation of a performance

parameter is linear with time. The temperature dependence is

taken to be an exponential function that defines the probability

of occurrence.

The Arrhenius equation is:

A = exp −E (T −T )

(1)

A J1 J2

k(T )(T )

J1 J2

Where:

A

EA

T J1

T J2

k

≡ Acceleration factor

≡ Activation energy

≡ Junction temperature (in K) at TA1 = 55°C

≡ Junction temperature (in K) at TA2 = 125°C

≡ Boltzmann's constant = 8.62 x 10−5 eV/K

L = X2 105

(2)

2T

Where:

L

X2

≡ Failure rate in %/1000 hours

≡ The tabular value of chi-squared distribution at

a given confidence level and calculated degrees

of freedom (2f + 2, where f = number of

failures)

T

≡ T Number of equivalent device hours

= (Number of devices) x

(number of test hours) x

(acceleration factor)

Another method of expressing failures is as FIT (failures in

time). One FIT is equal to one failure in 109 hours. Since L is

already expressed as %/1000 hours (10−5 failure/hr), an easy

conversion from %/1000 hours to FIT would be to multiply the

value of L by 104.

Example

A sample of 960 pieces was subjected to 1000 hours at 125°C

burn-in. One reject was observed. Given that the acceleration

factor was calculated to be 34.6 using equation (1), what is the

failure rate, normalized to 55°C, using a confidence level of

60%? Express the failure rate in FIT.

Solution

For n = 2f + 2 = 2(1) + 2 = 4,

X2 = 4.046.

Then L = X2 105

2T

(%/1000H)

=

X2 105

2 (s.s.)(test hrs)(acc. factor)

=

(4.046) 105

2(960)(1000)(34.6)

Because temperature dependence on power dissipation of a

particular device type cannot be ignored, junction temperatures

(TJ1 and TJ2) are used instead of ambient temperatures (TA1

and TA2) . Also, thermal resistance of a particular device

cannot be ignored. These two factors cannot be accounted for

unless junction temperatures are used. We calculate junction

temperatures using the following formula:

= 0.0061 %/1000H

Therefore, FIT = (0.0061)(104) = 61

TJ = TA + (Thermal Resistance) x (Power Dissipation at TA)

From the high-temperature operating life test results, the failure

rates can be predicted at a 60% confidence level using the

following equation:

3

Share Link: