Maker Pro
Maker Pro

MOSFET failure rates to MIL-HDBK-217F

T

Tom Faloon

Jan 1, 1970
0
MOSFET failure rates to MIL-HDBK-217F

Re failure rate predictions for MOSFETS.
I am using MIL-HDBK-217F (Section 6.4)

My main problem is with "Pi A" calculations.

What is meant by Rated Output power Pr ? Does it mean rated volts X rated
current ? Or what ?
e.g if calculations are for a MOSFET rated at 100V, 60A, to switch a 24V,
30A load, (ON/OFF), then How do I calculate "PI A" ?
If I use rated Volts X rated current it gives a very poor result.

e.g. Lp = Lb X Pt X Pa X Pq X P e


Given
Lb = 0.012,
Pt = 4.2 (Tj = 110 Deg)
Pa = 10 (for Pr = 100V X 60A = 6KW)
Pq = 8 (Plastic)
Pe = 9 (GM)

Lp = 36.3 This is a poor result for just one component ! It only need a
few of these on a PCB for their failure rates to dominate the calculations!

Is my Pa value calculated correctly? or just plain wrong!


Many Thanks,


Tom Faloon
 
M

mook johnson

Jan 1, 1970
0
217 is inherently weak in this area since the voltage and current ratings of
the device cannot be taken into account.

BTW,

Mil-hdbk-217 lists on its first pages that it is only to be used as a guide
and NOT to predict reliability of a product. There is also text that says
that there are NOT to be contractual requirements to meet a particular
mil-hdbk-217 calculated value for reliability.

Since I was put into a situation that a 217 analysis was highly desired, I
used the power dissipation of the mosfet.

My argument was that it appears that 217 assumes the FET will be used in an
amplifier application where Power dissipation of the Mosfet is approximately
= output power.

In my application I was switching 5 V and 100mA with a 200V 10 amp Mosfet
(don't ask why so overrated...long story)
 
R

Roger Gt

Jan 1, 1970
0
X-No-Archive: yes
: 217 is inherently weak in this area since the voltage and
current ratings of
: the device cannot be taken into account.
: BTW,
:
: Mil-hdbk-217 lists on its first pages that it is only to be used
as a guide
: and NOT to predict reliability of a product. There is also text
that says
: that there are NOT to be contractual requirements to meet a
particular
: mil-hdbk-217 calculated value for reliability.
:
: Since I was put into a situation that a 217 analysis was highly
desired, I
: used the power dissipation of the mosfet.
:
: My argument was that it appears that 217 assumes the FET will be
used in an
: amplifier application where Power dissipation of the Mosfet is
approximately
: = output power.
:
: In my application I was switching 5 V and 100mA with a 200V 10
amp Mosfet
: (don't ask why so overrated...long story)


I didn't follow the nature of you problem, however the calculation
SHOULD include the Stress level. That is the actual use vs the
Rated maximums.

I frequently use The Handbook to predict probable like expectancy.
My predictions are historically pretty close to what is
experienced.

No I haven't the time or inclination to teach you (or anyone else)
how to use it. Keep reading, it is all in the Handbook!


:
:
:
: : > MOSFET failure rates to MIL-HDBK-217F
: >
: > Re failure rate predictions for MOSFETS.
: > I am using MIL-HDBK-217F (Section 6.4)
: >
: > My main problem is with "Pi A" calculations.
: >
: > What is meant by Rated Output power Pr ? Does it mean rated
volts X rated
: > current ? Or what ?
: > e.g if calculations are for a MOSFET rated at 100V, 60A, to
switch a 24V,
: > 30A load, (ON/OFF), then How do I calculate "PI A" ?
: > If I use rated Volts X rated current it gives a very poor
result.
: >
: > e.g. Lp = Lb X Pt X Pa X Pq X P e
: >
: >
: > Given
: > Lb = 0.012,
: > Pt = 4.2 (Tj = 110 Deg)
: > Pa = 10 (for Pr = 100V X 60A = 6KW)
: > Pq = 8 (Plastic)
: > Pe = 9 (GM)
: >
: > Lp = 36.3 This is a poor result for just one component ! It
only need a
: > few of these on a PCB for their failure rates to dominate the
: calculations!
: >
: > Is my Pa value calculated correctly? or just plain wrong!
: >
: >
: > Many Thanks,
: >
: >
: > Tom Faloon
: >
: >
: >
: >
:
:
 
L

legg

Jan 1, 1970
0
MOSFET failure rates to MIL-HDBK-217F

Re failure rate predictions for MOSFETS.
I am using MIL-HDBK-217F (Section 6.4)

My main problem is with "Pi A" calculations.

What is meant by Rated Output power Pr ? Does it mean rated volts X rated
current ? Or what ?

Book-rated power at the junction temperature of your application.
..
e.g if calculations are for a MOSFET rated at 100V, 60A, to switch a 24V,
30A load, (ON/OFF), then How do I calculate "PI A" ?

Actual power dissipated by the device can be calculated or measured
many ways - if the frequency is low enough, you may be able to largely
ignore switching losses.
If I use rated Volts X rated current it gives a very poor result.

You would want actual operating Vds x Ids at every instance if time,
integrated over a representative period. Niether of these values
should exceed a book rating.
e.g. Lp = Lb X Pt X Pa X Pq X P e


Given
Lb = 0.012,
Pt = 4.2 (Tj = 110 Deg)
Pa = 10 (for Pr = 100V X 60A = 6KW)
Pq = 8 (Plastic)
Pe = 9 (GM)

You have not used a correct specified (mfr book) Pmax, nor have you
derated this value for Tj, to get Prated in your application Tambient.

This effect of Tj on Prated is independent of the Pt multiplier.
Lp = 36.3 This is a poor result for just one component ! It only need a
few of these on a PCB for their failure rates to dominate the calculations!

The biggest handicap for this kind of calculation under mil-hdbk-217,
that includes use of non-hermetic ( plastic) power semiconductors,
will be the Pq multiplier, providing that junction temperature limits
are avoided by a good margin.

The biggest handicap using 217, in general, is any attempt to use
incandescent lamps, electrolytic capacitors, batteries or non-hermetic
electro-mechanical components - including motors and fans. The
tendency is to isolate these components into their own rated
subassembly, if their use is unavoidable.

The method of calculation is spelt out pretty clearly in the handbook,
and manufacturers are generally compliant in using similar identifying
terms and symbols, to assist you.

RL
 
H

Harry Dellamano

Jan 1, 1970
0
Tom Faloon said:
MOSFET failure rates to MIL-HDBK-217F

Re failure rate predictions for MOSFETS.
I am using MIL-HDBK-217F (Section 6.4)

My main problem is with "Pi A" calculations.

What is meant by Rated Output power Pr ? Does it mean rated volts X rated
current ? Or what ?
e.g if calculations are for a MOSFET rated at 100V, 60A, to switch a 24V,
30A load, (ON/OFF), then How do I calculate "PI A" ?
If I use rated Volts X rated current it gives a very poor result.

e.g. Lp = Lb X Pt X Pa X Pq X P e


Given
Lb = 0.012,
Pt = 4.2 (Tj = 110 Deg)
Pa = 10 (for Pr = 100V X 60A = 6KW)
Pq = 8 (Plastic)
Pe = 9 (GM)

Lp = 36.3 This is a poor result for just one component ! It only need a
few of these on a PCB for their failure rates to dominate the calculations!

Is my Pa value calculated correctly? or just plain wrong!


Many Thanks,


Tom Faloon
MOSFET fail <10PPM. Most failures are caused by: Engineering, Production,
Final Test and improper use.
Put that in your 217 pipe and smoke it!
Harry
 
L

legg

Jan 1, 1970
0
You have not used a correct specified (mfr book) Pmax, nor have you
derated this value for Tj, to get Prated in your application Tambient.
This effect of Tj on Prated is independent of the Pt multiplier.

My error here. Pmax is not derated here.

The Pt multiplier is the ~ derating.

The method of calculation is spelt out pretty clearly in the handbook,
and manufacturers are generally compliant in using similar identifying
terms and symbols, to assist you.

RL
 
T

Tom Faloon

Jan 1, 1970
0
I didn't follow the nature of you problem.....

My problem was
What is meant by Rated Output power Pr ?
(As specified in MIL-HDBK-217F Section 6.4)
when applied to a MOSFET

I do not think it is stated clearly in MIL-HDBK-217F
The term 'output power' does not have a clear meaning for me when applied to
a MOSFET used in a switching application, and you will not find the term in
any MOSFET data sheet.

Also it says rated OUTPUT power. That suggests power delivered to the load,
I do not interpret it as meaning power dissipated in the device.

So what does the handbook mean by Rated Output power Pr, when applied to a
MOSFET used to switch a heavy load ?
A clear definition anyone?

All the other stuff in my original question was meant to illustrate the
point, but only served to cloud the issue! Sorry, and thanks to all who
replied.

Tom
 
R

R.Legg

Jan 1, 1970
0
Tom Faloon said:
My problem was
What is meant by Rated Output power Pr ?
(As specified in MIL-HDBK-217F Section 6.4)
when applied to a MOSFET

I do not think it is stated clearly in MIL-HDBK-217F
The term 'output power' does not have a clear meaning for me when applied to
a MOSFET used in a switching application, and you will not find the term in
any MOSFET data sheet.

Also it says rated OUTPUT power. That suggests power delivered to the load,
I do not interpret it as meaning power dissipated in the device.

So what does the handbook mean by Rated Output power Pr, when applied to a
MOSFET used to switch a heavy load ?
A clear definition anyone?

All the other stuff in my original question was meant to illustrate the
point, but only served to cloud the issue! Sorry, and thanks to all who
replied.

This is probably a typo in 217F. The two tables for Application Factor
and Power Rating Factor have been joined accidentally, and the power
rating multiplier has also been dropped from the general failure rate
formula.

Section 6.4 Application Factors, for LF Power Silicon Fets, on page
6-8 defines Pr within the table as Rated Output Power. The term
'output' does not appear in application factor tables for other power
semiconductors (ie bipolar transistors)where Pr is used - but does
appear in application tables for HF(>400MHz)devices of both bipolar
and mos types.

Definition of the 'output power' term is clarified in section 6.7 on
page 6-12.

RL
 
T

Tom Faloon

Jan 1, 1970
0
Many thanks,

Your theory makes sense, and I am inclined to agree with you.

My only doubt is that, if these really are typos in MIL-HDBK-217F,
then they are fairly serious ones, and I would expect them to have
been documented somewhere. (No mention in Notice 2 for example) ??
 
L

legg

Jan 1, 1970
0
Many thanks,

Your theory makes sense, and I am inclined to agree with you.

My only doubt is that, if these really are typos in MIL-HDBK-217F,
then they are fairly serious ones, and I would expect them to have
been documented somewhere. (No mention in Notice 2 for example) ??

If a viable solution is extractable from within the same doc as the
typo, I expect that there is no great rush.

In the last version of 217 that I ever used (a hard copy of rev D,
twenty years ago) there were no power mosfet device identified - just
standard silicon or GaAs Fets.

At that time, I simply used the power stress multipliers from the
derating tables for Group 1 Transistors (bipolar Si or Ge).

Though this produced degraded reliability figures, the general effect
was insignifigant considering quality factor multipliers, which are
extremely unforgiving.

RL
 
T

Tom Faloon

Jan 1, 1970
0
OK,
I am happy with that. THank you.

Ref your comment above.
The Power rating factor given for Power FETs in 6.4,
actually agrees pretty well with the formula given for Bipolars in 6.3
It is an even better match if the power is raised from 0.37, to 0.3766!
These figures are obviously derived from the same source as those for bipolar!


Pity indeed about quality factor multipliers
(and Environmental factors.) Both can ruin otherwise good results.
However, if they reflect reality, they are doing their job!

It is a pity there is no stress factor given for FET's.
That can sometimes be a life saver in bipolar calculations!
I don't know if the omission of voltage stress factor for
FET's does reflect reality, but I don't suppose MIL-HDBK-217F will
ever be updated again !

Anyway, thanks again for your help.

Tom Faloon
 
Top