(ASQ Reliability Review, Vol. 24, No. 1, pp 1823, March 2004)
Revised: March 9 2005 to open workbook in a separate window
November 12, 2006 to add resistors to library and update links
June 4, 2007 to include references to Notice 2 workbook
March 17, 2008 to update links. Thanks to David Dylis.
May 14, 2008 to link MILHDBK217 Update Survey
The Excel workbooks (www.fieldreliability.com/MH217F1.xls and www.fieldreliability.com/MH217F2.xls) implement the partscount MTBF prediction method in MILHDBK217F, Notices 1 and 2. Save and modify the workbooks to suit your needs. Add parts’ failure rates or substitute observed failure rates for the MILHDBK217F failure rates in the library spreadsheets. Add your part descriptions and failure rates. Try it on your next product or electronic subassembly, free of charge. Contact me for help.
Why make MTBF predictions? They’re required: practically everyone in reliability has to make MTBF
predictions for comparisons. They’re also useful for making agespecific
reliability predictions. Credible agespecific reliability prediction
requires some relative indication of old and new product reliabilities. At the
time reliability predictions are needed, all you may know about products are
their MTBF predictions and the field reliability of older, comparable products.
Years ago people hypothesized the constant failure rate
model for electronics parts made to military standards observed to have constant
or bathtubshaped failure rates. Presumably, culling or burnin removed infant
mortality, and the parts’ useful lives were short enough to avoid wearout.
Diligent people collected data and used statistics to estimate constant
failure rates and regression to estimate the pfactors
and stress factors according to acceleration models.
MILHDBK217
[1] standardized MTBF prediction, under the assumptions of series systems, statistically independent
parts, and constant failure rates. Unfortunately, many parts don’t have
constant failure rates. Some have infant mortality. Some deteriorate; such as
motors (dirt, lubricants, and bearings), some capacitors (electrolytic), and
ICs (electromigration and other physical and chemical processes). For these
parts, MILHDBK217 uses agreedupon constant failure rates under the
assumption of maintenance or preventive replacement.
Flash: Your taxpayer dollars are at work.
According to the May 2008 ReliaSoft Reliability Hotwire, “The Defense Standardization Program Office (DSPO) has tasked the
Crane Division of the Naval Surface Warfare Center (NSWC) with
executing a revision of MILHDBK217... In order to achieve the best
possible results, Crane has established a Web page to solicit input from the
reliability community.” (http://www.crane.navy.mil/sd18/ReliabilityFeedback.htm)
A Google search for “MTBF prediction software” yielded:
· RELEX (www.relexsoftware.com/MTBF/)
· RELEX (http://src.alionscience.com/prism/)
·
217PLUS
(http://quanterion.com/RIAC/ProductsAndServices/Products/217plus/index.asp)
·
Instant
MTBF (www.ereliability.com/)
· RelCalc (www.tcubed.com/)
· RELEX (www.lambdacorp.com/)
· ITEM TOOLKIT (www.itemsoft.com/mil217.html)
· Reliability Workbench (http://www.isograph.com/workbench.htm)
The Internet page www.enre.umd.edu/tools.htm listed other MTBF prediction alternatives. Despite ambiguous claims, the commercial reliability prediction software predicts MTBF, not reliability. MTBF is the mean value of system life. Reliability is the complement of the cumulative probability distribution of system life, P[Life > t], for age t. Commercial MTBF prediction software is usually limited to series system structures with independent parts and constant failure rates. To deal with redundancy and dependence, refer to [2], www.fieldreliability.com/MTBFMan.htm, and www.fieldreliability.com/ReliMan.htm.
Commercial MTBF prediction software attempts to link generic handbook descriptions with COTS descriptions and part numbers. A user told me that one commercial library linked his part descriptions and part numbers about 25 percent of the time. To use the Excel workbook described in this article, you have to associate your parts’ descriptions with MILHDBK217F descriptions or add them to the library of parts and failure rates.
You need Microsoft Excel to run the workbooks, and you may need Microsoft Internet Explorer access them. The workbooks contain four spreadsheets, named MH217F1 or MH217F2, PCBA, Pi, and Sorted.
The library spreadsheet, MH217F1 or MH217F2, contains MILHDBK217F and Notice 1 and 2 part descriptions and failure rates (table 3) and a few others. Not all MILHDBK217F parts are included, because some seemed obsolete. Please don’t change the entries in the library spreadsheet. Add your own part descriptions and failure rates at the end of the table. The lookup formulas in the PCBA spreadsheet accommodate additions up to row 250 of the MH217F1 MH217F2 library spreadsheet.
The PCBA spreadsheet contains two tables, table 1 for pfactors and table 2 for your bill of materials. The PCBA table 1 contains pulldown alternatives for pfactors and space for entering the temperature. Select the pfactors and environment temperature that suit your product. Under each pfactor, select the values that describe the environment, learning, and so on that fit your product described. The numbers to the right of the pfactors are the ranks of the selections used to look up the values. The PCBA table 2 contains generic part descriptions, looks up failure rates and pfactors, and does the MTBF prediction computations.
The Pi spreadsheet contains temperature acceleration factors
and pfactor values. Technically, MILHDBK217 allows pfactors to differ for every part in a system, but this workbook assumes that
all parts have the same pfactor
values. Override this assumption if you wish by entering pfactors, for each part, in the appropriate
columns of the MH217F1 spreadsheet.
The Sorted spreadsheet contains the PCBA spreadsheet table 2
data. You’ll have to resort it after changing the BOM in the PCBA spreadsheet.
Add or delete rows of the Sorted spreadsheet to match PCBA table 2, then
sort.
Table
1. pfactors,
ranks, and environment temperature
P_{Q}, quality factor 6 
3 
P_{Q}, quality, discrete semiconductor 6 
19 
P_{L}, learning factor 6 
1 
P_{E}, environmental factor 6 
8 
40 
<temp 
PCBA table 2 has space for your bill of materials, base
failure rates, pfactor, temperature
factors, and their products. Its columns
contain:
1. Your part descriptions.
2. Qty¾the number of identical parts used in your product. Enter a number for every part.
3. Generic Name¾the MILHDBK217F part description you choose as closest to describing your part.
4. FITs¾the MILHDBK217F base failure rate per billion hours per part for a part with the generic part description in column 3.
5. pfactor¾the numeric values corresponding to the pfactors chosen from table 1.
6. Stress and temperature factors.
7. Total device failure rate¾the product of the base failure rate, the pfactors, and the temperature factor. At the bottom of that column, you’ll find the column sum and its inverse, the MTBF prediction.
Table
2. Bill of materials, generic names, and base failure rates
Part Description 
Qty 
Generic
Name 
FITs 
C:0603, 470PF, 10%, 50V 
264 
Capacitors,
Discrete, Fixed, Ceramic 
3.6 
C:C, X7R, 0.01 UF, 50V 
251 
Capacitors,
Discrete, Fixed, Ceramic 
3.6 
C:C,0.1UF,16V,20%,0603 
225 
Capacitors,
Discrete, Fixed, Ceramic 
3.6 
ASIC,
LOOM, L2B 
4 
MOS
MicroProc 32 Bit 
190 
C:C,1UF,+/20%,SM,1206 
30 
Capacitors,
Discrete, Fixed, Ceramic 
3.6 
RPK,51,1%,12
ISO,24QSOP 
36 
Resistors,
Networks, Thick or Thin Film 
2.3 
DIO,
SCHOTTKY 
3 
Diodes
Schottky Barrier and Point Contact 
27 
C:OE,820UF,4V,ESR0.012 
6 
Capacitors,
Discrete, Fixed, Alum, Axial Lead, <400uF 
24 
C:C,1UF,20%,10V,X7R,0805 
46 
Capacitors,
Discrete, Fixed, Tantalum, Solid, Hermetic 
1.5 
IC,MT48LC4M16A2TG10,10NS 
2 
MOS DLA
10000 Gates 
49 
RNET,4.7K,8
BUS,10P 
27 
Resistors,
Networks, Thick or Thin Film 
2.3 
OSC,106.25,85PS,50PPM 
1 
Crystal
Oscillator 
32 
Open the Excel workbook www.fieldreliability.com/MH217F1.xls or www.fieldreliability.com/MH217F2.xls. Tab to the PCBA spreadsheet.
Copy part descriptions and counts from your bill of
materials, and paste them into the first two columns of table 2. Change my
entries as necessary to accommodate your bill of materials. Be sure to include your parts count in the
Qty column.
Select the generic name in column 3 that
describes each part best; i.e., choose the MILHDBK217F generic name
that most closely matches your actual part’s description. See the MH217F1 (or 2) library spreadsheet
for available alternative names. Copy the generic name from the MH217F1 (or 2)
spreadsheet, and paste it into the Generic Name column of the PCBA spreadsheet.
The PCBA spreadsheet looks up the MILHDBK217F base failure rate
corresponding to the generic name in the MH217F1 (or 2) spreadsheet library.
You can add your own parts to the failure rate library for parts not in
MILHDBK217F.
Recalculate the
workbook (click F9). Table 2 shows the system failure rate, the sum of the part
failure rates times part quantities in failures per million hours, and the
system MTBF prediction in hours. Resort the Sorted spreadsheet and add rows if
necessary. Click on the “Save As…” button to save the result to your computer.
Table
3. Library of base failure rates
Section 
Part Type 


5.0 MICROCIRCUITS All Defaults provided
with 
Generic
Name for Lookup 

DIODES 

6.10 
General Purpose
Analog 
DIODES
General Purpose Analog 
6.10 
Switching 
DIODES
Switching 
6.10 
Fast
Recovery Power Rectifier 
DIODES
Fast Recovery Power Rectifier 
6.10 
Transient Suppressor/Varistor 
DIODES Transient Suppressor/Varistor 
6.10 
Power
Rectifier 
DIODES
Power Rectifier 
6.10 
Voltage
Ref./Reg. (Avalanche & Zener) 
DIODES
Voltage Ref./Reg. (Avalanche & Zener) 
6.10 
Current
Regulator 
DIODES
Current Regulator 
6.20 
Si lmpatt (<= 35 GHz) 
DIODES Si lmpatt 
6.20 
Gunn/Bulk
Effect 
DIODES
Gunn/Bulk Effect 
6.20 
Tunnel and
Back 
DIODES
Tunnel and Back 
6.20 
PIN 
DIODES PIN 
6.20 
Schottky
Barrier and Point Contact (200 MHz 5 frequency:5.35 GHz) 
DIODES
Schottky Barrier and Point Contact 
6.20 
Varactor 
DIODES
Varactor 
6.10 
Thyristor/SCR 
DIODES
Thyristor/SCR 
MTBF prediction for series systems with independent parts and constant failure rates is as easy as addition and multiplication, and the spreadsheet implementation is natural. MTBF predictions are useful for predicting agespecific reliability, credibly. Collect field data and make agespecific system reliability predictions by using the ratio of old and new system MTBF predictions. See Credible Reliability Prediction [2] and www.fieldreliability.com for help and for software for doing more than MTBF prediction.
Not everything in MILHDBK217F has been implemented. Contact me at pstlarry@yahoo.com for additions, corrections, and changes, or if you are unable to access the Internet workbook. Send me your library additions, and I will incorporate them in the Internet workbook.
[1] MILHDBK217F 1995, “Reliability
Prediction of Electronic Equipment,” Revision F, December 1991, Notice 1, 10
July 1992, Notice 2, 28 February 1995; www.sre.org/pubs/
[2] George, L. L., Credible Reliability Prediction, ASQRD Monograph, May 2003, www.asq.org/reliability/pub/index.html
DISCLAIMER:
Because this software is free of charge, there is no warranty for the software, to the extent permitted by applicable law. Except when otherwise stated in writing the copyright holders and other parties provide the program “as is” without warranty of any kind, either expressed or implied, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. The entire risk as to the quality and performance of the software is with you. Should the program prove defective, you assume the cost of all necessary servicing, repair, or correction. [paraphrased from Web & Software Development, A Legal Guide, 4^{th} edition, Stephen Fishman, Nolo Press, 2004]