MTBF Prediction Workbook

(ASQ Reliability Review, Vol. 24, No. 1, pp 18-23, 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 MIL-HDBK-217 Update Survey

 

The Excel workbooks (www.fieldreliability.com/MH217F1.xls and www.fieldreliability.com/MH217F2.xls) implement the parts-count MTBF prediction method in MIL-HDBK-217F, Notices 1 and 2. Save and modify the workbooks to suit your needs. Add parts’ failure rates or substitute observed failure rates for the MIL-HDBK-217F 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 age-specific reliability predictions. Credible age-specific 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.

History of MTBF Prediction

Years ago people hypothesized the constant failure rate model for electronics parts made to military standards observed to have constant or bathtub-shaped failure rates. Presumably, culling or burn-in 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 p-factors and stress factors according to acceleration models.

MIL-HDBK-217 [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, MIL-HDBK-217 uses agreed-upon 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 MIL-HDBK-217... 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)

Commercial MTBF Prediction Software

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.e-reliability.com/)

·        RelCalc (www.t-cubed.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 MIL-HDBK-217F descriptions or add them to the library of parts and failure rates.

MTBF Prediction Workbooks

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 MIL-HDBK-217F and Notice 1 and 2 part descriptions and failure rates (table 3) and a few others. Not all MIL-HDBK-217F 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 p-factors and table 2 for your bill of materials. The PCBA table 1 contains pull-down alternatives for p-factors and space for entering the temperature. Select the p-factors and environment temperature that suit your product. Under each p-factor, select the values that describe the environment, learning, and so on that fit your product described. The numbers to the right of the p-factors 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 p-factors, and does the MTBF prediction computations.

The Pi spreadsheet contains temperature acceleration factors and p-factor values. Technically, MIL-HDBK-217 allows p-factors to differ for every part in a system, but this workbook assumes that all parts have the same p-factor values. Override this assumption if you wish by entering p-factors, 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 re-sort 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. p-factors, ranks, and environment temperature

PQ, quality factor 6

3

PQ, quality, discrete semiconductor 6

19

PL, learning factor 6

1

PE, environmental factor 6

8

40

<-temp

 

PCBA table 2 has space for your bill of materials, base failure rates, p-factor, 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 MIL-HDBK-217F part description you choose as closest to describing your part.

4.      FITs¾the MIL-HDBK-217F base failure rate per billion hours per part for a part with the generic part description in column 3.

5.      p-factor¾the numeric values corresponding to the p-factors chosen from table 1.

6.      Stress and temperature factors.

7.      Total device failure rate¾the product of the base failure rate, the p-factors, 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,MT48LC4M16A2TG-10,10NS

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

 

Instructions

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 MIL-HDBK-217F 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 MIL-HDBK-217F 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 MIL-HDBK-217F.

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

 

Conclusions and Recommendations

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 age-specific reliability, credibly. Collect field data and make age-specific 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 MIL-HDBK-217F 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.

References

[1] MIL-HDBK-217F 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, ASQ-RD 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, 4th edition, Stephen Fishman, Nolo Press, 2004]