In the simplest sense, a reliability analysis is a means for calculating the failure rate of an electronic device based on the type and quality of the components included in the design. It can be considered an important indication of the overall quality of a product. More specifically, it can be used to indicate the theoretical period of time one can expect a particular device to operate before it experiences a failure of some kind. As such, reliability is commonly quantified in terms of “Mean Time Between Failure” (MTBF). MTBF can be defined using any unit of measure of time. It is commonly given in units of hours or years. Although more than one defined method or ‘standard’ exists for performing a reliability analysis, this discussion will specifically address the method defined by the Military Handbook MIL-HDBK-217F, Notice 2.
If you’ve ever attempted to perform an MTBF reliability analysis per the Military Handbook ‘MIL-HDBK-217F Notice 2’, you may have discovered that it can be difficult – seemingly impossible, in some cases – to reconcile the method – published in 1995 – to current technology and components. Now, more than 20 years later, much of the method is completely outdated and inadequate for the qualification of current designs. In fact, the handbook doesn’t even include analysis guidelines for some commonly used component types, such as ferrite beads and mechanical fasteners. Here, I will provide some general insight into: the basic concepts of this particular MTBF analysis method, tips for locating or deriving failure rates for specific components, and some of the different reliability data formats and equations for conversion between formats.
Before beginning an analysis based on MIL-HDBK-217F Notice 2, there are two critical device parameters that must be defined: ‘use temperature’ and ‘use environment’. This is usually specified in the project requirements. These values factor significantly into MTBF calculations, as will be explained in a moment.
The failure rate modeling equation for each component type in the handbook follows the same general structure. A base failure rate is defined for each component type. This base rate is then modified by various pi factors, the subset of which varies by component type. The pi factor values are dictated by variables including, but not limited to, part quality, use temperature and use environment. Pi factor tables for each component type and associated variable(s) are also provided by the handbook. While the equations for some components are more complex than others, the failure rate is largely based on these primary factors.
As an example, let’s calculate the failure rate for a fixed inductor, assuming the part quality is ‘MIL-SPEC’, the use environment is ‘Ground Mobile’ and the use temperature is 25 degrees C (for simplicity, this example neglects temperature rise in the calculation of temperature). The base failure rate for a fixed inductor is given as 0.000030. The temperature factor at 25 degrees C is 1.0. The quality factor for a MIL-SPEC fixed inductor is 1.0, and the environment factor for the Ground Mobile environment is 12.0. Given these values, the failure rate for this component is calculated as follows:
Of course, not all MTBF calculations are so straightforward. So, in cases with more complex calculation, there are much simpler ways to obtain a component’s failure rate. For example, the complexity of modern day integrated circuits leads to an equally complex method for deriving a failure rate for such components. Fortunately, a number of IC manufacturers provide MTBF data for the parts they manufacture. For example, Linear Technology (www.linear.com) includes MTBF data for many of their integrated circuits on the product page for each part. For many of their parts, the MTBF data are as easy to find as the general data sheets. Other manufacturers, such as Freescale (www.freescale.com), treat MTBF data as proprietary. But, although the data are not published, failure rates can be easily obtained by submitting a service request via the Freescale website.
Unfortunately, not every manufacturer provides reliability data. And, due to the age of the document, not every component can be easily modeled (if at all) using the guidelines provided in the handbook. As an example, the document includes no method for calculating the failure rate of a ferrite bead. However, there is another reference available that can help to address such issues. An ANSI-approved standard (ANSI/VITA 51.1 – 2008 (R2013)), published by VITA, provides updated guidelines for applying the methods defined in MIL-HDBK-217F Notice 2 to modern day components. It also provides guidance on how to model parts, such as ferrite beads, that are not included in the handbook. This helpful resource is available from the VITA online store.
Whether calculated or obtained by other means, reliability data are not always presented in a single, standardized format. There are minor differences between the units of measure for failure rate, FIT and MTBF. Failure rates are often given as ‘Failures per Million Hours’ (FPMH). Failures in Time, or ‘FIT’ rates are defined in ‘Failures per Billion Hours’ (FPBH). For our purposes, MTBF is typically expressed as a number of hours, years or some other measure of time. The relationships between these different data formats are illustrated in the conversion examples below.
FPMH = FPBH*10-3
FPBH = FPMH*103
FIT = FPBH
MTBF = 109 / FIT
The following list provides a means for locating reliability data from some popular component manufacturers, as well as a general reference for estimating MTBF for mechanical components:
- Linear Technology: MTBF data for many individual products are provided in the ‘Reliability Data’ section of specific product pages
- Texas Instruments: MTBF data can be obtained from the ‘Reliability Estimator’ found on the TI website
- Micrel: MTBF data are provided on the Micrel website
- Microsemi: Reliability data are provided on the Microsemi website
- Intersil: Reliability data are provided on the Intersil website
- Freescale: Data are provided via email in response to “service requests” submitted through the website
- Fairchild: Reliability data for individual products are provided in the ‘qualification support’ section of specific product pages
- Maxim: Reliability data for individual products are provided in the ‘design resources’ section of specific product pages
- Coilcraft: MTBF data are provided on the Coilcraft website
- Wurth: MTBF values can be calculated as described by the Wurth document titled “WES_FIT – Reliability Data” Revision 1.10, issued 2015-06-18; the PDF document can be downloaded after selecting “Reliability Data” from the “Category” drop-down menu here
- Miscellaneous: This website provides guidelines for “Failure Rate Estimates for Mechanical Components”