Libris Britannia 4

home *** CD-ROM | disk | FTP | other *** search

/ Libris Britannia 4 / science library(b).zip / science library(b) / ELECTRON / H478.ZIP / ERP.ZIP / MANUAL.DOC < prev next >

Wrap

Text File | 1991-05-29 | 42KB | 943 lines

RELIABILITY PREDICTOR Version 1.0 Copyright 1990 Ian Thompson-Bell CONTENTS 1. Introduction 1.1 Getting Started 1.2 How to Register 1.3 Files on the Disk 2. Reliability Fundamentals 3. Using Reliability Predictor 3.1 Setting default environment and quality 3.2 Entering components 3.3 Results 3.4 Help 4. Reference data 5. Further Reading 1 1. INTRODUCTION RELIABILITY PREDICTOR is a low cost software package that allows a rapid reliability calculation of an electronic system to be made. You need only decide the default component quality and the environment, enter the quantity of each type of component and RELIABILITY PREDICTOR does the rest. 1.1 Getting Started First you need to make a backup copy of RELIABILITY PREDICTOR or copy it to your hard disk. Just follow the instructions below. 1.1.1 Single Disk Drive PC First format a blank disk. Place your RELIABILITY PREDICTOR disk in drive A and type DISKCOPY A: A: at the DOS prompt and follow the on screen instructions. RELIABILITY PREDICTOR will be copied to your blank disk. Always use this copy. Keep your original RELIABILITY PREDICTOR disk in a safe place. To start RELIABILITY PREDICTOR place the backup copy in drive A: and type ERP. 1.1.2 Dual Disk Drive PC First format a blank disk and place it in drive B. Place your original RELIABILITY PREDICTOR disk in drive A, type DISKCOPY A: B: at the DOS prompt and RELIABILITY PREDICTOR will be copied to your blank disk. Always use this copy. Keep your original RELIABILITY PREDICTOR disk in a safe place. To start RELIABILITY PREDICTOR place the backup copy in drive A: and type ERP. 1.1.3 Hard Disk PC First get into the root directory by typing CD C:\ at the DOS prompt. Then create a new directory for RELIABILITY PREDICTOR by typing MD RELY. Place your original RELIABILITY PREDICTOR disk in drive A and type COPY A:*.* C:\RELY\*.* and RELIABILITY PREDICTOR will be copied to your hard disc. Keep your original RELIABILITY PREDICTOR disk in a safe place. To start RELIABILITY PREDICTOR change directory by typing CD C:\RELY - followed by the ENTER key and type ERP. 2 1.2 How to Register RELIABILITY PREDICTOR is a 'Shareware' product. You may distribute copies of the full ERP disk but don't give out modified versions or copies of the printed manual. If you find RELIABILITY PREDICTOR useful, you are encouraged to buy a registered copy. Your registration fee of 20 pounds will provide the following benefits: 1. The latest version of RELIABILITY PREDICTOR and free updates for 12 months. 2. An up to date set of component reliabilty databases including an emulation for the MIL-HDBK-217E parts count method. 3. CRELON, a utility which allows you to create your own RELIABILITY PREDICTOR database files from simple ASCII text files. 4. Telephone support for your technical queries. 5. A reasonable chance that features you request will be added to the package. To register, send your remittance of 20 pounds to: I.Thompson-Bell 7, Cockhall Close Litlington Nr. Royston Herts SG8 0RB 1.3 Files on the Disk Your RELIABILITY PREDICTOR disk contains the following files: ERP.EXE - the RELIABILITY PREDICTOR programme itself *.REL - RELIABILITY PREDICTOR's database files DATABASE.DEF - an ASCII file listing the database files used by your version of RELIABILITY PREDICTOR MANUAL.DOC - this manual READ.ME - an ASCII text file giving last minute information about upgrades and how to use RELIABILITY PREDICTOR *.HLP - RELIABILITY PREDICTOR's help files 3 2. Reliablity Fundamentals Great changes have taken place in electronics technology over the last twenty years. The introduction of solid state devices and integrated circuits has significantly reduced the cost and increased the complexity of most electronics products. As a result of the much more widespread use of electronics in all sectors of industry and commerce, the reliability of electronics systems has become increasingly important. It is therefore necessary for engineers concerned in the design of electronics systems or components to understand the factors which affect reliabilty, how it can be measured or calculated and what steps they can take to improve it. 2.1 What is Reliability? Before we can talk about the reliability of an electronic system we first need to define the term 'reliability'. The standard definition of reliability is: ' the probability that the system will operate to an agreed level of performance for a specified period, subject to specified environmental conditions ' For example, we might say a certain PC Clone has a reliability of 95% over a 100 hour period at an ambient temperature of 25 degrees centigrade. Another type of PC Clone might have a reliability of 90% over a 150 hour period under the same conditions. The difficulty with this definition of reliability is that we have to specify a time period. This makes it difficult to compare two products and decide which one is more reliable. For example, which of the above two PC Clones is the more reliable? It is impossible to tell from the above definitions of reliability. What we need is a measure of reliability which does not require us to specify an operating period. Such a measure is MEAN TIME BETWEEN FAILURES (MTBF). If a system operates for a total life of 't' hours during which it fails 'n' times then its MTBF is simply: MTBF = t/n Notice that to measure the MTBF we have to operate the sytem for its total life - not much good if you want to know in advance how reliable it is likely to be. However, MTBF is a concept which can be applied to any system and in general the higher the MTBF the more reliable the system. MTBF is therefore very useful in COMPARING the reliability of different systems. 4 2.2 Component Reliability MTBF really only applies to systems consisting of many components. A similar measure for the components themselves is MEAN TIME TO FAILURE (MTTF). Thus if 'n' components were life tested until they failed and the times to failure were t1,t2,...tn then the MTTF is given by: MTTF = (t1 + t2 + ... + tn)/n We might think that component manufacturers would be able to give us MTTF figures for their components but unfortunately this is not generally the case for two reasons: - life testing can only be conducted on a batch of components so strictly speaking the MTTF only applies to that batch and not necessarily to components of the same type that we might purchase. - the MTTF of most electronic components is so long that it is impractical to carry out this type of life test. To overcome these problems manufacturers carry out accelerated life tests at high temperatures. This makes components fail more quickly so that results can be obtained in a more reasonable time. The MTTF at normal temperatures can then be worked out assuming we know exactly how it varies with temperature. Another way of finding out the MTTFs of components is from the maintenance records kept by users of large quantities of electronic products. Such users include telephone companies, the military and the rental companies which purchase large quantities of consumer electronics e.g. televisions. Some of these organisations even publish their results so other people can use them in assessing reliability. However, there is one big problem. THE PUBLISHED RESULTS DO NOT AGREE WITH EACHOTHER!! Even for apparently identical components under identical conditions these authorities can differ by a factor of TEN in the reliability values they give. The reasons for this are discussed in the next section. Finally, it is worth noting that in general the MTTF is not used in reliability calculations. It is more normal to use its reciprocal the FAILURE RATE i.e. FAILURE RATE = 1/MTTF 5 2.3 Environment and Component Quality Our original definition of reliability included the words ' subject to specified environmental conditions ' and it is the difficulty in knowing exactly what environment a particular component has ACTUALY been subject to that is one of the most significant causes of the sort of descrepancies noted above. The environmental factors which can affect a components reliability include temperature, humidity, vibration, ambient pressure, radiation and the presence of contaminants such as dust, oil, acid and so on. In addition, the electrical stresses which the design inflicts upon the component also significantly affects its reliability. With all these variables its no wonder that there is such a variation in published values. The other problem with published data is that you have no real idea if the components you intend using are at all the same as the ones used to create the figures. It is well known that certain types of components are more reliable than others. For example, it is kown that carbon film resistors are more reliable than carbon composition ones and that ICs housed in ceramic packages tend to be more reliable than the same devices housed in plastic packages. Furthermore, it is also well established that a proportion of components tend to fail much earlier than the majority ( known as infant mortalities) and that the failure of these can be hastened by accelerated life testing. Manufacturers therefore set up screening processes, which include accelerated life tests, to allow them to produce ' high reliability ' components for the more demanding applications. Quite often the screening process is specified by large purchasers of components who subsequently publish achieved reliability figures!! So unless the components you use go through the same screening process the chances are they will have a different reliability. 6 2.4 The Answer The problem seems to be that we can't trust the published data because there seems to be little relationship between the way it was obtained and the way components will actualy be used in our own application. However, all is not lost! RELIABILITY PREDICTOR makes use of the fact that a small number of factors have a much greater effect on reliability than any others. These factors are: - temperature - vibration - component quality - component stressing It uses the way these factors affect the reliability of different types of components to provide a database of component failure rates which is capable of providing a reasonable estimate of the likely reliability of the entire system at the DESIGN STAGE which is where it can be used to best effect in improving reliability. To do this it doesn't just calculate the overall system reliability but it also shows you just which components are contributing most to the failure rate so you can alter the design to make it more reliable. Just how it is able to to this is explained more fully in Section 3 Reference Data. For now all you need to know is that RELIABILITY PREDICTOR has built into it failure data from a wide range of sources and many man years of experience in the design of electronics products. 7 3. Using RELIABILITY PREDICTOR RELIABILITY PREDICTOR is extremely easy to use. When first started it displays a welcome message. Pressing return causes it to load in its database and display the opening menu. This allows you to select which one of the six basic operations you want RELIABILITY PREDICTOR to carry out. The current selection is highlighted and you can change this using either the cusor up/down keys or with a mouse. Once you have highlighted the operation you want, just press ENTER to select it. No special setup is necessary to use a mouse. Any mouse which inserts cursor key tokens into the keyboard buffer will do.If one like this is present and its driver is loaded, RELIABILITY PREDICTOR will make it available to you. The bottom line of the screen is reserved for status messages. These show which keys are active and what each one does. HELP is always available simply by pressing F1. HELP gives details about the choices available to you at any time along with reminders about the keys available. The following sections describe each of the six basic operations of RELIABILITY PREDICTOR which can be selected from the opening menu. 3.1 Setting Default Environment and Quality Two of the most important factors which influence the reliability of a system are the environment and the quality level of the components used. Usualy the whole system is subjected to the same environment and all components are bought to a common quality standard. For these reasons RELIABILITY PREDICTOR allows you to specify the default environment and component quality. These default values apply to every component in the system. Component quality can be over-ridden on an individual component-type basis as a common way of improving reliability is to use better quality components in particularly senesitive circuit areas. Environment is selected first by using the cursor keys or a mouse to highlight the one you want just as in the opening menu. Confirm your selection by pressing RETURN. The component quality options are then displayed. Make your selection just as for environment. Once both the default environment and component quality have been selected you will be returned to the opening menu. 8 3.2 Entering Components This is where you give RELIABILITY PREDICTOR the information it needs about YOUR design so that it is able to predict its reliability. The first menu allows you to select the CLASS of components to enter. Component classes include, for example, resistors, diodes, RAMs, microprocessors and so on. To select a class you first highlight the one you want, using the cursor keys or the mouse, just as with the other menus, but this time TWO columns of classes are available. Just use the cursor left/right keys or the mouse to move from column to column. Note: to leave the Enter Components menu and return to the opening menu just press ESCape. When you have highlighted the component classs you want, press RETURN to select it. You will now be presented with a screen which lists all the available components in that class. Notice that the CLASS NAME is repeated at the top of the screen. This screen has three columns. The left most column lists the component types in the selected class. You cannot select or make changes to this column. The other two columns show the quantity and quality of each component type. You can move up and down these columns or from column to column using the cursor keys or the mouse just as before. However, because you want to enter values rather then make a selection, a flashing cursor is used rather than highlighting to indicate the current position. The centre column is for the quantity of each component type. This is initialy set to 0000 by RELIABILITY PREDICTOR. To enter the number of components of a particular type in your design simply move the cursor to the relevant position in the quantity column and enter the number. Notice that the cursor automatically moves one space to the right after each number is entered. You can, of course, use the cursor left key to back space at any time to make a correction. It is important to remember that the significance of each digit is preseved. This means that to enter 1000 you just put a 1 in the left hand digit. To enter 1 you just put 1 in the right hand digit. To zero and entry, just put the cursor on the left hand digit and press 0 four times. The right hand column is for component quality. This will initialy show the default quality for each component type as either a capital C, I, or M to indicate Commercial, Industrial or Military quality respectively. To change the component quality level for a particular component type, simply position the cursor under its quality letter and type the letter corresponding to the new quality level ( C, I or M ). RELIABILITY PREDICTOR accepts upper or lower case letters. 9 When you have entered all the information about a class of components just press ESCape to return to the initial Enter Components menu. You can then select another components class to work with. Note: you can return to a previous component class at any time and all the information you have entered so far will be preserved and shown on the screen. This makes it very easy, in combination with the RESULTS option, to see the effect of changing component quality levels or quantities. 3.3 Results The Results option allows you to see a summary of the results at any time or to separately display or print a more detailed report. Every time the Results option is selected, the results are re-calculated using the latest component numbers, quality levels and environment and the summary results are displayed. The summary results show: i) The total predicted failure rate and the MTBF. ii) The percentage of failures due to the components in each CLASS. This allows you to see at a glance which types of components are contributing most to the overall failure rate. iii) The default Environment and Component Quality. iv) The assumed ambient temperature. The summary results are useful in answering 'what if' questions. You can return to the opening menu by pressing ESCape, alter the default environment, quality level or just the quantity or quality level of an individual component, and return to the summary results to immediately see the effect of the changes you have made. Alternatively, to obtain more detailed results, press F6. 10 3.3.1 Detailed Results The Detailed Results menu allows you to select one of three types of report: The Summary Report is the same as the one shown on the Results screen with the addition of a list of the Top Ten least reliable components. This makes it extremely easy to identify the components which are contributing the most to the overall failure rate. You can then go back to the Enter Components menu and change, for example, the quality level of a particular component or its type i.e. you might like to see the effect of changing from carbon composition to metal film resitors. Each time you select Results from the Main Menu, the MTBF is re-calculated so you can quickly see the results of any changes. The Short Report provides an additional separate report for the components in each class showing the quantity, quality and individual percentage contribution to overall failure rate of each component type. This makes it easy to see which components are likely to dominate the overall failure rate once the Top Ten components have been dealt with. The Complete Report also provides an additional report for the components in each class but provides more detail than the Short Report. The failure rate for each individual component type (taken directly from the database) is given in FITS along with the actual failure rate contribution for the quantity of that type of component used. This information is particularly useful in guaging the effect of the default environment on different types of component. This is because RELIABILITY PREDICTOR takes account of the fact that different components' failure rates change by different amounts as you change from one environment to another. (See Section 4) Finaly, any of the above reports can be sent to a printer connected to the parallel port. To enable printing, highlight the Turn Printing ON selection and press RETURN. The Status Line at the bottom of the screen will change to show that Printing is ON. Any report now selected will be automatically sent to the printer as well as being displayed on the screen. You will notice that the Turn Printing ON menu selection has now changed to Turn Printing OFF. To disable printing, highlight the Turn Printing OFF selection and press RETURN. The status line at the bottom of the screen will change to show that Printing is OFF. 11 3.4 Save Data and Load Data It is often useful to be able to save the results of a RELIABILITY PREDICTOR session and retrieve it later. In this way a design that has been modified can have its reliability re-calculated with the minimum of effort and standard types of designs, which may be used several times with minor modifications e.g. power supplies, microprocessor sub-systems, can become the basis of a library of standard designs. The Save Data and Load Data functions allow you to do this. To save a design, highlight Save Data and press RETURN. You will then be asked for a filename. Enter any valid DOS filename preceded by and optional path and press RETURN. If no path is specified the file will be saved in the current directory. Note that if the file already exists it will be overwritten without RELIABILITY PREDICTOR issueing a warning. To retrieve a design highlight Load Data and press RETURN. Again you will be prompted for a filename. Enter the name of the file you require, press RETURN and RELIABILITY PREDICTOR will load it. Any design currently in memory will be overwritten and lost. Note, you must load a file with RELIABILITY PREDICTOR running with the same database.def file that was in use when the file was saved or the results will be unpredictable! See Section 4 for details of how to edit or create your own version of database.def. 3.5 QUIT When you have finished using RELIABILITY PREDICTOR simply highlight QUIT, press RETURN and you will be returned to DOS. 12 4. Reference Data RELIABILITY PREDICTOR uses a database which consists of a list component class names and an associated file of failure rate data as the basis of its calculations. When run, RELIABILITY PREDICTOR reads an ASCII file named database.def which contains this list. Database.def looks something like this: * Bipolar Digital ICs * DIGICSB.REL * MOS Digital ICs * DIGICSM.REL etc The file consists of pairs of lines. The first line is the name of a component class EXACTLY as it will appear on the Select Component Class menu. The second line is the name of the file containing the the names of the component types in that class and their individual failure rate data. When run, RELIABILITY PREDICTOR loads and stores all this information in memory. The files containing failure rate data are stored in an encoded form for fast loading. Each file contains the following information about each component type: - Component Name - Base failure rate for each of three quality levels - Multiplication factor to account for each of three possible default environments. The failure rate for a particular component type is calculated by multiplying the base failure rate for the selected quality level by the multiplication factor for the selected default environment. This means the failure rate used for a particular component type can be one of NINE values depending one the chosen quality level and default environment. This level of detail is necessary because the failure rate can vary considerably with these factors even for similar components in the same class. RELIABILITY PREDICTOR can use an alternative database to database.def. To do this just start RELIABILITY PREDICTOR by typing ERP myfile.ext at the DOS prompt where myfile.ext is the name of your own database file. You can create your own database file using a simple text editor. Simply follow the format for database.def given above. Of course, you will have to choose your component name/failure rate files from the same set used by RELIABILITY PREDICTOR, but this does allow you to specify a database consisting only of the the classes of components you commonly use. For example, you will not often use both bipolar and MOS microprocessors in a single design and you may rarely use relays or opto-electronics. 13 For those people who have their own failure rate data, a utility called CRELON is available to registered users which allows you to enter your failure rate data into an ASCII file and then convert it to the format used by RELIABILITY PREDICTOR. This same utility was used to create the component name/failure rate files in database.def. If you want to modify or extend the database provided with RELIABILITY PREDICTOR, the original ASCII source component name/failure rate files and a variety of data on more specialised components is available to registered users. 4.1 Component Data To get the most out of RELIABILITY PREDICTOR it is essential to know which component name in which class to use for each of the components in your parts list. The demo version of RELIABILITY PREDICTOR includes 24 classes of components in its database.def file. Wherever possible, RELIABILITY PREDICTOR uses industry standard descriptions of individual component types and in most cases it should be quite obvious which component type applies. However, to avoid confusion, the following sections provide some guidance on selecting component types and classes. In all cases, PL is used to indicate devices in plastic packages and CE to indicate devices in ceramic packages. No distinction is made between leaded and surface mounted devices. 4.1.1 Bipolar Digital ICs This class covers all the popular TTL/ECL families and bipolar ASICs. To convert a particular device type to a number of gates range use the following: - SSI = 1 - 100 gates - MSI = 101 -1000 gates. Many of the popular databooks give a gate count for SSI and MSI devices. - ASICs will usualy be defined directly in gate count - LSI/VLSI devices can be determined by their approximate computing power or complexity. For example a device with a computing power or complexity similar to that of an 8-bit microprocessor will have a gate count around 5000. 4.1.2 MOS Digital ICs This class covers all the popular CMOS families including 4000 series and 74HC devices. To convert a particular device type to a number of gates, use the same rules as for bipolar digital ICs. 14 4.1.3 PLAs and PALs The device data sheets should state the number of gates in the device and of course whether it is a bipolar or MOS device. If gate count is not given, calculate it as follows: - invertors, AND gates, OR gates and output buffers = 1 gate. - D-types, J-K flip-flops and the like = 5 gates. 4.1.4 Microprocessors At present, RELIABILITY PREDICTOR only differentiates between 8,16 and 32 bit microprocessors in bipolar or MOS (or CMOS) technologies. A separate component name/failure rate file covering the more popular devices is planned. 4.1.5 ROMS/RAMS The database supplied covers all popular types and sizes up to the limit of current technology. The data for MOS PROMS is also applicable to EPROMs, EEPROMs and EAPROMs. 4.1.6 Linear ICs This covers both MOS and Bipolar types. Most OP-AMPs and similar linear devices e.g. linear regulators and SMPS control ICs will lie in the 1 - 100 transistors range. Most A/D and D/a devices will lie in the 100-300 transistors range. Few if any linear devices will contain more than 300 transistors. 4.1.7 Transistors Power or high power devices are those dissipating 10 Watts or more. Other types are self explanatory. 4.1.8 Diodes Silicon high power devices are those carrying average currents in excess of 1 Amp. Other types are self explanatory. 15 4.1.9 Opto-electronics The database covers mainly descrete devices. Little data is available for complex devices e.g dot-matrix displays. For multi- digit displays, enter the number of digits in the appropriate 7 segment column. 4.1.10 Resistors The base failure rates for the resistors included in the demo version of database.def are based on resistor types defined in MIL-HDBK-217E which have been modified to take account of the three quality levels available with RELIABILITY PREDICTOR. The handbook refers to two basic classes of resistors - ones of established reliability and ones that are not. The demo database only includes data for resistors which are NOT of established reliability i.e. non-military types. A brief description of each type is given below. Composition RC Refers to ordinary carbon composition resistors Film RL and RN Both these types refer to metal film resistors. RN refers specificaly to high stability types. Film Power RD Refers to metal film resistors rated at 2W or more. Film Network RD Refers to commonly used film resistor networks. You should enter the number of NETWORKS used, not the number of resistors in the network. WW Accurate RB Refers to the very accurate wirewound resistor types sometimes used in precision electronics where resistance values accurate to 0.1% are required. WW Power RW Refers to leaded wirewound resistors rated at greater than 3W and often contained in an insulating cement box. WW Chassis Mounting RE Refers to the very high power resistor types rated at up to 25W and often housed in an aluminium casing. 16 WW Trimmer RT Refers to the normal type of multi-turn wirewound trimmer. WW Precision RR Refers to precision multi-turn low power wirewound potentiometers. WW Semi-precision RA/RK These two types are listed together because they have very similar failure rates. They refer to normal wirewound single turn potentiometers. The RA type is a low operating temperature range type ( up to 70 degrees C ) whereas the RK type operates over an extended temperature range ( up to 150 degrees C). WW Power RP Refers to single turn wirewound potentiometers rated at 2W or more. Non-Wirewound Trimmer RJ Refers to normal multi-turn trimmers of any type other than wirewound. Composition RV Refers to normal single turn non-precision cermet or carbon film potentiometers. Non-Wirewound Precision RQ This type refers to multi-turn cermet or carbon film potentiometers. 4.1.11 Capacitors The base failure rates for the capacitors included on the demo version of database.def are based on capacitor types defined in MIL-HDBK-217E which have been modified to take account of the three quality levels available with RELIABILITY PREDICTOR. The demo database only includes data for capacitors which are NOT of established reliability i.e. non-military types. A brief description of each type is given below. Paper CP/CA Paper capacitors are not used much nowadays but they are included for completeness. Both types have similar failure rates. The CA type is intended primarily for RF bypassing and the CP type is a general purpose capacitor. 17 Paper/Plastic CH This type can be used for all types of plastic capacitor including Polystyrene, Polyester, Polypropylene and Mylar which are suitable for use at an operating temperature up to 85 degrees C. Mica CM This type covers all commonly available Mica types which are suitable for use at an operating temperature of up to 125 degrees C including Silvered Mica. Mica CB and Glass CY This types find very little use today and are included only for completeness. Ceramic CK This type covers all general purpose types of Ceramic capacitor including Metallised Plate, Monolithic, Disc and Feed-through types suitable for use at operating temperatures up to 125 degrees C. Tantalum CL This type covers NON-SOLID Tantalum types only. The more common solid Tantalum (bead) types are approximately twice as unreliable as these. Aluminium CU/CE The CU type covers aluminium oxide electrolytic capacitors and the CE type the dry electrolyte aluminium types. Variable Capacitors Three types of variable capacitor are included in the demo database - ceramic, air trimmer and piston types. The first two types are of similar reliability and can be used for most variable capacitor types. The piston type is about four times as reliable as the others and can usualy be used for such types as UHF tubular trimmers. 18 4.1.12 Transformers/Inductors The demo database includes audio, pulse, power transformers along with RF transformers and coils both fixed and variable. Power, ceramic, ferrite and SAW filters are included here along with various types of discrete and and crystal filter types. 4.1.13 Switches/Connectors The demo database contains a number of self explanatory switch and connector types. Simply enter the quantity of each type used. Interconnect assemblies are simply cables. Again just enter the number of cables used. The same applies to the three types of PCB included in the demo database. However, the four types of connection, hand-solder, flow-solder, wire-wrap and crimp, are treated differently. You need first to add up the TOTAL number of each type of connection and enter this number. The TOTAL includes all connections made to connectors, cables and by components to the PCB. For most designs, this number will be in the hundreds and can easily exceed a thousand. RELIABILITY PREDICTOR can cope with up to 9999 connections of each type. 4.1.14 Miscellaneous The demo database includes motors, alternators, batteries and various other components under the miscellaneous heading all of which are self explanatory. 19 5. Further Reading Many books have been written on the subject of Electronics Reliability Prediction, but there are two I would particularly recommend as an introduction to the topic. They are: 1. Reliability and Maintaiability in Perspective by David J. Smith published by Macmillan. This not only discusses reliability prediction but also contains a number of useful chapters covering such topics as project manangement, product liability, contracts and software reliability. 2. Electronic Equipment Reliability by J.C. Cluley also published by Macmillan. This book concentrates on the definition of reliability, the mathematics involved in calculating and predicting it and the factors affecting component reliability.