

NAVSO P3634: Sneak Circuit Analysis: A Means of Verifying Design Integrity 
 

3.3 Estimating the Cost of SCA
As with many analytical efforts, cost estimating for an SCA can be a difficult task, since a number of factors contribute to the complexity of the cost estimating process. Variations in SCA techniques can lead to significant differences in estimates for the analysis. Further, differences in the methods of tailoring the analysis (i.e., selection of partitioning schemes, application of different techniques and different levels of analysis to certain subsystems) can cause wide variations in cost estimates for what is apparently the same job. Other factors such as accessibility of the data base, government coordination and review, travel, independent technical consultant services, frequency of reports, and specific contract data requirements will also affect the price estimated for the SCA.
SCA FOR HARDWARE SYSTEMS
Several methods for developing cost estimates for SCA have been proposed. In all cases, the parameter of prime interest is system complexitythe more complex the system, the greater the SCA cost.
An SCA may be conducted at two levels of detailthe system or the black box. At the system level, emphasis is on the interconnections between subassemblies. The blackbox level of analysis examines the individual subsystems for sneak conditions. Obviously, the depth to which the analysis is carried out will affect the cost of the effort. In the following paragraphs, estimation procedures for the blackbox and system levels of analysis will be addressed.
BlackBoxLevel SCA
One method of cost estimation is based on a study of 111 projects in which SCA was used. A sample of these projects was presented in Table 3 . From the cost data accumulated and a knowledge of the number of
component parts in the system or equipment analyzed, a costcomplexity
relationship was developed. It is shown in Figure 4. While this curve yields
only "rough order of magnitude" estimates, it does provide useful inputs to
the decision process. The curve in Figure 4 assumes some generalized mix of
hardware component types. When the distribution of parts by type is known, a
more refined estimate may be made by use of information in Table 4. These data
indicate the cost associated with the analysis of one part of a particular
type and a tolerance level for each cost figure. From this information, SCA
cost estimates can be made for the total number of parts of each type, as
shown in the sample calculations in Table 5. (See Reference 1).
FIGURE 4. ROUGHORDEROFMAGNITUDE COST ESTIMATE FOR BLACKBOXLEVEL SCA
TABLE 4. SCA COST FACTORS FOR DIFFERENT PART TYPES
Part Type 

Weighting Tolerance ($ per part) 

Weighting Factor Tolerance ($ per part) 

Resistors, Capacitors, Coils Relays, Transistors, Switches SmallScale Integrated Circuits (SSI) MediumScale Integrated Circuits (MSI) LargeScale Integrated Circuits (LSI) Generalized Component Mix (Used when actual component mix is not known) 

29 79 164 284 468
94 

± 8 ± 11 ± 14 ± 14 ± 25
± 19  
TABLE 5. SAMPLE CALCULATIONS OF SCA COSTS FOR DIFFERENT PART
TYPES
Part Type 
Number of Parts 
X 
Weighting Factor 
= 
Component Cost 
Cost Tolerance 

Resistors, Capacitors, Coils 
400 
X 
$ 29 / Part 
= 
$ 11,600 
± $3,200 

Relays, Transistors, Switches 
200 
X 
79 / Part 
= 
15,800 
± 2,200 

Small Scale Integrated Circuits (SSI) 
150 
X 
164 / Part 
= 
24,600 
± 2,100 

Medium Scale Integrated Circuits (MSI) 
100 
X 
284 / Part 
= 
28,400 
± 1,400 

Large Scale Integrated Circuits 
50 
X 
468 / Part 
= 
23,400 
± 1,250 

TOTALS 




$ 103,800 
± $ 10,150 

System and BlackBoxLevel SCA
Another method of SCA cost analysis which can be used is based on a
relatively quick and simple method for developing a "rough" cost estimate for
systemlevel and blackboxlevel analysis. For a given SCA task, a few general
rules are used to help yield an orderofmagnitude estimate of the required
man hours. This method of calculation assumes that experienced SCA personnel
will perform the analysis. The general rules are listed below.
For the blackbox level of analysis
 Each of the individual elements (e. g., resistors,
capacitors, transistors, switches. integrated circuits) to be treated is
assigned one to two hours of engineering effort, depending on the complexity
of the device.
 Each input/output connection (circuit board to circuit board, and
boxtobox) is assigned two hours of engineering effort.
 The hours are summed to obtain a rough idea of the
effort involved in a blackboxlevel analysis in which a detailed
examination is made of each circuit board.
For the system level of analysis
 The estimate made for the blackbox effort is
divided by two. (This accounts for the fact that all input/output interfaces
are considered but only some of the internal elements are modeled.)
 The cost estimating techniques presented above are applicable to
hardware analyses using the network tree approach. If a different approach
to the SCA is anticipated, the estimate would have to be based on the
individual tasks to be performed.
SOFTWARE SNEAK PATH ANALYSIS
Cost estimation for a software sneak path analysis also depends on
complexity of the subject of the analysis. In this case, the measure of
complexity is the number of executable lines of assembly language
instructions. One procedure sometimes used, suggests a cost of $10.00 for each
line of executable instructions.
PERIOD OF PERFORMANCE
Also of interest to the manager is the time required to
perform an SCA. The time required is obviously related to the extent of the
analysis, and therefore, to the cost of the analysis. In Reference 2, a
relationship between SCA cost and the duration of the SCA was developed. The
result is shown in Figure 5 to provide a basis for estimating the time
required to perform an SCA.
FIGURE 5. SNEAK ANALYSIS
PERIOD OF PERFORMANCE
BENEFITS VS. COSTS
Whatever the method of cost estimation, and whether
the subject of analysis is hardware or software, costs must be carefully
weighed against the possible benefits that system programs can derive from
sneak circuit analysis. As called out in Section 3.1
, these benefits are
 Detection of potentially serious system problems
 Discovery of design oversights
 Discovery of documentation errors
 Reductions in systemchange costs and program
disruptions
 Improvements to system reliability and safety
 Reduction in testing and analysis requirements
 Benefits to other analyses
The time and effort devoted to an effective sneak circuit analysis can
result in significant positive effects on a projecteffects that might not be
achieved as costeffectively through any other
approach.




 
 