3. CAR DOOR SYSTEM LIN BASED HARNESS – CASE STUDY
3.2. Case Study – Main Outcomes
This work reported in this submission two and described in this chapter developed the
preliminary design to cost process. This can be summarised as a process for
comparing a hardwired original electrical architecture with a LIN based alternative
and ascertaining a cost target for adding LIN communications capability to electrical
architecture nodes. Key steps were:
1. develop an appropriate wire model for the wire type (intra- and inter-zone in this
case)
2. ascertain the number of wires and types that are to be added or removed (i.e. by
signal substitution analysis)
3. analyse the sensitivity for estimating the target add-on nodal cost to be achieved
for the LIN candidate to be the same price or lower than the hardwired original
The methodology of the analysis presented focused on a comparison of the
costs of the bill of materials of two different methods of electrical system integration,
in this case hardwired and LIN based architectures. It is important to note that it was
not a methodology for cost estimation but a methodology for comparing two
architectures. Cost estimation itself would require the designers to go a lot further
of the research presented was to establish a methodology with which candidate
electrical architectures can be realistically compared on a cost basis at a very early
stage of design. Wire cost was investigated and this led to the identification of a
target LIN node cost that should be met to help ensure that the LIN architecture will
be of lower cost than the hardwired alternative. Historical wire cost data was studied
but there was no data available on the LIN nodal costing or some of the other costs
that might affect the choice of architecture. Therefore a number of areas of further
work were recommended.
The architecture cost comparison methodology presented did not take account
of other costs such as non-recoverable software development costs associated with
LIN, cross platform portability (which would amortise the costs across a number of
platforms), vehicle assembly, warranty, weight savings (which will ultimately reduce
costs for both the customer in terms of fuel consumption and the manufacturer in
terms of CO2 based taxation). Making a justification for the use of LIN over a
hardwired architecture, based on the cost of bill of materials alone, appeared to be
very challenging and as it resulted in very low target nodal costs. Therefore there is
motivation for research into the effects of other factors on both LIN and hardwired
architectures.
Use of Process
A key point about the use of the process developed is it cannot be used to predict the
cost of an alternative architecture based on a particular harness. However it does
have two other applications; estimation of target nodal cost or estimation of
If the target nodal cost for adding LIN communications (CNode) is not known,
the process can be used to estimate the target CNode that must be met for the LIN
candidate architecture to be the same cost as the hardwired original. This was the
process followed for the mass vehicle wiring harness manufacturer case study that
was described in submission two.
It can also be used to estimate whether a move from hardwired integration to
LIN would result in a cost reduction or increase. This is the case if CNode is known,
e.g. if legacy ECUs are already available and are to be used in a new architecture
design. Use of the wire model information and CNode values will give delta values
from the architecture equations.
It was recommended that further work should include an investigation of how
target nodal cost can be achieved and how it varies with changes in architecture and
signal requirements. The aim of this recommendation was to help ascertain whether
the target nodal cost values obtained within this submission are realistic and
achievable (e.g. 1.61 to 1.9 Euros). It was stated that it would be interesting to
understand the cost of FETs, relays, switches and the size of application software and
how this affects the target cost of a network node. This recommendation gave
motivation to the work in submissions three and four since they helped uncover some
of the cost implications of LIN communications and embedded software generally.
Submission three investigated the relationship between microcontroller ROM
and RAM capability and microcontroller unit cost. Data from the semiconductor
manufacturer Microchip for their PIC16 and PIC18 microcontroller families was
investigated. It was found that there was a relationship between cost and ROM/RAM
capability with the implication being that an increase in the LIN communications
software ROM and RAM requirements. Therefore an increase in the LIN
communications embedded software ROM and RAM requirements would result in
increased microcontroller cost. One other output from submission three was that it
was discovered that there tends to be a linear relationship between the ROM and
RAM capability across a microcontroller family. This therefore provided the basis of
a microcontroller capability assessment process for assessing how well it is suited for
a particular piece of embedded software (e.g. in this case a LIN communications
stack). This is described further in chapter 5.
The relationship between LIN communications requirements and LIN
communications software stack ROM/RAM requirements, was explored further in
submission four. It was found that it was possible to estimate ROM/RAM
requirements of the LIN communications stack from information such as number of
nodes, signals, messages etc. using regression models. The process for this is
described in chapter 5.