Advertisement

Maintaining military superiority in an era of cost constraints

NSC.OCT–National Semiconductor–rp

Maintaining military superiority in an era of cost constraints

An industry study says that commercial devices are not a panacea for
burdensome financial reporting requirements and outmoded specifications

BY RICK CASSIDY National Semiconductor Corp. Santa Clara, CA

Historically, semiconductor technology has been one of the most
important contributors to our overwhelming superiority in defense
electronics. Now, the U.S. may be on the threshold of an era in which
forced economies coupled with outdated government procurement practices
could hobble the ability of the semiconductor industry to support future
defense systems. The Government Procurement Committee of the
Semiconductor Industry Association (SIA/GPC) recently published a white
paper on the subject. The study examined four key issues: cost accounting,
offshore manufacturing, military IC specification; and the potential use
of commercial components in military systems.

Cost accounting Current acquisition laws make it impossible for a
supplier to integrate commercial and military product manufacturing. A
body of restrictive financial reporting requirements fails to acknowledge
the close relationship between commercial devices and processes and their
military derivatives. The core problem is that the term “commercial” is
often misused. It is seen by many to include only the limited-temperature,
plastic-encapsulated devices used in commercial electronics. For military
systems, that definition must be updated. Any device containing a die
designed for, or provided to, nondefense applications should be recognized
as commercial, regardless of what percentage of sales go to defense. A
commercially derived device must be considered commercial even when
additional processing or testing is performed to certify it for military
applications. If acquisition laws could acknowledge that truth, suppliers
would no longer be forced to prepare needless, expensive certified cost
and pricing data. Other changes are needed as well, the white paper says.
The threshold for reporting cost and pricing data must be permanently
raised to $500,000. We The semiconductor industry believes it would make
sense to raise that limit to $1 million for a three-year period to
determine what the impact of such a change would be. This, and other
procurement changes relating to the use of commercial products now being
implemented by DOD, must be done in a manner that allows existing
contracts to be modified so all future subcontracts can be issued under
the new guidelines. Where price justification is required, the
determination of “price reasonableness” must be based on realistic
criteria. Value to the government must be part of that criteria, and the
pricing history of other part types with the same or similar functions
must be considered. Where there are no established market prices,
commercial price analysis techniques should be used, since those are the
techniques used to establish prices in the commercial market for the same
part. This financial streamlining would eliminate a major burden from IC
suppliers. That burden has had two historical effects: it has either
increased the overhead (which must ultimately be absorbed into military
component prices) of those who complied, or driven from the market those
who could not.

Offshore manufacturing Economic pressures have made it difficult for IC
manufacturers to maintain military-only assembly, test, and inspection
facilities in the U.S., the white paper says. The industry has become
truly global. On the commercial side (and frequently on the military side
as well), the industry faces increasing competition from Southeast Asia
and Western Europe. To remain competitive, U.S. firms have been forced to
move first commercial and then non-JAN (Joint Army Navy) Class B military
production offshore. In a very dramatic recent decision, DOD has agreed to
allow the assembly and test of JAN MIL-M-38510 products to be moved
offshore to prevent the diminishment of onshore capabilities from bringing
that important program to an end. Now allowed, that move must be
implemented quickly and smoothly. DOD has also begun a transition from a
system based on QPL (qualified parts list) to one that is QML (qualified
manufacturer list) based. MIL-I-38535 will be the basic specification for
that change. This will provide improved cost effectiveness with no
reduction in reliability. One major advantage that specification offers is
that it permits “best manufacturing practices” rather than artificially
imposing other methods that are no more effective. With this transition, DOD
will issue no new MIL-M-38510 detail specifications. One issue remains
unresolved. Currently, neither QPL nor QML products can be assembled from
dice fabricated in offshore plants. There are no military part numbers
whose volume can by itself sustain a modern high-volume fab module.
International pressures have forced U.S. manufacturers to locate at least
a portion of their fabrication facilities in foreign sites. In some cases,
manufacturers who wish to participate in military business have been
forced to maintain two separate sites for the same product–a practice
which certainly, but unnecessarily, increases costs. The offshore
facilities operate to exactly the same standards as their U.S.
counterparts. Offshore wafer fabrication must become an allowable option
for military ICs. As the cost of establishing and maintaining
wafer-processing facilities increases, a failure to make such an allowance
could drive some suppliers from the military marketplace. Obviously,
completing this offshore transition will require additional education for
all concerned personnel, particularly in the area of “best manufacturing
practices” as they relate to military components. SIA has committed its
resources to the support of that education, knowing that the end result
will be increased value for each dollar spent on military-grade ICs.

Military IC specification Although ICs represent less than 1% of the DOD
budget, they have been the key to the system performance and reliability
that have made U.S. defense technology the best in the world. Several
commercial-based specification initiatives have been proposed. While these
aim at cost savings, their net effect could be to increase the cost of
military system reliability. Recommendations (some coming from within DOD
itself) to eliminate the military IC specification system could raise
overall costs as manufacturers find that lower-priced commercial devices
require additional tests and screens before they can be considered
suitable for the broader temperature ranges and harsher environments that
typify military applications. SIA/GPC strongly supports the continued use
of military-grade ICs produced by suppliers employing “best commercial
practices” in accordance with MIL-I-38535 or a commercial equivalent. But
QML's advantages resolve only part of the problem. MIL-I-38535 does not
provide a standard-parts numbering system, and the IC industry has no
structure for administering such a system on its own. The purpose of
product data sheets today is to establish competitive advantage, not to
provide any standardization between manufacturers. It is not unusual for
two devices from different manufacturers to bear the same “industry part
number” even though one might provide slightly different functions than
the other, or provide the same functions through different pins on the
package. The SMD (Standard Military Drawing) system, currently
administered by DESC (Defense Electronics Supply Center), does provide
standardization, even when it uses specifications that are prepared by
industry. The SMD system must be included within the QML approach if we
the semiconductor industry isare to properly specify ICs for
military/aerospace applications, and if we the semiconductor industry
isare to ensure that those ICs are standardized across a number of
potential suppliers. Without SMDs, we the semiconductor industry would
regress to the cumbersome, expensive proliferation of contractor-prepared
specifications of the early 1980s. SIA/GPC publicly affirmed that product
supplied under a QML system would cost roughly the same as commercial
devices tested to equivalent specifications in equivalent volumes. But
SIA/GPC added two provisos: cost compatibility would only occur if an SMD
system continued to provide standardization, and if the current
infrastructure remained in place to provide test and assembly capability.
Previously, a conference of government representatives, F-22 systems
contractors, and IC suppliers had concluded that QML coupled to the SMD
parts-numbering system was the most cost-effective IC procurement approach
for that program. It is also important that MIL-STD-883 remain the basis
for military IC screening, says the SIA study. Now at revision C,
MIL-STD-883 has contributed to theour technical predominance of the U.S..
Over 85% of the ICs used today in military systems are screened to
MIL-STD-883 specifications. Other industries (such as medical electronics
and commercial avionics) have affirmed its value by adopting it as their
standard as well. Its retention will continue to provide a screening
methodology for those smaller “niche” IC suppliers who may not wish to
pursue, or be able to afford, QML.

Military use of commercial ICs “Commercial” is a poorly defined word,
despite the Packard Commission's recommendation that DOD use “commercial”
technologies. In the IC world, there are no uniform commercial electrical
specifications. Manufacturers can change published specifications almost
at will. They alter their fab processes regularly to improve yields. They
can change molding compounds, lead finish, soldering specifications,
marking permanence requirements, and many other critical characteristics
without notifying anyone other than their own QA departments. This
commercially acceptable practice is not allowable within current military
specifications. The most critical issue with commercial parts is that
their electrical testing is limited. Many parameters are not tested. Those
that are, are tested only at 25 degreesC. The remaining parameters and all
parameters at the temperature extremes (typically 0 degreesC and +70
degreesC) are “guaranteed by design.” No commercial ICs are guaranteed
over the -55 degrees to +125 degreesC military temperature range, which
falls well outside any manufacturer's definition of commercial. Over the
years, the IC industry has developed an extensive infrastructure to
provide the special tests, quality inspections and customer technical
support required by the military/aerospace market. This infrastructure has
already been greatly reduced by the move of JAN offshore and by the
primarily commercial nature of MIL-STD-883-compliant products. If the OEMs
who assemble military systems buy the same commercial-grade ICs used by
non-military customers, that infrastructure will cease to exist in its
present form. Manufacturers will have no incentive to maintain it. But
more than just infrastructure would disappear. Many military-specific
products would be discontinued. Future products and technologies might
prove unsuitable for future military applications because they had not
been designed for military temperature and environments. Use and support
of the SMD system would wither. SMD part numbers would become unavailable,
and all references to them in current systems parts lists would have to be
converted to the individual IC manufacturers' part numbers that OEMs would
then be forced to use. But the harsh military environments and the wide
temperature ranges associated with them would not go away. OEMs would need
to perform substantial additional testing to provide the commercial parts
with a “pedigree” acceptable for those environments. The cost of that
testing and of the associated redocumentation of parts could be greater
than the entire current DOD IC budget.

The package question There is one other catch to “commercial.” Most
military ICs are now provided in hermetic metal or ceramic packages. Most
commercial ICs come in plastic packages whose reliability in high moisture
environments has been proven inferior to that of hermetic packages. In
addition, many plastic packages offer heat dissipation capabilities
inferior to those of hermetic packages–a critical factor at the elevated
temperatures associated with military applications. Until a substantial
database can be developed to support their use in harsh military
environments, plastic ICs should only be used in benign environments (such
as offices), in protected environments, and in areas where maintenance is
easily performed (and where redundancy can be incorporated into the system
design). Their use might best be supported through the implementation of a
military-grade plastic processing program–for example, MIL-I-38535 allows
such a program–with electrical specifications tied to SMDs.

Automotive-grade ICs Some proposals have suggested the use of automotive
ICs for military purposes. However, there is no “standard” automotive IC.
Each auto manufacturer maintains several screening flows, each tied to a
different set of applications. Those flows focus on large volumes of a
small number of device types, many of them custom designed for unique
automotive functions. Their reliability goals aim more at the minimization
of customer inconvenience than at the elimination of failures. The low
prices typically associated with automotive ICs are based on enormous
volumes–typically millions of a given part type per year. Screened in the
volumes typical of military applications, they would be priced comparably
with military parts. One key point is that almost none of the parts now used
in the dozens of military systems currently in production are available
from any of the limited lines of automotive products.

SIA recommendations and actions SIA/GPC believes that SIA, DOD, OEMs,
and the individual armed services must immediately establish a monthly
working group whose 18-month objective will be to complete and begin
implementing an IC procurement program. That program would include among
its elements an expedited implementation of the already approved transfer
of MIL-M-38510 assembly and test to offshore plants. Also important (and
now under consideration at DOD) is the allowability of offshore
fabrication for military dice. The working group would also push for a
quick decision on whether a military or a commercial version of
MIL-I-38535 would define the long term system of supply for both hermetic
and plastic ICs. MIL-STD-883 would continue to serve as the document
around which changes to the system would be developed, and the Military IC
Applications Handbook update now in process would have to be completed. A
partnership must be established between industry and DESC to maintain a
device specification program equivalent to today's SMD system. That, and
any other actions taken, should support a rational approach to the use of
commercial ICs in defense applications. SIA/GPC has already begun working
with the Defense Microcircuit Planning Group to further some of these
objectives through the preparation and submission of recommendations to
DOD management.

Advertisement

Leave a Reply