The vulnerabilities of COTS Components

I read Wilson Dizard`s article on the Navy`s testing of COTS boards with some interest. The article, entitled "COTS boards OK`d for helicopters; but enclosures still need work," which appeared in the September 1997 issue of Military & Aerospace Electronics, certainly proved the viability of COTS boards in harsh applications. However, it does need two critical footnotes.

To The Editor:

I read Wilson Dizard`s article on the Navy`s testing of COTS boards with some interest. The article, entitled "COTS boards OK`d for helicopters; but enclosures still need work," which appeared in the September 1997 issue of Military & Aerospace Electronics, certainly proved the viability of COTS boards in harsh applications. However, it does need two critical footnotes.

First, there has been a blind rush in some sectors to limit the definition of COTS to commercial-grade components. The testing in this case was done on boards populated with devices which were screened by the board manufacturer for extended temperature. While the board may be COTS (it is available off the shelf), the components on it are not. They cannot be replaced with the manufacturer` s off-the-shelf devices without risk. In some cases, the risk may be so near zero as to be negligible, but no one other than the device` s manufacturer can make that determination. Many IC firms now state that they will not guarantee or warranty any ICs used outside the limits (including temperature) for which they have been specified and tested, regardless of what additional testing the user or the user` s supplier may perform. I won`t cite all the reasons for this, but it is a valid position. If extended temperature operation is needed, the parts used should be those certified by their manufacturers as capable of operation at that temperature for extended periods. No one-time test can accomplish that.

Also, the fact that the conduction-cooled enclosure`s cold wall reached temperatures as high as 97 degrees Celsius has two implications. First, the temperature at the device will be higher than it is at the cold wall (possibly by as much as 25 degrees, depending on the power of the ICs and the conduction curve of the enclosure). Second, device heating in a convection-cooled enclosure may be worse. Someone might suggest this can be controlled by restricting the operating periods of the equipment, but, when the need for reliability becomes greatest (in a real combat situation), it will be impossible to curtail operating time without risking the mission, the platform, and our uniformed personnel. This clearly underlines the wisdom of the time-honored 125 C maximum operating temperature specification for devices used in combat aircraft (and in any other equipment in applications where cooling is difficult). There are COTS ICs which meet that requirement. The Navy`s testing proves they should be the components of choice.

I support acquisition reform. The increased use of COTS will give DOD more for its procurement dollar. In the `70s, when I championed COTS (through standardization), I was attacked by many of the same people who now tout upscreened parts as COTS. I do not support cutting corners in a futile attempt to pursue bad definitions of COTS. I do not support jeopardizing our personnel and their missions as a result of that bad definition.

Jim Martin

independent consultant

Scotts Valley, Calif.

More in Home