GM Recall: the Ignition Switch

Ignition switch. The two words that cause heart palpitations and cold sweats among GM management, engineers and lawyers. The one small part that will ultimately cost GM over $9 billion. What is it, and what went wrong?

The Valukas Report begins with the following sentence: “In the fall of 2002, General Motors personnel made a decision that would lead to catastrophic results – a GM engineer chose to use an ignition switch in certain cars that was so far below GM’s own specifications that it failed to keep the car powered on in circumstances that drivers could encounter, resulting in moving stalls on the highway as well as loss of power on rough terrain a driver might confront moments before a crash.”

GM’s intent with the new ignition switch it began designing in 1997 was to have a less expensive and less likely to fail part that it could use across multiple vehicles (referred to as “corporate common”). The ignition switch was designed to communicate the position of the key to other vehicle components through a new “electrical architecture” that was supposed to be more reliable.

GM engineers known as Design Release Engineers (DREs) drafted specifications for the ignition switch and then solicited bids from potential suppliers. The specification included a target “force displacement curve specifying 20 Newton-centimeters as the torque needed to turn the ignition from Run to Accessory.” In other words, the 20 N-cm measured what GM described as its target for the amount of force required to turn the key from the Run position to the Accessory position.

Ray DeGiorgio, a DRE, took over responsibility for the ignition switch sometime between fall 1999 and March 2001 (accounts differ), and he was the primary point of contact between GM and Eaton, the supplier it selected to develop and produce the switch. (Eaton was later acquired by Delphi.)  DeGiorgio was the third DRE assigned to the switch, as the other engineers – including the original author of the specification – were assigned to different positions.

DeGiorgio worked with Eaton/Delphi to update and refine the ignition switch specification (which included all technical parameters for the switch). According to the Valukas Report, early torque tests performed by Eaton/Delphi indicated that the prototype switch it was working on did not meet the torque requirements in GM’s specification, even though DeGiorgio had changed the torque specification to be within +/- 5 N-cm (meaning anything from 15 to 25 N-cm was acceptable). Nevertheless, DeGiorgio designated the specification as “finalized” in early 2001, meaning that no more changes would be made to the switch design prior to production. According to Valukas, “there is no question” that DeGiorgio knew he was approving a design that did not meet GM’s specifications.

The new ignition switch was intended for the upcoming Saturn Ion and Chevrolet Cobalt (and ultimately several other vehicles as well).  The Ion was to be launched first, so GM first tested the switch in a prototype Ion. In GM’s vehicle development process, issues were noted through the Problem Resolution Tracking System (PRTS). Each issue is coded with a severity level, which determines the urgency with which the issue is addressed. PRTS records on the switch included DeGiorgio’s note that the electrical design had “failed miserably.” DeGiorgio also signed one e-mail to Delphi “tired of the switch from hell.”

Delphi, the supplier, also conducted tests during the development of the switch. Delphi’s testing in late 2001 and early 2002 demonstrated that the ignition switch consistently failed to meet the torque specifications. One Delphi engineer brought this to DeGiorgio’s attention, with the observation that Delphi could increase the torque, but at the risk of causing other issues with the switch’s design and perhaps leading to a delay in getting the switch into production. DeGiorgio’s response: “maintain present course.”

Despite the fact that the ignition switch had miserably failed Delphi’s torque tests (with results as low as 4 N-cm), GM put the switch through its Production Part Approval Process (PPAP) in May 2002. According to Valukas, “PPAP is the process through which parts are tested, validated, and ultimately released for production. The PPAP package should contain, among other things, two key authorizations: one from the supplier and one from GM.” GM then indicates its approval of a part by signing off on Form 3660.

No one has been able to locate a complete PPAP package for the ignition switch.

Nevertheless, aware that the switch did not meet the torque specification, DeGiorgio approved the switch for production, and it was installed in Model Year 2003 Saturn Ions beginning in the fall of 2002.

Next up: the Saturn Ion and Chevy Cobalt in production.

See Part One of the GM Recall Series: Cast of Characters

See Part Two of the GM Recall Series: the Valukas Report – What It Is (and What It Isn’t)

See Part Three of the GM Recall Series: My Article in Texas Lawyer