10.2.1 Examples

Most people are familiar with static electric discharges on cold winter days or other especially dry conditions. These discharges, although only unpleasant to humans, may be fatal to modern electronic products; further, weak discharges that are undetectable to humans can still damage electronic products.^[1] What follows is a selection of ESD-related events associated with consumer electronic products, illustrating the importance of the problem and the variety of guises in which it may appear.

• A new product was developed one spring, and volume shipments began the following summer. That winter, however, field failures, in the form of random access memory (RAM) errors, were noted in temperate regions. The problem was believed to be ESD-related, and became serious when the RAM errors could not be duplicated in the manufacturer's laboratories, because the problem could not be corrected until the product was made to fail under laboratory conditions. Although many attempted design improvements were tried in the field, the field failures did not stop until spring. It became apparent that the ESD test procedure as performed by the manufacturer at that time was inadequate, and a task force was organized to identify improvements. The recommendations of the task force, including an environmentally controlled ESD testing room and a new testing procedure based on internationally accepted standards, were adopted after the new test accurately duplicated not only the field failure of this product (which was cured by a bypass capacitor on a RAM integrated circuit [IC]), but also the warranty failures of several other products. (ESD had not even been considered as a possible root cause of the failures in the other products.) By the next winter, products of modified design were in the field, and there was no seasonal rise in ESD-related failures. The cost of the new ESD testing room and equipment was more than repaid by the improvement in the manufacturer's reputation among its customers.

• During development of a product, several problems were uncovered during ESD testing. It was noted that the switching voltage converter IC on a microcontroller circuit board would go into a test mode when an ESD discharge reached a circuit board runner through a gap in the housing around the "5" key on the product's keypad. An added capacitor and diode to the runner cleared this problem. It was next noticed that the RAM memory locations storing trim values for analog circuits would become corrupted when a discharge occurred on the serial port contacts used to program the microcontroller, through another housing hole. Placing a diode between the two contacts and placing spark gaps (pointed circuit board runner shapes) between the runners to the contacts fixed the RAM problem; instead, a discharge to the serial port contacts would now cause the microcontroller to reset. Making the serial port runners thin to increase their inductance and modifying the ground metal beneath them, fixed this problem.

Now, after a discharge to the serial port contacts, the product's real time clock (RTC) appeared to stop — but the product would continue to work normally in every other way. It was found that the microcontroller had two programming bits that controlled the RTC increment rate. The RTC could be set to increment either once per second, once per minute, once per hour, or to not increment at all. The ESD event was somehow corrupting these bits so that the RTC would not increment, and, therefore, appear to stop. Working with the designers of the microcontroller, the product engineer assigned to this problem determined that the RTC was supplied power through a particular supply pin, the V_rr pin. A 150-Ω resistor in series with the V_rr pin (made possible due to the very low current drain of the RTC) fixed it. To speed testing, special product firmware was written during this part of the investigation that constantly checked the value of the sensitive bits. If a change were noted, a light-emitting diode (LED) on the product would turn on.

10.2.2 Failure Modes

• Long-term failures and latent defects. Long-term failures, by definition, do not become apparent until some time after the ESD event.^[2] The most common long-term failure involves CMOS ICs, which may have a shorter operating life after an ESD event, due to gate oxide degradation. The noise figure of radio frequency (RF) transistors may also increase over time after ESD exposure. It has also been demonstrated that ESD events during manufacturing processes produce latent defects in Schottky diodes used in microwave transceivers, leading to a loss of sensitivity after a few months of operation.^[3]

• Short-term failures. Short-term failures become apparent immediately following the ESD event. They can have a confusing, application-or market-specific taxonomy; in particular, the distinction between catastrophic ("hard") versus noncatastrophic ("soft") failures can be vague. Although one may debate the classification of a particular product failure, the point to remember is that all failures are failures, from a quality and customer satisfaction point of view (the only point of view that matters). It is rare to find a user interested in entering into a lengthy discussion with a manufacturer on the precise categorization of a product failure due to ESD.

  • Catastrophic failures. A catastrophic failure is, simply put, any failure from which the network node cannot be made to recover by action taken over the air or at the user interface. This includes transceiver parametric failures, such as loss of receiver sensitivity, and failure of any internal sensors and actuators. Typically, catastrophic failures involve physical component damage, and require the replacement of hardware (e.g., integrated circuits) to return the network node to its proper operating condition. These types of failures are those most commonly encountered when reading the ESD literature.

  • Noncatastrophic failures. Noncatastrophic failures, also called system upsets, include such transient phenomena as inadvertent turn-off, improper operation of the transceiver (partial or complete transmission of inappropriate messages, for example), improper operation of any internal sensor or actuator, low battery indication, or system reset (which can include loss of microcontroller state, loss of volatile memory, and system turn-off). Of these undesirable events, system turn-off is particularly undesirable. If the user (or network, if automatic network monitoring is in place) does not notice the ESD event and its effect on the node, the user may not turn the node back on and so miss future messages. Because such future messages could be to inform the user of a fire alarm, breaking window, or other safety-related event, the loss of future messages may not be "noncatastrophic." Even if the wireless sensor network is not performing such a critical function, if the node suffering the ESD failure is the master of a star network, or performing some other centralized function, loss of the node can lead to a failure of the network. Noncatastrophic failures should be considered just as seriously as catastrophic failures.

  • Data loss. Loss of data falls in the gray area between catastrophic and noncatastrophic failure. The loss of messages queued for transmission, unprocessed sensor data, or received data not stored in nonvolatile memory, although unfortunate and undesirable, probably falls in the "noncatastrophic" category. The loss of factory tuning values for the transceiver or calibration values for a sensor, which renders the network node inoperable, probably falls in the "catastrophic" category. The loss of security keys needed to communicate with other network nodes via symmetric-key cryptography may or may not be considered catastrophic, depending on whether or not a mechanism exists by which the user may enter keys into the node. (A failure of a memory chip itself, so that it is no longer capable of storing information, is clearly a catastrophic failure.)