Scan for ESD-induced errors
By Martin Rowe, Senior Technical Editor -- Test & Measurement World, 8/1/2006
An electrostatic discharge (ESD) from a person touching an electronic device or system can produce thousands of volts and several amperes, which can easily damage or destroy an integrated circuit (IC). A damaged device is often easy to find with a visual inspection, but ESD can produce secondary effects that are difficult to troubleshoot. A team consisting of Professor David Pommerenke and two students at the University of Missouri-Rolla's EMC Laboratory (www.emclab.umr.edu) developed a tool that can measure the impact of secondary electromagnetic interference (EMI) effects from ESD. The scanning system can help engineers find where and how secondary ESD events cause bit errors in digital systems (Ref. 1).
 |
Figure 1. A PCB scan indicates areas that are susceptible to secondary ESD effects. Red indicates highest susceptibility.
Courtesy of University of Missouri-Rolla. |
An ESD event produces radiated EMI in the form of E-fields and H-fields. The fields can then couple into printed-circuit board (PCB) traces and IC pins, turning back into current. Current in a PCB trace can produce a voltage sufficiently large enough and long enough to cause a bit error without damaging an IC. A PCB with an embedded processor, for example, may continue to function after the bit error, or it may crash, requiring a restart.
Finding the point or points where induced current couples into a system is difficult because radiated fields can travel anywhere within a system's enclosure. To identify EMI-susceptible areas on a PCB, the UMR scanning system scans a probe across a PCB, producing localized EMI and tracking the PCB's response. The system can use an E-field probe, vertical and horizontal H-field probes, and a direct-contact probe to inject a localized disturbance into the PCB. The E-field and H-field probes produce fields from an ESD pulse and radiate them into PCB traces and IC pins, whereas the direct-contact probe injects an ESD pulse directly to the point of interest. As the system scans the probe across the PCB and produces EMI or ESD, software monitors the effects of the disturbances and plots them on a map using color to indicate the intensity of an error (Figure 1). The system is, in effect, the opposite of a radiated EMI scan that measures radiated emissions from a PCB.
 |
Figure 2. A TLP generator produces a repeatable pulse for ESD testing.
Courtesy of University of Missouri-Rolla. |
To create the localized EMI or ESD, the system generates a series of transmission-line pulses (TLPs) as it scans across a PCB. A TLP generator produces rectangular pulses with a 900-ps rise time and a peak voltage of up to 5000 V. TLP pulses hold their peak level longer than the Human Body Model (HBM) pulses produced by handheld ESD simulators (Ref. 2). TLP generators also produce more repeatable pulses (Figure 2). Measured at a 50-ohm terminated trace, the pulse is about 200 ns wide at the half-amplitude point.
The researchers discovered that ESD-induced EMI can affect a system even on signal lines that carry slow digital signals such as status lines. PCB designers usually use short traces to carry high-frequency signals such as clocks and serial data streams, but they may use longer runs for slowly changing signals. Unfortunately, the longer the trace, the better it acts as a receiving antenna for stray EMI. The research team learned this lesson when a scan revealed a trace that, when subjected to pulse EMI, caused soft errors to occur. They fixed the problem by adding an R-C filter to the trace. The filter removed energy from the ESD-induced EMI. It also slowed the wanted signals, but not enough to affect system operation.
| REFERENCES |
- Pommerenke, David, Jayong Koo, and Giorgi Muchaidze, "Finding the root cause of an ESD upset event," DesignCon 2006, International Engineering Consortium, Chicago, IL. web.umr.edu/~davidjp/paper/Pommerenke_DesignCom2006.pdf.
- Rowe, Martin, "TLP testing gains momentum," Test & Measurement World, September 2002. p. 37. www.tmworld.com/2002_09.
|
