TLP ESD Testing of Electronic Components

Leo G. Henry

A review of the scope requirements in the recently published standard practice document raises questions.

Since the release of the ESD Association's Transmission Line Pulse (TLP) ESD Standard Practice (SP) document in September 2003, the questions being asked about TLP testing have shown a simplistic trend.1 Design, test, and product engineers, as well as failure analysts and technicians are asking some basic questions. One such question revolves around the oscilloscope requirements for not only recording the waveforms from the TLP tester (initial pulse verification), but also for recording the electrical parameters from the device under test (DUT).

This standard practice document (ESD SP5.5.1-2003) provides some guidance for the scope requirements. The document, "Electrostatic Discharge Sensitivity Testing Transmission Line Pulse (TLP) Component Level," defines the TLP as a rectangular current pulse formed by discharging a transmission line. It further defines the TLP test system as one that applies a rectangular pulse to a DUT and allows measurement of the device's electrical characteristics during a pulsed state. These electrical characteristics include the current, voltage, and leakage (IVL) data from the DUT at the various pulsed voltages.2

In an earlier article, TLP is referred to as an engineering characterization tool or a development tool.3 It can be shown to represent or simulate most, if not all, of the many ESD events and models like human body model (HBM), machine model (MM), and charged device model (CDM). TLP testing of integrated circuits (ICs) first appeared in the literature when Maloney and Khurana presented their unique approach to testing the ESD susceptibility of the IC core's protection structures.4 Their two-pin stress-testing technique was similar in some ways to that of the existing test method in the HBM standard. However, the way the HBM test was developed limited it to a pass-or-fail test. The data obtained from TLP stress are more useful to design and product engineers for development and design purposes. TLP uses rectangular-pulse testing (RPT) or square-pulse testing (SPT) to simulate the energy in an exponential HBM test pulse.

The Specifications

The measurement requirements in the SP document are as follows: The oscilloscope (o-scope) must have a minimum single-shot bandwidth of 500 MHz. The current probe must have a minimum bandwidth of 1.0 GHz. The pulse width is specified as 100 nanoseconds wide, and the TLP tester rise time range is from 0.20 to 10 nanoseconds for both the voltage (open load) measurements and current (short load) measurements. The document also specifies that a sampling rate of 5 GSa/sec for the 500-MHz scope is sufficient because 100 data points exists in a 20-second measurement window of the pulse. In this case, the document is referring to the flattop area of the current or voltage pulses from the DUT. This flattop area is used to collect the I-V data. The sampling rate must be such that it captures enough data points in the measurement window to secure accuracy. If the sampling rate is insufficient, aliasing will occur; that is, the number of data points sampled is too low for the waveform to be faithfully reproduced.5

Bandwidth

The bandwidth specification is important because the scopes measure the rise time, the voltage (and/or current), and the pulse width. If the bandwidth of the scope is not adequate to properly reproduce the pulse applied to the input of the scope, errors will be introduced. These errors will occur not only in the time domain, but also in the voltage and current measurement of the pulse height.6

Figure 1. Operating bandwidth region and -3-dB point of the o-scope.

The bandwidth specification tells the frequency range that the scope is capable of measuring accurately. The bandwidth tells the frequency at which the displayed signal reduces to 70.7% of the applied signal. This 70.7% point is the well-known -3-dB point referenced in many texts (see Figure 1). The amplitude of the signal is attenuated by approximately 30% at this frequency (BW) point. As the frequency of the input pulse increases above this frequency point, the ability of the scope to accurately respond decreases as is shown by the rolloff in Figure 1. It follows then that an o-scope cannot accurately display input pulses with rise times faster than the specified rise time of that oscilloscope.

The Rise Time

Rise time is a parameter that reflects a storage instrument's ability to accurately record fast pulses. It is another way of describing the useful frequency range of an oscilloscope. The rise time, Tr, of the scope, if a Gaussian response is assumed, is determined as follows:

or

So, a 500 MHz BW o-scope will have a theoretical rise time of 700 picoseconds (0.700 nanoseconds), and a 3.5 GHz BW o-scope will have a theoretical rise time of 100 picoseconds (0.100 nanoseconds). A simple, but well-established equation is used to define what the o-scope actually measures, and what is observed on the screen or window of the scope.

It is important to note that the oscilloscope rise time includes the rise time of the current or voltage probe used.6 The probe should ideally be nonintrusive and should not introduce resistive, capacitive, or inductive loading to the circuit being measured.7 This probe's Tr must be considered because it can be a source of large error. The general rule of thumb is that the probe's Tr must be faster than the Tr of the o-scope.8

Error Calculations

An example of the error associated with the o-scope measurement is as follows. Consider the 2001 ESDA HBM standard test method (STM) document, ESDA-HBM-STM-5.1.9 This STM document specifies that the fastest HBM rise time to be measured from the tester is 2 nanoseconds and the slowest is 10 nanoseconds. Using Equation 3, the measured value is 2.83 nanoseconds.

where S is the input signal pulse and O is the o-scope. If a 175 MHz (2 nanoseconds Tr) o-scope is used to measure the 2-nanoseconds input pulse, the result is a 41% error (0.83/2.0) in the rise time. However, if the higher bandwidth 500 MHz (0.700 nanosecond) o-scope is used to measure the same 2.0 nanoseconds Tr, then the measured Tr = 2.12 nanoseconds, and the error is reduced to 6%. If a 10-nanoseconds input pulse is then measured using the same 500 MHz o-scope, the measured Tr = 10.02 nanoseconds, and the error is further reduced to 0.25%. The trend is obvious and expected. The slower the input signal Tr, the lower the error in the measurement.

Table I shows that 500 MHz scope meets the 2-10 nanoseconds rise time specification per the 2003 ESD DSP5.5.1 document. The error in the measurement is 6%, and it is lower for the whole range. However, below this 2-nanoseconds minimum, the errors (22%) are huge as is seen for the 1-nanosecond Tr. The scope's 264% error at the SP's 0.20 nanosecond Tr should not even be considered. The limit of the scope is clearly demonstrated.

Table I compares the errors associated with the rise times and relevant scope used to measure the same input signal. It is clear that the 350 MHz and the 175 MHz o-scopes are not the best suited for the 2-10 Tr measurements and probably should not be used to measure the voltage and current from the DUT. The 1.0 GHz BW o-scope is also adequate for the 2-10 nsec range specified in the document, but even this apparently high bandwidth o-scope has a huge measurement error (102%) at the 0.2-nanosecond rise time.

The last section of Table I shows the change (decrease) in error as the rise time of the scope chosen gets faster. It appears that the higher bandwidth 5.0 GHz o-scope is the one ideal for measuring the rise time of the 200-picosecond input signal. It is helpful to create a table for the range (10 to 0.20 nanoseconds) specified in the new TLP document. The acceptable level of error is dependent on the needs of the individual.

Table II. The errors associated with the 2-nanosecond measurement.

Table II shows the error associated with the measurement of the 2-nanosecond input signal. Column 5 shows the actual error, and column 4 shows the general rule of thumb for the ratio of the o-scope rise time to the signal rise time. The table clearly shows that the 500 MHz scope provides the lower limit for acceptable rise time measurement accuracy.

Rule of Thumb Considerations

In the literature as well as in manufacturers' (e.g., TEK, HP and LeCroy) technical notes, the rule of thumb ratio ranges from 3:1 to 5:1; that is, the Tr of the measuring system or scope is three to five times faster than the Tr of the event being measured (the input signal).10 The 5:1 rule has a maximum allowable error of 2% and avoids most of the rolloff region shown in Figure 1. The 3:1 rule has a maximum allowable error of 5%, but includes some of the rolloff regions described in Figure 1.

Table III. The bandwidth required for the 5:1 (2%) rule-of-thumb ratio.

Another way to present the same idea is as follows: for a 5% error, the Tr system is <1/3 Tr of the signal being measured, and for a 2% error, the Tr of the measuring system is <1/5 Tr of the signal being measured. Table III indicates the best (theoretically) o-scope to use to measure each rise time in the range as specified in the ESD SP5.1.1 document.
Table IV shows the best o-scope to use to measure each rise time in the range of rise times as specified in the 2003 document.

The rule of thumb changes depending on the technical note or published article being cited. When the 5:1 ratio with the 2% error is recommended as the minimum, the 3:1 ratio with the 5% error is the recommended maximum error. In fact, Tables III and IV are only of major concern if the intention is to calibrate the TLP tester rise times in addition to verification of the SOLZ (SORZ). When possible, it is best to leave the calibration to the manufacturer.

Furthermore, some of the scopes mentioned in Tables III and IV do not actually exist commercially. They are included in the tables simply to provide the theoretical rule-of-thumb relationship. Therefore, the closest bandwidth would have to be used (e.g., an existing 500 MHz for the theoretical 525 MHz).

Table IV. The bandwidth required for the 3:1 (5%) rule-of-thumb ratio.

The SOLZ/SORZ acronyms represent the short-open-load/short-open-resistor-zener elements to be tested. These elements are tested so that corrections can be identified for the errors associated with the parasitic resistances in the TLP system or tester. 11 The zener diode is used specifically to verify the correctness of the TLP system voltage error, and the load-resistor is used specifically to verify the system's current error.

Conclusion

This article presents some guidance for selecting an appropriate o-scope for TLP measurements. It also points out that the apparently simple question as to which o-scope is most suitable does not have a simple answer.12 There are three distinct TLP measurements: the measurement of the DUT's I-V and leakage (IVL) data, the I-V data measurement for the SOLZ/SORZ verification data, and the waveform data for the calibration and verification of the TLP tester rise times. The new TLP document addresses the bandwidth of the scope for IVL data collection and for SORZ data, but it does not specifically address the rise time measurement and the associated error issue. Fortunately, the 500 MHz bandwidth is sufficient to measure rise times down to 2 nanoseconds. However, this bandwidth is inadequate to measure any rise time below 1 nanosecond (see Table I) because the errors at this level are above the acceptable limits.

The rise time is important for devices that are sensitive to ?V/?T or ?I/?T changes. The duplicating of HBM behavior requires 2-10 nanoseconds per the TLP standard, but engineering development may actually require <2 nanoseconds Tr.13,14 Data do exist that show changes and differences for Tr <5 nanoseconds for both HBM testing and TLP testing.14,15 Therefore, it is important that the o-scope used to measure the rise time parameter have sufficient bandwidth to ensure that the errors in the measurements are at an acceptable minimum level.

References

1. ESD SP5.5-TLP, 2002, "Standard Practice for Electrostatic Discharge (ESD) Sensitivity Testing, Transmission Line Pulse (TLP) Testing--Component Level," ESD Association, Rome, NY.

2. J Barth et al.,"TLP Calibration, Correlation, Standards and New Techniques," in Proceedings of the EOS/ESD Symposium (Rome, NY: ESD Association, 2000), 85-96; and IEEE Transactions on Electronics Packaging Manufacturing 24, no. 2, (2001): 99-108.

3. LG Henry, "All Types of ESD Testing Are Not Created Equal," Compliance Engineering 20, no. 2 (2003): 22-27.

4. TJ Maloney and N Khurana, "Transmission Line Pulsing Techniques for Circuit Modeling," in Proceedings of the EOS/ESD Symposium (Rome, NY: ESD Association, 1985), 49.

5. Tektronix Application Note, "An Introduction to Digital Storage," 1990.

6. Hewlett-Packard Application Note, "BW and Sampling Rate in Digitizing O-scopes," 1992.

7. Tektronix Technical Brief, "The ABCs of Probe," 1998.

8. Tektronix Application Note, "The XYZs of Oscilloscopes," 1992.

9. ESD STM-5.1-HBM, 1999, "Standard Test Method for Electrostatic Discharge (ESD) Sensitivity Testing, Human Body Model–Component Level," ESD Association, Rome, NY.

10. Tektronix Technical Brief, "High BW Transient Capture," 1991.

11. Barth Electronics Inc. Application Note, "Evaluation of the TLP System by the Use of Known SOLZ Elements to Determine the Pulse Measurement Range, Accuracy, and Resolution."

12. Hewlett-Packard Technical Note, "Ten Steps to Choosing the Right Oscilloscope," 2001.

13. S Voldman et al. (TLP-5.5 Working Group), "Standardization of the Transmission Line Pulse (TLP) Methodology for Electrostatic Discharge (ESD)," in Proceedings of the EOS/ESD Symposium (Rome, NY: ESD Association, 2003), 372.

14. M Chaine, J Davis, and A Kearney, "TLP Analysis of 0.125 µm CMOS ESD Input Protection Circuit," in Proceedings of the EOS/ESD Symposium (Rome, NY: ESD Association, 2003), 70.

15. LG Henry et al., "Transmission Line Pulse ESD Testing of ICs--A New Beginning," Compliance Engineering 18, no. 2 (2001): 46-53.
 
Leo G. Henry, PhD, is chief engineer for ESD-EMI-TLP Consultants (Fremont, CA). He can be reached at 510-708-5252 or leogesd@ieee.org.