Evaluation Of The TLP System By The Use Of Known SOLZ Elements
To Determine Pulse Measurement Range, Accuracy and Resolution

Introduction:
Experienced ESD designers have used TLP beginning in 1985 to analyze the electrical characteristics of their protection circuits’ designs. Precision TLP Test data, measured primarily on wafer, is now used throughout the world to provide the pulse characteristic data needed to optimize the design and implementation of ESD protection circuits. As transistors in IC's continue to decrease in size and operating voltage, the design of efficient sub-micron protection structures requires an accurate measurement tool to optimize the design and minimize the amount of silicon real estate used for ESD protection.

ESDA Standard Practice for TLP
Many home-made TLP systems have been made since the TLP method was first described in 1985.[1] Since then, a large number of technical papers have described test results of many structures designed with  different types of TLP systems. However, the precision of these TLP systems has not been determined so that users can determine test data accuracy. Over time, operators of different TLP testers learn how to interpret the results in each system to minimizing measurement errors inherent in each system. This process takes significant engineering skill and time to be able to use the test data to speed their designs. Acquiring such experience is costly. To assist TLP users in the industry, the ESDA has recently formed a new Working Group to create a "Standard Practice Document" for TLP. Many different ESD designers experienced in TLP are combining their knowledge into suggested operating parameters for this document. Their experience learned in collecting test data from their systems is bringing together a wide base of different operating parameters learned during the use of many different homemade and commercial systems. When this document is completed, it will inform the ESD community how to best use a TLP system and interpret the data for their particular needs. It will hopefully also be able to describe a process that can "standardize" the reporting of TLP data and provide a method to identify the level of accuracy.

Until the ESDA Standard Practice Document is finished, we suggest the following Evaluation Procedure to provide detailed methods and technical guidance for new TLP users, and for those who are considering its use. This TLP evaluation procedure explains methods that can identify the basic accuracy characteristics of different measurement systems. When these "testing the tester" procedures are used throughout the industry, the accuracy of data collection will be known. We provide a method for TLP system analysis, which can be trusted to determine the accuracy of data collection. This fundamental information is important and useful to ESD protection designers.

The evaluation procedures described here can be performed on any commercial or homemade TLP system. The goal of this evaluation process is to provide a well-defined measurement method that we use as the tool to determine the accuracy of a TLP system. It also provides knowledge of TLP test data consistency, and a basic understanding of how these test procedures evaluate the capabilities of time domain measurement equipment. This information is helpful for TLP users or buyers alike, because it was developed to measure the accuracy and capabilities of the system. This is important because it will be used to measure how well their circuits meet the electrical parameter designs. Knowing the precision of the measurement data provided by any system allows the designer to more precisely gage the effectiveness of each silicon circuit produced. Because each revision to a wafer is very expensive, accurate and repeatable test data can minimize the number of revisions, which is critical to produce cost effective designs. Achieving the best possible test data on each revision minimizes the effort and number of steps required to create high levels of ESD protection. Performing this type of evaluation of a TLP system is valuable either when considering purchase of a TLP system, or when verifying the operation, precision and measurement drift of an operating TLP system.

SOLZ testing
The most precise method to evaluate a TLP system is to measure known electrical elements with precise electrical characteristics. A simple analysis of this TLP data provides clear information on the amount of deviation from known electrical values. Each reference element is placed at the "device test terminals" and the test data from each is recorded. These reference elements "test the tester" to determine how closely the measured test data matches the reference element. During our development of the first commercial system, we found four reference elements that are stable, repeatable, and affordable, for these tests. These elements are a "Short", "Open", "Load", and "Zener" diode. The SOLZ acronym is made from these four TLP test elements. To encourage making these tests on all TLP systems by any ESD designer at any laboratory, we are providing a kit of these four elements in a handy DIP package, with an instruction sheet and the measured parameters, free of charge to any one who requests it.

Short and Open Circuit Testing of a TLP System
The following plots are similar to what you will find when evaluating a TLP system:

 

      

         Raw I-V "Short" measurement data plot                                 Raw I-V "Open" measurement data plot.

Notice that the raw data from a TLP system when measuring a zero ohm (Short Circuit), in the left plot shown above, measures the internal series resistance in the TLP system. The resistance derived from the slope of the line (V/I) in this case shows an internal loss in the system of about 0.2 ohms. Variations from a straight line is the amount of "noise" inherent in any system. The voltage scale here is magnified greater than that used for device testing to clearly identify the system errors. We correct for that internal series resistance by removing its value in the data reduction and software analysis plots. These measured points should be a vertical line with minimum voltage deviations, because no voltage can exist across a Short Circuit. The corrected Short Circuit measurement is shown below on the left plot.

Notice also that in the raw data when using a TLP system to measure an infinity ohm (Open Circuit), in the right plot shown above, there is some small amount of internal shunt resistance losses in the TLP system. In this case, the shunt resistance calculated from the resistance slope is approximately 12 K ohms or about 86 micromhos. We also correct for that internal shunt resistance by removing it in the software so that the measured points of an Open Circuit becomes as close to a horizontal line with minimal current deviations as possible. The corrected Open Circuit measurement is shown below on the right.

         Corrected I-V Short Circuit data plot                             Corrected I-V Open Circuit data plot.

Load Resistor Testing of a TLP System

The purpose of measuring the Load element, or Load resistor, is to compare the voltage to current data by analyzing the slope of the plotted data. It should be a straight line with the V/I slope determining the measured value of resistance. For TLP testing of ESD protection on integrated circuits, it is best to use a low value resistor, which simulates the "on" resistance of a modern protection device. Other resistance values can be measured; but testing the slope of a 50 ohm resistor in a 50 ohm system is hardly meaningful because ESD protection circuits rarely produce dynamic resistance values above 10 ohms at high current levels. Also, measuring the V/I slope of a resistor that is equal to the TLP system impedance is the least demanding test possible for any TLP system. Measuring a 2 ohm, or 5 ohm resistor provides a much better evaluation of a TLP system because these values are much closer to those found in modern protection circuits. Evaluating this resistance region in a TLP system where most of the test data on ESD protection devices will be measured also provides information on the linearity of the system that will not be demonstrated with short or open tests.  The three resistance values of zero ohms, infinity ohms and 2, or 5 ohms provide the maximum amount of information on any TLP system capabilities. An example of the I-V data plot for a 5 ohm Load resistor after the Short and Open correction values are included in the software analysis, is shown below.

5.00 ohm resistor slope for an A/V and V/A corrected TLP system

Zener Diode Testing of a TLP System
The final element to use when evaluating a TLP system is the Zener diode. It provides a precisely defined voltage point on the voltage axis to identify the accuracy of the TLP voltage data.

    

               Corrected TLP data plot of 10 Volt                                       Corrected TLP data plot of 10 Volt

Zener diode at low currents                                                 Zener diode at high currents

The Zener (voltage) calibration combined with the Load (resistor) plot can provide information on the accuracy of the TLP current data. Two I-V plots of the same Zener diode are shown above at two different current and voltage levels. The left plot is shown at low current levels where the average Zener voltage can be computed by averaging a number of data points. This Zener diode had an average voltage of 7.32 volts from 0.3 mA to 30 mA. Comparing this to the measured DC Zener voltage at 1 mA of 7.342 volts shows the exceptional accuracy of the pulsed voltage measurement was better than 1% at this amplitude.

Continuing the testing too much higher current levels will amplify the TLP system I-V characteristics of the Zener diode as shown in the plot above at the right. A simple Zener diode has a smooth dynamic resistance plot. This test simulates the I-V characteristics of many ESD protection circuits and in this case shows a dynamic resistance characteristic of about 0.48 ohms. This plot can identify abrupt variations in the TLP measured pulse voltage as the current is increased to its maximum level.  Shifts in the voltage can be due to slightly different sensitivities in the digitizer as ranges are changed in either the voltage or current scales.

A Zener diode of any breakdown voltage can be used to evaluate the voltage measurement accuracy; but we have found those in the 5 to 20 volt range to be the most useful. Using diodes with sharp knee characteristics allows the turn on voltage point to be precisely identified at low current. TLP testing of the diode creates very little heating in the silicon and allows the dynamic resistance of the diode to display dynamic resistance to very high currents with deviations from heating.

When measuring the DC characteristics of these diodes, the power dissipated in it should be kept below 20 milliwatts to prevent the turn on voltage from increasing as the temperature increases from self heating. Limiting the dissipation in the Zener to 20 milliwatts allows it to remain near room temperature and limits the error in turn on voltage to less than a few millivolts. DC power dissipation of 20 milliwatts is produced by 2.0 milliamps of current through a 10-volt diode. A 10 millivolt error in turn on voltage for a 10 volt Zener diode, keeps the DC calibration error below 0.1% which is sufficient for evaluating the accuracy of the measured voltage pulse in any TLP system. Measuring a 5 to 15 volt Zener diode above a few milliamps with direct current (DC) causes heating and a shift in the voltage turn on point.

TLP measurement a of a 10 volt Zener to 10 amps only causes an average power dissipation of 10 microwatts for a 100 ns long pulse, and does not cause any measurable shift in the voltage from heating. Pulse testing a Zener diode to the maximum current level does provide a measure of its dynamic resistance slope at high currents. This high current test can provide additional data on a TLP system that is even more descriptive than the 2 ohm or 5 ohm resistor measuring explained above. 

The low current pulse data plot of a 10-volt Zener diode can be used to precisely identify the voltage accuracy of a TLP system. Measuring the DC voltage at 1 mA and 2 mA of a Zener diode with a modern digital voltmeter provides accurate verification of voltage points to evauate a TLP system.

When the Zener diode example shown above is pulsed to 10 amps, the slope of the "on" region indicates that it has about 0.48 ohms of dynamic resistance. (As a note of information, we have not had any Zener diodes fail when pulsed to 10 amps).

Accuracy of TLP Voltage and Current Data
The Zener diode provides a voltage comparison method to measure the TLP voltage accuracy. Once the TLP raw data is corrected for internal series and shunt resistance losses, the Load resistor provides a comparison method to determine the accuracy relationship between the voltage and current. Voltage divided by current equals resistance (V/I = R). These two measurements together therefore, provide a method to determine accuracy of current measurement. 

Dual Polarity TLP Systems
If the TLP system under test does not use a balanced connection to the DUT where the leads can be reversed without being affected by capacitive ground current; but instead uses both a positive and negative pulse source, the above procedures must be used to test the measurement characteristics of both polarities. This is necessary because the oscilloscope, or digitizer can have different accuracies and linearities when the baseline is set high and negative pulses are measured below the baseline.

Conclusion
It is our hope that other manufacturers of TLP systems will adopt this SOLZ method to evaluate the fundamental accuracy of TLP systems. In addition, identifying SOLZ evaluation data will provide designers with the accuracy of their TLP test data when it is shared within their own organization or published for readers throughout the industry.

Any comments or suggestions regarding this application note are welcome.

The Engineers at Barth Electronics

References
1. T Maloney, EOS/ESD Symposium Proceedings, 1985, pp 49-54
(revised May 15, 2003)

Return to Designer's Corner.