A Pulsed-measurement Instrument for Device Testing
A recently developed pulsed-measurement instrument that measures the RF large-signal and DC characteristics of discrete GaAs FETs , high electron mobility transistors and integrated devices in GaAs MMICs
A Pulsed-measurement Instrument for Device Testing
GaAs Code Ltd.
Linton, Cambridge, UK
For many discrete GaAs FETs, high electron mobility transistors (HEMT) and integrated devices in GaAs MMICs, the large-signal IV characteristics of the device at RF, microwave or mm-wave frequencies are very different from the DC characteristics. This phenomenon, which is known as dispersion, may also occur in some HEMTs used in other materials systems as well as heterojunction bipolar transistors (HBT) and silicon bipolar junction transistors (BJT). In GaAs FETs and HEMTs, dispersion may occur due to charge changes in deep levels (or traps) or from self-heating during electrical operation. In HBTs and BJTs, self-heating is the main mechanism. A recently developed pulsed-measurement instrument provides the means to measure the RF large-signal and DC characteristics of all these device types.
Pulsed IV measurements directly provide the correct device characteristics needed for circuit design. Such measurements are also useful in helping to develop dispersionless technological processes and on-line process checks during the manufacture of discrete devices and ICs. In addition, the ways in which the IV characteristics change with regard to bias point and the length of the pulses provide valuable insights into the physical processes that cause dispersion and lead to improvements in large-signal models of devices for use in nonlinear circuit simulators.
Instrument Features and Specifications
The pulsed-measurement instrument performs measurements under conditions that are representative of the physical conditions that exist in practical RF, microwave and mm-wave circuits. The instrument is designed to measure FETs, HEMTs, HBTs and BJTs as well as diode devices, and is driven by a PC running Windows-compatible software developed specifically for these measurements. The speed of the host computer is not important because all of the time-critical routines involved in the testing procedures are handled by the instrument's three internal processors. The minimum requirements for the host computer include a 386 (or better) processor operating at 33 MHz with Windows 3.1, '95, '98 or NT; VGA graphics (640 ¥ 480 pixels with 16 colors); an RS-232 serial communications port; and at least 512 kB of free memory.
The unit has been designed as a desktop instrument rather than as a piece of laboratory equipment, so it is compact enough to be readily portable (13.5" X 12.0" X 4.5"). The unit is highly cost competitive as well because it does not utilize expensive commercial pulse generators or power supplies. All controls for the instrument are accessed through a soft control panel displayed on the screen of the host computer. The panel provides a means of setting up all the parameters for device tests. An emergency stop button ensures that the current test routine can be aborted instantly.
The supplied software allows a series of tests and measurements to be performed after which the results can be viewed graphically on the computer screen. The measured data also can be written to files that can be used subsequently by other software packages. Special versions of the hardware and software are available on quotation.
The instrument features two output ports. Port 1 (collector or drain port) features an impedance of 10 W, maximum voltage range of -10 to +10 V (constant or pulsed) and current limit of 0.5 A. Voltage and current measurement resolution are 10 mV and 1 mA, respectively. The operating mode for port 1 is pulse or DC sweep in constant voltage steps, and the pulses are synchronous with the gate or base port pulses.
Port 2 (gate or base port) features an impedance of 50 W and voltage and current ranges of -5 to +0.8 V constant or pulsed (FETs, HEMTs and diodes) and -10 to +10 mA (bipolars), respectively. Voltage and current measurement resolutions are -5 mV and 10 mA, respectively. Pulse duration is 100 ns to 1 ms and pulse separation time is 500 ms to 1 s. The operating mode for port 2 is pulse or DC sweep setup in constant voltage or current steps. Pulses are synchronous with port 1 pulses. Both output ports are SMA sockets and feature internal relays to disconnect the output socket from the electronics.
The Basic Operating Principle
Figure 1 shows the basic operation of the pulsed-measurement instrument. The device is DC biased at a steady point X. The bias point may be set anywhere on the ID (VDS,VGS) characteristics. The pulsed characteristics then are measured by pulsing both the drain- and gate-source voltages synchronously from the bias point to the point where the current is to be measured. The dwell time of the pulsed voltages at each point (pulse length) can be as low as 100 ns. The duty cycle is low enough to prevent self-heating of the device. As a result, any self-heating that occurs is due entirely to the power dissipated at the set DC bias point.
Typical Measurement Results
Figure 2 shows the dramatic differences that can exist between the device's DC and RF large-signal (pulsed) characteristics (in this case, for a commercial GaAs MESFET used in mobile telephone power output amplifiers). The DC characteristics are the curves that are flat at low currents and droop at higher currents. The upward sloping characteristics are the pulsed set measured about the bias point X (VDS = 3 V, VGS = -2 V) and constitute the large-signal RF characteristics the FET will follow when used as a class AB amplifier operating from a 3 V supply.
Not all devices exhibit dispersion at all bias points. For example, a contemporary pseudomorphic HEMT can be almost totally nondispersive.
Figure 3 shows an example measurement of the drain current transient for a commercial GaAs MESFET illustrating how the drain current evolves from the instantaneous (pulsed) value to the steady-state (DC) value. The data relate to a GaAs FET at VDS = 0 V, VGS = 0 V for operation as a cold mixer, and pulsed to VDS = 3 V, VGS = -0.5 V. The transient is complex and does not have a simple exponential decay. An initial rapid fall occurs, which is approximately exponential with a time constant of a few microseconds, followed by a partial recovery where the drain current increases slightly, finally decaying with a time constant of several hundred microseconds.
One of the most revealing experiments made possible by the pulsed-measurement instrument deals with the transient behavior changes with respect to bias point and the point being pulsed to. Even on the same device, the transient behavior can vary widely.
The pulsed-measurement instrument is useful in production testing as well as in analyzing and designing nonlinear circuits, developing nonlinear device models, monitoring device processes and optimizing device structures. The application of pulsed measurements in circuit simulation has long been known.1 However, the accuracy of nonlinear device models can be improved in various ways using results from pulsed measurements of IV characteristics. Manual to fully automatic selection of RF IV vs. DC characteristics is possible in recompiled models running in commercial nonlinear circuit simulators that accept full user models (for example, Libra). Contract services for upgrading nonlinear device models are available.
Developers of GaAs and III-V heterostructure technologies utilize pulsed IV measurements for process diagnosis, device design and optimization, process control and production testing. During processing, different etches, dielectric films and thermal processing steps can lead to damage to semiconductor surfaces. The difference between pulsed and DC IV measurements provides a useful diagnosis of the surface damage introduced by a given sequence of processing steps. As a result, pulsed IV measurements provide a basis for differentiating between different process options.
Even a well-designed process will introduce some surface damage and potential device performance sensitivity to deep levels. Designing devices for reduced sensitivity to the deleterious effects of deep levels involves trade-offs with respect to on-resistance, channel current, output conductance, breakdown voltage, complexity and manufacturability. The difference between pulsed and DC IV measurements is an indication of the sensitivity of a particular device structure to deep levels and thereby provides a basis for differentiating between different device designs.
Pulsed IV measurements are also useful for process control and production testing. When a production process goes out of specification, information is required as soon as possible to diagnose the cause of the problem. Pulsed IV measurements provide diagnostic information that is a useful supplement to other information gained from DC IV characteristics, electrical measurements on test structures and physical measurements. Pulsed IV measurements also are used in production testing when the observed DC-to-RF correlation is poor. Because pulsed IV measurements eliminate the low frequency responses to self-heating and charge exchange with deep-level traps, they provide a much better correlation to RF behavior.
A new low cost, compact instrument has been developed that permits measurement of RF/microwave device large-signal IV characteristics utilizing a Windows-based PC for control and display. The instrument is particularly useful for characterizing GaAs FETs, HEMTs and MMIC integrated devices, leading to improvements in large-signal device models for nonlinear circuit simulation.
1. W. Struble, S.L.G. Chu, M.J. Schindler,Y. Tajima and J. Huang, "Modelling Intermodulation Distortion in GaAs MESFETs Using Pulsed I-V Characteristics," 13th Annual GaAs IC Symposium Technical Digest, Monterey, CA, October 20-23, 1991, pp. 179-182.
GaAs Code Ltd.,
Linton, Cambridge, UK
+44 (0)1223 894900.