Technical Support
Feedback
Introduction
As IGBTs grow increasingly important in the power semiconductor market, new test methods are essential for characterizing products and for production. The modular test system described here targets static and dynamic measurements of power transistors and addresses the needs most critical to IGBT designers and to buyers of such equipment. We first present the circuit diagrams required for IGBT testing and define all parameters, then conclude with an example of control-system selection to illustrate the system's basic architecture.
IGBT Test System Requirements
In recent years the electronics industry has continued to advance power devices and systems, and the number of IGBT devices used in the power electronics industry has grown rapidly. Manufacturers from semiconductor to power product companies all seek better efficiency and lower cost across every step of the industrial process, from product development through final quality control, and every step must be tightly controlled. Testing has become correspondingly important for all these steps and must be accurate and reliable. In production, test equipment must offer a variety of features to meet different test needs. R&D engineers need to know device characteristics to support their designs; production personnel need fast, easy-to-use tools to control their manufacturing processes; and quality control personnel need to check the characteristics of the products they assemble and collect results for statistical analysis.
(1) Versatility
Modern test equipment must not be limited in the maximum current and voltage it can supply. Today's IGBT products can carry 1200 A with blocking voltages up to 3300 V, but before long IGBT products will reach 4500 V and 2000 A. No one can predict how far voltage and current will go in the future. Moreover, today's IGBT test equipment must also be able to test new devices such as the MCTS (MOS Controlled Thyristor) and the IGCT (Integrated Gate-Commutated Thyristor).
Test equipment must therefore be evaluated on its ability to adapt to and accommodate likely future developments, in both hardware—such as voltage and current generators—and software.
IWATSU's CS-3000 offers users the best solution
(2) Ease of Handling
Beyond the features above, test equipment should remain easy to operate and efficient. The best way to achieve this is for users to be able to operate the equipment without special training. Users familiar with a transistor curve tracer will find it even more convenient. The test system must also accommodate frequent changes in the type of semiconductor device under test (often several times a day); a new device type may require a different jig, more complex gate drive control, or snubber protection. A modular system meets these needs: a basic test system plus interchangeable test units enables measurement across a range of products.
(3) Safety
Because an IGBT test system can deliver very high output voltages, it poses a potential hazard to operators. When working with high voltage and high current, operator safety must be considered—the equipment must meet the safety standards of the various countries in which it is used.
During testing, hazardous areas should be protected by sliding door interlocks, all parameters should be set within the safe operating area, and all protective measures should be controlled by hardware rather than software. Clearly, time is a critical factor on a production line (and in R&D testing), so the safety system must be convenient and affect test speed as little as possible.
Static Tests
The purpose of these tests is to provide detailed device characteristics so designers can accurately predict device behavior under steady-state conditions. This helps users select the best device for their application and further supports proper design of the equipment connected to the semiconductor device, such as voltage clamp units, gate drives, and cooling systems.
(1) Collector-Emitter Breakdown Voltage VCEs
Measured at a specified collector current IC with the gate shorted to the emitter, the voltage across collector and emitter is VCEs (Figure 1). The collector-emitter breakdown voltage of an IGBT increases with junction temperature (typically 0.7 V/°C for a 600 V IGBT).
(2) Collector-Emitter Leakage Current ICES (also called collector cut-off current)
The collector current at rated collector-emitter voltage with the gate shorted to the emitter is ICES (Figure 2). ICES is typically measured at 25 °C and at the maximum operating junction temperature; collector leakage current also increases with junction temperature. It is therefore important to limit current flow during testing and avoid thermal rise.
(3) Collector-Emitter Saturation Voltage VCEsat
VCEsat (Figure 3) is the collector-emitter on-state voltage at a specified collector current, gate-emitter voltage, and junction temperature. VCEsat is a very important characteristic because it determines conduction loss. At high collector currents the test pulse must be very short so that excessive loss does not occur.
(4) Diode Forward Voltage VF
VF (Figure 4) is the forward conduction voltage of the freewheeling diode in an IGBT module at a specified current and junction temperature.
(5) Gate Threshold Voltage VGEth
VGEth (Figure 5) is the gate-emitter voltage at a specified collector current with the gate shorted to the collector. When the gate-emitter voltage is below the threshold, the IGBT is off; the gate threshold voltage is therefore the gate-emitter voltage at which the IGBT turns on and conducts a specified collector current. VGEth decreases sharply as junction temperature rises (−11 mV/°C).
(6) Transconductance gfs
Transconductance gfs (Figure 6) is the ratio of collector current to gate-emitter voltage at a specified collector current. Transconductance expresses the gain of the IGBT. Because it is derived from two measurements made under clearly defined, tightly specified conditions, the accuracy of the test equipment strongly influences the result.
A second method adjusts the gate-emitter voltage to a specified collector-emitter voltage (VP) and uses the following relation:
△ VGE = △Vp
The forward transconductance of an IGBT increases with junction temperature because, at a fixed gate voltage, increasing collector current raises the die temperature through thermal runaway. For this reason, IGBTs are not recommended for use as linear amplifiers.
(7) Gate Charge (QGE, QGC, QG) and Gate Capacitance (Cies, Coes, Cres)
QGE (Figure 7) is the charge delivered by the drive circuit to raise the gate-emitter voltage to the level that sustains a specified collector current. QGC is the charge delivered by the drive circuit that allows the voltage across the gate capacitance to fall from a specified value to the final on-state value. QG is the total gate charge, the sum of QGE, QGC, and an additional component related to the gate "overdrive" voltage. Because some unknown stray capacitance exists in the circuit, calibration pulses are required. Stray capacitance is measured before each test and used to correct the gate charge/capacitance values. This stray capacitance varies with wiring and ambient temperature conditions.
Gate charge and capacitance specifications are very useful when planning gate drive circuits and determining gate drive losses.
(8) Gate-Emitter Leakage Current IGES
IGES (Figure 8) is the gate leakage current at a specified gate-emitter voltage with the collector shorted to the emitter. This test can be performed with either positive or negative gate-emitter voltage. The measured current is very small, so the pulse must be maintained for at least one power-supply cycle integration time to avoid errors caused by current absorbed by the gate capacitance. The measurement must be made only after the gate voltage has stabilized.
9. Wafer-Level ON-WAFER High-Voltage/High-Current Probe Station Test System
ON-WAFER Testing
For most IGBT users, testing is not simply a matter of firing a power pulse at a single transistor and reading the result; they are more interested in how these devices behave in the circuits of their application designs. For this reason, we must extend measurement as far as possible to the half-bridge configuration within IGBT modules. Figure 9 shows the switching characteristics: current flows through the load and the DUT. The purpose of this switch pair is to allow measurement of one IGBT and the freewheeling diode of the other IGBT at the same time. The timing of the switch action and what is to be tested.
The collector current pulse required by the IGBT is provided by the gate drive circuit, as shown in Figure 10. As in many other applications, the diode dI/dt is generated by the IGBT itself, and the waveforms in Figures 11 to 13 occur just as they do in real situations. In an actual design, the connections between all power modules should be as short as possible and use bus bars, keeping all internal stray inductance below 50 to 100 nH. This system can also perform switching measurements with either resistive or inductive loads, including gate characteristics. The method allows fast testing of single or dual devices and can be used for characterization, production, and final inspection. The main unit of the test system is a digital storage oscilloscope connected to specially designed voltage and current probes; the greatest benefit of this design is that it can measure a very important parameter: switching energy. The values obtained by this method are far more accurate than those from other test equipment on the market.
Of particular concern in this test system is the gate drive circuit design: inductance should be as low as possible, because it can produce parasitic oscillation with the capacitance at the IGBT junction. The leads from the gate control circuit to the emitter side must be connected to separate terminals to avoid coupling with the collector-emitter main circuit.
The maximum gate-emitter rated voltage VGEs given in the datasheet (mostly ±20 V) must not be exceeded, since doing so may cause IGBT failure. In a production-line test system, the most common approach is to connect a voltage protection device (Zener diode) between gate and emitter. The wiring between the gate generator and the UUT must therefore be designed to limit waveform reflections in the cable to a very small level.
(1) Switching Times and Switching Losses
For an IGBT, switching times depend on the gate-emitter voltage and the collector waveform. They are determined by the internal capacitances, the parasitic inductance in the module, and the internal resistance of the gate drive circuit. 1. Turn-on delay time td(on)
Because of the internal resistance of the gate drive circuit and the IGBT's gate-emitter capacitance, the gate-emitter voltage VGE rises less steeply than the voltage produced by the gate drive circuit. When VGE reaches the threshold VGE(th), collector current begins to rise. td(on) is the time from the gate drive circuit voltage to the start of collector current rise in the IGBT; both parameters are taken at the point where they reach 10% of their final steady-state value.
2. Current rise time tr
tr is the time between IC reaching 10% and 90% of its final steady-state value. In practice, because of diode recovery current, the IGBT current overshoots its output current, so the time for IGBT current to reach its peak is longer than the time to reach 90% of output current.
3. Turn-on time ton
ton is defined as the sum of turn-on delay time td(on) and rise time tr.
4. Turn-on switching energy Eon
Eon is the total energy dissipated when the IGBT turns on. In the inductive-load case, a preload current flows into the diode of the opposite IGBT before turn-on.
The peak collector current in Figure 11 is caused by this diode reverse recovery current, and this peak current must be included when calculating turn-on power loss. At the end of turn-on, the collector-emitter voltage VCE has usually not yet fallen to its final saturation value VCEsat, so this must also be taken into account when calculating turn-on loss.
When the gate-emitter voltage falls below the threshold, the IGBT turns off. To prevent spurious turn-on through Miller capacitance during testing, the gate circuit is usually switched to 0 V or a negative voltage.
5. Turn-off delay time tdoff
tdoff is the time from the instant the gate-emitter voltage falls to 90% of its final steady-state value to the instant collector current IC falls to 90% of its final steady-state value.
6. Fall time tf
tf is the interval during which collector current falls to 10% of its final steady-state value.
7. Turn-off time toff
toff is defined as the sum of turn-off delay time and fall time tf.
8. Turn-off switching energy
Eoff is the energy dissipated when the IGBT turns off. This energy is measured over an interval: Eoff is calculated from the point at which the test VCE voltage reaches 10% over a period of several μs. Because of the influence of tail current during turn-off, a relatively long time window should be considered (typically 5 μs).
9. Diode reverse recovery energy Erec
The main effect of diode reverse recovery is that it increases the energy required to turn the IGBT on. Some manufacturers add this value to the IGBT's Eon specification without separately stating or specifying other parameters to represent the diode's switching loss.
(2) Short-Circuit Testing
One advantage of the IGBT over other power transistors is its ability to withstand large short-circuit currents. The short-circuit test circuit and its test waveforms are shown in Figure 14. Under the specified conditions, the short-circuit current can rise to 10 times the device's rated current. The test equipment must therefore deliver a very large current in a very short time (mostly 10–50 μs) at a relatively low collector-emitter voltage.
Programmability
To illustrate control of the modular test system, we take a typical static VCEsat measurement as an example. The required equipment is the circuit shown in Figure 3 (VCEsat test). We assume the device under test is properly placed alone in the test position and that ambient temperature is controlled within the required specification (manual or automatic handler mode).
As noted above, this test must be performed in a very short time to avoid raising the temperature of the semiconductor device. Specifying each test type therefore implies the need for very accurate timing. As Figure 15 shows, the first element to act is the gate drive circuit, which ensures the semiconductor is on throughout the test period; then the current generator supplies a current pulse whose peak is related to the specified collector current. The pulse slope is affected by the generator's characteristics and by the connection to the downstream DUT, particularly by stray inductance in the wiring, so for accurate control of current rise and output stability it is best to avoid other unnecessary interference. Once the collector current has stabilized at the specified value, the collector-emitter voltage of the conducting IGBT can be measured accurately. The semiconductor device under test is placed at the output called DUT and connected to the associated generators (IPL2000 is a 2000 A current generator and TRCg is the gate driver) and to a measurement setup consisting of a programmed interface and relays. Each unit is controlled via electrical signals by a so-called function module (FM). For the current generator, for example, several IPL2000 power modules can be connected in parallel and controlled by a single function module. Before testing, each module is assembled via a fiber-optic serial bus; when this combination is to be executed for a test, the generator modules immediately prepare to output the voltage or current required by the test, and the function module sends a "Ready" signal over the fiber-optic bus, at which point the test can begin.
The synchronization pulses shown in Figure 15 are transmitted to each function module over separate fiber-optic lines. Each function module can generate basic parameters to produce a pulse waveform that keeps the generator in linear output. The control unit consists of an industrial PC with a CPU, a fiber-optic distribution board, a timing board that generates synchronization pulses, and a GPIB bus controller board.
As with the general modular concept, the test system software is divided into independent units. Each software module, called a server, controls one hardware unit, as shown in Figure 16—for example, the IPL server, GPIB server, and so on. These servers communicate with one another via messages called IPL (inter-process communication).
IPCs contain test information: that is, the maximum current and slope of the IPL2000, the measurement voltage range process, and the pulse duration of the timing board. More general test information, like the other independent units, is contained in the program as a so-called test entity—a test object that immediately reveals which test units accompany the test.
The complete set of tests and test parameters required on a production line can be edited by the user into a sequence program for execution, and the test results can be stored in files compatible with the user's database.
In practice, several servers and test entities must operate concurrently so that the operating system can execute several tasks simultaneously and in real time. Each server is independent, so software can be updated in stages without modifying the entire software package. To facilitate update, maintenance, and development, the PC section should have an Ethernet card or modem so users can easily load new software versions or new software development tools. A multi-user operating system also allows remote control of the tester, which is very helpful for later inspection or maintenance.
