- Buyers Guide
Increasing RF Device Test Throughput with Better Instrument Coordination
Knowledgeable use of the programmability of today’s multi-function instruments can go a long way toward reducing test costs and improving manufacturing productivity.
Testing speed is important for all electronic components, but it is vital for low price two- and three-terminal devices such as diodes and transistors. Before RF tests can be conducted, the devices must be tested for DC operation. For diodes, that includes forward voltage, reverse breakdown voltage and leakage current. For transistors, it includes the various junction breakdown voltages, junction leakage currents, DC beta (hfe), and collector or drain characteristics. This article will show that choosing the right test equipment and setting it up appropriately can greatly speed these tests.
While tests can be done by stacking up a collection of digital multi-meters (DMM), voltage sources and current sources, this takes up substantially more rack space than a system built with all these functions in one unit. There are multiple sets of commands to learn, and system programming and maintenance become complicated. Additionally, trigger timing becomes more complex and triggering uncertainty increases, while coordinating the operation of separate instruments increases the amount of bus traffic required, again decreasing throughput. The first part of the solution is to combine several functions in one instrument. A source-measure unit (SMU) combines a precision voltage source, a precision current source, a voltmeter and an ammeter in one instrument, saving space and simplifying integration. The second part is to eliminate communication delays between the instruments and the control computer.
Minimizing Communications Overhead
When high speed communications between instruments and computers became available, it led to widespread automation of test systems using a general-purpose interface bus (GPIB) (IEEE-488 bus) link to deliver commands to control each step of a test. While this was a considerable advance over what went on before, it has a significant speed penalty. First, GPIB has considerable communications overhead. The other disadvantage of GPIB for real-time test control comes from what is generally at the other end of the line — a PC running Windows.“ Windows has significant and unpredictable latencies in responding to communications, which makes close synchronization of multiple instruments in the same test setup nearly impossible with the PC as the sole controller.
The solution to this problem is to preconfigure the instruments using GPIB and then let them execute the tests themselves. Many of today’s instruments have source memory list programming available, which allows up to 100 complete test sequences to be set up to run without PC intervention. Each test can contain different instrument configurations and test conditions, and can include source configurations, measurements, conditional branching, math functions and pass/fail limit testing with binning capability. Some units can operate in DC mode or pulse mode with varying parameters and timing (integration, delays, etc.), making it possible to slow down more sensitive measurements and speed up others to optimize overall timing. With the instruments basically running themselves, the role of GPIB is to download the test program before the test and upload the results to the PC afterwards, neither of which interferes with the actual testing.
To make a simple current-voltage (I-V) sweep, an SMU outputs a series of voltages while making corresponding current measurements, as shown in Figure 1. At each voltage step, the SMU will first source a voltage. The voltage change in the circuit will induce a transient current, so using an appropriate amount of delay time between sourcing and measurement is critical to overall measurement integrity. At different ranges, the instrument will adjust the delay time automatically to produce the optimal results. However, adding extra elements to the test circuit, such as long cables or a switch matrix, will change the circuit’s transient characteristics. Longer test times are usually necessary for high resistance devices. In these cases, additional delay time defined by the user will be needed to maintain the measurement integrity.
The first example involves one test instrument, a device handler and a PC. Note how the use of internal programming speeds up the test by eliminating most of the GPIB traffic.
Production testing of diodes involves a qualification step to determine the polarity of the diode under test, followed by measurements of forward voltage, reverse breakdown voltage and leakage current (see Figure 2). The forward voltage, VF, is the voltage appearing across the diode at some specified value of forward current. It is measured by passing that value of current through the diode and measuring the voltage across it. The reverse breakdown voltage, VRM or VBR, is the value of reverse voltage at which the current suddenly increases without limit. It is measured by forcing a specified reverse current and measuring the resulting voltage drop across the diode. The voltage reading is compared to a specified minimum limit to determine the pass/fail status of the test. The leakage current, IR, also sometimes called reverse saturation current, IS, is the current that flows when a reverse voltage less than the breakdown voltage is applied to the diode. It is measured by applying a specified reverse voltage and measuring the resulting current. A program is written to set up the diode tests in the memory location of the source/ memory instruments. The tests are then executed from one trigger over the IEEE bus. The instrument steps through the memory locations without computer intervention.
If the diodes arrive at the test setup with unknown polarity, a component handler can rotate them if necessary before testing (see Figure 3). The test steps are as follows:
- 11. The operator indicates to the PC that a diode production lot is in place and ready for test.
- 12. The PC preconfigures the tests that the SMU will perform on each diode via GPIB.
- 13. The SMU waits for the Start of Test trigger from the handler.
- 14. When the first diode is in position, the handler sends a Start of Test trigger signal to the SMU, indicating the first diode is ready for testing.
- 15. The SMU executes a polarity test. If the diode is in forward polarity, the SMU proceeds with functional tests (step 6). If in reverse polarity, a signal is sent to the handler to turn the device and return to step 4.
- 16. Once the diode is in forward polarity, the SMU runs diode functional tests in the order stored in source memory, makes pass/fail determinations and saves data for each test — Forward Voltage Test, Breakdown Voltage Test and Leakage Current Test.
- 17. The SMU sends an overall pass/
- fail code and End of Test signal to the handler and simultaneously sends test data to the PC via GPIB.
- 18. Steps 3–7 are repeated for the remainder of diodes in the lot.
- 19. The SMU returns to the idle state. The operator installs a new lot of diodes in the handler.
- 10. Steps 1–9 are repeated as required.
Note that the GPIB communication occurs only before and after the actual testing.
RF Power Transistor Tests
While there are many types of RF transistors available, the heterojunction bipolar transistor (HBT) will be used as an example. Analogous tests apply to other devices. Since transistors are three-terminal devices, two SMUs are generally used. Figure 4 shows two SMUs connected to the device, the first between the HBT base and emitter, and the second between the collector and emitter. To acquire collector family curves from the HBT, the base SMU is set to output current and measure voltage. The collector SMU is set to sweep voltage and measure current. After the first base current is set, the collector voltage is swept while the collector current is measured. The base current is then stepped up and the collector voltage is swept again while collector current is measured. This process is repeated until all the collector I-V curves at the different base current levels are acquired.
Synchronizing the Instruments
Since it is desirable for both instruments to be preprogrammed (and thus avoid GPIB delays), all the instruments in the setup should operate synchronously. At first this would not appear to be a problem. For example, if several SMUs all have the same firmware and are programmed with the same test parameters, the execution times for each step should be the same. The difficulty comes from the memory location recall and auto-ranging steps, which take an indeterminate amount of extra time (see Table 1).
In cases like this, an external, dedicated trigger controller should be used to make sure that the measurement occurs at the same time for multiple instruments. It is especially useful when a test system is built using equipment from different manufacturers or even products from the same manufacturer with different methods of triggering. The trigger controller takes the guesswork out of learning the nuances involved in operating each piece of instrumentation.
The process works as follows. Although the specific examples refer to Keithley instruments, analogous methods can be used with equipment from other manufacturers.
- The trigger controller outputs a trigger that is received by all instruments (source input).
- A Source Memory location is recalled from memory.
- The source output is enabled on all instruments.
- Each instrument performs the user-defined delay.
- Each instrument outputs a trigger to the controller once the delay operation is complete.
- The trigger controller waits for a trigger output from each instrument (delay output).
- The trigger controller outputs a trigger that is received by all instruments (measure input).
- Each instrument begins the measurement operation.
- Each instrument outputs a trigger to the controller once the measurement is complete.
- The trigger controller waits for a trigger output from each instrument (measure output).
- Go to Step 1 to begin the next test.
Figure 5 shows the result of synchronization of the triggering.
Specific Transistor Tests
Two important breakdown voltages are commonly measured for an HBT. The first is the collector-emitter breakdown voltage, which can be measured with the base open or shorted. Figure 6 shows the setup to measure the collector-emitter breakdown voltages with the base open (BVCEO or V(BR)CEO), while Figure 7 shows the setup with the base shorted (BVCES or V(BR)CES). The next breakdown voltage is the collector-base breakdown voltage (BVCBO or V(BR)CBO), which is commonly measured with the emitter open. Figure 8 shows the test setup. In these measurements, the SMU sweeps the voltage across the HBT while simultaneously measuring current. The current will remain fairly constant until the breakdown voltage is reached, at which point the current increases suddenly.
Other parameters commonly measured on RF power transistors are collector-emitter sustaining voltage, BVCEO(sus) or VCE(sus), collector-emitter breakdown voltage with reverse bias applied to the base-emitter junction, BVCEV or BVCEX, and emitter-base breakdown voltage with collector open, BVEBO (see Figure 9).
Junction Leakage Current
Characterizing the off leakage current of the device is also very important, because this leakage current will waste power while the device is not operating and will shorten the operating time of a portable, battery-powered device. The most often measured leakage current parameter is the collector cutoff current, ICBO, which is measured between the collector and base, with the emitter open. The base reverse bias leakage, also called emitter cutoff current or emitter-base cutoff current, IEBO, is another important leakage current measurement. It demonstrates the base leakage when the device is turned off.
DC Current Gain
The DC gain of an RF power amplifier is very closely linked to its RF gain. It can be measured directly and quickly by sourcing a base current and measuring the corresponding collector current, as shown earlier. Another technique that is often used is to sweep the base and collector voltage while simultaneously measuring the base and collector currents. After the measurements are complete, the base and collector currents are plotted on a semi-logarithmic scale to produce a so-called Gummel plot (see Figure 10). A number of useful parameters can be extracted from the Gummel plot, including DC gain (ß), base and collector ideality factors, series resistances and more.
Test throughput can have a significant effect on profitability for manufacturers of semiconductor devices. Knowledgeable use of the programmability of today’s multi-function instruments can go a long way toward reducing test costs and improving manufacturing productivity.
Mary Anne Tupta earned her BS degree in physics/electronic engineering and her MS degree in physics from John Carroll University, Cleveland, OH. She is currently a senior applications engineer at Keithley Instruments, where she has served in the company’s applications engineering department since 1988. She may be contacted via e-mail at firstname.lastname@example.org or by phone at 440-248-0400.