This literature was published years prior to the establishment of Agilent Technologies as a company independent from Hewlett-Packard and describes products or services now available through Agilent. It may also refer to products/services no longer supported by Agilent. We regret any inconvenience caused by obsolete information.HEWLETT For the latest information on Agilent's test and measurement products go to: PACKARD www.agilent.com find products Or in the US, call Agilent Technologies at 1-800-452-4844 (8am-8pm EST) Efficient Microwave Bias and Test Using the HP 4142B Modular DC Source/Monitor Application Note 1205 Improve microwave semiconductor device quality while reducing test costs by integrating precision DC measurements into MMIC test systems. Table Page of Contents Introduction ................................................................................................................................................. Using SMUs to Reliably Test Microwave Semiconductors 1.1 Special Testing Needs of GaAs and Microwave 1.2 Versatile 1.3 Using the AFU for Active Biased S-Parameter SMU Simplifies 1.4 A Complete Application MMIC Bipolar Devices ............................................... Test Setups ........................................................................................ Measurements ................................................ DC/RF Wafer Probe Test System ................................................................ .............................................................................................................................. Techniques 2.1 RF with Precision 2.2 Safely Applying 2.3 Advantages 2.4 Fixturing 2.5 High Speed Testing of Monolithic Cover photo: Shown is a configurable Company. DC/RF wafer ....................................................................... DC Bias Compensation Bias to Microwave ................................................................................. Transistors ......................................................................... of DC Pulsed Bias .................................................................................................... Considerations MMIC test evaluation When Using the HCU.. ...................................................................... Amplifiers system available is possible with ............................................................................ from Hewlett-Packard a single connection. 1 2 2 3 6 7 9 9 12 13 14 .16 Introduction Overview The ability to provide DC and RF in both continuous and pulsed modes are important for testing gallium arsenide (GaAs) and other microwave devices. For test system efficiency it is necessary for DC hardware to first provide DC functional test and then act as bias supplies during RF testing. This is particularly true with the emerging monolithic microwave integrated circuit (MMIC). High test costs continues to be a predominant barrier to high volume utilization of MMIC technology. DC instrumentation is required which is designed to integrate well with automated RF instrumentation. Conventional power supplies and voltmeters provide awkward solutions which can damage high frequency devices or produce erroneous results if not carefully used. The precision DC parametric measurement capability and flexible configuration of the HP 4142B make it ideal for comprehensive MMIC DC characterization and production test. Design features of the source/monitor unit (SMU) make it particularly well suited for testing microwave devices which can be easily damaged by static or power supply overshoot and glitches. Wide measurement ranges, excellent sensitivities in both current and voltage modes, and high throughput are just a few reasons the HP 4142B is the instrument of choice for MMIC applications. 1 Chapter 1 describes why the HP 4142B SMUs provide the best total solution for DC testing microwave devices. A complete MMIC test station is proposed which integrates precision DC test with RF test for single touch-down wafer characterization. Chapter 2 extends concepts of using SMUs for precision microwave test by providing application details. A high-volume benchtop test system is described, capable of measuring 10,000 units per hour for both DC and AC limits. The HP 4142B Modular DC Source/Monitor The HP 4142B is a high speed, highly accurate computercontrolled DC parametric The HP 4142B Modular It’s modular applications. SMU design is particularly instrument capable of complete characterization of today’s premier semiconductor devices such as GaAs MMICs. &digit precision voltage and current measurements are obtainable over a wide range from 4 pV400 V/20 fA-20 A. Pulsed bias, synchronized staircase sweeps, and very high speed spot measurements are features which make the HP 4142B ideal for testing GaAs and silicon devices. For the basic principles of HP 4142B operation, refer to application note “High Speed DC Characterization of Semiconductor Devices for Sub pA to 1 A” (Application Note 356). Further detail is found in “Techniques and Applications for High Throughput and Stable Characterization” (Application Note 356-l). DC Source/Monitor well-suited for multi-pin MMIC Using SMUs to Reliably Test Microwave Semiconductors 1.1 Special Testing Needs of GaAs and Microwave Bipolar Devices The performance of high-frequency active devices is primarily due to extremely low junction capacitances (sub pF) and small transit times between device terminals (~10 ps). To obtain low capacitances, the control electrodes are significantly reduced in size compared to conventional lower frequency devices. In the case of GaAs FETs, sub-micron fabrication technologies keep the gate contacts to the active device region confined to as small an area as possible. With GHz Ft bipolar devices, the base region is kept extremely thin using very shallow ion-implanted diffusions. These scaled down devices make DC testing a particular challenge. First of all, the small sizes reduce current and voltage handling capabilities, making microwave devices extremely sensitive to electrostatic damage. Breakdown voltages between terminals can be as low as 10 volts. Extreme care must be taken to test in completely anti-static environments, and then to bias the device “gently” without any voltage or current “glitching.” Even non-catastrophic damage may occur due to improper DC biasing. In this case, the device may be weakened so as to fail upon subsequent testing, or worse yet, result in a field failure. Although individual transistor bias conditions may be modest, when combined to form ICs, bias currents can rise quickly. 2 necessary to separate the intrinsic transistor characteristics from extrinsic packaging or probing environment to understand device behavior. Precision DC parametric extractions are required to verify initial models used in circuit simulations. Special attention must be given to eliminate residual resistances (kelvin contacts), residual capacitances (driven guards), and to properly isolate device terminals from one another. Designs usually are optimized for controlled impedance loading (50 ohm) and useful RF output may be only a small percentage of bias power applied. MMICs may typically require heat sinking to dissipate 10 watts or greater. To reduce heating effects, DC pulsed biasing is highly desirable, or in some cases, mandatory. GaAs and microwave bipolar semiconductor performance is quite sensitive to parasitic capacitances, stray inductances, and lead resistances. Computer models of RF transistors emphasize DC parasitics which can have dramatic effects on RF performance. It is usually Figure 1. Shallow junction depths enhance microwave performance but make MMICs particularly susceptibly to static and biasing damage. Reslstor 3 And finally, many microwave devices must be tested both at DC and RF at the wafer level. The end product may be a hybrid involving many discrete chips. Cost to remove bad chips from hybrid packages may be in the hundreds of dollars. Thorough screening is now possible using special microwave probe stations. This allows end users to integrate chips into cost effective hybrids which require no-rework. Figure 2. Power fast pulse testing failure. GaAs FETs require to avoid heating For more information on the special testing needs of microwave GaAs and bipolar semiconductor devices, refer to the application note “Role of DC Parametric Test in High Speed Digital and Microwave Semiconductor Component Manufacturing” (Application Note 339-20). Figure 3. The equivalent of four instruments, SMUs can source voltage and measure current or source current and measure voltage. 1.2 Versatile SMU Simplifies Test Setups HP 4142B Modules Four We will now take a closer look at how the modular SMU architecture of the HP 4142B make it particularly well-suited for MMIC applications. Chose the Right Performance MMICs may require eight or more pins biased simultaneously and measured over widely different current or voltage ranges. The open architecture of the HP 4142B accommodates a complete family of modules. Select up to eight modules per mainframe to build a cost effective tester. Chose from 5 digit, 20 fA and 4 pV resolution for low level characterization such as electrode resistances, resist&ties, leakage currents, etc. Extract high power parameters using the HPSMU (1 A high power SMU) or HCU (10 A high current unit). See Table 1 for specifications. Minimize thermal drift with the HP 4142B’s pulsed output. Perform quick analog current or voltage searches using two SMUs and the AFU (analog feedback unit). Simplified SMU Circuit Instruments in One A SMU combines all of the instrumentation required to make constant or pulsed I-V measurements on a wide variety of semiconductor devices. Figure 3 shows a simplified circuit diagram of a SMU. A SMU acts simultaneously as a current source and voltage monitor or as a voltage source and current monitor. Switching between the two modes is done programmatitally. You don’t have to change your fixture wiring to switch from one test configuration to the next. With an SMU attached to each electrode of an active device, virtually any DC parametric test may be performed with one device insertion. This increases test throughput and test integrity. Test development time is also reduced because the engineer is freed from worrying about integrating source and measurement functions. Simple force/measure commands take care of ranging, compliance, wait times, and synchronizing more than one SMU. Diagram 4 Guarded Kelvin Connections and Stable Ground Reference SMUs have separate force and sense terminals which calibrate out residual cable resistances. This is critical when making low voltage or high current MMIC measurements. Significant error would occur if cabling, bias network, and network analyzer DC line resistances were not taken into consideration. In addition, active guarding is used to minimize effects of cable capacitances and noise when making very low level current measurements. The HP 4142B mainframe contains a special Ground Unit (GNDU) which is a high current SMU set to 0 volts. This further eliminates noise by referencing all measurements to an active ground terminated as close as possible to the device under test (DUT). RF system ground is usually very noisy and can produce significant DC parametric errors of >lO mV. The GNDU is built into the HP 4142B’s mainframe. This saves you from sacrificing one regular SMU as a stable ground reference. Table 1. HP Model 4142B plug-in DUT Figure 4. The stable ground reference (GNDU) and force-senseguard SMU configuration eliminate system noise and cable errors modules number Voltage range HP 41420A HPSMU flO0 pl to 5!00 v* HP 414218 ~100pvto~100V MPSMU Current range k5OfAtokl Measurement resolution A k50 fA to k100 mA Accuracy V I 40 pV, 20 fA 0.05% 0.2% 40 uV, 20 fA 0.05% 0.2% HP 41424A VS/VMU *l mV to *40 V St0 mA to k100 mA 4 pv, 20 PA 0.05% 0.3% HP 41422A HFU k200pvt0~10V klOOnAtoflOA* 40 pV, 20 nA 0.5% 0.5% HP 41425A ACU *Note: Searches 400V max using two HPSMU channels for a specified current or voltage on one SMU by controlling or 20A max using two HCU channels the voltage output of another SMU. 5 Flexible Compliance and Filtering SMUs are inherently well suited for testing MMIC devices which are easily weakened or damaged by excessive bias conditions. SMUs are multi-range instruments with up to 10 current ranges and 5 voltage ranges. Compliance resolutions are the same as setting resolutions. And programming the measurement range automatically sets the compliance range. There is no need to coordinate these two functions as with separate sources and monitors. Each SMU provides a low-pass filter at the DAC output. This further limits the possibility of damaging the device and increases source setting accuracy. The filter (optionally programmed ON or OFF) reduces current or voltage overshoot to less than .03% of the maximum range value. SMU settling time is only slightly longer so for most MMIC applications the filter switch should be set to ON. Filter at DAC Output OFF I-,)’ , ,” ON Figure 5. The DAC output filter limits voltage or current spikes when changing value to l/30 of unfiltered value. Wide variety of sweep conditions SMUs can provide all of the sweeps expected of a semiconductor curve tracer, and much more. Table 2 illustrates a few of the variety of waveforms easily programmed by the HP 4142B. Sweeps include single fixed-level sweeps, linear or logarithmic staircase sweeps, and single and double pulsed sweeps. Command software is tailored for synchronizing SMUs for typical semiconductor applications. For example, the double pulsed mode allows rapid plotting of family curves where measurements from two device electrodes are required. Programmable hold and delay times ensure that your source/measurement sequence is coordinated exactly as you desire. Table 2. Output Waveform available 0 / I 3 0 0 1 5 I 0 I 0 0 0 l 0 2.0 I I 0 0 0 0 6 1.3 Using the AFW for Active Biased S-Parameter Measurements Small signal S-parameter measurements using network analyzers is the industry method of choice for evaluating RF performance of microwave devices. Active devices such as GaAs FETs or BJTs require “glitch-free” and stable DC biasing during RF measurement. Using two independent sources, such as linear power supplies, for external bias during S-parameter test can result in inaccuracy and even damage the device under test. Thermal heating causes bias point drift which can lead to run away destruction of sensitive microwave devices. This section shows how active feedback may be placed between two SMUs of the HP 4142B to precisely and safely hold bias conditions during S-parameter measurement. The Analog Feedback Unit (AFU) modulates the output voltage of one SMU while monitoring the current or voltage of the second. Target currents or voltages are found using rapid ramp integrations (10 ms) and precision analog searches (10 ms typically). Control over the slew rate of the search ramp can be used to eliminate overshoot, a common way discrete and MMIC devices are damaged. The AFU is a modular plug-in for the HP 4142B which requires no user connection. Internal connections allow it to be programmatically placed within the feedback paths of two specified SMUs. Figure 6 shows how the AFU would be used to set bias conditions for a bipolar device. S-parameters, in this case, are usually specified at a constant collector current. The AFU monitors collector current and ramps the DUT base voltage from a specified start voltage and ramp rate. Negative feedback is used to reliably settle the collector current to the precise target value after the ramp integration. See figure 7. TI >To Figure 6. Thermal drift during bias. The active feedback unit (AFU) holds collector voltage and current constant. ] ./- ~I- "CE final value V bE tlOms+ Search SMU Figure 7. Simplified AFU Operation. A programmable ramp followed by an analog search applies bias quickly and safely. GNDU Sense SMU 10ms - 7 Figure 8 shows a setup using the AFU biasing method with the HP 4142B for making reliable S-parameter measurements immune from thermal drift error. Later in the application note an improved circuit is shown which includes kelvin bias compensation to correct for voltage drops in series with the DUT terminals. Network S parameter Test Set (HP 85046A) HP 41428 Figure 8. S-Parameter measurement configuration using the HP 4142B AFU. 1.4 A complete MMIC DC/RF Wafer Probe Test System Test costs can often be the single most prohibitive factor limiting availability of competitively priced MMICs. Tests are often exotic and manual in nature. Usually multiple test setups are required for each unit tested. There is an industry need for solutions which integrate DC and RF testing in an efficient manner, preferably at wafer level. Thorough testing early on in the MMIC process can help avoid costly or time consuming mistakes. More accurate test data is now being required to support simulation models during the design phase. In production, as chip volumes increase, reduction of high MMIC costs require faster (higher throughput) testing using integrated test systems. Figure 9 shows an integrated test solution for MMIC test which is made possible, in part, by the robust feature set of the HP 4142B. Through the use of Analyzer bias “tees” and central test sets such as the HP 85110A shown, precise DC and RF characterization is performed with a single device connection. You benefit with this approach with, l l l Reduced setup and test times l l Eliminate pad damage due to multiple probing l l l Increase correlation accuracy between DC and RF data Reduce test development support costs l l Precision extraction of DC parameters for modeling High throughput DC functional test in production Integration of up to 8 DC channels in one 11” high rack space External triggering synch to RF test instrumentation and Safe biasing using active feedback (AFU) Advanced DC waveforms like staircase sweeps and dual pulse and The test system shown provides DC test and S-parameter, noise figure, distortion, power, pulsed DC/RF and frequency translation measurement capability. Features of the HP 4142B which enhance this systems capabilities include, l SMU design eliminates matrix switching, complex fixturing For more detailed information concerning multi-parameter MMIC test systems, refer to “A complete microwave test system integrated for you by Hewlett-Packard” (publication number 5952-1749). This solution is referred to as the SCMM (single connection multiple measurement) test system. Instrumentation is optionally tailored for your requirements. A test shell contains all instrument control software and provides a unified user interface for turnkey DC/RF test. Application Techniques 2.1 RF With Precision Bias Compensation DC As mentioned earlier, the ability to combine DC and RF test with a single connection is critical for production MMIC test. The HP 4142B with its flexible SMU architecture allows virtually any measurement to be made without re-connecting the device. Key to integrating DC with RF signals is the bias network (tee) and RF test set. This section will review techniques for increasing system accuracy and throughput by using kelvin (remote) sensing to compensate for bias tee and RF test set resistive losses. 9 MMICs above 500 mA both pulsed DC and remote sensing becomes mandatory to solve this problem. Overheating of bias tee coils and significant resistive channel drops of >lOO mV error would otherwise occur. Probe contact resistance during wafer evaluation can be another source of resistive error. In some cases multiple pad connections are available or designed in for remote sensing right down to the chip level. Bias Tees Bias tees are precision broadband components for supplying DC bias to the center conductor of a coaxial RF line which can be connected to the microwave DUT. A tee-junction is made to the DC source through a low pass filter, consisting of a loss choke and a bypass capacitor. See Figure 11. RF test sets are boxes containing RF coaxial switches and couplers which allow complete S-parameter characterization of a two port DUT with a single connection. A bias tee is included in each RF port. Compensating Table 3. Bias Tee Specifications RF Connector Port Match (Min Return Loss) DC error Bias tees and fuses in each test set channel can add >l ohm of uncompensated series resistance to each DUT RF line making precision DC measurement impractical. For biasing of I Insertion (Max) Loss Max Bias Voltage Max Bias Current (Continuous) Bias Ports DC Resistance (Bias to RF port) Compensating for DC error with remote sensing not only improves system accuracy, but can increase test throughput by up to 5 or 10 fold. Kelvin remote sensing can eliminate the need for calibrating out fixture losses which can involve complicated look-up tables of correction factors. Figure 11. Bias network with two BNC ports used for DC force and DC sense. 10 Kelvin Sensing RF Test Sets with Bias-Tl Bias-T2 RF Port 1 Ideally, a dual-coil (force/sense) bias tee is desirable, which can be placed at the DUT RF-in and RF-out connections. Table 3 lists several models of HP bias tees which can be used to bias the DUT external to the RF test set. Bias tees with two DC ports can be used as shown in figure 12. The high DC resistance of the bias choke coil in the force (F) line does not affect the programmed voltage monitored by the high impedance sense (S) line. By calibrating the network analyzer at the fixture interface, RF impedance error from the external bias tees can be removed during S-parameter measurement. The GNDU (0 volt SMU) of the HP 4142B should also be connected with force/sense terminated as close as possible to the DUT. Proper use of the GNDU removes noise and resistive losses due to ground paths which can add up to >lO mV of typical system error. The point at which the GNDU force/sense lines are terminated becomes the reference point for all SMU measurements. Keeping this point at the DUT fixture is necessary to effectively reduce cable and instrumentation chassis ground losses. This is especially true for wafer probing applications. RF Port 2 SMUl GNDU SMU2 Figure 12. Adding external bias tees to achieve full kelvin sensing at the DUT. Use of the HP 4142B Shorting Bar Disconnect the shorting bar on the front of the HP 4142B when the GNDU is being used with a network analyzer. This floats the circuit common of the HP 4142B and prevents the GNDU from being grounded to the network analyzer chassis. It will also eliminate the possibility of any ground current loops flowing between the HP 4142B chassis and the network analyzer chassis. Circuit Common r-0 Figure 13. HP 4142B chassis shorting bar disconnected at the front panel. 11 Triax to Coax Adapters Simplify Bias Hookup The driven guard connections used to insure accurate low current measurements (<lnA) usually are not necessary in microwave applications. By using a triax to coax adapter which leaves the inner shield (guard) disconnected, simple SMU hookup is possible with standard 50 ohm BNC coaxial cables. Figure 14 shows a particularly convenient way to fixture to the DUT when using the HP 11509B bias network. Coaxial cables connect directly to the HP 11509B bias/bias sense terminals and serve as a solid ground return path to the SMU circuit common. The GNDU is not needed in this application due to the low impedance return and excellent shielding afforded by the coax cable. HP 11509B HP 115098 in Bias Sensing Bias Bias to adapters I F SMUl ’ I Bias Sensing Use of the proper triax adapter is important. The recommended adapter for SMU force and sense connections is the Trompeter Elec. Inc. part no. ADBJ20-E2PL75. This adapter connects the triax center lead and outer shield while leaving the inner shield floating, as shown in figure 15. I G-3 TRB Plug Figure 15. Triax to coax adapter floats guard and simplifies SMU connections for MMIC applications. Figure 14. Using RF ground as circuit common for the HP 4142B. BNC Jack 12 GaAs FETs 2.2 Safely Applying Bias To Microwave Transistors 1. Ground the gate to source. 2. Apply bias approximately equal to or less than the pinch-off voltage between gate and source. Some FETs are unable to handle Idss without damage. 3. Apply the desired drain-tosource bias. 4. Decrease (N-channel) or increase (P-channel) gate The following procedures provide the safest turn-on sequence for devices which are very sensitive to current and voltage spikes. These procedures are easily adopted in automated systems using several SMU channels of the HP 4142B. Bipolar bias to obtain the desired drain-to-source current. 5. Remove bias in the reverse order. Transistors 1. Ground the base to the emitter. 2. Apply desired collector-toemitter bias with an SMU in voltage mode. 3. Apply base-to-emitter bias through an SMU in current mode to desired quiescent point. 4. Remove bias in the reverse order. SMU2 Dual-gate FETs are biased by first applying forward bias to the second gate (closest to the drain). Next, apply reverse bias to the first gate. Lastly, bias the drain. GNDU SMUl Figure 16. Measurement connections GaAs FET. ! Safe 2’0” 30 &DSuB 40 50 60 2 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 bias sequence for GaAs FET susceptible ALL FROM "EP4142-DRV" OPTION BASE 1 ASSIGN @Hp4142 To 717 CON @Bp4142 INTEGER Gate,Drain I Drain=1 Gate=2 Vdsl6.0 Vgsginch=-2.5 Vgsl=--8 vgsa=o I Init-hp4142 Ch-sw-on Force-v(Gate,Vgs-pinch,O,.Ol) Force-v(Drain,Vds,O,l) Force-v(Gate,Vgsl) Measure-i(Drain,Idsl) Force-v(Gate,VgsZ) Measure-i(Drain,IdsZ) Force-v(Gate,Vgs-pinch) Force-v(Drain,O) Zero-output Ch-sw-off Gm=(Ids2-Idsl)/(VgsZ-Vgsl) END Figure 17. Measurement program I Source 1 Drain I Gate for to Idss : GNDU : Channel : Channel damage 1 2 1 1 I 1 Connect SMU channels to FZT Vgs-pinch set with 10 mA compliance Set Vds after setting Vgsqinch Raise Vgs above Vgs-pinch 1 1 1 I Return Vgs to Vgs-pinch Remove Drain bias before Remove remaining bias Disconnect safely biasing a GaAa for FET. Gate bias Figure 17 gives a programming example of a safe turn-on procedure for a microwave GaAs FET with very low gate and drain break down voltages. The HP 4142B control software includes easy-to-use commands which take care of all setup and measurement requirements. 13 2.3 Advantages of DC Pulsed Bias The clean pulsed waveforms of the HPSMU (1 A max) and the HCU (10 A max) allow precision testing at high currents by reducing thermal drift in the DUT and overheating of bias tees (rated at 500 mA max continuous DC). The 100 ps minimum pulse width of the HCU is still well within the cutoff response of the bias tee lowpass filter response to allow both accurate DC and RF characterization in one fixture. 100 ps Pulse And with a 1% duty cycle, heating can be reduced to below 10 degrees C. This becomes significant with MMIC devices where heating can change electrical parameters such as HFE or Vth by 0.5% per degree C. Less device heating correlates to greater system accuracy and repeatability. FET threshold voltages are quite sensitive to thermal drift. Figure 19 shows excessive drift when measurements are performed with a non-pulsing DC power supply. I 5 - - z----------- 4 - HCU 0 Width -50 0 Junction 50 100 Temperature 150 (“12) b 0 Figure 18 illustrates how junction temperature can vary widely with different pulse widths. The HCU can operate down to 100 ,us pulse widths. (degree Power supply V Figure 19. The HCU allows pulsed down to 100 ps and 1% duty cycle to reduce device heating. C) Synchronized 4 ‘/ 100 vs 1 ms Swept Pulsing Figure 20 shows a MMIC traveling wave amplifier which electrical can be characterized as a high power FET. Synchronized dual pulse sweeps allow quick generation of family curves such as Id vs Vds. Notice the simplicity of software control statements. Use of supplied HP 4142B software driver routines simplifies development of custom MMIC test algorithms. See figure 21. 10ms Pulse width Figure 18. Pulsed measurements improve system accuracy by reducing thermal drift. I ORAIN BIRS I cR.F. R.F. Figure OUTPUT INPUT - 20. A MMIC travelling wave amplifier is tested as a power FET. 14 10 % it"0 s: ifi 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 1 High current pulsed sweep with ! LOADSUB ALL FROM "BP4142-DRV" OF'TIONBASE 1 ASSIGN @Hp4142 TO 717 COM OHp4142 INTEGER Base-,Collector,Vc-steps DIM Ic( 101) synchronous pulsed : GNDU : HPSMU Channel 2 : ACU Channel 5 Emitter ! Base-=2 Collector=5 Vc-start=.1 Vc~stop=10 bias ! Base I Collector Vc-steps=101 1c_comp=lO Ib=5.E-2 ! Init-hp4142 Ch-sw-on Setgiv(Collector,l,0,0,Vc~start,Vc~stop,Vc~steps,2.E-4,2.E-2,0,Ic~comp) I Set Zero-output Ch-mu-off END Figure 21. The HCU operated tandem with standard SMUs for synchronized pulsed testing. pulsed voltage source in 2.4 Fixturing Considerations when Using the HCU To allow full 10 A sinking of ground currents without resistive losses, a special active ground is supplied in each HCU channel. The HCU is really two SMUs in one box. Its floating design keeps ground currents internal to the HCU. The RF system ground and the DC ground reference (GNDU) are isolated from these currents, which increases noise immunity and extends the current sinking capability of the HP 4142B mainframe. See figure 22. sweep 1 Set pulsed current bias source f Trigger both sources synchronously and 1 store swept measurement data in Xc(*) Dpulse_i(Base~,O,O,Ib,2) Dsweepgiv(Collector,Z,O,Ic(*)) II II I \ vs / GNDU Figure floating allows to sink currents 1oA. DUT 22. Its design the HCU pulsed up to s 1 S 15 HP 160!38A/E3 Test Fixtures To simplify device hookup of packaged parts to the HCU, SMU or other HP 4142B channels, the HP 16088A/B test furtures are recommended. These fixtures terminate all coax and triax channel cables properly in a dark shielded box. Standard and customizable sockets provide safe and accurate hookup to any device pin using quick, jumper-wire connections. The 16088B supports an internal three-input matrix. This allows the full range of HP 4142B resources to be applied to a device pin. Low Inductance Figure 23. HP 16088A test fixture and accessories Cabling Sense Low Normal coaxial wiring capable of handling 10 amp currents would add 5 pH inductance to the DUT. Oscillation would be very likely due to typical capacitive inputs of MMICs in the order of 1 pf. By coupling the high force . and low force wires with a special twisted and shielded design, inductive effects cancel one another to allow ~0.1 pH in a 1.5 m cable. Such specialized cabling is an integral factor allowing the HCU to retain the same &digit accuracy and resolution as the standard SMUs in the HP 4142B product family. Sense High Force High Force Low Figure 24. Coupling force high and force low currents in the same cable greatly reduces cable inductance. 16 2.5 High Speed Testing of Monolithic Amplifiers Features l An application well suited to the HP 4142B modular DC source/ monitor is high speed production testing of microwave components such as the HPMA-200 series of single stage bipolar gain blocks. Gains of 12 dB at 1 GHz and bandwidths of 3 GHz are typical. Such components are typically priced at less than $1 each and must be 100% tested for DC functionality and RF gain/flatness specs. Automatic testing of up to 10,000 units/hour is required to make production economically feasible. This means DC bias, RF sweep and data gathering must be completed in as little as 200 ms. l l l 3 dB bandwidth: 2.4 GHz DC to 11.6 dB gain at 1 GHz Cascadeable block 50 ohm gain Low cost surface mount plastic package Figure 25. HPMA-0211 Silicon Monolithic Amplifier The AFU Increases Speed and Stability SMUl Figure 26. Typical amplifier bias configuration. SMU2 Test The automatic feedback unit (AFU) and two SMUs of the HP 4142B are designed to bias such high-frequency devices in as little as 25 ms without overshooting or spikes. As seen in figure 26, SMU2 monitors the amplifier current. The AFU controls SMUl as determined by the monitored current. This method contributes to a stable bias point, free of thermal drift. The AFU slew rate is programmable from 0.5 Vfsec to 50 kVf set without spikes, thus controlling bias overshoot. 17 Fully Automated High-volume Test Station Using a bias tee such as the HP 11590B with bias sensing, reliable measurements to f500 mA are possible. And using a fast scalar analyzer such as the HP 8757, at least 10 RF readings are possible in less than 50 ms. A complete block diagram of a production-line bench test station is shown in figure 27. The DUT is packaged in a plastic four-lead surface mount configuration (SOT143). Commercially available handlers easily move the SOT package in and out of a strip-line fixture. This fixture is designed to test S-parameters over the required 3 GHz bandwidth of SOT amplifiers such as the HPMA-0211. HP 11590B Bias Network 3: f:: SE The sample code in figure 28 shows a testing sequence which quickly searches for the desired bias condition using the AFU. The HP 4142B then releases control of the HP-IB bus and allows the HP 8757 to apply RF input power over a range of frequencies and extract RF out power levels at channel A. The HP 4142B has the ability to store entire test sequences in memory and then execute them on command, once for each new device tested. This eliminates the time that would be required to program the HP 4142B over the HP-IB bus each time a control command is issued. ‘I HP 8757E Network Analyzer HP 11590B Bias Network Figure 27. Using the HP 4142B to achieve 10,000 unit per hour throughput for DC and RF test. 1:: Programmed TMemory F%thtir Increases Test Speed I, I i:: 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 310 320 330 340 350 I Using programed I to quickly bias memory and analog feedback an amplifier for RF test APTION BASE 1 ASSIGN @Hp4142 To 717 ASSIGN @Hp8757 M 714 OUTPUT $Ilp4142;"*NST" INTEGER Input,Output DIM A$[311 I Input=1 Output=2 1 1 1 t Ground Input Output ICC Ibias ICC==,05 vcc-10 1 OUTWT QAp4142;"ST";l OUTPUT QBp4142;'CN";Input,Output OUTPUT QHp4142;"ASV";Input.0,2,100,.1 OUTPUT QEp4142;"hVI";Output,Vcc,Icc,. OUTPUT QHp4142;"i4SM";1,4,.0002 OUTPUT QHp4142;"HM";6 OUTPUT QHp4142;"XE" OUTPUT QHp4142;"END" OUTPUT QLip4142;"RU";l,l ENTER QHp4142;AS Iblas=VAL(A$[4,15]) OUTPUT QHp4142;"DI";Input,16,Ibias OUTPUT QFlp4142;"DV";Output,12,Vcc I OUTPUT QHp8757;"SC";Start_posn OUTPUT QHp8757;"OC" ENTER QBp8757;Value-a S21(Start~freq*lO)=Value-a Using -_ programmed memory can reduce test times below 300 ms for a complete DC and RF test of monolithic amplifiers. The test cycle includes DC leakage, DC current, RF gain and RF gain flatness tests. : : : I : RF Ground Plane SMU channel 1 SMU channel 2 Target output current Input bias 5 I I f 1 1 1 I 1 f Store program in memory Connect channels Set search SMU channel Set sense SMU channel leg feedback lnteg. time Analog search meas. mode Trigger search End of program in memory Trigger program in memory ! Bold amplifier I eettings for I Begin BP Test, Figure 28. Programming ~~~~dp&o~~~hmeasurement. bias RF test 521