Reducing test costs in high-volume component production using test sequencing instruments

Tightening profit margins are driving component manufacturers to reduce production costs, including test costs. Using instruments with embedded test sequencers can help.

Increased demand for testing

As product complexity increases, the cost of production test increases. Higher complexity means additional test functionality, which increases equipment cost and space. For example, integrating analog, digital, and even RF circuits on a monolithic integrated circuit means higher density and pin count. Higher pin counts require more test channels to maintain acceptable throughput.

Expanding the number of test channels is only part of the solution. The current practice of focusing functional-level testing on products at the end of the production line also needs to change. Failures found there are costly to manufacturers. A better solution is to move more testing to the front of the line. This way, bad components can be removed earlier in the production process.

In order to take more effective test means to improve profit margins, manufacturers need to consider new test paradigms and ways to build test systems. In many cases, new test technologies and instruments can be used to reduce system and test operation costs and improve the performance of existing test benches. The new generation of SMUs (source measurement units) with advanced embedded test sequencers and data communication capabilities can achieve this.

These features of SMUs allow for compact and economical systems to be built for fast, multi-channel component testing. They are also highly flexible and can be modified quickly and easily as test requirements change. Systems designed with these instruments can take on more testing earlier in the production cycle and help reduce the high cost of end-of-line testing.

Typical measurement requirements

At first, it may seem impractical to create an instrument-based system that fits the needs of different component manufacturers. The DUT (device under test) may be a simple 2-pin or 4-pin device such as a diode, LED , or transistor. These tests require very simple source-measure procedures with fast transient response to generate accurate IV curves on two channels. Since the components are manipulated very quickly (less than 100 milliseconds per part in some cases), instrument speed is critical.

For components like resistors and RC networks, multi-pin transient voltage suppressors, and EMI filter arrays, all individual components must pass testing before an array is passed. Therefore, parallel multi-channel testing is required to achieve high throughput.

As the complexity of components increases, the number of tests and channels also needs to increase. For semiconductor devices, test equipment should be able to adapt to wafer-level testing because packaging is expensive. In the early stages of production, static and leakage currents of various complex low-power devices need to be measured at the nanoampere level. In addition, simple DC measurements are required for all DUTs on the wafer to check basic functions. Because there are tens of thousands of DUTs on each wafer, fast multi-channel testing must be adopted.

A common need in the production of all these components is to apply voltage or current to the DUT using a repetitive test sequence, measure its response, compare the measurement results to acceptable limits, and make a pass/fail decision. The basic design of SMUs makes them well suited for this type of testing. However, production test engineers need to carefully consider the differences in instrumentation and test system architecture to select the best test instrument for the task at hand and try to anticipate future testing needs.

For multi-DUT testing or multi-channel testing of more complex devices, parallel channel IV systems are used to increase test throughput. However, test speed may still be limited by the instrument, application, or DUT settling time. Limitations of existing parallel channel systems include continuous channel hopping (i.e., all parallel channels cannot be measured simultaneously), slow measurement range changes, and low data communication speeds.

Scalable multi-channel systems are designed for more complex components and test scenarios. They often include different instruments for various test functions. SMUs are often a core component, and broadband instruments (signal generators, oscilloscopes, spectrum analyzers, etc.) are often added externally. The two most common architectures are integrated functional testers and IV test systems with open API (application programming interface) instruments.

Open API means that independent instruments are installed into a customized test system, which is usually achieved by the user or system integrator. In comparison, a functional or parametric tester is a completely pre-installed (turnkey) system, in which most of the hardware and software integration has been done for the user before delivery. The disadvantage of this system is that it is relatively expensive. Open API systems provide users and system integrators with a highly flexible solution and have the potential to achieve lower costs.

Figure 1: TSP-based multi-SMU test system

Using the latest SMU to achieve higher throughput

Once the test fixture is loaded, most of the test time is consumed by the following time periods:

1. Signal source applications, including voltage or current transients;

2.DUT stabilization time;

3. Measure and change the range when necessary;

4. Trigger delay;

5. Data communication;

6. Procedure execution, including pass/fail and packing decisions;

7. Test fixture movement and/or electrical switching time.

By converting stand-alone instruments to an integrated SMU test system, trigger delays and data communication time between stand-alone instruments and PC controllers are reduced. Some SMUs have program memory that can run up to 100 predefined tests, and can use or not use a PC for limit comparisons, conditional program branching, etc. This reduces slower GPIB traffic and PC latency.

In a single-channel system, it is relatively easy to achieve test time improvements using an SMU with test program memory; however, in a multi-SMU system, reducing test time is much more complicated due to the difficulty in managing multiple triggers and test sequencers.

Because of this, older generation SMUs have sequencers that are command-prompted only, meaning they store multiple GPIB commands to be executed with a single call from the PC. These SMUs typically do not have the logic to perform limit testing or make instant pass/fail decisions. They also do not have a DUT control interface, so there is a lot of GPIB traffic. In addition, many older designs do not allow for parallel channel testing—channels are accessed sequentially, so throughput improvements are limited.

The latest SMU designs (sometimes called smart SMUs) have test script processors (TSPs) and high-speed control buses that address these issues. This allows simple programming to run complex and high-speed test programs that have been downloaded and stored in the instrument. This is achieved using advanced resource sharing methods across multiple SMUs.

For example, in the Series 2600 System SourceMeter instrument from Keithley Instruments, a master-slave arrangement allows parallel measurements of all channels. This architecture also allows easy expansion of multiple channels. Therefore, test engineers can take advantage of other SMU functions in a variety of test applications. These features and functions include:

1. Memory for 100s high-speed embedded test programs;

2. Voltage and current pulse and sweep capabilities;

3. 4-quadrant IV operation;

4. Wide source/measurement dynamic range (1uV to 200V; 1pA to 10A);

5. 6.5-digit digital resolution;

6. High-speed instant pass/fail testing;

7. Digital I/O for trigger management and component control interface.

In addition, the smart SMU has the function of pulse and low-frequency arbitrary waveform generator, which can be applied to each channel. It simplifies complex testing by providing universal analog I/O pins for multiple applications.

By adopting these features and changing the way test systems are programmed, it is possible to greatly increase throughput. Instead of relying solely on PC-based control, the SMU's test sequencer and program memory can control most tests. In Keithley Instruments' Series 2600, throughput is further improved by adopting the SMU's test script processor, high-speed sequencer, and fast control bus (TSP-Link). Multiple SMUs can be connected together according to the number of channels required (up to 128) and used like a single instrument. TSP-Link technology uses a low-latency 100Mbps serial bus to enable multi-channel IV scanning between multiple SMUs.

These smart SMUs also feature wide dynamic range and seamless range switching. This is very important when measurements are to cover a wide range, as adjusting the range can consume a significant amount of source-measure time. Figure 1 depicts this type of multi-SMU system. Its throughput is comparable to a mainframe-based system.

Other cost savings

Smart SMUs make it easy for system designers to integrate units and shorten software development time. For example, Keithley Instruments' TSP provides an intuitive command language similar to Basic as a simple programming interface. Using TSP and Test Script Builder software, it is easy to create complex test sequences to control multiple SMU channels as a single entity. Evolving test requirements can be easily adjusted with minimal SMU hardware changes.

Rack space and hardware also increase test system costs. The high-density design of new SMUs uses a space-saving 2U half-rack form factor. This allows multi-channel systems to remain in a test rack during the transition to high pin counts. This type of rack-and-stack system eliminates the large hardware and overhead costs typically associated with a mainframe system. Reusable SMU hardware and simpler software development will further reduce costs when changes need to be made to the test system.

Conclusion

Keithley Instruments' Series 2600 represents the next generation of SMUs that meet the needs of high-pin-count device production and multiple DUT testing for fast throughput and cost-effective automated test systems. These instruments make it easy to add new features and capacity to existing test benches and further reduce the cost of new test benches. They provide convenient expansion capabilities, simpler system integration, and a small test bench footprint, reducing test system development time while increasing flexibility, performance, and reliability.

author:

Andrew Armutat

Product Marketing Department

Keithley Instruments

appendix

Scalable instrumentation for fast and accurate IV testing

Keithley Instruments' Series 2600 System SourceMeter instruments provide electronic component and semiconductor device manufacturers with a cost-effective, scalable, high-throughput solution for precision DC, pulse, and low-frequency AC source-measurement testing. Each SourceMeter instrument is both a highly stable DC power source and a true instrument-grade 5.5-digit digital multimeter. In operation, these instruments can act as a voltage source, current source, voltmeter, current source, and ohmmeter.

In IV functional test applications, Series 2600 instruments test 2 to 4 times faster than competing solutions. They also offer higher source measurement channel density and lower cost of ownership. The analog/digital converters provide simultaneous current and voltage measurements in less than 100μs (10,000 readings/s), and the source measurement sweep speed is less than 200μs per point (5,500 points/s). This high-speed source measurement capability, coupled with advanced automation features and time-saving software tools, makes the Series 2600 SourceMeter instruments ideal for IV testing of a wide range of devices.

The Series 2600 instruments use an innovative technology that allows economical creation of multi-channel IV test systems without sacrificing test throughput. TSP-Link is a high-speed system expansion interface that can be used to connect multiple Series 2600 instruments in a master/slave configuration.

Once the connection is established, all instruments in the system can be programmed and operated under the control of the master unit as if they were installed in the same chassis. TSP-Link provides unlimited flexibility to increase or decrease the number of test system channels, depending on the application requirements.

Series 2600-based systems can run high-speed embedded test scripts on the main unit's Test Script Processor (TSP). This is another new technology these instruments have. Test sequences are processed and run on the embedded computer in the instrument rather than using an external controller, eliminating delays caused by GPIB traffic congestion.

TSP test scripts can achieve up to 10 times the throughput of equivalent PC-based programs operating over GPIB. TSP test scripts can be run from the front panel or through the system's GPIB interface. A single TSP test script can control all SourceMeter channels in the system and acquire data from any instrument connected to the TSP link, supporting up to 64 Series 2600 instruments.