Design Article
Tell us What You Think
We want to know what you thought about this Design. Let us know by adding a comment.
Optimizing functional-test throughput in PXI-based automated test systems
Alan J. Lesko, Agilent Technologies
10/19/2010 2:46 PM EDT
To ensure ongoing competitiveness, most manufacturing engineers are looking for new and better ways to optimize test throughput while ensuring measurement integrity. Manufacturing targets can be very aggressive and two key factors — testing methodology and instrument selection — can make or break a test-time budget.
The selected test methods must be efficient, focusing on validating functionality of the device under test (DUT) while gathering parametric data to help improve the manufacturing process. Also, instrument selection must consider both speed and measurement integrity. PXI test platforms form a good basis for high throughput functional test systems. This article will discuss test methods and instrument characteristics that will help ensure successful implementation of future test systems.
Improving functional test
In manufacturing, test plans take various forms. A true functional test places the DUT into an electrical environment that emulates the actual application. Connections are loaded and driven as if the DUT was in the final application and the unit is closely observed to verify correct performance.
Unfortunately, a true functional test can be extremely slow and therefore may be impractical in many instances. An alternative test method called DUT-assisted test (DAT) utilizes a communication link to the DUT combined with built-in test commands to isolate and exercise specific portions of the device. This approach yields greater flexibility during the test plan and enables a very fast and efficient testing. Best-in-class manufacturers use DAT methods to achieve high test coverage and a throughput advantage.
Exploring an example
To help understand key concepts in DAT, let’s consider the manufacturing test plan for a high-volume automotive engine control unit (ECU). A review of the test plan shows that it is very transaction intensive, with many DUT setups and measurement transactions over the course of the test. Figure 1 shows a Pareto chart breakdown of the individual transactions.
This ECU has a large pin count (more than 150 pins). To meet capacity targets, the total test time for this ECU is expected to be an impressive 15 seconds.
It is easy to see that low test system latency (i.e., the time delay of each programming transaction) is very important. Even a few milliseconds of latency for each programming transaction would significantly impact total test time—and may even force the addition of a second tester to meet capacity targets. The old adage “time is money” is very true in this case.
Fig 1: ECU test plan transactions.
Examining the Pareto chart
The top three items in figure 1 highlight the key areas to address in pursuit of the aggressive throughput targets for the example ECU.
Switch activations are the #1 item in the Pareto with 400 switching statements. Because relay technologies vary dramatically in speed, proper selection of the relay switch is critical to achieving high throughput. Armature relays can carry large currents but may take tens of milliseconds to switch. High-speed reed relays can switch in hundreds of microseconds while still carrying up to 500 mA of current.
Fortunately, the majority of switching statements are used for low-level measurement signals and other low-current routing. For low-current switching, it is common to use a high-speed reed relay matrix or multiplexer.
The selected test methods must be efficient, focusing on validating functionality of the device under test (DUT) while gathering parametric data to help improve the manufacturing process. Also, instrument selection must consider both speed and measurement integrity. PXI test platforms form a good basis for high throughput functional test systems. This article will discuss test methods and instrument characteristics that will help ensure successful implementation of future test systems.
Improving functional test
In manufacturing, test plans take various forms. A true functional test places the DUT into an electrical environment that emulates the actual application. Connections are loaded and driven as if the DUT was in the final application and the unit is closely observed to verify correct performance.
Unfortunately, a true functional test can be extremely slow and therefore may be impractical in many instances. An alternative test method called DUT-assisted test (DAT) utilizes a communication link to the DUT combined with built-in test commands to isolate and exercise specific portions of the device. This approach yields greater flexibility during the test plan and enables a very fast and efficient testing. Best-in-class manufacturers use DAT methods to achieve high test coverage and a throughput advantage.
Exploring an example
To help understand key concepts in DAT, let’s consider the manufacturing test plan for a high-volume automotive engine control unit (ECU). A review of the test plan shows that it is very transaction intensive, with many DUT setups and measurement transactions over the course of the test. Figure 1 shows a Pareto chart breakdown of the individual transactions.
This ECU has a large pin count (more than 150 pins). To meet capacity targets, the total test time for this ECU is expected to be an impressive 15 seconds.
It is easy to see that low test system latency (i.e., the time delay of each programming transaction) is very important. Even a few milliseconds of latency for each programming transaction would significantly impact total test time—and may even force the addition of a second tester to meet capacity targets. The old adage “time is money” is very true in this case.
Fig 1: ECU test plan transactions.
Examining the Pareto chart
The top three items in figure 1 highlight the key areas to address in pursuit of the aggressive throughput targets for the example ECU.
Switch activations are the #1 item in the Pareto with 400 switching statements. Because relay technologies vary dramatically in speed, proper selection of the relay switch is critical to achieving high throughput. Armature relays can carry large currents but may take tens of milliseconds to switch. High-speed reed relays can switch in hundreds of microseconds while still carrying up to 500 mA of current.
Fortunately, the majority of switching statements are used for low-level measurement signals and other low-current routing. For low-current switching, it is common to use a high-speed reed relay matrix or multiplexer.
|
|
|
|
|
Navigate to related information