Determining Multioutput Power Supply Unit MTBF with LabVIEW and PXI

Author(s):
P. Kannan - Soliton Technologies Private Limited
Ganesh Devaraj, Ph.D. - Soliton Technologies Private Limited

Industry:
Medical/ Medical Instrumentation, Electronics

Products:
Compact FieldPoint, FieldPoint, PXI/CompactPCI, LabVIEW

The Challenge:
Developing a full-featured, flexible, and reliable test system to determine the mean-time-between-failure (MTBF) of a multioutput power supply unit.

The Solution:
Developing, in 10 weeks, a highly configurable, scalable, and feature-rich reliability test system for DC and AC power supplies using the National Instruments PXI platform and LabVIEW Real-Time software.

Developing a Precise, Versatile System
Our customer, a leading global manufacturer of medical equipment, wanted the MTBF of a sophisticated multioutput power supply unit to be determined precisely through reliability testing. The objective was to develop a system that would reduce the cost and time needed for the test. Additionally, the investment in the system had to be protected by making it versatile to handle different types of power supplies, modular, and scalable.

The power supply unit (PSU) had three outputs: low-voltage DC, high-voltage DC, and single-phase AC power, each rated in the kilowatt range. Four such units had to be connected to the system for simultaneous testing. Each of the three outputs would have to be subjected to its typical loading pattern for a specified number of cycles (as high as a million) while being monitored for failure. Different criteria could be set up to denote the failure of the PSU, for example, a voltage dip beyond a specified value under peak load.

Our primary task of the test system was to apply the load as required for each of the 12 outputs and monitor the voltages and currents (24 channels). A set of critical parameters are calculated from the waveforms for each cycle and recorded. At the end of the test, the results from all of the units tested had to be consolidated and the MTBF calculated. All the units may not actually fail during the test, but by extrapolating the trend in the values of the critical parameters, the failure time of these units are predicted and used for the MTBF calculation.

Using LabVIEW Real-Time to Build the System
Because the system had to run without interruption for many weeks at a time, we chose to build the system on the NI LabVIEW Real-Time platform. To provide the scalability to test multiple units, we chose the 18-slot PXI-1006 chassis. We designed the hardware and software to be modular so the user could start and stop the tests on each unit without interrupting the other units under test.

For each output, we designed a loading pattern that simulated the stresses experienced during a typical loading cycle, while reducing the cycle duration to shorten the overall testing time. The loads were primarily capacitive and resistive, and so we also designed a “load box” with a network of capacitors and resistors. In order to simulate the loading pattern precisely, the system switched various capacitors and resistors in and out of the load circuit with a timing precision of 10 µs. This was accomplished using the NI PXI-6534 high-speed digital pattern I/O cards along with switching MOSFETS with a response time less than 50 ns. Two cards were used, with the 32 lines on each card configured into two groups of 16 outputs each and connected to the four load boxes. The pattern output provided extremely precise timing synchronization in the switching with completely independent control of each group.

The system had to measure the voltages and currents at 100 kS/s to faithfully capture the transients. From each PSU, the six channels were connected to a NI PXI-6070E high-speed data acquisition card. The throughput from the card was 600 kS/s and the combined throughput from all four cards was 4.8 MBPS. We used high-bandwidth 100 KHz Hall-Effect transducers for isolation and signal conditioning. The system also had to monitor numerous alarm status lines from the PSUs for which a PXI-6508 96-channel digital I/O card was used. We installed thermocouples in the quite large load boxes to monitor their temperatures at various locations and raise an alarm and instruct the PXI controller to turn off the unit. The thermocouples were connected to FieldPoint thermocouple input modules and a FieldPoint RT controller was programmed to handle this task, reducing the load on the main (PXI) controller.

The main challenge was processing the 4.8 MBPS of data in real time, storing the computed parameters for each cycle, and, when any irregularities were detected, storing the entire cycle’s waveform to disk. We accomplished this by judiciously spreading tasks that did not occur each cycle over multiple cycles and using the cooling-off periods. For each PSU, a separate loop was launched to enable independent monitoring and control. The system provided the status of each PSU on the run screen shown on the monitor of the display PC connected to the PXI RT system. A safety loop monitored the alarms from the PSUs and from the FieldPoint RT system monitoring the temperatures and switched off any unit in alarm.

Meeting the System Requirements with LabVIEW Real-Time
We met the very demanding requirements for this system in terms of performance, reliability, versatility, scalability, and safety, along with all of the productivity-enhancing features, and delivered in a period of 10 weeks purely because of the high-performance hardware products and drivers from NI and the ease of programming of a real-time system provided by LabVIEW Real-Time. We estimate that we reduced the development time and system cost by a factor of three due to our choice of hardware components and development tools and produced a system that met and exceeded all of our customer’s requirements.

For more information, contact:
Ganesh Devaraj, Ph.D.
Managing Director
Soliton Technologies Private Limited
www.solitonautomation.com

Browse All Case Studies »