As part of the Quiet Technology Demonstrator 2 (QTD2) project, Boeing flight tested new technologies intended to reduce noise generated by its aircraft. Measuring the improvement these technologies provide required a flexible, accurate, and scalable test system to perform phased array acoustic imaging during the tests. We needed a distributed system architecture with the ability to expand to up to 1,000 channels or more while still maintaining tight timing and synchronization between channels.
Phased Array Data Acquisition and Analysis
To flight test new technologies for quieter operation, we conducted research at a facility in Glasgow, Montana. We used an array of microphones to acquire noise data, which we then processed into noise-level maps showing from where and at what frequencies the noise was generated and how loud it was.
By overlaying the noise-level maps with a visual image, we could assess the effectiveness of noise reduction technologies, identify opportunities for additional noise-source reductions, and distinguish between engine and airframe sources.
Using NI tools, we could then validate several advanced noise-reduction concepts, including chevrons on the engine exhaust ducts, new acoustic treatment for the engine inlet, and aerodynamic fairings for the main landing gear.
Previous System Limitations
During the first stage of the QTD project in 2001, we deployed a VXI test system that was limited in both channel count and channel bandwidth. The system required a centralized data architecture that required us to co-locate all the VXI chassis for synchronization, necessitating long cable runs from the microphones to the data acquisition system – about 10 miles of cable per 100 channels of data acquisition.
In addition to the channel and architecture limitations, we faced challenges including time delays when synchronizing instruments across multiple VXI chassis, significant cost per channel, and significant time required for data retrieval. We wanted to deploy a new system in the second stage of the project (QTD2) that would address these issues.
NI System Solution
Using the flexibility and modularity of PXI, we were able to create a scalable system with virtually unlimited channel-count capability. In addition, by taking advantage of NI timing and synchronization cards, we could distribute the data acquisition hardware into the microphone array, decreasing cabling by nearly 80 percent while maintaining within one degree of phase match between channels.
To collect the data, we used the National Instruments PXI-4462 dynamic signal acquisition module, which provided acquisition rates up to 204.8 kS/s. We used eight PXI chassis, each containing the NI PXI-4462, PXI timing and synchronization cards, and fiber-optic connections. With the timing and synchronization cards, we distributed the acquisition clock and start trigger to every data acquisition channel in the system.
Each fiber-optic card linked a PXI chassis with a National Instruments PXI-8350 server-class machine running Windows XP and NI LabVIEW. We were able to separate the chassis from the controlling computer by up to 200 meters with the fiber-optic link. We connected each NI PXI-8350 controller through Gigabit Ethernet to one central host computer for faster post-acquisition data recovery to the host computer and other systems used for data processing and analysis. With increased performance and an unlimited, distributed architecture, we reduced the cost per channel by more than 50 percent compared to our previous system.
The Phased Array Flyover Test
We outfitted the test facility with more than 600 ground-based microphones arranged in a custom spiral pattern spread over the end of the runway in a 250-foot-wide by 300-foot-long area. We acquired the noise of a 777-300ER as it flew overhead and immediately retrieved and processed the data to get an acoustical image of the airplane. A data processing computer cluster connected to a host computer via Gigabit Ethernet analyzed the data in real time.
During a typical test cycle, the aircraft flew over the microphone array approximately every six minutes. The system was able to upload the previously acquired data and be ready to acquire more data within that window. During the test sequence, we conducted more than 300 acquisition events, yielding 78 minutes of flyover results – more than 1 TB of data.
Hardware System Architecture
To create a system that is scalable to 1,000 channels, the NI system architecture uses multiple PC-based controllers and PXI chassis. In this architecture, a master chassis controls timing and triggering while slave chassis distribute clocks, control local acquisition, and store data to disk. A host computer controls the configuration of all the PXI systems, provides the user interface for software setup and control, and receives all the data from each PXI system. A master PXI chassis controls timing and triggering while slave chassis receive the timing and triggering signals, acquire data locally, and store it to disk. We could transparently and remotely control the PXI systems with the PXI-8350 1U rack-mount, server-class controllers bundled with a fiber-optic link, giving us the flexibility to distribute the dynamic signal acquisition devices in several clusters around the microphone array with the device controllers located in a trailer up to 200 meters away.
Based on commercial off-the-shelf hardware, Serial ATA drives configured in RAID 0 installed in the PXI-8350 let us stream all channels directly to disk at full sampling rate. This modular system gives us the framework to easily scale channels as needed to reach a higher channel count, or to divide up the system for lower-channel-count applications.
Software System Architecture
We developed the system completely in LabVIEW. We were able to directly reuse or easily adapt code and designs from other Boeing developers and from the NI Web site. Even with the LabVIEW learning curve, one person developed the entire application in less than six months.
By taking advantage of a carefully chosen software architecture and the modular nature of PXI systems, we simplified the process of scaling the system. We clearly demonstrated this when, midway through development, we needed to add 128 channels to our system. It only took about two hours to scale the system from 320 to 448 channels – from unpacking and plugging in the input modules to making a two-minute update to a configuration file.
Timing and Synchronization
We used National Instruments PXI-665x timing and synchronization control modules to provide tight synchronization among modules in a single chassis and to extend timing and synchronization to multiple chassis. Using a combination of NI PXI-6653 master modules with NI PXI-6651 slave modules allowed all the PXI chassis to operate using the same clock. Cables distributed the timing signal throughout the system, allowing up to 200 meters of chassis separation while still maintaining tight synchronization among the dynamic signal acquisition devices. With this architecture, we could match all 448 channels spread over eight chassis within one degree at 93 kHz.
Dynamic Signal Acquisition
Looking ahead during our data system selection process, we knew that we needed a system we could use for a broad range of applications, from full-scale tests to scale-model tests in a wind tunnel. We also needed a system that had higher sampling rates and a larger dynamic range than our existing system. To meet these needs, we selected the PXI-4462 dynamic signal acquisition module with four simultaneously sampled input channels and 93 kHz bandwidth.
For full-scale tests, the frequency of interest is typically no higher than 11.2 kHz; however, higher sampling rates are required for wind-tunnel tests with scale models as small as 1:20. With 24-bit, sigma-delta analog-to-digital converters, we could measure signals as low as 1.25 microvolts. With the integrated electronic piezoelectric (IEPE) integrated current source for sensors provided by the PXI-4462, we achieved a 30X cost reduction and greatly reduced the complexity of the transducers for certain applications.
Using NI software and hardware, we were able to create a high-end, low-cost system that could distribute the acquisition system across multiple chassis, tightly synchronize all channels, provide high channel count with full bandwidth on all channels simultaneously, and allow virtually unlimited channel-count expansion. With this new system, not only were we able to improve capabilities of the individual acquisition channels, but we also achieved a 5:1 reduction in the amount of cable required and cut the cost of microphone systems by 30:1 for flyover test applications.
Boeing Aero/Noise/Propulsion Laboratory