A customer selected Green Mountain Research to design an instrumentation system to remotely operate and monitor mechanical, optical, and electronic components of missile systems. Technical requirements included analog signal acquisition, analog signal generation, live data and video integration, functional control via remote relays, and capabilities for HIL performance characterization. The customer chose us for this project based on our successful implementation of an earlier system using the NI PXI platform with LabVIEW Real-Time.
Distributed System Architecture
The CompactRIO platform not only met our requirements, but it was the only solution that could meet the following criteria:
- Offers FPGA determinism and real-time connectivity in a single package
- Can simultaneously perform a variety of interdependent tasks, including analog input, analog output, data processing, digital input, and data streaming
- Can be reprogrammed for unique and changing test requirements
- Seamlessly integrates with IP networks, VISA instruments, and remote user interfaces
- Is mechanically and thermally sturdy
- Meets efficient mass and volumetric parameters
However, one CompactRIO chassis could not achieve two of the project objectives single-handedly: high-channel-count data acquisition and remote tactile switch control. Thus, we implemented multiple chassis, as shown in Figure 1. The MXI-Express expandability of the NI cRIO-9082 controller proved to be an ideal feature to increase our channel capacity with a single controller. We chose an NI 9159 MXI-Express RIO expansion chassis to construct a high-throughput system with a total capacity of 50 NI C Series I/O modules. We installed additional NI cRIO-9076 controller chassis at remote locations and connected them to the instrumentation system via Ethernet. We used these remotely located controllers for tactile switch operator input via digital I/O.
The MXI-Express four-chassis CompactRIO system, shown in Figure 1, is the core component of the network-distributed instrumentation system. This system is the sole hardware interface to the devices under test (DUTs) and can operate up to four missile seekers simultaneously. It includes the following:
- 56 synchronous and 12 asynchronous 24-bit analog input channels (NI 9229 modules)
- 128 relay channels (NI 9481 and NI 9485 modules)
- 32 digital input lines (an NI 9403 module)
- Four synchronous analog output channels (an NI 9269 module)
A key differentiator between this platform and our previous missile-testing instrumentation systems is the true differential and isolated circuitry of the analog input channels offered by NI C Series I/O modules. In addition to the ±60 V input range of the NI 9229, we can use these characteristics to operate DUTs with floating-ground potential, and without external signal conditioning. The result is a system that performs with a lower level of electromagnetic interference and has fewer components to maintain.
The element that connects the hardware interface to the power sources and the remotely located user is Ethernet. Connectivity is implemented between real-time controllers, instruments, and operator GUIs by TCP/IP, UDP/IP, and NI-VISA drivers. The following tasks occur via the Ethernet interface:
- Live data streaming of analog input data (collectively, ~125 Mbit/s) to client workstations
- Live data streaming of operator control commands from remote real-time systems to the relays
- Determination of modes of operation, which are input at the client workstations and forwarded to the CompactRIO controller
- Voltage control and current polling of network power supplies using VISA
Thus, Ethernet forms the foundation upon which the distributed system architecture is based. Advantages of using Ethernet include near-universal compatibility of the employed protocols and ease of adding more instruments, clients, or real-time controllers. A disadvantage of using Ethernet is its nondeterministic performance. We have overcome this limitation by isolating deterministic time-critical requirements to real-time and FPGA cores while optimizing networking parameters to use the minimum possible bandwidth and maximum uptime.
We established a software architectural plan that takes advantage of the deterministic capabilities of the CompactRIO real-time platform to perform time-critical tasks while allowing users to interact with controls and output data streams on network-attached Windows workstations. The tasks that demand the highest level of determinism and lowest latency throughput are analog signal acquisition, signal generation, certain signal analysis routines, and HIL simulation. We programmed these processes with the LabVIEW FPGA Module across four individual FPGAs to operate at predetermined clock rates and to deliver steady streams of raw and reduced data to the cRIO-9082 real-time controller through the built-in DMA first-in-first-out memory buffer capability. We used LabVIEW FPGA Express VIs to provide digital signal-processing functionality, which significantly reduced development time and effort. Assistance from National Instruments applications engineers and field engineers through the Standard Service Program also proved instrumental in completing this project quickly.
Processes that require a high level of determinism but involve floating-point calculations or that are too computationally expensive to deploy on the FPGAs were programmed with the LabVIEW Real-Time Module to run within Timed Loops on the cRIO-9082. We configured these processes to run on core 1 of the dual-core controller. This core was designated for all deterministic operations, while core 0 was designated for all nondeterministic network routines. The ability to divide application tasks between two physical processor cores generated excellent performance and stability. We performed a methodical parametric study to determine packet sizes and transmit and receive frequencies that result in low latency and reliable data availability to network clients. We also developed algorithms that can simultaneously transmit data to multiple network clients or receive and arbitrate command inputs from multiple network controllers. Since network threads are inherently nondeterministic, we implemented the capability to automatically reconnect between the cRIO-9082 and cRIO-9076 real-time controllers and the Windows workstation clients. Although the OSs are different, a substantial portion of the source code that we developed for real-time usage was also usable in the Windows applications.
The Windows applications that are part of this instrumentation suite are shown in Figure 3. With these applications, the user can monitor the status of the real-time controller, visualize and analyze signals, and issue functional commands to the DUTs. Since the common element throughout the instrumentation system components is the IP network, we integrated a third-party data-archiving application by simply receiving TCP/IP packets in a DLL.
Deployment and Implementation
The missile test platform that we developed not only meets all initial requirements but also expands for additional requirements in the future:
- The Core i7 processor of the cRIO-9082 is currently being used at only 40 percent capacity.
- The MXI-Express link can expand to accommodate up to three additional MXI-Express RIO chassis if needed.
- The network architecture can accept additional real-time control systems or Windows clients.
CompactRIO performance, customization, and expansion potential make it an excellent long-term, scalable investment. Green Mountain Research continues to customize and add capabilities to the missile test platform while operating and maintaining the current configuration.
Albion W. Knight IV
Green Mountain Research, Inc.