Smart power management
Being smarter with available energy usually consists of profiling component energy use in the system and identifying where the biggest savings can be made to significantly impact overall power usage. As an example, the biggest consumer of power on most wireless remote monitoring equipment is the radio used to synchronize with the network and transmit data. Knowing this, system designers try to turn on the radio as little as possible, meaning they must either decrease the module sampling rate or turn to more advanced methods to decrease the communication with a host system, such as using measurement nodes that they can program. Programmable nodes can perform logic, such as sending a sample back only if it exceeds a threshold or performing averaging over longer periods of time and then transmitting only the average value. Getting smart at the node level by using selective sampling and sample averaging in a remote monitoring application can result in dramatic power savings.
Engineers trying to detect the presence and concentration of oil in the Gulf of Mexico, for example, could program the node to return samples only when the samples exceed a certain minimum threshold and, furthermore, to return an average measurement of the concentration over the course of 10 minutes instead of once a minute. Nodes not making impactful measurements would never turn on their radios, and those making impactful measurements would be 90% more efficient while maintaining the desired measurement frequency. Such drastic power savings in the monitoring and data acquisition methods used in remote extreme applications are very valuable because engineers can start using that power to perform other critical tasks like implementing control.
Many extreme remote systems these days perform control in addition to data collection. Control systems need, in general, higher data acquisition rates, appropriate outputs in addition to sensor inputs, and additional processing power for more advanced calculations. These additional requirements, and the introduction of real-time controllers and field-programmable gate arrays (FPGAs) to meet them, can make implementing an embedded control system a daunting task. Two specific areas for embedded control that are receiving a lot of attention for their potential to simplify the process are system modeling/simulation and the conversion from a prototype to actual deployed hardware.
Prototyping a control system for extreme environments can be difficult because it’s hard, or near impossible, to replicate the conditions under which the system will be deployed. This makes physical testing and prototyping costly and time-consuming and limits the number of design iterations. Yet if system designers are trying to design a robust control system that is going to operate correctly under extreme conditions, they need to extensively test it.
As computing power increases and modeling software improves, more and more embedded designers can prototype their control systems in software and define simulation environments in which to iterate through initial prototypes quickly without incurring the traditional costs associated with scale testing. Control design and simulation, as the process is commonly called, is resulting in more robust controllers and an accelerated development timeline because engineers can run a controller through much more rigorous test regimes under any combination of simulated conditions.
Engineers also can combine control design and simulation with virtual prototyping (collaboration between a 3D modeling software containing a hardware model and a high-level development language containing a system controller model). This results in a more efficient physical design because engineers can test many more iterations and combinations of materials and designs for the physical prototype in conjunction with the control algorithms in software before moving to an actual physical prototype. With the combination of control design and simulation and virtual prototyping tools, engineers can design systems to tighter tolerances and with smaller margins of error, both in the control response of the system and in the physical design.
Figure 2. Simulation of a wind turbine in a graphical development tool for control design and simulation
Click on image to enlarge.
Control design and simulation and virtual prototyping techniques can’t fully replace physical testing because the results are only as good as the mathematical models the simulations are based on, and many simplifications and assumptions are made in models of real-world systems. Engineers are still a long way from confidently predicting the precise and complete behavior of complex real-world systems with a software model. Yet these models do provide greater visibility through the design process so system designers can be much more efficient with the physical prototyping and testing they do perform. The more problems that can be caught and fixed while still prototyping in software, the more effective and efficient the design process will be.