NASA Fair Rover (see figure) landed on Mars in January 2004 with a brother or sister, Spirit. This was going to be a short run, but Opportunity has overtaken the life of his project – although the latest dust storm on Mars could cut this short run. Part of the reason Opportunity lasted longer was due to the pre-engineering done by NASA scientists, which was necessary because they were dealing with the unknown. Mars is famous in general, but the need for autonomous operation was determined by the distance between Earth and Mars.
The design and construction of tools and systems that go beyond basic specifications is often desirable and sometimes by design. For example, the “headroom” in terms of computing power and memory is often increased for devices that may have a long capacity for future upgrades. The additional facilities allow changes to be made which will normally require these facilities. This approach was previously less common with electronic devices, as most of them were remote. Not so these days with the Internet of Things (IoT), where over-the-air connections and updates (OTA) are possible.
NASA ‘s Opportunity rover has survived far beyond its planned voyage that began in 2004.
Consider costs
The problem is that overengineering comes with costs on both the design as well as the implementation side of things. It can take longer to create because design tends to be more expansive and, of course, additional resources usually cost more initially and over time. Similarly, these costs could exceed the very purpose of the design. A low cost wireless sensor would be inconvenient if it weighs a ton.
Cost is not limited to the physical aspects of design. Consider an application that uses only 8-bit data. Would a 32-bit microcontroller be a desirable platform to implement? It is inefficient to store 8 beats in a 32-bit word, and pack and unpack data is above them. Of course, the detail will be unique in terms of application; one such application may be more efficient on an 8-bit micro and another would work better with a 32-bit micro.
For example, controversial neural networks (CNNs) are in high demand these days with machine learning embedded in a number of applications. These often use 8-bit data or even fewer beats, but also involve thousands of computers per feather. Efficient hardware design typically revolves around a lot of data even though the individual items are small, making wider buses and tables better. Overwriting can guarantee wider data widths. However, if this makes very few calculated improvements while significantly increasing costs in terms of power, performance, and so on, that is not always a good idea.
Reliability and Security
The idea of being above engineering covers all aspects of design. Reliability is critical in systems such as aircraft, where catastrophic results occur when the systems fail. High standards are obviously part of a design specification, but engineers usually go over “just to be safe. “Self-driving cars and semi-autonomous vehicles make such design requirements more common.
Ultimate security is becoming a more common requirement, especially for IoT devices – including self-driving cars. Here, overengineering can occur, although security is often neglected rather than over-built. Part of the challenge in this aspect of design is what real engineering is, as it is more difficult to determine the impacts and costs.
For example, what does an increase in the number of hits in encryption cost in terms of performance and power consumption? What would be the payoff in terms of a higher level of security? Similarly, is there another security cover, such as using a hypervisor, to be an advantage or an unnecessary overindulgence?
Things become even more complicated when the whole system is considered. For example, a board designer who is more familiar with the digital side can specify the voltage regulator spectrum that significantly exceeds the system requirements, as they may not be familiar with the specific requirements.
It is often unknown why designers choose technologies like SLC NAND flash memory over QLC NAND flash because they believe that the reliability, speed and longevity of their predecessors are essential. for the application without knowing what those requirements really are. That’s not good if machines get into the field and fail early. The flip side is that they could live as long as Opportunity, but it could cut into profits.
The unfair aspects of the problem arise because of its own costs, life, etc. Values given to renowned car engineer Ferdinand Porsche are, “The perfect race car crosses the finish line in the first place and immediately falls apart. ”
Some appliances, such as cars, have key parts that are easy to replace, such as tires or fuel. Unfortunately, most systems have many parts that are not so easy to replace, and these tools are often useless if a necessary part fails. The typical example is an IoT battery-powered device that only works while the battery is running. This can be 10 minutes or 10 years. Tools at these edges can be useful and by design. For example, a device that needs to turn off temperature for a few minutes when thrown in a watt of molten metal is comparable to a temperature sensor designed to monitor ambient room temperature for decades.
Getting Edgy
So how close to the edge are you drawing?
I learned programming back when collector was common. There are still collectors, but the average programmer rarely goes down to this point. I developed a digital camera system for small robots that used an 8-bit SX microcontroller. It was fun to pack so much activity into the small amount of available memory. The time for reading data from the camera chip was coded so that the instructions were in line with the camera data. The only time the system waited for synchronization was at the beginning of a data line from the camera.
Eventually, the memory filled to the point where I could only disable parts of the system because the debugger required two bits of primary memory. That was enough for a jump table entry, where I could discuss one set of code at a time using conditional assembly to build the test systems. The final version used both of these measures. Fortunately, the potential was so great that it was possible to conduct independent experiments.
Engineers and programmers are often constrained by these types of boundaries, but that’s usually the exception rather than the rule, especially these days when so many options are available. System design is often replaced by overexuberant overengineering. These are obvious, but it’s a reminder that developers need to be aware of the impact of the choices they make while designing a system. This can be as fundamental as choosing a programming language, choosing a processing platform, the type of concrete used, or whether plastic or paper will affect the operational capability of a long-term solution.
Engineers often think they are “outside the box.” Such a philosophy usually leads to new and useful approaches to problems. However, it should not lead to undesirable and undesirable transgender solutions.
