When considering a flash storage media that should satisfy the rigorous demands of an industrial environment, engineers may want to pay particular attention to temperature, industrial levels of testing, life-time support, and long-term availability...
Another observation on system level is also that the flash memory controller is as crucial as selecting the memory itself. Together, controller and flash define the system quality. Hence, both need to be considered as a whole. In this article, we’re going to investigate why not all forms of memory subsystem, SSD or USB flash drives are created equal, and some of the critical factors engineers need to consider when making a buying decision for industrial equipment designs.
For consumer electronics, operating temperatures might vary slightly. Still, in most cases, we’ll be using our computer, for example, in ambient temperatures in the low 20°C. For consumer electronics generally, the permissible temperature range is typically from 0°C degrees up to 40°C, or sometimes as high as 60°C. However, within the industrial domain, the temperature and the amount it varies can be more extreme. The quoted industrial temperature range is from -40°C to +85°C.
While all components and modules designed to be used in industrial applications will meet the extended operating temperature range, it is important to remember that the temperatures stated are for the ambient environment. Internally, at the junction, the temperatures will be much higher. Testing these devices at ambient temperatures of, say, 125°C, gives greater confidence in their reliability. To mitigate the impact of high ambient temperatures, engineers should aim to select components that have a lower operating power. This will help keep the internal temperature low and have a positive impact at the system level.
Attempting to use flash memory and a controller in temperatures they are not designed for is going to introduce concerns of potential failure at any stage. Failure is not a term popular with industrial engineers. Halting a production and restarting can be an expensive incident. A sudden failure of equipment in an uncontrolled way might also cause significant damage that is costly to fix and may result in further downtime.
You might ask why you should not purchase a 32 GB USB flash drive for, say, $10 in a retail shop, but instead buy another one looking very similar for $30? We will try and answer this question in more detail but you can already note that for industrial applications every module is tested with a more demanding use-case in mind and ensuring a constant high quality of products for a long life-time.
Industrial-grade reliability starts with design fabrication.
Making sure that a memory system is designed to withstand industrial temperatures is, perhaps, the end of what is a long design process that commences with industrial qualities at the foundation of the hardware. For example, at Hyperstone, our flash memory controllers are designed for industrial applications already on intellectual property (IP) block level. The IP design approach is far more stringent than for consumer devices with respect to reliability and fail safety.
Proven testing techniques are applied to stress-test the IP. This ensures that the design includes a safety margin with respect to critical aspects, such as clocking and timing. For example, voltage drop analysis is applied, which contributes to the definition of industrial quality.
When temperature is one primary consideration, factors such as hardware timing are very important. It requires design-time and possibly chip-area to make sure that a chip operates reliably at different temperatures. Years of experience led to an evolution of product features and built-in reliability. Each new controller continues to incorporate industry use-case knowledge and application specific support gained over many years.
Firmware development is another critical element. It needs to be approached with the same industrial mindset and run in parallel to the hardware design, closely linked to it. The firmware used in flash controllers must include futureproofing in terms of its flexibility. Industrial systems have a typical operating lifetime of a decade or more, normally in very demanding environments. By following development techniques that ensure the firmware is developed in a highly structured way leads to greater code coverage during verification. This, in turn, results in higher quality firmware that is as reliable as the hardware it runs on.
Finally, testing is a fundamental part of delivering systems that conform to challenging industrial use. In addition to temperature cycling, tests for sudden power failures are part of an intensive framework to validate that products operate as intended. The sudden power fail test simulates a complete loss of power. To maintain data integrity, when the system powers back up, you need to be sure that not data has been lost. At Hyperstone, for example, we test each new product and firmware through intensive power cycling.
Another vital consideration when using flash memory is data retention which is heavily dependent on operating temperature. Data is lost from flash over time, and the higher the ambient temperature, the greater the rate of data loss. The loss occurs whether the equipment is active or not – another reason why temperature is an important consideration. Typically data retention characteristics for flash memory are quoted at 25°C. In an industrial setting, it would be fairly common to see equipment experiencing 60°C or so, and this has a profound impact on the data retention. For example, if you increase the temperature of a memory system from room temperature to 60°C, the data retention rate likely reduces by a factor of 20. That sounds like a lot. It is a lot. Data may be lost after 6 months instead of 10 years. The choice of flash as well as refresh features managed by the controller can protect your data.
Ensuring long-term product support
Only a few suppliers really live-up to providing long-term product support. Your selected memory and controller supplier should be with you every step of the way, from initial design concepts through to in-life product. This last point is particularly relevant for industrial applications where the in-service life may be ten or even twenty years rather than two or maybe maximum five years for consumer products. For example, Hyperstone’s dedication to longevity is exemplified by still supporting CF card controllers that were introduced in 2003.
Suppliers need to engage with the customer as early as possible in order to design best to their applications’ requirements. For example, at Hypestone, our first question might be: ‘What is your Use Case?’ Together with customers we can use tools such as our Use Case Tracker to assess precisely how an application is accessing and stressing a storage media. We can determine specific parameters, such as access patterns, write and read speed, access frequency that are required to optimize the firmware, choose the right configuration and capacity and thereby maximizing reliability and optimizing cost.
Later during a product’s service life, problems may be encountered that can only be analyzed by the controller vendor. Debugging interfaces and protocol features integrated into controllers and firmware architecture help vendors like Hyperstone analyze faults should they occur. In 99.9 % of cases, we can see what happened and take action to prevent it from occurring again. Try doing that with a consumer flash controller!
When considering a flash storage media that should satisfy the rigorous demands of an industrial environment, engineers may want to pay particular attention to temperature, industrial levels of testing, life-time support, and long-term availability. Aligning the configuration to the use-case with the system can save money and headache. Otherwise you might end-up with a great solution that does not fit your application.
— Steffen Allert is VP Sales at Hyperstone.