Many thanks to Keith Curtis of Microchip, who was great guest. With his backgorund in CNC machines, RF applications, and even gaming, Keith was able to provide a practical perspective to our radio show.
I fully agree with the thrust of the article. I am always impressed with the capabilities available on-chip with the new MCU's. These devices also make it quicker and easier to implement a design. One of the real skills that are often overlooked is the ability to match the device to the application. Not only is it important to utilize the capabilities given the designer by the vendor, but it is important to match the MCU to the application. I have seen these situations where a particular MCU was used when something less fancy (and expensive) would have sufficed.
I have had the chance to see several MCU's used fully and appropriately. The only issue i have noticed is life expectancy. I work in aviation where operating temperature has large swings from cold to hot. the life of an IC is truly tested in these conditions. It is probably the reason why MCU's have had such slow integration in automotive and aviation. Not claiming i have done any research on this. This is just my opinion.
That's an important observation, Ervin. In most of our discussions about MCUs, we talk about the hardware in terms of clock speed, # of bits, (and now number of cores), etc. On the software side, we talk about the application, whether it's real time, how do you write it (in C or with the new graphical drag-and-drop software tools). But rarely do we discuss thermals and environmental factors. These, as you write, are hugely important when it comes to aerospace apps. The same thing applies (perhaps to a slightly lesser degree) in automotive. Also any embedded app in a harsh environment (out in the field) will need to be robust enough to handle temp and humidity swings, may require encapsulation etc. So this is all very relevant stuff.
It is always possible to come up with a design less satisfactory and more expensive than the one that you chose. The assertion that an A/D converter is needed when reading speed is about the weakest that I have heard. Just read pulse period time and calculate speed. A count and a divide a constant. Of course using some simpler sensing mode that goes right to an input and does not need any conditioning would often be the best choice, if it would do the job adequately. But to read speed with a capacitive touch input you wil still need to have something that varies capacitance at the required resolution, and assure that that something runs true enough to not produce a cyclical error in the reading due to being off-center. In addition the relative effect of electrical noise in the area must be accounted for.
For some products, continued availability of a particular controller can be a serious issue, while for the product intended to be obsolete and out of production in six months continuing availability is much less of a concern.
But in all cases there is a great deal of merit in considering the various means of achieving the required sensing function. Tradition is not always the best guide. In addition, clearly understanding the performance requyirements is also vital.
I fully agree with William. In all cases there is a great deal of merit in considering the various means of achieving the required sensing function. Tradition is not always the best guide. In addition, clearly understanding the performance requirements is also vital.We got to get the NEED statement spot on.
Also the biggest issue to the life of a MCU that i did not mention is failure mode. It is difficult to predict how an MCU will fail. The worst part is that an MCU can continue to switch and mimic operation even in a failure mode by passing wrong data, or having a pwm continuously operate damaging other devices or falsifying information. So while MCU's give a simple solution to everyday life, in flight critical, life critical, or main safety systems they are a high risk device to use. There is some background noise in the market for a market ready MCU that is capable of checking if each internal component is working. The check circuit is built in. and supposedly this MCU has only one Failure mode. Inoperable pins and peripherals are off. No power out, high impedance setting. I have not looked into this as of now, however the moment i hear something other than rumors i will have a starter kit.
Texas Instruments has a new family of Hercules "safety critical" MCUs that use ARM Cortex cores. Each processor includes two ARM processors that operate in locksetp, cycle by cycle. Built-in self-test (BIST) operations demand no software, so you don't use code space for this capability. Memories and buses include parity for error detection and communications provide cyclic-redundancy checks. TI has several inexpensive development kits and small evaluation modules for each of the three device families. For more information, go to:http://www.ti.com/lsds/ti/microcontroller/arm-safety_hercules/tms570ls_rf4/getting_started.page. TI aims these provessors at industrial, medical, transportation, and similar products. --Jon
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.