Yes, semiconductors do go away. A company called Reticon--acquired years ago by EG&G--made a programmable filter I liked very much. No replacement. Likewise, Intersil manufactured a dual-slope (integrating) amalog-to-digital converter. That went away, too, although Maxim made the chip, too, but I haven't checked recently. There are some secondary-market companies that have stocks of older ICs for companies that need replacements.
There's a balance between choosing recent semiconductors that have some history and longevity vs. picking the latest-and-greatest devices that might be a "flash in the pan." Engineers and product designers should always ask a supplier about the longevity of a device or product line.
Yes, it is true that the 8051 is still around, and in quite a few variants, as well. But try to find an 8047. And of course the *86 is still around. But neither of those would be considered a "small" processor, I don't think. Also, it looks a bit like many of the newer crop of small devices are constantly changing, hoping to find a configuration that is what everybody wants. The less popular versions don't lkast as long. Sort of like the flat-pack CMOS ICs of a few years back.
Of course there are also those devices that go away because of yield problems, although that may be less common than a few years back. Presently it seems that Analog Devices has stopped producing one of the audio compressor devices, and a few other audio chips, for some reason. So products do go away.
Hi, William. Yes, watch you don't get cut on the "bleeding edge" of new technologies. That said, the venerable 8051 continues to live as does the Z80 family. I believe Zilog still has variants of the Z80 family and Rabbit Semiconductor (now part of Digi International) has modules based on the Z80 architecture. Of course the x86 architecture continues to roll on, too.
Long after whatever MCU that you choose has gone out of production and is not available anywhere, thye same analog ICs will still be available from multiple makers and distributors stock. So if the anticipated product life is measured in days or months, then choose the MCU approach. But if it is a product with an expected lifetime of years, then make it out of sustainable parts. (A new expression?) Cutting edge stuff often causes bleeding.
Hi, AndyT. The "memory" connector is actually the connector for a TI MCU "ControlCard," already in place. The odd perspective of the image places the upper edge of the ControlCard along the same line as the far edge of the small motherboard. Look again and you'll see a board plugged in. The MCU has a lot of memory. The connector lets engineers and programmers use different types of ControlCards.
I have to disagree, 8 bits are still very much alive, especially when combined with a good compiler and plenty of memory, which is now pretty cheap. I was looking into crunching some color graphics a while ago and did two designs; one with an arm and the other with a SiLabs 8051. The 8051 was faster and cheaper in this application.
In most of the small appliances and sensors I work on, an 8 or small 16 bit machine is still my first choice. It's hard to justify a full 32 bit core when I only need 4k of code.
Hi, Island_Al. Yes, plenty of life left in 8-bit MCUs for a wide variety of uses. A few days ago I sketched out a neat circuit for model-railroad enthusiasts that would use an 8-bit PIC in an 8-pin package--and an assembly-language program.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This discussion will examine what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.