SparkFun Forums

[MJ]

All,

I read and hear so much about how "useless" 8-bit microcontrollers have become with the advent of much more capable microprocessors that can easily handle 32-bit calculations and have tons of program memory available. A recent article on Embedded.com alludes to the near-term disappearance of the 8-bit guys we know and love, and we all have seen to be quite capable. Granted, we're not inverting a 4000x4000 matrix, but A/D, PWM, UART, etc with a decent chunk of processing space can do a lot.

I guess I've got two questions...
1) What's your take on the future of PICs (and friends in the same category) given that processing power is becoming cheaper and smaller?
2) Do you know of examples in which PICs have been used for large scale production of consumer-grade products? With order quantities in the 10s of thousands available, that's appears beyond the hobbyist level.

Thanks,
Jeff

[e]

I really dont see an end to 8-bit processors in any short future. Many projects and products just dont need all the features or speed as a 32 bit floating point processor or advanced processors.

As far as I know many PICs and other 8-bit processors are used in consumer products, in fact I doubt they would still be around with just sales to hobbiests...

Just my take.

[SOI_Sentinel]

Cost will always be the driving factor. ALWAYS. Many basic designs (say, your toaster) use 4 bit microcontrollers because they're a few cents (or maybe just a few tenths of a cent) cheaper than a PIC12CXXX. Those situations are what the PIC10's were designed to try to fit into. When you're talking about a volume of millions a year, those few rounding errors will add up.

Now, what I do see is the end of 8 bit microcontrollers as something you'd settle on. If you have a lot of 16 or 32 bit numbers you have to crunch on, you'll choose a 16 or 32 bit processor instead of working with a now only slightly cheaper 8 bit processor that runs slower. Why would I build a brushless motor controller around a PIC18F processor when I can use a similarly priced dsPIC and get both better speed and smarter control because it has the native 16 bit number capacity to run the numbers?

As for using PICs, here's a few more generic things:

http://channels.lockergnome.com/news/ar ... late.phtml

Otherwise, I haven't found any specific examples, although them having top spot in the micrcontroller market does tell me they're somewhere... anyone want to help me tear apart some random new car and see if they're in the electronics? You bring the car.

[sparky]

Good note about the car. Last I heard, the Lexus had 240+ microcontrollers. 6 alone in the Anti-lock brake system.

Micros will always be around. Who would ever dream of sticking an ARM into a TV remote? There are far more applications where a low-end micro can handle the job then there are customers wanting to pay an extra $2 at Wallmart.

ARMs are wonderfully powerful chips that will bridge the gap between the embedded world and the portable electronics world. What used to require a PC and a few PICs could now theoretically be encapsulated into one ARM with ethernet/wifi.

The power of the micro is more than just $/MIP performance any more. It's simply amazing what you can squeeze out of batteries these days. ARMs and FPGAs, while being absolute speed demons, still consume quite a bit more power than 8/16 bit micros. MSP430s are the king when it comes to low power, but PICs are a close second. If you need battery life, micros are the only way to fly.

I like to think that microcontrollers are best for students first learning hardware. I took a class in college called "Computers As Components" where we learned the 8086 architecture. It was nearly a complete waste of time. The architecture was far too old and convelluded - you needed an entire personal computer to attempt to learn about interrupts. Why not just get a PIC to blink an LED? It's a much better way to learn about registers and hardware. IMO.

Great post/question,
-Nathan

[phalanx]

Another way 8 bit controllers are used is to off-load processing tasks from your more powerful master CPU. One example would be a distributed motor control system with a graphical user interface. There are several PICs with motor controllers built into them which can be setup to run a motor in any way the programmer sees fit. Through its parallel slave port, any number of PICs can be addressed by the master CPU rapidly and efficiently. The master CPU can issue new running instructions to the PIC controllers (which can run without supervision since the PIC is an intelligent device) and that frees up substantial processing time for graphics display. At the same time, the PICs could be serving as additional clock sources, communications modules, and as a user interface to the system.

And of course, they cost much less than the big 32 bit controllers. Even a few cents savings per device on a 10,000,000 unit order can save you hundreds of thousands of dollars!

-Bill

[MJ]

In the spirit of answering my own question I found this article, which explains the importance of embedded systems in major systems, namely aerospace applications. But they specifically mention that BMW distributes much of its computing capability among some 80 microprocessors (albeit with some difficulty).

http://www.embedded.com/showArticle.jht ... D=57700545

Jeff

[Guest]

I'm not sure who is saying "useless", but it really depends on what you doing. If you are making something with limited production, you don't want to fool around with the idiosyncrasies of the earlier 8-bit machines, If you are going to produduce 1,000,000 units/year, you are going to want to spend weeks, even months, trying to cram the application into a chip that costs 50 cents less. That's $500,000 in a year!

I think that it's fair to say that for hobbyists, if there as any doubt, you want to use the more expensive part, because the money involved is squat, and chances of success are better. I don't fool with the 16f series, unless I am looking at making a number of units. On the other hand, some people just just want to see how much they can squeeze. Depends on what you are trying to do.

I don't think that microchip would have rolled the 10f series out last year if the low end was about to disappear.

[MJ]

I found another article explaining a specific application of a microcontroller--and it's a 16F88 to boot, (pun intended, if you read the article).

http://www.embedded.com/showArticle.jht ... D=21700389

Engineers (mech-e's nonetheless) put an F88 in the heel of a shoe to measure gait/pace/force/etc using a Hall effect sensor and use the information to adjust the firmness of the heel cushion.

I fully believe these little guys are all over the place and aren't going away; i just like to see specifically where; partly to reassure me that my little hobby could pay off, and also just out of curiosity.

Jeff

[MJ]

Alright--I'm just talking to myself now, I think; but here's that shoe mentioned above (on the embedded.com site) presented at a technology show. Still cool; tiny little processor guy adapts to your running style.

http://news.bbc.co.uk/2/hi/technology/4337519.stm