Choosing the right microcontroller
Posted by HSPalm at
I use 8-bit microcontrollers, some 16-bit (TI MSP and Atmel xMega) for hobby projects.
Size of memory are not something I consider much. Most of my project ideas rarely exceed 4K in code size, unless it has some heavy math or bad delay loops or something.
First of all, packaging. Is it gonna be a prototype, or am I ordering custom PCBs (DIP vs SMD)? Then I start counting all the I/O I need so to not put in a overly big part in my design. When I have found the package with the correct pin count, I start to look closer to the project idea itself and look for peripherals needed. Timers, PWM, ADC and serial buses are what I use the most, but many times I don't need either of them.
I almost always use Atmels AVR, I just feel so stuck with them. I know what's wrong when things are wrong, and I know exactly where to begin when I want to expand my knowledge on them (wanna learn I2C? PID loops? just check out the atmel app notes!). I know AVRs are cheap compared to most 8-bit micros, so I gladly pay for them where they're priced reasonably.
And by the way, digi-key, farnell, rs components and all the other major distributers charges huge money for shipping to Norway, so availability in sites like taydaelectronics.com and futurelec.com is a huge plus.
Replies
-
Agree that familiarity with the architecture and its tools, and having those tools to hand, is a huge plus.
With a familiar architecture, I will usually have some sort of hardware already to hand on which to do preliminary tests/experiments - without having to obtain and/or design anything specific.
That's why I like ARM and 8051 - because then I'm not tied to any particular manufacturer, and have a huge range to choose from.
Then it's finding the right peripheral mix, number of IOs, memory, etc...
Note that Digikey do free shipping to the UK for orders over £50 - none of this unavoidable £15 surcharge nonsense from RS and Farnell...!
-
Engineer and hobbyist requirements run by different ways, nevertheless many features and performances may be similar both markets. Now days there are a lot of similar solutions products that may be used so the selection is probably the compromise between lots of matters, not only the very best chip for certain application.
I believe we have to take into consideration the programming language we are going to use. If we plan to use C language the chip has to be one designed for that purpose. There are C compilers for almost any chip but efficiency is improved a lot when the chip is C compatible.
Another consideration is supplier support, especially for hobbyists. Support may be critical in project success, now days all suppliers are supporting excellent forums.
Price may become one of the most important considerations for large scale production, a supplier that have a wide portfolio of products is interesting to allow choosing the cheapest chip that match with the requirements and programming do not need lage modifications from chip to chip, in that case the alternatives is reduced a lot.
Other considerations, like memory, I/O’s, USART, PWM, etc. are covered widely by most of suppliers. Code solutions available like IP stack, zegbee, etc. has to taken into consideration but in that case most suppliers are quite similar.
It is very valuable test board and accessories for evaluation and initial project develop at low cost. VGA display, touch screen, WiFi, etc. accessories for evaluation boards is very important for professional applications. Once we have our platform probably it is no more need, but new projects we have to contain costs as low as possible and shot time to market, availability of evaluation HW that may cover almost of our needs probably is the key for the final chip selection. If we have something that allow start coding that’s all I need.
Third part supplier, HW and SW, for the selected family chip is important too because allow certain flexibility. FreeRTOS supported chip has to be taken into consideration for certain applications if we use that RTOS. There are other RTOS that support few chips so migration in that case makes impossible if new chip is not supported.
8, 16 or 32 bits?, Clock MIPS?, DSP support?, 8 bit chips may be OK for many applications, but the 16 and 32 bits new chips not only are faster they provide more memory, I/O’s, internal RTC, fast HW CRC generator, etc. so for certain applications we better use a 16 or 32 bits chip because it saves external HW and in some cases SW like CRC generator. In chip DSP availability is very good not only because is faster than I2C port, it is cheaper too and saves space in our board reducing parts to mount.
If you want a trade mark or device proposal I believe Microchip covers and exceeds all consideration I have described, but there are other too.
-
A microcontroller uses the same techniques to shrink the entire computer to a single chip (or very small module.) CPU, Memory, and I/O all in a little package as small as a grain of rice. Just connect up power and it starts doing its thing; computing and talking to the world.
First of all check for number of ports you required.Speed of a microcontroller dosent matter because we won't use the entire efficiency of the microcontroller.
Then after go for the timers, are they programmable or not.Check weather your microcontroller is ISP or On board programmable. Then go for how much memory you required. There are microcontrollers like ATMEL's 89c51 which are having ondboard 64k flash ROM and 2k RAM.
-
A question for electronics design engineers and hobbyists alike: When choosing a microcontroller for an electronics design project, what factors are most important to you and why?
-
Im impressed, I need to say. Really not often do I encounter a blog thats both educative and entertaining, and let me tell you, you will have hit the nail on the head. Your concept is outstanding; the difficulty is something that not enough persons are speaking intelligently about. I'm very completely satisfied that I stumbled throughout this in my seek for one thing relating to this.
Life Insurance upland Free Porn ooporphyrin