-
HELP! What MCU (greater than 8-bits of smarts) is easiest to graduate to?
Posted by SuzyC on 26 Aug, 2014 15:44 -
Maybe I need to start a new post: Which processor is the easiest to graduate to from an 8-bit to start doing some highly educated management of my MMMMM's?
I have succeeded in controlling on/off speed/direction of my MMMMM's but I don't have any memory or processing power to do anything else!!
Right now, as a basis for comparison I have the following "conveniences:"
(1) I can edit, compile and program the 16F886 chip in less than 60-seconds. I simply unplug the DIP chip and pop it into my programmer and bingo! That's very important to me, since MCU hardware programing is always full of unexpected surprises!
(2) 5V supply. I cannot help but find it so easy to interface to and do A2D accurate work with and have a high noise-immunity.
(3)I can work with a CCBB (Common Chinese BreadBoard), makes it so easy to construct, modify any idea I have.
(4) I can make multiple PCB's each using a SOIC packaged version of a PIC MCU to make small modules; so each instantiation is something quite small enough and compact to fit into a tiny space of my robot.
(5) I can always aford'm find and buy a dozen, a large supply of (10x cheaper than Arduino) PICC chips when I blow the #!**$ out of one and still have enough money to buy a new dress at the end of the month. ( I have a large collection of somewhat zorched PICC chips, more than a dozen, in my parts drawer.) So this is what fits my tight budget that relies on salvaging parts form discarded electronic equipment and then buying new only what is necessary, practical and cheap.
What I don't have is any way to simulate my code. I have to just carefully and tediously examine the unexpected result, study my scope readout and then look over and over my code, then maybe a few hardware spike scope markers, that is all I have to try to make any sense out of whatever I am doing when it almost always does not do what I think it should be doing. This does certainly slow down my catching up to Japan in robotics.
Right now, if I attempt to use anything more than a few 7-segment approximations of display, and even using only a measly
16 chars of feedback, I quickly exhaust a great deal of the 8-KWords of program memory.
Right now, if I try to do something interesting with my MMMMM's I cannot because I have run out of memory. Any attempt at a more than a very few floating point calculations to unleash the power of my hare-brained Al Gore Rhythms results in my program memory, processing speed and R/W ram space being sucked dry.
Ok, so I've tried the ATMEGA2560, its got more RAM, but with the ATMEL IDE 6.x. It take s'many mouse clicks here and there, and so more than a few minutes to make a minor edit of the source, exit, compile and download the code to the MCU. I'll have gray hair if I have to work that slow! I also don't like the bulk of the Arduino board, it doesn't fit into the tight corners of my robot!
So what works well, is better, I donno?? DO YOU?
p.s. If you don't know yet what a MMMMM is: look at my previous post. It's my moronic mutl-MCU modular approach to multi-motor management. -
An ATMEGA1284 has 16K and available in a DIP suited to breadboarding. Despite the popularity of Arduino for beginners, doing significant software projects using tiny RAM is actually a fairly advanced skill.
If you want more RAM and speed, then some sort of 32-bit ARM is going to be the way to go. You would find moving to an ARM with 64K or 128K hugely liberating. Maybe consider an Arduino Due or Teensyduino platform. -
Thanks Gardner,
But I don't know from beans about 32-bit programming.
You say try an ARM..I say "Which ARM?"
I have yet no idea how I might go about learning to setup the hundreds of registers in some 32-bit MCU in a C++ place to wrap my head around, seems so far so far a distant goal.
I have perused a few ARM offerings by ATMEL, but haven't found anything yet that promises to be always available, cheap, modular (DIP or break out board adaptable to common breadboards), low power consumption, small footprint, doesn't take an ARM and a leg to buy a compiler-IDE-debugger-simulator-chip programmer that works easy and well and cheap. I even see on this same page of this topic forum the post about a guy complaining about not having any availability of an ATMEL MCU. Atmega32U4 Availability Problems
Doesn't 32-bit programming require that every mcu instruction uses 4x the program memory than an 8-bit MCU? How do I learn to graduate from an 8-bit char variable mentality to the 32-bit world of the ARM?
Which ARM does best work to graduate to that has the economy of programming time, as well as the cost conveniences I have described above? -
I have yet no idea how I might go about learning to setup the hundreds of registers in some 32-bit MCU in a C++ place to wrap my head around, seems so far so far a distant goal.
This is where the low-barrier to entry platforms -- Arduino and Tennsyduino are going to help you. You don't have to know or learn ANYTHING to get a hello world project going. They are aimed at school kids and artists, and are very easy to get started with.
The Teensyduino is on a little board that you could stick some headers on and stick it into a breadboard, no problem.QuoteHow do I learn to graduate from an 8-bit char variable mentality to the 32-bit world of the ARM?
You will continue to write C or C++ programs in the usual way. Nothing much changes there. -
Quote
any memory or processing power
Needing more memory is easy to solve - you can always get a bigger chip.
Needing more processing power requires (or may require) considerable uplifting. A few options for you, each with their own significant challenges:
1) ARM: a lot bigger chips and far more powerful peripherals. Much steeper learning curve and a new set of tools and a new approach to programming. Rarely they are 5v compliant.
2) PIC24 (including the dsPIC chips too): more processing power, slightly steeper learning curve, and similar tools. Marginally more complex peripherals. Many are 5v compliant. Unknown future.
3) advanced 8051: some firms, like Silabs, make amazingly fast 8051 loaded with features. The high end ones surpass many ARM chips in terms of processing power. New tools to learn.
Personally, I find 5v tolerant to be a non-issue for ARM chips. However, the toolchains can be a challenge for a newbie. But it is highly dependent on the individuals involved.
-
Maybe I need to start a new post: Which processor is the easiest to graduate to from an 8-bit to start doing some highly educated management of my MMMMM's?
Why would you mess about unplugging the chip ? I have succeeded in controlling on/off speed/direction of my MMMMM's but I don't have any memory or processing power to do anything else!!
Right now, as a basis for comparison I have the following "conveniences:"
(1) I can edit, compile and program the 16F886 chip in less than 60-seconds. I simply unplug the DIP chip and pop it into my programmer and bingo! That's very important to me, since MCU hardware programing is always full of unexpected surprises!
I start to get annoyed when my compile/program cycle exceeds about 10 secondsQuote(2) 5V supply. I cannot help but find it so easy to interface to and do A2D accurate work with and have a high noise-immunity.
That ship has long sailed - you will severely constrain your options clinging onto an obsolete standardQuote(3)I can work with a CCBB (Common Chinese BreadBoard), makes it so easy to construct, modify any idea I have.
PIC is the only option for higher-end parts in DIP - PIC24 and PIC32 are available in DIP, SOIC, SSOP and QFN. And I mean actually available
(5) I can always aford'm find and buy a dozen, a large supply of (10x cheaper than Arduino) PICC chips when I blow the #!**$ out of one and still have enough money to buy a new dress at the end of the month. ( I have a large collection of somewhat zorched PICC chips, more than a dozen, in my parts drawer.) So this is what fits my tight budget that relies on salvaging parts form discarded electronic equipment and then buying new only what is necessary, practical and cheap.
Most have some 5V tolerant pins if you need it - there are one or two PIC24s that run on 5V.
Many peripherals and devtools will be pretty familiar if you're already using 8 bit PICs
-
Take a look at the mbed library and its supported parts.
You don't have to use C++, most people refuse this, because it generates larger and slower code.
However, if your design can improve by using specific C++ features such as templates and inhering, it might help you a lot. (objective programming is not limited to C++)Quote5V supply
You should forget 5V operating antiques, all chips nowadays do 3 Volts or less with *most* of the pins 5 V tolerant if Vcc applied. (analog excluded)
Find some LDO's.QuoteI simply unplug the DIP chip and pop it into my programmer and bingo!
You're joking? These days we use SWD or JTAG in circuit debuggers. DIP is also antique.Quote(4) I can make multiple PCB's each using a SOIC packaged
NXP still has some soic parts, but lqfp seems the most popular.Quote(5) I can always aford'm find and buy a dozen,
This is an issue with 32 bit chips, or at least those used on demo-boards, they're quite expensive on the consumer market.
Fortunately, ST and TI will provider samples. You can basically forget buying NXP parts in low quantities. wvshare sells loose parts.
The loose parts easily cost half your demo board. -
Consider doing some projects with what you know -- 8-bit AVR ICs -- but using 3.3v and SMD parts. Wade in the shallow end for a while to show yourself that it's not that scary.
I have bins full of through-hole parts, but as soon as I started using SMD, worlds opened up. Don't deprive yourself of that. Same goes for 5v. You will be shocked how rarely you need 5v I/O when you start using 3.3v. Most of the time, you're interfacing with parts that are either perfectly capable of 3.3v I/O, or have been level-translated to work at 5v in the first place.
Also, get used to using in-circuit programming. It should be easy and reliable. If it's not, you're most likely doing something wrong. No biggie, it's just a knowledge gap, and that's entirely within your power to fix. So resolve it and you'll be far better off. Take some time to figure out what has been tripping you up. Most of the time, you end up learning through troubleshooting -- which can be annoying when you're trying to accomplish something else. So, make it a point to accomplish this instead.
After you've gotten the hang of those changes, then start looking into new micros. Too much at once is a recipe for frustration. Take attainable steps. A journey of a thousand miles, and all that. -
Quote
I find 5v tolerant to be a non-issue for ARM chips.
Take a look at the various STM32 copter control boards and see for yourself how little you need for 5v in today's environment. -
Thanks Gardner,
But I don't know from beans about 32-bit programming.
You say try an ARM..I say "Which ARM?"
I had the same dilemma recently and settled on NXP M0 ARMS. One strong point for me was the availability of a simple to install tool chain. They packed Eclipse and all the plug ins and tools into a single and free package LPCExpresso that installs on Mac, Window and (I think) Linux http://www.lpcware.com/lpcxpresso. You can user everything interactively from the GUI or set your own makefile if you prefer.
For example get this board http://www.nxp.com/demoboard/OM13014.html (~$15), hook it to your computer and you can start edit, run and debug (full feature debugger).
(mini rant: I hate complex to install tool chains, especially if I need to download different parts from different web sites or compile the source on my machine (which typically fails with cryptic error message). Single package install is the way to go.). -
Quote
(2) 5V supply. [including A-D range]
You may be out of luck. Even the faster 8bit chips are going away from 5V.
(Hmm. The Cypress PSoC4 (4200) seems to have an IO section that is pretty capable of 5V operation, including the A2D section.
It's had some discussion here as being particularly "interesting", but I don't think I'd call it "easy to graduate to.")
-
Thanks Everyone!
Thanks especially to Westfw for the idea of using Cypress.
What I would really like to work with would be a single MCU that would fit on a tiny break-out PCB that would plug into my breadboard and ultimately into my MMMMM modules and yet could still re-programmed or maybe even debugged in my final tiny PCB modules.
I looked up ARM/Cortex on Wikipedia and was surprised to learn that ARM type chips are made by STM, TI, Cycpress et al.
So much hardware/Development IDE's to choose form.
But to just look at one arm of the problem:
I looked at the Cypress ARM chips available in large quantity at Farnell, there are 100+ to choose from that support +5V and range in price from a little more than a buck to over a hundred each. The availability is very enticing.
I am especially confused about which chip to start with, which chip that would have the cheapest debugging, programming and adaptability to a breadboard and also then to actually fabricate a small PCB, that is putl the ARM/Cortex chip directly into a tiny PCB that would fit into my MMMMM robot that still would have the hardware support for re-programming it? What support components/chips would be necessary for that to be possible???
How do I get a leg-ahead in this chip to choose to use selection enigma?
Farnell has 4000, 4100, Arm and ARM Cortex versions in stock, all in flat pac types of packaging that would require some way to put the target MCU into a breadboard or break-out board.
I also see that the PSoc IDE is a free download. What hardware do I need to program or maybe(if it isn't costing an ARM and a leg) a debugger?
-
Maybe the hardest part of starting with ARM is making a choice. From my own experience I don't believe getting an ARM going is harder then getting started with any other microcontroller. It is a matter of reading the documentation while being anxious to get going (like having to wait for someone to finish their dinner prayers).
Everyone has a preference for a certain brand and everyone has a preference for a certain IDE/toolchain. When I started with ARM I looked into devices from several manufacturers, tried making boards with several chips and stuck with the brand which suited me well. Maybe you should do the same: get developments boards from several brands, try several IDEs and see what is the easiest one for you to work with. Getting some hands on experience with various devices allows you to make a well informed decission on what is good for you. -
Thanks Zapta!
I clicked on the link you posted and got this chip pop up at Farnell. It is cheap enough and fast but I couldn't find a link to buying the development board. When I clicked on "order" I got the link to the chip itself at Farnell.
NXP LPC11U12FBD48/201, MCU, 32BIT, CORTEX-M0, 50MHZ, LQFP-48
also LPCXpresso Board for LPC11U14 is quite inexpensive and uses JTAG.
Ummm I wonder what JTAG is and how to I work with it? I have a ATMEL AVR MARKII Main uni usb to 6-pin header made for the ATMEGA2560 and other ATMEL chips to directly bypass the Arduino bootloader and download my programs. But would NXP/TI or Cypress support this USB intergace??? What's a mother to do??? -
Thanks again, nctnico
I appreciate you suggestion, but I have been fooling around with MCU chips as a hobby ever since I broke my leg in the early 90's. I had some encouragement from a boyfriend who gave me a computer complete with the software to do it in the mid 90's. I found this new hobby much more challenging and interesting than knitting winter wool caps for my friends in Florida! My first MCU development boards I got from my boyfriend allowed me to learn about computers and I was working with Z80 and 8086 chips in machine language and even some assembly, but several years later many PIC chips became available and I was quick to try to work with one.
Since then I have worked with just two PIC chips, the 16F87x and the 16F88x. I think I am almost an expert about them, but there is still a lot to learn. I learned how to program in BASIC and then with C all by myself with the help of the internet and a few books. It's been quite a few years since I got my first LED's to blink.
Your suggestion "Maybe you should do the same: get developments boards from several brands, try several IDEs and see what is the easiest one for you to work with. Getting some hands on experience with various devices allows you to make a well informed decission on what is good for you." is a bit more more than a mom could handle!
I've already spent more than 10-years playing around with just two models of PIC chips!
| would likely spend the rest of my short-enough life going through all these different approaches trying to figure which one next to try, bankrupt my piggy-bank, and not get anything done! By the time I evaluated all the possibilities I would have discovered that the chips they were intended for were already obsolete and my poor MMMMM's would be plagued with rust! -
I don't believe getting an ARM going is harder then getting started with any other microcontroller.
It is different. 8 bit toolchains tend to hide the initial boot assembly code through options. While with ARM you (or the ide) have to do this manually. And now you have to enable all peripherals yourself, if do it bare metal. Most people use the drivers supplied by the vendor, since peripherals are more complex.QuoteWhat hardware do I need to program or maybe(if it isn't costing an ARM and a leg) a debugger?
Most hobbyists use the demo-board programmer, since genuine debuggers often cost 3 digits.
JTAG is an test interface where each pin can be controlled by the jtag tester. It is "abused" by mcu vendors as programming interface. Atmel isp programmers don't work for ARM. Get a segger or ulink if you want a single device for all targets.QuoteFarnell has 4000, 4100, Arm and ARM Cortex versions in stock
Yes, 50 per family or something. For development its usually the flagship chip which is chosen (the best one, as on the demo-boards). Which is scaled down when working to a production design for cost reduction. As beginner you should focus on Cortex M chips only.
If you're looking for a plug&play experience as Arduino you should definitely look at mbed. Or an STM32 with ChibiOS, but the latter might complicate things a bit too much I think.
-
Thanks Jeroen3!
I appreciate your suggestions, even though they don't agree with others as which ARM to shake a stick at first!
I am not looking for an Arduino experience, I wanna learn about how to program these chips register by register to get my thingees to work!
I'll take a look at your suggested chips. -
Thanks a million Jeroen3 for your many comments and help!
Which Cortex M chip?
Could you elaborate a little whyfor I should post my bet on Cortex M chips? -
dsPIC30 family if you really must use 5v.
then migrate to dsPIC33F and dsPIC33E families when you need 3v3 and a bit more grunt.
Learning curve isn't that steep if your used to microchip stuff already. -
Hey, I found a one page, a little lotta stuff available on ebay regarding programming the Cypress ARM
Can anyone tell me if any of it is anything worthwhile/a bargain/stay clear of?
http://www.ebay.com/sch/i.html?_from=R40&_trksid=p2050601.m570.l1313.TR0.TRC0.H0&_nkw=%09CY8CKIT-002&_sacat=0
-
Thanks FCB!
I don't quite understand why DsPIC30 are a great choice, what is this choice quite all about?
Are DSP just some hopped-up mcu's optimized for analog-digital signal processing. Why this insteada ARM?
Is the the IDE software available for the Dsp's free and does it work, easy to work with? What about hardware needed to program these chips or debug them?
-
I would also recommend ARM Cortex-M. Cortex-M is the CPU core series for microcontrollers. Many manufacturers use the same CPU core, the just plug their own peripherals in the same chip next to the ARM CPU core. So once you get used to them, you just need to learn the new peripherals when switching to another ARM Cortex-M vendor's chip.
* I would suggest starting with something that has good vendor support to get the first projects going. Compared to 8-bit controllers, the 32-bit ARM Cortex-M solutions usually do not hide away the low-level stuff - the array of interrupt vectors, initial value of stack pointer, linker script (memory mapping) etc. When using a well-supported chip, you just create a new project in the IDE and you get everything prepared to you (of course you can change things if you need). When using not so well supported chips or third-party tools (ie not the IDE+toolchain from the vendor but for example separate gcc+eclipse+gdb), you need somewhat more work to get the cpu going and it might be too much for the first project. Reasonable selections would include NXP, Atmel, SiLabs (former EnergyMicro), Cypress. NXP LpcXpresso supports Windows,MacOS,Linux; others are Windows-only if anything has not changed recently.
* Do not be afraid of the code that gets automatically added to your project. For 8-bitters it is usually hidden, for 32-bit controllers everything comes into your project. Compared to AVR, the clock settings are not configured by fuse bits anymore; the cpu starts running the code and the code itself configures the clock settings and switches clocks while running. All of that is inserted into auto-generated code. For start, just ignore it and write your code next to it. Later you can explore what gets done and try to do something different if needed.
* The vendor-given code usually includes the "driver" code for most of the peripherals. You may use it to talk to the HW. You may ignore it and write to the registers yourself (and use the vendor's code as a reference). Or you may modify it if you need to. Do not be afraid of that code. You may not like it; you usually don't. Usually it is too complicated; it provides functions to configure everything that you'll never use etc. Things can be done in multiple ways, you do not need that chained-descriptor-dma-controlled-uart-code just because the vendor does it that way. In most of the cases, the simple polling loop or IRQ-based approach works fine.
* Hardware-wise it would be good to choose a board with built-in debugger. Most of the vendors now offer boards that have the chip will be using and in addition to that the debugger. ARM microcontrollers have standardized JTAG or SWD debug interfaces. The PC to JTAG/SWD debug interface is on those boards, so you'll get all the HW you need at once. Depending on the board, it may have a couple of LEDs or something else as well. Everything is powered from the debugger USB connector, so you just plug the debugger USB and you are ready to upload the "blinky" example project. Later on when you buy separate microcontrollers of the same manufacturer to build your own boards or simple breakout boards for breadboarding, you can detach the connection between the debugger and the uC on the evaluation board and use that debugger to debug another chip on another board. (NXP has even boards that can be cut in half to physically split the debugger half). Usually the debuggers from one uC vendor do not work on chips of another. You can also get a more expensive generic professional debugger (jlink etc), but then you might run into HW and SW compability issues that you don't want for your first project.
* Actually, it is possible to program most of the ARM Cortex M chips without a debugger at all. Most of the chips come with built-in bootloader. Usually it depends on some bootstrap pin of the uc; depending if it is low or high when powered up, the uc starts the code or bootloader. Depending on the chip, the bootloader uses an UART or the USB of the uc (that you usually bring out to a connector anyway for most of the projects); using software provided by the vendor you can just program the chip via that interface. This is usually used for manufacturing; for development it is good to use a debugger that is integrates well to the IDE.
So the simplest way to get started would be:
* buy an LpcXpresso board (doesn't matter which... but unless you need to try a specific uC, get the biggest one you can... the price difference is almost nothing but you'll get 10x more flash space and many more peripherals to play with. For example, 15xx and 17xx are nice to play with.)
* install lpcXpresso software
* connect the board via USB to the PC
* create a new example project (use the blinky demo, for example)
* press the debug button to upload to uC
If you need more HW, get an Embedded Artists Base Board; the lpcXpresso plugs into that and you'll have plenty of things to play with.
-
I'm not quite sure you really need a debugger. In 25 years I never run into a problem which could only be solved by using a debugger. When it comes to ARM controllers I use the ones from NXP and a serial port to program them. Life can't be any simpler.
-
Gadzillion Thanks, dgtl
I am in AWE! Thank you so much for using some of your valuable time to give me so much helpful information.
Your reply is very comprehensive and I've just read though it once and I am very impressed by your detailed reply!
I will print it out and study it very carefully. -
Thanks again nctnico,
Your advice on debuggers is very valuable.
I am trying to make up my mind, what little is left of it. So many possibilities to choose from.
You say things are simple, could you please reply with info about the exact hardware/software NXP chip to use to make this so?
Where do I buy the parts, the NXP chips? Ebay?
I found over 200 different NXP ARM chips on Farnell.com.
How can I make a choice?