Preamble: after working a couple of years as application developer, the world of the software engineering became more obscure than it was before. The reason is that the real stuff is hidden under zillions layers of abstractions: OS, frameworks, etc. The young generation is deprived of the pleasure of working with PDP-like machines where all programming was done via electrical switch toggling. Another problem is the ephemeral nature of modern programming languages. Once there was Python 2.x, now it is deprecated and there is Python 3.x which in its turn will be deprecated in a couple of months. Idem for other languages. ANSI C looks like the Pyramid of Cheops: it was there in 70's and I don't doubt it will be there after the Sun will become a red dwarf.
It seems that now the only way to understand the interaction between the hardware and the software is to play with embedded development. From the pedagogical point of view physical chips are very handy because they allow to tackle the most difficult part of C language, namely pointers. When coding in OS environment, */& notation is still very confusing because it refers to some location somewhere inside of the virtual memory. And before you will got the understanding of what is the virtual memory, you have to read a couple of monographs about OS development, etc. You may find it stupid but I do really want to know which transistor is holding my bit right now. At least, I can wire physical pin voltage to programming abstractions.
Currently I am working with Atmel chips and WinAVR package because of numerous textbooks and accessible hardware. Though all books promise to teach AVR coding using plain C, the reality is that all pointers are hidden behind macros like PORTA, DDRB, etc. All code examples include header file 'io.h' which in its turn refers to other header files specific for a given chip like 'iomx8.h'. So far, I cannot find any macros definition in these headers. The code to increase the voltage on the physical pin 14 on Atmega168 looks like
DDRB = 0x01;
PORTB = 0x01;
Fortunately, Microchip site provides some basic documents where it is stated, for example, that if I want to rise the voltage on the physical pin 14, I need to follow these steps:
unsigned char *ddrB;
ddrB = (unsigned char*)0x24; // the address of ddrB is 0x24
*ddrB |= 0x01; // set up low impedance/ high current state for the transistor 0
unsigned char *portB;
portB = (unsigned char*)0x25;
*portB |= 0x01; // voltage on
*portB &= ~(0x01); // voltage off
Unfortunately, this is the only info I got after one week of the lurking. Now I am going through USART programming and the things become more complicated with all these UBRR0H, UCSR0C. Since provided header files don't contain macros definitions for any register, where else can I find it?
A similar question was asked several years ago: accessing AVR registers with C?. However, provided answers were somewhat useless, besides the clue that GCC itself can map some mythical PORTB to real physical locations. Could someone describe the mechanism behind the mapping?
From a memory-mapping standpoint: The general purpose registers, special function+I/O registers, and SRAM share non-overlapping ranges a single address space, as described in datasheets for various processors in the AVR series. All of your pointers will reference this memory space, unless annotated as pointers to PROGMEM (which will cause different instructions to be emitted). The reference will be made without any sort of virtual memory mapping.
For example, the ATtiny 25/45/85 has the following map shown on page 18:
Your linker is aware of this memory map and will place variables accordingly. For example, a global variable declared in one of your compilation units will end up in an address above 0x0060 in the example device described above, so that it ends up in the SRAM.
From an instruction encoding standpoint: Although there is one address space, there is special functionality reserved for certain important regions. For example, IN and OUT instructions have six bits in their instruction encoding which can be used to directly refer to one of the 64 addresses within [0x20, 0x5F).
The IN and OUT instructions are unique in their ability to load and store to a fixed address encoded directly in the instruction, since the normal load and store instructions require an indirect load with the 'Z' register being loaded first.
As a result, when the compiler sees memory operations to a fixed I/O register, it may generate these more efficient instructions. However, a normal load/store via a pointer will have the same effect (although with different numbers of clock cycles required). For extended I/O registers that didn't fit into the first 64 (e.g. OSCCAL on an atmega328p), normal load/store instructions will always be generated.
Short answer - hidden away in the included headers from Atmel are a collection of macros that create pointers to the register locations. If you want to see any of the source, as well as additional necessary headers like interrupt.h, they are in WinAVR-20100110/avr/include/
Here's a brief overview of the process:
Your Makefile defines the device to be used, and then passes it has a definition to the compiler.
DEVICE = atmega2560
...
-D__$(DEVICE)__
You then include io.h, which automatically includes the necessary headers based on your device:
// In main source file
#include <io.h>
// In io.h
#include <avr/sfr_defs.h>
// ...
#elif defined (__AVR_ATmega2560__)
# include <avr/iom2560.h>
// In sfr_defs.h
#define _MMIO_BYTE(mem_addr) (*(volatile uint8_t *)(mem_addr))
#define __SFR_OFFSET 0x20
#define _SFR_IO8(io_addr) _MMIO_BYTE((io_addr) + __SFR_OFFSET)
// In iom2560.h
#include <avr/iomxx0_1.h>
// Other device specific definitions
// Om iomxx0_1.h
#define PINA _SFR_IO8(0X00)
// Other device family shared definitions
So if you unroll all of that, what you get is a volatile pointer to the register address. When ever you use PINA in your code, the preprocessor replaces it with all of the expanded macros:
PINA
_SFR_IO8(0X00)
_MMIO_BYTE((0X00) + __SFR_OFFSET)
(*(volatile uint8_t *)((0X00) + 0x20))
Which specifies that PINA is a pointer to a volatile 8-bit memory address of 0x20. The internal chip architecture then maps that address to the appropriate peripheral register whenever it is accessed.
Different devices have different register addresses and offsets. If you want to define your own, you'll need to check out the relevant datasheet. For most AVR chips, there is a section towards the end titled "Register Summary" that lists all of the register addresses and names of the individual control bits. In my experience (for AVR, at least), the names of the registers and bits found in the datasheet are exactly what they are defined as in the io.h files.
Also notice the use of "uint8_t" rather than "char." It's common (and highly encouraged) to use the bit-width specific definitions found in <stdint.h> to specify signed/unsigned and 8/16/32 bit variables whenever appropriate. Since AVR is 8-bit, any use of 16 or 32 bit (or float) variables will require multiple clock cycles for each operation. In this case, stdint.h should have:
typedef unsigned char uint8_t
In ARM Cortex-M4F MCU (TM4C1294NCPDT specifically), to deal with interrupts (GPIO interrupts), one of the steps to get the interrupts working is to clear the I BIT.
I searched a lot but I couldn't find any useful information about that, could anybody please tell me where to find that bit exactly and how to edit it if I need some special procedures?
And that will be great if I have been told where to find that information exactly after the explanation please (to learn how to answer myself on any other questions).
The CMSIS provides a standard cross-vendor software interface to Cortex-M based devices. The CMSIS defines a number of functions for interacting with the NVIC and PRIMASK including the intrinsics __disable_irq()/__enable_irq()
The ARM Cortex-M interrupt system is quite complicated and very well thought. It consists of CPU registers and a tightly coupled interrupt controller (NVIC). Interrupts are prioritized and vectored. There is no single interrupt-enable flag as for smaller 8/16 bit MCUs.
For each interrupt, there are two ARM-core instances to gate the event to the CPU: The CPU PRIMASK register (single bit), which can be seen most similar to the classical interrupt-enable flag. Second is one enable bit in the NVIC. For these, there is an ARM standard in the CMSIS headers. These provide functions __enable_irq() and __disable_irq() for the PRIMASK bit. The peripheral interrupt itself has to be controlled by NVIC_EnableIRQ(IRQn_Type IRQn) where IRQn is the interrupt number as defined in the MCU-specific header file.
Finally, there are most times also interrupt enable bits in each peripheral module as know by the smaller MCUs.
Note that to have an interrupt pass through all gates have to be open (all bits set to "enable"). Use the CMSIS functions to manipuate the bits. They very likely will not take more instructions than a hand-crafted version.
Edit:
There is no actual need to fiddle yourself with assembler or the registers. Just use the CMSIS functions, you can very likely not do better yourself, but possibly worse. That's actually the intention of CMSIS.
(end edit)
Start reading in the reference manual for the MCU and the vendor's homepage. That should provide references and app-notes for the device. You also should read the technical reference manual, architecture reference manual from ARM. Actually, just have a close look at all related documents there for the CPU (M4 for you). These are for free, some require registering.
For the NVIC, you should not access it directly, but using the CMSIS header files as provided by TI for exactly this MCU (the headers require some device-specific settings). If not available,you can get them from ARM, but have to provide the device-specific settings yourself (they are few and are given in the MCU's reference manual).
As the ARM Cortex-M4 has multiple interrupts, you need their symbolic names to enable/disable. These have to be defined in the MCU header which defines all peripheral modules, too (there might be multiple such headers). The names end with _IRQn, just search for that.
To use the Cortex-M4 you should read the documents given, or you can try with a good book. However, as this is no tutorial site, nor is it allowed to recommend books, please search yourself.
OK, the easiest answer for my question is:
To use " CPSID I" or " CPSIE I" inline assembly code which will set or clear the PRIMASK (I) Bit respectively. (of course that will work just in privileged mode).
And both assembly instructions are equivalent to __disable_irq() and __enable_irq() functions in CMSIS respectively.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 9 years ago.
I'm trying to familiarize myself with the embedded field, but also have limited resources in terms of time and equipment to buy.
What's a good language to wrap my head around embedded, without investing too much time leaning an embedded-specific language? I'm most familiar with PHP, Java, Actionscript, but unfortunately know very little C. I remember reading somewhere that someone used PERL to program embedded systems, but not sure if that's really possible.
Can learning be done without needing to buy chips, etc. via simulators or such?
Can someone recommend a simplified roadmap to show how one would get sarted? I'm a little unsure where to even start.
You need to know C (but every programmer needs to know C !)
Most of these platforms have a simulator/emualtor, but since the point is to learn real applications and real problems ( which are all to do with real world timing issues) then you want a real board.
You probably also want an oscilloscope (a very cheap n'th hand slow analogue scope will do) and have some idea how to use it.
Easiest way in is probably Arduino, perhaps more professional but a little harder is launchpad MSP430
There are a few embedded programming lessons that carry over from one platform and style to another, but it is really a broad field. Different processors can require very different tactics, and different applications can dictate both different firmware design tactics and different microcontrollers. Here's some stuff to get you started....
msp430
Texas Instruments has several very inexpensive USB development kits which they call EZ430 and are based on their MSP430 family of micro-controllers. The simplest one has an msp430 f2013, which has 2K of flash program space, 3x128 bytes of usable user flash (another 128 byte page exists, but it's special), 128 bytes of RAM (yes, 128 bytes but it's enough for lots of things), and 16 CPU registers (some of these are special purpose like the Stack Pointer, Instruction Pointer, Status Register, and maybe one or two more). MSP430s also have several memory mapped special function registers which are used for configuring and controlling the built in peripherals. MSP430s are von Newman processors, so everything lives within one address space. These cost about $20us for both the programmer and a removable tab (pc board) containing the msp430 f2013. For about $10us you can get 3 replacement tabs with msp430 2012, which is pin compatible with the 2013 (mostly) and has a few different peripherals. These tabs have an LED, a button, and several large vias (holes in the pc board) which are connected to the pin of the processor. These vias are easy to solder wires into even if you have never soldered before -- due to capillary action the vias just suck the molten solder up and while it's hot you can just jab the end of your wire in there.
They also have a couple more similar kits with 802.15.4 radios. Even if you aren't interested in the radio you may still be interested in these because their programmer also has a UART pulled over from the removable tab and are compatible with the tabs used on the other kits mentioned above. These kits also contain at least one extra programmable board and a battery pack for it. (one of these kits may contain more, but I don't have mine with me right now, and not going to look it up).
They also have a kit that has a programmable watch as the target platform. I've never had one of these, but they have a display, accelerometers, and several other cool things, but this may overwhelm you for your first project. I'd suggest one of the previous kits to get you started with MSP430s.
You can get free C compilers and development environments for MSP430s in the form of IAR's Embedded Workbench kickstart (4 kb program space limited ) IDE, Code Composer Studio (also limited program size, but higher limit, I think), and gcc/gdb for the MSP430. IAR's kickstart is pretty easy to get started with quickly, though it's not perfect. You may find that you have to shut it down, unplug your USB EZ430, restart IAR, and plug back in to get it going again. Or maybe some different order will work better for you.
TI also provides many examples in badly named files (all of their downloadable files go out of their way to be badly named). Be warned -- similar MSP430s may have different device control register interfaces for similar peripherals, which can be confusing. Make sure that any document or example you are reading really does apply to the microcontroller you are using.
other small systems
There are many many other processors families and kits that you can go with, and you should probably at least know a little bit about them.
AVR -- Atmel's 8/16 bit Harvard architecture. Harvard refers to separate address spaces for code and working memory. It has 32 8 bit registers, some of which may be used in pairs as 16 bit registers. It's a very popular and pretty cool processor. Some of the smallest ones only have registers with no extra RAM, which is scary. Atmel also has an AVR32 which isn't at all the same as the AVR. Unless you make use of an existing bootloader capable of loading your new code you will need to get a JTAG unit for these.
8051 -- This is old as the hills and a pain in the butt to use until you finally understand it. It is an 8/16 bit processor, with many more limits on how you go about doing 16 bit math and only has 1 pair of registers which can act as a pointer. It has 3 separate address spaces (stack, global memory, and code) and lots of odd (compared to other architectures) features. The low level stuff might not mean much to you if your are programming in C except that very simple C operations can turn into much more code than you thought they would. You don't want to start on one of thise, most likely.
propeller -- Parallax's very interesting multi-core processor which is very unlike other processors. It has several cores which act mostly independently and can be used to simulate peripherals or do more traditional computational tasks. I've never used one of these, though I'd like to. Just never had a task that seemed to fit it. They have their own high level language to program them as well as the processor's assembly language.
larger systems
After you get out of the 8/16/24 bit processors you start to blur the lines between embedded and desktop level programming, even if it is technically embedded.
AVR32 -- There are 2 main versions of these. One is a Harvard architecture and the other is von Newman. The von Newman version is essentially a better ARM than ARM, but it's not as popular as ARM. As near as I can tell it was designed with "run Linux" in mind, though not tied to it in any crazy way. You used to be able to get cheap development boards for these and code is often almost as easy to load as copying files from one PC to another, though you will probably make use of uboot and tftp to do some work. JTAG is only needed when you mess up the boot loader. I think all of these have support for native JAVA acceleration. www.AVR32.org
ARM -- The most popular embedded processor. There's many versions of these. Some don't have an MMU (memory management unit) and some do. There's too much to say about them. Some version have native JAVA acceleration, though I think that the ARM lords don't freely tell all of the details of how to use it, so you have to find a JVM which knows how to use it. Many vendors make them, including Atmel, Freescale, Intel, and many others.
MIPS -- A very RISC processor. The RISCiest.
There are many others.
Programming styles
I could write 3 books on this but the general rule is make things as simple as the application can let you. An exception to this is that if you can easily make use of an operating system you might want to make use of it if it simplifies your task.
The first thing you need to know when answering this question is "WHAT IS" an embedded system? A GENERAL definition would be a computer system which is dedicated to a single specific purpose. This doesn't limit the type of hardware you can use, as matter of fact "Embedded PCs" have been used for years. The QNX realtime OS has existed since the early 80s and been used in industrial PCs for embedded applications for years. I've personally used in control systems for steel mills XRAY thickness gages. On the other hand I currently use TI DSPs without any OS support and only using 256K of Ram. Another example would be the key fob for your car. Old ones used to use a PIC microcontroller from Microchip. (That's actually the company's name.)
Some people call the IPhone an embedded system, but due to the fact you can load applications to do just about anything I tend to say it's a palm top computer with phone capabilities. An OLD DUMB cell phone that is just a phone, not PDA is an embedded system. That's just a bit of philosophy.
As a general rule there are a handful of concepts you need to grasp for embedded systems programming, and most of them can be explored on a PC.
EDIT:
The REASON why C or C++ is recommended is C itself was designed to do systems programming. C++ maintains all it's advantages but adds capabilities for OOP programming. Some ASM maybe required in some systems. However a lot of chip vendors, such as TI, provide tools basically make it possible to do your entire system in C++.
:END EDIT
ALOT of simple embedded systems look more or less like this:
While(true) // LOOP FOREVER... There is no command prompt
{
// Typically you want I/O to occur on fixed "timebase."
wait(timerTick);
readDigitalIO(&dioStruct);
readAnalogIO(&aioStruct);
// Combine current system state with input values
// and do some useful calculations. (i.e. Analog input to temperature calc)
Process(dioStruct,aioStruct,&CurrentState);
// This can be a serial output/audio buzzer/leds/motor controller
// or Whatever the system REQUIREMENT call for.
driveOutputs(CurrentState);
// The watchdog timer resets your system if it gets stuck.
petWatchDogTimer();
}
There is nothing here that you can't do using the PC. (Well a PC that still has a parallel port anyway. Which is a more or less just a DIO port.) On a simple system without a os this might be all there is. On a RTOS based system you may have several task that all look somewhat simalar to this, but pass data back and forth between the tasks.
The interesting parts come when you have to interface to the hardware on you're own, the first job I had out of college was writing a device driver for a data acquisition board under QNX.
Basic concepts of dealing with hardware, or device drivers (Which you can experiement with by hacking Linux device drive code which is freely avaliable.), most hardware looks to the programmer like just another memory address. This is called "Memory mapped I/O." What does this mean? Lets use a serial port as an example:
// Serial port registers definition:
typedef struct
{
unsigned int control; // Control bits for the port.
unsigned int baudDiv; // Baud rate divider.
unsigned int status; // READ Status bits/ Write resets fifos;
char TXdata; // The head of the hardware TX fifo.
char RXdata; // The tail of the hardware RX filo.
} serRegs;
// Using the volatile keyword to indicate the hardware can change the value
// independantly from the software.
volatile serRegs *Ser1 = (serRegs *)0x8000; // Hardware exists at a specific location in memory.
volatile serRegs *Ser2 = (serRegs *)0x8010; // Hardware exists at a specific location in memory.
// Bits bits 15-12 enable interupts and select interupt vector,
// bits 11-8 enable,bits 7-4 parity,bits 3-0 stop bits.
Ser1->status = 1; // Reset fifos.
Ser1->baudDiv = CLOCKVALUE / 9600; // Set the baudrate 9600;
Ser1->control = 0x1801; // Enable, 8 data, no parity, 1 stop bit.
// Write out a "OK\r\n" message; (Normally this would be a loop.)
Ser1->Txdata = 'O'; // First byte in fifo Transmission starts.
Ser1->Txdata = 'K'; // Second byte in fifo still transmitting first byte
Ser1->Txdata = '\r'; // Third byte in fifo still transmitting first byte
Ser1->Txdata = '\n'; // Fouth byte in fifo still transmitting first byte
Normally you would have a function or an interrupt handler to handle TXing the data, but for example I wanted to point out that the hardware is working while the software keeps going. Basically hardware works like, I write a value to an address and "STUFF" happens independantly of the software. This is perhaps one of the key concepts for embedded programming, how to make the computer effect a change in the real world.
EDIT:
If you actually want to get a cheap board, the current trend from Micro developers is to put a dev kit on a usb thumb stick. This page has info on several, ranging from 8 bits upto ARM architectures: http://dev.emcelettronica.com/microcontrollers-usb-stick-tool
The Cypress PSOC was one of the first to do this with the "FirstTouch Starter Kit." The PSOC is a very unique part in that it has a micro controller and "Configurable analog and digital blocks" that allow you to plop down a ADC, serial port, or digital I/O using a gui and automatically configures you're C app to use it. The PSOCs also are avaliable in DIP packages which makes them easy to use on a prototyper's breadboard.
Picture your embedded controller sitting in a switched-off circuit...
The Vcc power is applied and the reset circuit asserts reset signal.
Clocks have reached running speeds and voltages stabilized, so reset is de-asserted.
Now your controller sets its instruction pointer to the "reset vector," which is physical address 0xE0000000 on this particular chip. The controller fetches the instruction at that location.
Interrupts are disabled, and the first order of business is to initialize registers such as the stack pointer. On some chips, there are flags bits (e.g., x86 direction flag) which need to be cleared or set.
Once the registers and flag bits are set up correctly, it becomes possible for interrupt service routines to run. By now, we must have run code to about location 0xE0000072 when we get to the code which enables interrupts by first toggling some GPIO pins to the external interrupt controller, then enables the CPU interrupts mask.
At this point, the equivalent of "device drivers" are running in the form of interrupt service routines. Assuming the C environment has a library which matches the interfaces of these routines' data structures, by now our boot-loader code can jump to the main() function of some C object code.
In other words, the code which brought us from power-on to main(), and which handles the low-level I/O, is written in the assembler peculiar to the chip you choose. This means that if you want to be versatile at embedded programming, you must know how to implement assembly code starting at the reset vector.
The reality is that hobbyist embedded programming doesn't allow time for implementing all the ISRs and the boot-loader code. For this reason, many people use standard software frameworks available for specific chips. Others use custom-language chips such as the BASICstamp. The BASICstamp is an embedded chip which hosts a BASIC language interpreter on-board. The interpreter and all the ISRs are pre-written for you. The BASIC environment gives you the ability to control I/O pins, read voltages, everything you could do from assembly with an embedded controller, but a bit slower.
As for the language, C is probably the most important language to know. From Java you should be able to adapt, but just remember that a lot of high level Java will not be available to you. Loads of textbooks out there but I'd recommend the original C programming book by Kernighan and Ritchie http://en.wikipedia.org/wiki/The_C_Programming_Language_(book)
For a good introduction to embedded C you could try a book by Michael J Pont:
http://www.amazon.com/Embedded-C-Michael-J-Pont/dp/020179523X
As for the embedded side of things you could start with Microchip, the IDE is OK to develop in with a reasonable simulator, and the c compilers are free for the slightly limited student editions c18 and c30 compilers, the IDE installer will also ask if you want to install a 3rd party HI-TECH C compiler which you could use. As for the processor I'd recommend selecting a standard 18 series PIC such as the PIC18F4520.
http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en019469&part=SW007002
Whatever the chip manufacturer, you have to get to know the datasheets. You don't have to learn it all at once but will need it to hand!
Embedded, like most programming, tends to revolve around:
1) initialising a resource, in this case rather than data being from a data store it is from computer registers. Just include the processor header file (.h) and it will allow you to access these as ports (usually bytes) or pins (bits). Also micro-processors come with useful resources on the chip such as timers, analogue to digital converters (ADCs) and serial communications systems (UARTs). Remember that the chip itself is a resource and needs initialising before anything else.
2) using the resource. C will allow you to make data as global as possible and everything can access everything at every time! Avoid this temptation and keep it modular like Java will have encouraged you to (though for speed you may need to be a little looser on these rules).
But they do have an extra weapon called interrupts which can be used to provide real-time behaviour. These can be thought of a bit like OnClick() events. Interrupts can be generated by external events (e.g. buttons or receiving a byte from another device) and internal (timers, transmissions completed, ADC conversions completed). Keep interrupt service routines (ISRs) short and sweet, use them to handle real-time events (e.g. take a byte received and store it in a buffer then raise a flag) but allow background code to deal with it (e.g. check a received byte flag, if set then read the byte received). And remember the all important volatile for variables used by ISR routines and background routines!
Anyway, read around, I recommend www.ganssle.com for advice in general.
Good luck!
The scope of embedded computing has grown very broad, so the answers somewhat depend on what kind of device you're aiming at. On one end, there are 8-bit controllers with only a few KB of memory, usually programmed entirely in assembly or C. On the other end, processors such as those in your router are fairly powerful (200 MHz and a few MB of RAM is not uncommon) and often run an OS like Linux, which means you can use pretty much any language, though C and Java are the most common.
It's best to buy a real chip and experiment. Most of the work involved is usually in getting to know a device and how to interface with it, so using a simulator kind of defeats the purpose.
What's a good language to wrap my head around embedded, without investing too much time leaning an embedded-specific language?
As everyone will suggest: C. Now depending on how deep are you going to dig into your platform of choice, you may also need some assembly, but don't be scared about that: typically you'll use just a little.
If you are learning C, my personal suggestion is: work as you would do in assembly; the programming language won't give you much abstractions, so think in terms of memory management. When you've learnt how to do it move up towards abstractions and live happy.
C++ is also popular on embedded platforms, but IMHO is difficult unless you know well how to program in C you can also understand what's under the hood of its abstraction.
When you feel confident with C/C++ you can start to mess with embedded operating systems. You'll notice that they can be totally different from your OS of choice (By example not all operating systems have a C standard library, processes and splitting between userspace and kernelspace).
You'll learn how to build a cross compiler, how to mess with linker scripts, the tricks of binary formats and a lot of cool stuff.
For the theoretical point of view there's a lot of stuff as well: if you study Computer Science you can get a master's degree in embedded systems.
Can learning be done without needing to buy chips, etc. via simulators or such?
Yes: many operating systems can be run on simulators like qemu.
Can someone recommend a simplified roadmap to show how one would get sarted? I'm a little unsure where to even start.
Try to get a simple operating system which can be run on emulators, hack it and follow your curiosity. Don't be scared of messing with knotty code.
1) Most of the time for most and usually the lower end embedded system, you need to know C.
And I will still recommend you to get a vanilla development board to get yourself familiar with the work flow and tricky part of working with embedded system like debugging and cross compiling. You will run into trouble if you only rely on emulator.
You can try out The Linux Stamp, it is not expensive and is good for beginner but you do need some prior knowledge on Linux.
2) For high end embedded system, a good example is Smartphone from HTC (CPU speed can reach 1Ghz)or some other Android phone it run fast and you can even code Java on it.
C and the assembly specific to your chip.
No, you really need a real chip. Simulators aren't the real thing. You need to be able to deal with keypress jitter, voltage funkyness, etc.
The Arduino is the current fad for embedded hobbyists. I'm not a big fan of Harvard architecture, personally. But you will find oodles of help out there for it. I use an XCore for my thesis work and I have found it super easy to program multicore stuff. I would suggest starting with an AVR32 and going from there.
As everyone else is saying, you need to know C.
Have a look at AVR butterfly for a cheap development board.
Smileymicros have a simple kit with dev board and book:
http://www.smileymicros.com/index.php?module=pagemaster&PAGE_user_op=view_page&PAGE_id=41
The soon to ship Raspberry Pi board looks like an incredibly cheap way to get into this field.
Yup, Arduino would be the way to go. Agreed.. Cheap (about $20 to start) and has great API to get started with high level functions. C is a must though, can't avoid it. But if you can program in other languages you'll be all good.
My recommendation is to start shopping at http://www.sparkfun.com lots of examples to work from and helpful hints of what devices to buy.
(If your lazy see bottom for TL;DR)
Hello, I am planning to build a new (prototype) project dealing with physical computing. Basically, I have wires. These wires all need to have their voltage read at the same time. More than a few hundred microseconds difference between the readings of each wire will completely screw it up. The Arduino takes about 114 microseconds. So the most I could read is 2 or 3 wires before the latency would skew the accuracy of the readings.
So my plan is to have an Arduino as the "master" of an array of ATTinys. The arduino is pretty cramped for space, but it's a massive playground compared to the tinys. An ATTiny13A has 1k of flash ROM(program space), 64 bytes of RAM, and 64 bytes of (not-durable and slow) EEPROM. (I'm choosing this for price as well as size)
The ATTinys in my system will not do much. Basically, all they will do is wait for a signal from the Master, and then read the voltage of 1 or 2 wires and store it in RAM(or possibly EEPROM if it's that cramped). And then send it to the Master using only 1 wire for data.(no room for more than that!).
So far then, all I should have to do is implement trivial voltage reading code (using built in ADC). But this communication bit I'm worried about. Do you think a communication protocol(using just 1 wire!) could even be implemented in such constraints?
TL;DR: In less than 1k of program space and 64 bytes of RAM(and 64 bytes of EEPROM) do you think it is possible to implement a 1 wire communication protocol? Would I need to drop to assembly to make it fit?
I know that currently my Arduino programs linking to the Wiring library are over 8k, so I'm a bit concerned.
Since you only need to send data (which is simpler than receiving) and you can select your own protocol, it should not be a problem to fit the code in the available memory space.
I once created software for an industrial control panel that contained 8x14 segment LCD display, some LEDs, some buttons, a serial (I2C) EEPROM, and serial interface to the host. A 4 bit processor was used. The device did not have any serial interface, so both the RS232C interface and I2C bus had to be implemented in software. On top of that, there was Modbus protocol (which among other things requires CRC calculations some exact timing), and the application program.
The device had some 128 x 4 bits of RAM and 1kW, 2kW, 3kW or 4kW of ROM (10 bits per word). The size of the final program was about 1100 words, so it did not quite fit in the smallest device. I used Assembler, of course.
However, instead of using multiple microcontrollers, you could consider using a hardware solution.
You could use a sample and hold circuit. For that, you need an array of analog switches and capacitors and perhaps op-amps. Just issue a trigger to latch all the voltages into the capacitors. Then you can use as much time as you need to read the voltages with your master processor.
Update: Forgot to mention that there are ready-made sample-and-hold amplifiers that need very little or no external components. This is probably the easiest solution.
1k of program space should be plenty, considering that your protocol only needs to be complicated enough to send a single integer when tickled. Look into Manchester Encoding.
You can probably get away with using a C compiler that targets this architecture, but you'll have to create your own runtime environment and not rely on the one supplied with the compiler. That's doable, but I'm not sure if the additional work to essentially create your own mini-OS outweighs the productivity benefit of using C over assembler.
I've done embedded programming in similar constraints. I used Borland Turbo C (it was a long time ago) in the tiny model and obtained code that was hardly bulkier than I could have done in assembler, with a fraction of the effort. What I'm saying is: It's quite feasible and sensible to use C as a high level assembler.
Just like me, though, you will be facing the problem of providing C with a (tiny) runtime environment. Ideally, you will only need to set up the stack and a few registers. Also, you won't have room for the C library, so you will need to program any needed functions yourself.
Yes, probably, though if you know your compiler very well you might be able to get away with c.
What you could do, is use a compiler to emit any standalone functions you need based on c code, then glue them together with a little of your own. (You'll certainly have to do the c runtime setup yourself - stacks etc.)
You may consider upgrading to the ATTiny25. It is a more capable 8-pin AVR that includes Atmel's Universal Serial Interface. It is capable of doing 1-wire serial comms in hardware, given only a few bytes of software.
Why wouldn't you just use sample-and-hold hardware, rather than a pile of microcontrollers?
I designed a master-slave system recently using an AT90USB646 master and ATtiny85 slaves. Obviously I had a lot more memory to work with on the slaves, but what I wanted to share with you is this:
With regard to your communication protocol, bear in mind that the uncalibrated internal oscillator on the ATtiny13 has an accuracy of +/- 10%. This means you won't be able to use, e.g., RS-232 communications.
I used a variant of the Dallas 1-Wire protocol in my system. Including full support for slave enumeration etc., the C source code compiles into 1626 bytes.
Edit: Whoops, didn't realize the question is so old. Hopefully this may still be of some help.