Slashdot Mirror
Microcomputers for Homebrew Projects?
tengwar asks: "Way back when I was at university, I did a course on microcomputers which went into enough detail to design, build and program a Z80-based system - more or less state of the art at the time. Now that my lecture notes are firmly embedded in the Carboniferous layer, I'd like to have a go at doing this with a more modern chip, and I wondered what's available. I'm not brilliant at electronics, and I liked the way the Z80 peripheral chips integrated easily with the CPU. Obviously I'm not looking to just slot together the latest PC motherboard with the latest Pentium, but I'd need to go for something where the board design won't get too complicated, which probably rules out processors with full 32-bit external interfaces on space grounds. I'm not really concerned about performance, but it would be nice to be able to port a JRE to it rather than working entirely in assembler. Any thoughts on suitable starting points?"
I'm not sure if anything short of Pentium/Athlon would be state of the art enough, but I'd recommend the Atmel AVR. Firstly, there are lots of demo/examples with it and existing software archives exist.
This processor is used in some smart-sensor applications where you have distributed sensors.
Here's a 1998 EDN mag review and some simm circuit boards which make project computers.
One of the reasons you won't get modern CPUs to work on hand-made boards is the length of connections. With the z80, you can trim wires by hand and it'll still work quite happily; with faster CPUs, though, you need to be far more precise with the length of the wires / tracks connecting the CPU to memory and the like. At speeds of over a few MHz, even a mm difference is more than enough to screw things up...
Microcontrollers are extremely small, low-powered devices containing a CPU, some code/data memory peripheral interfaces. You might have seen them at work in your preferred game station mod chip...
I think Motorola, AMD and Intel also have a line of powerful 68xxx- and x86-, respectively, -based microcontrollers that might be closer to the capabilities of a PC processor.
There are also some chips that you can program in Basic (ugh) via a small interpreter in their boot code. It really depends on what you are planning to do.
PIC, or AVR generally is where you should start. There's really no way around working in either Assembly or C if you want to do anything serious with it.
To go to a more powerful CPU gets you into all the stuff you don't want to get into, like critical and complex board layouts.
There's BASIC compilers, but you aren't going to get very far before you run into some serious limitations if you want to do anything complex.
Good luck on the JRE thing, something like that is way too slow and bloated to run on a hobbiest CPU. I think sun made a chip that executed bytecode, but I don't think they make them anymore.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
We have made a few ARM Boards to teach undergraduate students. They work with ARM Angel or with our own debugger KMD. You can write code in asm or C or any even gcc front end (inc java).
The best feature for me is the huge FPGA's to play around with. Its quite easy to pick up any electronic equipment and plug it into it. Lots of fun and great experience gaining stuff.
The board total was around 100 GBP (inc board manufacture and mounting) and I think if you ask nicely the designs will be available.
Alternatively you could use something like a 6809 or an 8051 but then writing code for them is a pain. The best idea if you use someting old is to emulate a better processor. You can then run your favorite ARM/MIPS/x86 code on an emulation and forget the real system.
Mouse powered Chips, Open source Processors and Lego
You might try an Arm processor, many of which have great built-in features (like NIC, daq, memory management). You can get demo boards, and run linux + related gnu tools on them. ARM-based systems make great embedded/distributed systems (aka 'the future'), and are a useful to learn for the old-resume.
mx
Firstly, these are good, and cheap: http://www.oopic.com/
:)
They can be programmed with java, c and much more, yet are basic. They are made to work well with servos and such, and are very versitile (read up on vitural circuit)
For a more expensive option, the famous mini-itx boards are great, and robot sized heh http://www.mini-itx.com/
Starting at about the same price as a microcontroller (well, some)!! Java is here too
I suppose I'm more looking into cpus for robots, so maybe this is not ideal.
puts ("Python r0cks\n");
I've been using Jave modules for a while...both from DalSemi/MAXIM (the "TINI", http://www.ibutton.com) and Imsys (the "SNAP", http://www.imsys.se).
They provide a JRE environment (JDK1.1.3 in the case of TINI, J2ME CLDC 1.0 in the case of SNAP) and allow me not to worry about processor and memory design etc.
Right now, my SNAP module lives in a custom extuded aluminum enclosure I designed, and has a MMC interfaced for it (an extra 16Mbytes of removable storage), and a bunch of other things like temperature/humidity sensors, LEDs, an LCD, and dry contact sensors.
One reason I like both these modules is that they support I2C, SPI and 1-wire for expansion....meaning that you don't have to necessarily hang stuff off the CPU's data bus.
Good luck.
-psy
If you *really* want a JRE (which is generally not synonymous with performance, in the microcontroller world), check out the TINI from Dallas Semiconductor, here.
If you want to get into heavier duty gear (and available only in surface mount), you can look at things like the Patriot from PTSC, here. There are also several others that I've seen, but can't recall the name of. A little Googling should find those.
JStamp make native java hardware. Never used them though.
dominionrd.blogspot.com - Restaurants on
If you want a processor, and not just a microcontroller, but still want it fairly simple, you might want to check Motorola's Dragonball chips. It sounds like you want something with a low pin count, and simple ISA, so that'd be my first suggestion, followed very closely by ARM.
If you want a microcontroller, then maybe the 68HCxx series, or a RISC-ish PIC. The 68HCxx series is probably the most popular there, and you can easily find code and design samples for them.
--That's the point of being root, you can do anything you want, even if it's stupid.
i think the z80 still can be used. modern doesn't mean too much on that scale. for more complex things Zilog makes a pretty big line of various processors for different uses.
Look at PIC processors from MicroChip. They nice, not extremly powerful, and you can hardly get a JRE for it. But you should be able to put together a little board in your garage for about $20.. And there are som decent C compilers for those processors. Look into them. And there is the ARM processors ofcourse...
I had to get back into microcontrollers, after a long absence (since the days when the 6809 was state of the art).
I decided to go with the Motorola HC12. Unlike the HC11, it runs compiled C code with reasonable efficiency (gcc supported), it's highly integrated, and you have a simple execution model (not very many registers, etc.) when debugging by hand. It's reasonably capable, and almost trivial to design your own project board (clock frequencies within reason, few external components required, every kind of I/O pin conceivable).
The more powerful versions have huge amounts of internal FLASH (256K), but not quite as much internal memory as I would like (8K is typical of recent versions).
There are cheap modules available from Technological Arts in Toronto. They don't seem to be very active this year with new designs, but they continue to supply their product line to local colleges last I checked.
Anything more complex than an HC12 I think I would want some kind of kernel OS. The next step up the food chain, in our evaluation, was an Atmel chip.
Nuts, I thought the title was Home Computers for Microbrew Projects. So I am understandably disappointed with the non-beer focus of these comments.
You know, the Z80 is still in production, and your experience helps explain why. It's a useful, accessible CPU. I mean, why are you doing this? Fun, self-edification? If so, the Z80 suits your purpose as well as it every did.
If you want to go really hardcore, check out homebrewcpu.com. That guy basically made a processor out of TTL components. I was actually searching google about this very same thing the other day and came across that site(although not much else other than old homebrew computer clubs).
For $69 you can get a 33mhz 80186 board with 512k of flash and ram, a flash file system, and a full Borland development ide. You get 44 digital I/O lines plus the raw processor bus and a bunch of integrated peripherals.
"Eve of Destruction", it's not just for old hippies anymore...
You can download it for free here.
MSP430 series from Texas Instruments are quite good (and with ultra-low power consumption). They also have GNU toolchain available.
* Origin: XBase BBS (2:490/4100) Well the good old days may not return and rocks might melt and sea may burn.
Why not directly jump at designing a Java processor? SUN's PicoJava architecure is AFAIK publicly available, at least for educational use.
:o)
And as you are at it, you could as well design this as a PCI board to finally speed up Java on the desktop
Personally, I prefer Atmel AVR devices these days. atmel.com and www.avrfreaks.com will give you all the info you need. Z-80 is still around in different guises. Zilog has a few single board computers that can run C and other languages and they're pretty inexpensive for what you get. Can even get them with webserver code! If you want to build another Z80 machine, get back to me. I have tubes of Z80s, Z80-CTC, Z80-PIO (I think) 32k EPROMS and 32k SRAMS sitting in my basement that I will probably never use. Considered packaging them up for eBay sale, but got too lazy. Give me $20 for the lot (50-100 chips) and you can build Z80 based computers to your hearts content :-)
This is Engineering work, and Slashdot barely qualifies for basic programming advice.
Better I would think to start with an FPGA. I did a lab like this when I was in school - not that it was being taught in the course, just that I was bored with the labs so decided to make my lab reports more interesting. Take an adder (in my case it was an actual 74181) and show how it's incorporated into a "real" CPU. Then counters, memory, etc. You can introduce it a module at a time, build a lab project around it (like wiring the FPGA as a basic adder and flipping switches on the workbench) then use it as part of the next project to make a more complex ALU. Keep going piece by piece and you've not only made everyone relatively comfy with a soldering iron and proto boards, you've also prepared them for senior classes and CAD tools.
There's a bazillion premade CPUs on the market now that have several on chip peripherals. But every one of them in low quantity is relatively expensive at $20 or more. Why not just spend that on an FPGA (or even twice that) and teach some skills that will help qualify students for something better than "skilled labor" assembly line work?
Rabbit Semiconductor makes a Z80 compatible which has all sorts of interesting hardware on-board, like a PWM generator and two quadrature encoders.
-russ
Don't piss off The Angry Economist
Remember that for embedded stuff, you want low power consumption. And for that, you'll have a relatively slow (by modern desktop PC standards) CPU. A JRE will use a stupid amount of CPU horsepower just to run, so your actual embedded system will run like a bag of shit. Can you imagine if the ABS controller in your car used Java?
Learn C. It's pretty similar to Java, but is far more suitable for writing embedded controller software. Remember that you are going to be controlling things, not drawing widgets on a screen, so an OO language is not really necessary (or even desirable). Instead, you will be reading and writing IO ports, which will involve a certain amount of bare metal programming. Java won't really let you do this. You could use FORTH, which was invented for writing controller software for steerable telescopes.
Of course, if you must use Java, you could always compile to machine code instead of byte code. It's still not particularly great for embedded stuff, though.
www.arcom.com they do a range and some of them can run a jre and linux.
What ya wanna do is subscribe to Circuit Cellar the magazine that is today what BYTE magazine was back in the z80 days. Full of articles on using modern, small processors to do "stuff". They also run some great Design Contests supported by various manufacturers that get you a development board and software (and generally extra chips!) for free.
/frank
These days, at the low end (less than 100 MHz), there is seldom a need to wire a processor up to much of any peripherals. For example, the motorola Coldfire processors are basically 200+ Mhz 68000 (e.g. 66Mhz with single-cycle instruction execution compared to the 68000's 4-10 cycle instructions) with just about any peripherals you might ever want onboard. Not really sufficient for a JRE, but not bad for just about anything else. Also, they're roughly $10 each in quantity. Many other manufacturers are making similar types of chips these days:
Hitachi processors
Rabbit Semiconductors
Zilog
One of the problems you'll have to deal with if you want to build your own systems is that Wire Wrap is simply unusable in this day and age. Not only is it impossible to find a socket for somthing like the 256-ball BGA that the coldfire comes in, or the more standard 144-pin QFP packages, the speeds make it unlikely you'll be able to use that technology successfully. I've built fine-pitch boards in my garage using photosensitive PCBs, but the best solution is something like PCBExpress or ExpressPCB and get 2 or 3 3"x3" double-sided boards for $60-$80. Even so, building high-speed systems is not for the amateur; laying out a system using PC-133 SDRAM is not something you want to do without a bit of up-to-date layout knowledge. Good luck, hope this gives you some pointers to get started with!
And the worms ate into his brain.
Many of the current MCUs are ball grid array (BGA) devices, which make them pretty hard to work with if you're not a professional, but a few can be had in PLCC or QFP packages, which means you can get an adapter board or socket.
You can also buy preassembled demo/development boards (this is the route I'm taking) and wire-up anything the board doesn't include by hand. Most of the MCUs on the market will have 32-bit memory busses (though they may not support more that 25 or 26 address lines), so you can attach just about anything you want to them.
This is exactly the kind of thing that Steve Ciarcia (of Ciarcia's Circuit Cellar) used to do: building personal computers from microcontrollers. Most of his designs used Z80 based devices, which was fine back in the mid- to late-eighties. Now, however, you can do a fair bit better.
As for speed, I don't know exactly what you're looking for, but the ARM devices can be had in speeds from 50MHz to 400MHz, and the same is true of the PPC and MIPS devices. That may not seem like much, compared to a 2GHz Pentium, but it's really quite nice.
Some good resources: Digi-Key is a reasonable source for all sorts of parts, Atmel makes some nice MCUs, programmable logic, and Flash RAM, Cirrus Logic makes some ARM MCUs and networking chips (amoung other things), Sharp, Samsung, Motorola, and AMD all make nice MCUs, Cogent Computers builds some nice development boards, and EarthLCD has good prices on LCDs and has an ARM based board in the works.
Forgot one thing... You simply don't have to program any of these chips in assembler any more, free software will get you where you need to be. Most, if not all, are supported by GCC, and most (if not all) have a port of the remote side of GDB available. Write in C++, debug in GDB over a serial link, and you're doing pretty good. You can also get free operating system software for them, ranging from simple multi-tasking kernels (uCos, eCos, etc) to full-blown desktop style OSs with filesystems (e.g.e uCLinux). Good Luck /frank
And the worms ate into his brain.
In the 'bigger chip' category... Coldfires are fun. (Strong)ARMs are popular. Embedded PowerPC is more prevalent than you'd think. MIPS makes a showing in various guises. 486s and Geodes are great fun for some.
And SuperH seems to go overlooked, though it did crop up in the Dreamcast. Here's one 'DIY' project that used it:
http://www.azpower.com/mylinux/
And the guy explained why he liked it:
http://www.azpower.com/mylinux/why_superh.html
He also shows how FPGAs can be used. What ever happened to that project!? I wanted one!
It does seem like GCC for SH has been a bit.. 'variable' or 'neglected' of late. But now that the 3.x branch has settled, I gather it's improving/seeing maintainership again.
The Kaii - http://kaii.info/ was supposed to have been another SuperH design, but I haven't heard a peep about it, either. (They wanted to design SH boards in appropriate form-factors, port the OpenPDA platform of the Zaurus and friends to it, and license their designs to manufacturers cheap. I assume they were stymied by some of GCC's unsettledness.)
Meanwhile, if you want to start by 'reverse-engineering' an existing design, the "CommodoreOne" will be a well-documented design when it's finally released, also using FPGAs to good effect. Sadly, the project pages are perpetually down at the moment (unless someone knows where they've moved), but it's a simple uATX board, and will be produced/sold by Individual Computer - http://jschoenfeld.de/indexe.htm" - when they finally get things straightened out. Google may help if you care. There's another (unrelated) project out there for a similar but simpler design meant to resemble early arcade or home game systems' boards, but sadly, I can't remember the name of it.
A company called Ampro has these little 486 boards complete with Ethernet, PS/2, IDE and VGA connections and they are affordable.
It's all Politics
They're powerful (considering their cost, $3 to $12 each), and very easy to program.
http://www.avrfreaks.net/ is an excellent AVR resource.
In addition, you can program AVRs in C rather than assembler. There is a GCC version that targets the AVR platform, although my preference has been for Codevision AVR compiler/IDE/library set- It's $150 for a full version, but worth every penny (Mainly for the library set it comes with. No need to write your own I2C, One-Wire, or LCD interface routines, as CV-AVR has libraries for all of those).
retrorocket.o not found, launch anyway?
Depending on what you want to do, a DSP may be the answer.
TI makes a few good ones. Advantages over a MicroController? How about a decent Math package (fixed point or floating point), built in programmers, and a large array of I/O options? The DSPs I've been playing with have 3 General I/O registers, any pin of the 24 available to the GIO can be configured to be input or output.
Add on integrated ADCs and DACs, and a decent clock speed, and you've got a nice system.
Did I mention that you can find older development kits on ebay for $100?
Oh and if you get a nice commercial application out of it, TI will custom manufacture them for you in sufficient quantities, with your program masked on the silicon itself. Prices for these are pretty cheap; starting at about $12/unit last time I looked into it (about 2 years ago)
I am disrespectful to dirt! Can you see that I am serious?!
In my experience embedding processors in very tiny MEMS sensor packages (smaller than a postage stamp) the Microchip Inc. PIC processors are an excellent embedded processor with many low cost programming tools. The PIC16f876 is particularly neat with a flash reprogrammable ROM and in circuit programability. But the 8 pin designs pack quite a wallop as well.
The Javelin is java version of the basic stamp for the more hardware adverse, For low performance work the Basic stamp is excellent.
Finally the uClinux folks have a series of SIMM (Memory) form factor single board computers with the same power of a small Palm Pilot which run uClinux.
Sure they don't have the power of some of the beefier solutions mentioned, but they're much easier to get into.