Among old CPUs, the 6809 never got as much attention as some of its cousins. The Radio Shack CoCo used it and so did a construction article in Wireless World Magazine. Now [Dave] has reconstructed that computer on breadboards and it looks great. The files are on GitHub and there is even a series of videos about the machine. You can watch the first one below.
You can even read the original articles in the January 1981 Wireless World where the board used a 6802. The upgrade to a 6809 appears in the July 1981 issue. The magazine promised you could build the system for £100. Besides the 6809 there were only a few chips. A PROM, two RAM chips, A 6821 PIA, and a 74LS138 decoder for address selection. An MC1413 transistor array also allowed for a 7-segment display and a keypad along with a 7442 BCD decoder.
Apparently [Dave] had started a similar computer back in the 80s, and made changes to it to adapt to the Wireless World’s project memory map. It sounds like he didn’t finish it, but he found the old boards and decided to recreate it on a breadboard.
Like many computers of the day, the machine had a cassette interface. We really like the aesthetic of the 7-segment LEDs and the overall look of the build.
The 6809 did see use in some specific industrial and video game applications. There was also a New Zealand educational computer based on the 6809, along with a few other home computers like the SuperPET and the Dragon.
The use of breadboard power strips for data and address buses is the Real Hack (TM) here!
Yeah, I am so done with breadboards. They are all fun when new and never used, but quickly make unreliable contact to the wires. Someone once told me there is a brand that is the bee’s knees but I cannot source it in my country and quickly forgot in frustration. A spool of insulated wire and a prototyping board is the only answer.
I’ve only used the American ones that have a lifetime warranty. I’ve been using the same ones for many years, some for decades, and they’ve never had any trouble. The troublesome ones are the Chinese ones, which don’t use the right material or shape for the prongs. I’ve seen one or two YouTube videos on them, where the presenter takes them apart and shows the problems. Unfortunately I didn’t keep the links. I think Global specialties and Jameco still sell good ones, and the ones from BusBoard are excellent. See http://www.busboard.com/breadboards .
I don’t recommend solderless breadboards for digital work if there are fast parts involved though. “Fast” refers to the rise or fall times of the edges in nanoseconds. It does not refer to the MHz (although ones with fast edges are generally capable of a lot more MHz too). I’ve seen newbies have a lot of problems with designs whose connections are all correct and good but the inductance of the long wires and especially the ground connections between ICs gets them into trouble. (Note that increasing the size of the wire has almost no effect on inductance which is the real enemy.) I don’t think the 6809 ever got past a 2MHz bus speed though, which hints that its output edge rates probably aren’t very fast either, and any reasonable solderless breadboard layout will probably work.
I like this – the 6809 is a seriously neglected chip that should have survived a lot longer, but it was lost when the 68000 series was launched.
Just in case anyone was in the dark – the (Tandy/Radioshack) CoCo and the Dragon Data Dragon 32/64 were built on the same circuit design. The could even run each others’ programs.
The CoCo was a better build, but the Dragon seems to have attracted the hardware hackers in the UK.
Then you might be pleased to hear that the 680x was by far the most popular CPU for Pinball machines* in the late 70s/all through the 80s and that in Arcade machines the 6809 was more popular than the 6502 although I didn’t research that, it’s just from fixing literally hundreds of arcade machines.
*) Bally and Williams (#1 and #2 in the 70s/80s) used these pretty much exclusively. Atari pinballs also used the 6800 but that’s a footnote in history. Gottlieb (#3 in the 70s/80s) used the 6502 for pinball machines and Zaccaria (#4 for the most part at that time) used the Signetics 2650. And Stern just copied Bally’s MPU board.
I haven’t heard of the Signetics 2650 in years. One of my first computers was based on a 2650 evaluation board with extra memory added off-board. It was pleasant to code in assembly and I used it for several years. One thing I remember is that in an era where 64K of address space was normal, the 2650 could only address 32K.
The CoCo was basically just the reference design you could find in the Motorola data books under the 6883 ‘SAM’ part. This was a trinity of 6809 CPU, 6883 SAM, and 6847 VDG. Unfortunately this was aimed at being a cheap ‘teletext’ style terminal to use consumer televisions as displays, so the VDG only offered 32×16 text which was just dreadfully inadequate for basically anything even at the point in time that the CoCo launched. The 6809 itself rocked, and the SAM let you use 64K-bit DRAMs w/o an acre of MSI chips, so those were cool, but that wretched 32×16 display and the chicklet keyboard dragged the machine into the toy category.
Nonetheless, with the ‘EDTASM’ cartridge it sure was fun for a couple of months, and made learning 68K assembly painless later on.
I seem to recall that Radio Shack had the Coco as a Teletext or terminal before offering it as a general computer.
Did a load of work on 6809 in the day. It was ideal for the Forth we were using at the time as it has two stacks!
Yes – Forth was a shoe-in on the 6809.
I found a real live UNIX clone that ran on the coco. OS9. It operated identical to Unix at the time with installable drivers etc. (which DOS stole to great success). The 6809 had address independent code ability. When written properly you could move your whole code with one byte granularity.
I used that to actually get a hard drive running on my COCO. I tried to get FORTH to run under OS9 but it wasn’t compatible with address independent code.
OS9 actually lived on on the 68000 as OS9000. The instruction set for the 6809 was basically a subset of 68000 code with all the addressing modes.
I always thought that the 6809 was the most advanced 8bit processor for its time with its 16 bit internal instructions, but it came out too late, just as the 16 bit processors came out.
Interesting note; you couldn’t stop the 6809 clock to pause, it used dynamic memory type cells for its registers. You had to use the pause pin to pause execution.
Whilst I started with Z80 and and 6502, once I had my Dragon 32 it was 6809 that I spent most of my time using – largely through reference to this fine tome, which I still have, and in slightly better condition than the one featured here :-)
https://archive.worldofdragon.org/browse/downloads/Books/6809%20Assembly%20Language%20Programming%20-%20Leventhal%20%281981%29.pdf
Being able to write in 6809 assember (and MC) as an “in” into the fledgling arcade game industry, and was a much needed alternative to my then day job as a cobol programmer.
There is something weird with this video. Youtube presents it as “4k”, but it definitely is not in that resolution, or it’s compressed to 1080p bandwidth or whatever. I did like the view of the old fashioned wiring pen @21:00, but video resolution is really lacking here.
Also, when working with KiCad, be aware of small squares that indicate open wire ends. For example, @32:31 there is a wire drawn right through C3 which almost shorts out the capacitor (I’ve seen shorts created this way a few times on the KiCad forum). Just below that, near R4 there is also such a square of an open wire end.
The French Thomson family of “personal educational” computers was not really successful but it was certainly widespread back in the days. I now own a few MO5 and TO8 as well as the hardware reference book that explains the inside working of the machine, reading that as a teen was really a life changer. I can’t touch z80 or 6502 because I was used to the comfort of the 6809, but going RISC makes the 6809 now look as… artificially limiting. Times change !!!
The Heathkit ET-3400A Microcomputer Learning System had absolutely outstanding digital logic and microprocessor course materials with it. It used the MC6800 with a later MC6809 adapter to upgrade it to, in my opinion, the best 8-bit CPU ever made. Someone should create a cheap modern clone of the MC6809 version since ET-3400As in decent shape are becoming more expensive and the adapters are rare.
Emulation using an Arduino or other inexpensive computing module with adequate I/O to connect to external hardware for the breadboarding experiments would do since the goal would be to use the training materials instead of recreating the hardware.
The course materials can be found online with difficulty and in the vast majority of cases don’t come with the used ET-3400As being sold.
The simple architecture of the MC68XX CPUs is ideal to teach the low-level principles of operation of microprocessors lest anyone ask, “Why train with an obsolete CPU.” It’s the course materials that are the hidden gem.
Do you remember when Motorola offered the 6811 development kit to a lucky few for $68.11? It was an amazing price for a development kit at the time, before Arduino and RasPi brought prices down.
The 6809 was widely used in industrial control systems and I met them when I was working at CERN in the 90s
https://www.europhysicsnews.org/articles/epn/pdf/1991/02/epn19912202p40.pdf
Maybe some of these systems are still running.
Interesting!
A lot of old Moore Products industrial control instruments (the 352, 382, 324 etc.) used the 6809. The 352 was a wildly popular controller that captured something like 80% of the annual US sales in the digital signal loop control market at its high-point. I remember going to a DuPont plant in Texas and seeing one peeking out from a little corner .. Until I pointed it out, the instrumentation engineer at that site didn’t even realize that he had any Moore equipment .. Which was amazing because this guy was one of the most fastidious nerds I have ever met. I guess someone stuck the controller in one day, and it simply did its now quietly for a decade or two, calling no attention whatsoever to itself.
The Moore instrumentation devs used those designs long after the chip was pretty antiquated, because a) the stuff worked, so no reason to change, and b) they could write code for those systems in their sleep.
So, yeah .. a solid little micro that could serve as a reliable brain for simple industrial controllers, running 24/7 for decades.
We are still using the 6809 on our spinning frames producing fiberglass textiles….we have 22 frames with that processor running SwiftX, which is a Forth based language. We are in the process of upgraded these to Allen Bradley PLC’s, but they are still running strong since 1990. They are using the 6821 PIA to control Opto22 Modules…..Two inverters(20HP and 10HP) and a Servo combine to drive one hundred 20 pound Fiberglass packages at each doff. Each frame is 40 foot long and 8 foot tall…..pretty impressive that a processor from my high school days is still running industrial machines!
SWTP had a 6809. And by then they had an MMU made with a 7489 RAM, so it could get a lot of RAM.
Gimix had high end machines that used the 6809. They even had smart peripherals for serial and parallel with their own CPU.
The first available amateur packet radio TNC used the 6809. But later iteratikns went with the Z80 for some reason.
I recall a Motorola application note that showed how to use two 6809s together, alternating on the same clock.
You could do that interleave trick with the 6502 too – essentially dual porting any/all RAM!
The 6809 was too little and too late for the personal computer market. By the time it because available, people were poised to move on to 16 bit processors. As others here have pointed out, it did find a market in a variety of embedded applications.
I hate to say it, but that’s true. The 68000 was out. There was that three part article in Byte about the 6809, and all the work they did to analyze 6800 code. And then there’s the 68000, far different, but better in other ways.
One thing the 6809 had was Microware OS-9. Motorola went to them to create a Basic that took advantage of the 6809 advances. So they created BASIC09. But they decided it needed an operating system, so they made OS-9. Muoti-user, multi-tasking, and influenced by Unix.
I got a Radio Shack Color Computer in 1984 so I could run OS-9. I.actually bought the OS a few days before the floppy drive.
I’m sure OS-9 appeared in some of the embedded 6809s.
A lot more than some. :-) Ran OS9, OS9-68K and OSK (didn’t run OS9000) on several machines. Still want a Gimix ghost with the smart IO but they cost a fortune!
@Al, careful with your loads. The 6809 needed buffer chips. I recall that the load on those CPU to be about 2 chips.
There was a 6809 project in Kilobaud, maybe around 1981. I think by someone who worked at Motorola. But it was a full project, complete with floppy disk controller.
It was Kilobaud for November 1980, page 118, by Tim Ahrens.
I actually leveraged what I learned using OS9 to work on a UNIX mainframe system in a voice response system.
OS-9 was super to learn coding c, asm, basic, nameit… on unix based systems. C compiler was great. VDG was fun to play with (obviously it’s not on the breadboard otherwise it would look that clean…!). All the arcade games existed on OS-9 and also car race, ski race, flight sims, schematic designers, windows based interface, graphics text editors, paint, db, worksheets, … Amazing what we could do at the time with just 1MHz clock, 8-bit bus and 64kb ram. There were also hardware extensions, floppy disk, hard disk, everything. I’m surprised we never talk about the CoCo2 in these pages.
Before I used it to hook up a hard drive, I used OS9 to run 4 floppies, 2 x 5 1/2″ and 2 x 3.5″. All at once with no problems.
Very easy to use it to talk to a SCSI drive since SCSI’s had a sector buffer on them.
We designed a Unix/Uniflex computer using the 6809: idd.nl/sophie.php. Problem is that it has some test-instructions that make the CPU crash, so you can’t test new software on it. Btw i wrote an assembler converter that could translate 6809 code to 6811 code of about the same size and speed, so we prefered the 6811 for later (embedded) projects. But the 6809 is still very nice although overkill.
Notable among the personal computers using a 6809 are the Fujitsu FM-7, FM-77, and FM-77AV. Also, the 6809 (and OS-9) were used in the Fairlight CMI. The Ensoniq Mirage used a 6809 if memory serves, but I’m not sure what OS it had.
I keep thinking (and ‘branting about) how Motorola really missed opportunities all over the place — “by just this much”. I’m sure that they were trying to squeeze the design into the technology they thought they were capable of producing, looking for places to shave things to fit, and shaving one or two too many. Maybe everyone had similar problems, but I’m most familiar with the 68XX and 68000.
In an ideal world, Motorola might have recognized that the 68000 was actually their entry into the 32-bit market, more than it was into the 16-bit market. Maybe that would have allowed them to see the 6809 as their entry into the 16-bit market, to fill in the gap between the 6801 and the 6805. A 6809 with a few minor fixes would have prevented the dominance of the 8086.
Unfortunately, by the time the PC-XT came out, there was no competing with the intel crap. It annoyed me that the 8086 was just a 8080 with extra registers shoehorned in with the abominable segment registers, but the PC was just too big.
I was in the middle of upgrading my COCO beyond EGA standards, literally wirewrapping my design when the first $25 IBM-PC VGA board came out for the PC and I gave up.
I opened my Hewlett Packard 5350B microwave frequency counter, to make a small repair, and felt kind of nostalgic to see the familiar 6809 processor inside.