I was wondering about the 68xx family of chips and tried to put them in a family tree to keep them straight for myself based on Wikipedia and datasheets. Please let me know if I missed anything or you know anything I can fill in.

(Speeds listed are the highest I could find. I only listed two 65xx chips, although there are a lot more.)

Interesting! I didn't realise there were so many. I don't know if there's truly an ambiguity, or how to address it, but for example between 6805 and 6804, were the registers further reduced, or is "reduced" relative to 6800?

Also interesting that the address space went up as high as 23bit - I wonder what applications benefited.

When there are parallel descendants, I think it might be best to put them in time order, from oldest to newest, left to right, if you can. That would put 6502 to the right of 6800, and the (unrelated) 68000 right off to the right too. But would it swap any other columns? I don't know!

Wikipedia describes the 68HC11 as "descended from the Motorola 6800 microprocessor by way of the 6809", so the sub-tree that starts with the 68HC11 should probably be moved (if you trust Wikipedia).

Perhaps the existence or absence of the U register (and its binary footprint in the machine instruction encoding) are influencing the tree structure? Correct me if I'm wrong, but I believe that the U register was an evolutionary dead-end, last implemented in the 6309.

Mike B.

I think the 65816 was a Western Design Centre product not Western Digital.

The U register allows separate system and user stacks. Perhaps deemed not really necessary for micro-controller type apps of later processors. The 6809 didn't really evolve into a desktop workstation type processor.

Reduced relative to the 6805. Apparently the 6804 was meant to compete with low-cost 4-bit chips, so the ALU and data path were cut from 8 bits to 1 to reduce costs. I tried to list everything relative to the model it came from. So the 68RS08 is a reduced version of an enhanced version of an enhanced version of an enhanced version of a reduced version of the 6800. You can see why I wanted a chart

That's a good idea but finding the year they were introduced is the hardest part so far. You can go by the earliest datasheet you can find but that is not 100% reliable and a lot of them don't list a year. Does anyone know a better way to find out?

I saw that but I don't think it is right. The datasheets I looked at said it can execute 6800 and 6801 instructions, not 6809 instructions. The 6801 has a D register and 8x8 multiplier like the 6809, but if I understand correctly, some instructions were redesigned for the 6809 and the 68HC11 cannot execute those.

Looks like you're right. Thanks.

Thanks for the clarifications.

Things like the U reg appearing (and perhaps disappearing) are certainly notable changes to the architecture.

For introduction date, sometimes you can find new product announcements in a Google Books search. But indeed, there's the marketing launch, then availability of the first parts, and then volume - could be a year or more between them. So it's not easy.

I never heard of the S and U stacks being used for system and user stacks, despite the implication of the names. What OS was that?

AFAIK, the U stack was provided to be the Forth data-stack. Implementing Forth is a major hassle on most processors because they only have one stack and Forth needs two stacks. This is why Forth has a reputation for being inefficient. If a processor has two stacks though (as my TOYF does), then Forth can be efficient.

The 6809 was primarily used as a desktop workstation type processor --- it ran OS9 that was, by most accounts, the best OS available for 8-bit computers.
The 6809 was obsoleted by the 68000, that was better for desktop workstations --- the 6809 was not obsoleted by the 6811 because the 6809 wasn't used as a micro-controller all that much anyway.

I remember one electrical engineer (in 1994) who told me that he wouldn't use any 68' chip as a micro-controller because Motorola continually pulled the rug out from under him. He had used the 6809 as a micro-controller, but then Motorola said that it was obsolete and he had to rewrite his code for their 6811. He switched to the 8032 because it never changed. An argument can be made that it is not as powerful as the 6811, but he didn't care because it was adequate for his purposes. Besides that, he designed hardware --- writing software was somebody else's problem --- he didn't care if programmers complained that the 8032 lacked a lot of features (an indexed addressing-mode) that make their job easier, because making their job easy wasn't what he was paid to do.

Even today, the 8032 continues to go strong, despite the fact that it is 1980s technology. By comparison, Motorola has given up on processors altogether --- they spun off FreeScale, and now we have NXP, but they are maybe 1/10 of 1% of the market --- Motorola pretty much failed, and a big part of the reason was that they were too innovative, continually upgrading, leaving their customers behind...

Almost everything today is ARM Cortex. It hasn't really changed since it came out in the 1990s --- it was better than the CPU32 of the 1990s because it had less interrupt latency, whereas the CPU32 was mostly a desktop workstation type of processor --- micro-controllers have strict requirements regarding interrupt latency, but there is no requirement that the programmer's life should be easy (by most accounts, ARM assembly-language is much more difficult that CPU32 assembly-language).

Two stacks is really handy for Forth but I'm sure Motorola were thinking of operating systems more generally. Their market strength was in automotive embedded computing. Two OSes which spring to mind:

OS-9
Flex 9

(I don't know their internals or how they use the 6809's registers.)

@ Druzyek: some informations for your (pretty nice !) chart:

I have an "M6805 M146805 Family Microcomputer/Microprocessor Users Manual" (M6805UM(AD)) First Edition from 1980. The 146805 was the CMOS branch of the HMOS 6805.

A 4th edition "MC68HC908GP32 HCMOS Microcontroller Unit Technical Data" from 2000 (still Motorola, not Freescale).

A "Advance Information MC68HC11A8 HCMOS Single-Chip Microcomputer Technical Data" from 1985.

An Application Note AN942 "MC68704P2 8-Bit EPROM Microcomputer Programming Module" appeares in "MCU/MPU Applications Manual Volume 3" First Edition 1986, mine is a reprint from 3/88.

The "microcomputer components" book from 1979 (no information about the edition) states (pg.18) "with the expected availability before the end of '78, the MC6801 single-chip microcontroller...".
The same book (pg.23) "Introducing the MC68000... Motorola's Advanced Computer System on Silicon", and within this intro there appears a "Motorola's Microprocessor Evolution" chart where the 6805 is mentioned. There were full datasheets about the 6800, 6801, 6802, and 6808. A brief descriptions about the 6803 (a romless 6801) and a preview of the 6809. But no further word about that 6805!

The 6801 was the first 6800-family microcontroller (with ROM), the romless 6803 was released nearly at the same time. (Motorola was a second source for some Fairchild F8 (3870) and Texas Instruments TMS1000 microcontrollers at that time as well.) The 6809 appears maybe a few months before the MC68000 (that depends on whether you are counting XC devices (XC = MC but not fully fulfilling the specs) or not).

To my knowledge the (CMOS) 6301/03/09 are only produced by Hitachi. Motorola usually named CMOS devices MC14xxxx that days.

I think you should insert the 65C02 (and perhaps its derivates) between the 6502 (max speed 4 MHz in NMOS) and the 65C816.


@Hugh:
The 6809 was designed in parallel to the 68000. It was never considered to be a desktop computer - at least by Motorola. The 6809 was a lightweight but powerful processor for advanced control purposes where the outdated 6802 was too slow and a 68000 would be ways too expensive. But soon NMOS and HMOS technologies were outdated and HCMOS was "state of the art". Perhaps the monolithic design of the 6809 was the cause that it was discontinued. The then emerging families 68HC11, and HC08 were designed in a very modular fashion. Their various peripherals could easily added or omitted within the numerous devices that follows.

OS-9 was indeed a wonderful OS for the 6809 - and perhaps the most reason why the 6809 was a popular microprocessor (and desktop computer) in its time. Was it Microwares extraordinary restrictive policies that prevents OS9-68000 to become similar prominent ?



@BigEd:

To my knowledge Flex-09 was a straight port of Flex-6800 for the 6809 (I didn't find my Flex handbook, so I'm not sure about the 6800-Flex name). Flex-09 make no use of Y and U registers (as they simply didn't exist in the 6800).

EDIT: some typos corrected

Interesting note about Flex-09. It does seem to be very small.

Whereas uniFLEX seems to have been quite substantial:
http://www.retro.co.za/6809/UniFLEX/index.html

(A bit more info here and here.)

Quite a while ago I have been reader-only of your first reference. Flex-09 requires 8 KB (RAM) while Miniflex (sometimes called Flex-0) only requires 4 KB (RAM). That's something I wasn't aware so far. Perhaps sometime I can figure out the reasons what causes Flex-09 being that "huge".