Ask Hackaday: When It Comes To Processors, How Far Back Can You Go?

When it was recently announced that the Linux kernel might drop support for the Intel 486 line of chips, we took a look at the state of the 486 world. You can’t buy them from Intel anymore, but you can buy clones, which are apparently still used in embedded devices. But that made us think: if you can’t buy a genuine 486, what other old CPUs are still in production, and which is the oldest?

Defining A Few Rules

An Intel 4004 microprocessor in ceramic packaging
The daddy of them all, 1972’s Intel 4004 went out of production in 1981. Thomas Nguyen, CC BY-SA 4.0

There are a few obvious contenders that immediately come to mind, for example both the 6502 from 1975 and the Z80 from 1976 are still readily available. Some other old silicon survives in the form of cores incorporated into other chips, for example the venerable Intel 8051 microcontroller may have shuffled off this mortal coil as a 40-pin DIP years ago, but is happily housekeeping the activities of many far more modern devices today. Still further there’s the fascinating world of specialist obsolete parts manufacturing in which a production run of unobtainable silicon can be created specially for an extremely well-heeled customer. Should Uncle Sam ever need a crate of the Intel 8080 from 1974 for example, Rochester Electronics can oblige.

Thus when we’re looking for the oldest CPUs, it’s those available from regular production that we’re talking about, not old stock, not special manufacture, and not a core included in a modern part. What’s the oldest CPU in production today that an engineer from back in the day would recognise? To answer this question it’s necessary to delve back into the timeline of early microprocessors, and comb the world’s electronic suppliers in search of obscure chips.

Working Through The Years

A Motorola MC6800 advert from 1976.
A Motorola M6800 advert from 1976. The architecture survives in some of NXPs microcontroller ranges. Motorola, Public domain.

The first microporcessors appeared on the market in 1972, with Intel’s 4004 being generally accepted as the first of the breed. None of the early Intel chips are in regular production today, mostly we suspect because of the success of the aforementioned Z80. Its backward compatibility with 8080 instructions as well as its lower support chip count than the Intel part made it a huge seller, and the obvious choice for continuing development in that line. From the rest of the players in those first couple of years as far as we can see no others survive. This shouldn’t be too much of a surprise, as the microprocessor was a new, expensive, and fast-moving technology in which no one chip gained enough market share to attain longevity.

So with nothing from ’72 and ’73 remaining in production, it’s in 1974 that things become interesting. The TMS1000 mask-programmed microcontroller has long since faded away, but the Motorola 6800 series still appears in suppliers listings in the form of the Hitachi HD46800 and its derivatives manufactured by Renesas. A closer look shows these processors as being supplied through Rochester Electronics though, so sadly we’ve not found our earliest still-in-production chip. The NXP HC05 and HC08 microcontrollers may be the 6800’s distant descendant, but don’t count here because they’re not the same part.

Like a tired glam rock band unable to see punk on the horizon, we stumble into 1975. Here there’s a clear contender in the form of the aforementioned 6502. It was famously first sold at the Wescon electronics trade show in September of that year, and continues to be available in CMOS form as the WDC 65C02. Before crowning it though, there’s another contender. The RCA 1802 is variously listed as launching in either 1975 or 1976, and is listed on some suppliers websites as the Harris CDP1802. However as far as we can ascertain it was first on sale in early 1976, and yet again it’s revealed as coming via Rochester Electronics.

The Winner, And A Few Others

A two-page advert from 1975 for the MOS 6502
Winning since 1975, the 6502! MOS Technology, Public domain.

We have a winner then, in the 6502. Its popularity in home computers and more mundane appearances in a myriad control systems have kept it in production for nearly five decades, and its influence has permeated far and wide.

But before we leave this subject, it’s worth taking a look at some other famous microprocessor families to see whether any of their forebears remain from later in the ’70s.

It’s fair to say that the families we’ve already covered span most of the popular 8-bit lines from the mid-1970s, and by the end of the decade the attention at the cutting edge had moved to 16-bit designs. Of these, 1979’s Motorola 68000 ceased production in the mid 1990s, and the closest remaining family parts are NXP’s Coldfire range. For its 8086 rival we’ve talked about the much later 486 above, but surprisingly there’s a semiconductor company that still lists the an equivalent to the 16-bit Intel part in its product line. The Kvazar plant in Kiev, Ukraine (Google Translate link) has the KR1810VM86 Soviet 8086 clone, though sadly due to the war in that country we’re guessing they might have more important things on their plate at the moment.

Finally, it may be a surprise to some users of PIC microcontrollers to find that this family has its roots in a peripheral chip for a long forgotten General Instruments 16-bit system. The PIC1650 from 1976 was simply a Peripheral Interface Controller that contained a programmable core. This gave it a usefulness far beyond its original application, and while the original chip has long gone out of production there are still PIC chips in production that share the original PIC architecture.

When we were discussing this topic among Hackaday staffers we didn’t expect the 6502 to be the winner, as we were sure that a few others would have made it. But since our ruleset applied to volume-produced devices and the original chips rather than merely architectures, it’s hardly surprising that something so hugely popular would remain in production. Let’s hope it’s still on sale as it passes into its sixth decade!

header image: Pauli Rautakorpi, CC BY 3.0.

89 thoughts on “Ask Hackaday: When It Comes To Processors, How Far Back Can You Go?

    1. It’s not clear-cut. The WDC 65C02 has a modified pin-out from the old NMOS 6502, so it isn’t quite a drop-in replacement (it can be bodged in with a little effort). There was a drop-in version of the 65C02, the Rockwell R65C02, but that’s long out of production. The WDC part is still a 6502, though. It executes the same code (provided you don’t use undocumented opcodes) and uses the same bus protocol.

      I think a good contender for the title might be the 68020. It came out in 1984 and production was re-started a few years ago, apparently to supply military customers. Rochester sell brand new 68020s that are for all intents and purposes identical to the Motorola original.

      1. My dad implements and maintains security systems based on the 68k (I think they use the 68320) in a field where the customers are slow corporate behemoths that don’t want anything changed at all. They have ~real~ thorough testing and one day new products had a low change to misbehave under a very specific brown-out condition, although previously systems were able to recover to a safe state 100%. Every component was checked, replaced and rechecked. After months of investigations they narrowed it down to the 68k fabrication date.

        At first Motorola denied anything being different but after more back and forth it turned out that someone over there decided to slightly change the silicon after 30 years for reasons undisclosed (maybe patch some super obscure error?).

        The outcry was immense

        1. Sounds like when Zilog fixed a bug in their Z80, and there was much hue and cry from the financial software segment because some programs had been written to use the undocumented and buggy instruction – but in a way that worked.

          So Zilog put the bug back and IIRC also made it work the right way too. It’s been years since I first heard of this so I’m fuzzy on the details.

          Needs to be a HaD retro tech article on it.

          1. Yea – I’d love to get more details about this Z80 bug if anyone has them? I wrote a Z80 assembler (in assembler) with source editor and monitor back in the day, so pretty close to the Z80 bare metal.

        2. This is quite off-topic, but I want to share it anyway. :)

          I worked for a company that had something quite similar happen, although the company was not a slow corporate behemoth that didn’t want change ;).

          Company made devices with S3C2410 Samsung CPUs. Went through all of the development and verification phases. All fine, factory toolset was set up, and first testrun of 1000 devices came out great. And at the end gave the go-ahead to have a batch production of 100,000 devices.

          So, 100,000 devices went to the shops, and were sold like hotcakes. But a few days later the reports started rolling in… All customers had the problem that if they turned off the device (device was put into sleep mode), it wouldn’t turn on anymore. The only thing that would turn it on again was to disconnect and reconnect the battery (power cycle). But battery was built-in, so it was no solution for customers. Basically a 100,000 one-time use devices, already sold to our customers. Their Christmas presents ruined, etc.

          After lengthy investigation, it turned out that Samsung had made some small last-minute patch to the CPU without telling anyone. And then the same thing happened: first they denied it, making us do a full investigation, make a detailed report and basically prove beyond doubt that they had changed the silicon and that that was the cause of the problem. Samsung even made us go to court over it, I believe.

          I don’t know how it was settled in the end, but we ended up with running a new batch of mainboards with a fixed CPU, a new batch of brand new working devices, and setting up a refurbishment program. We took back all of the customer’s devices, gave them brand new devices, refurbished the faulty devices and sold those as ‘refurbished’ for a lower price.

          Also, because of Samsung’s inflexible stance towards us, even though we were a big customer for them, caused us to switch to Qualcomm for future devices. I think Qualcomm cpus were being manufactured by TSMC. Don’t know if something similar ever happened after switching to Qualcomm, but I think not.

          I won’t give a moral, just the story. Things like this happen, nobody is perfect. Maybe one moral: the bigger your supplier is, the more they’ll let you jump through hoops if they made a mistake.

          1. I have a very similar story. I was working for an appliance control OEM who had been using Hitachi 8 bit processors because there were some of the remaining who made uCs with VFD capable output drives (a big thing in appliances still, though this was in the mid 90s). Design and test went fine with the OTP parts. Did a small run of mask ROM parts and EMC gives it the thumbs up. Switch to production and turn the first 10K chips. They fail EMC. Turns out Hitachi switched to a die shrunk version of the processor between the first and second mask ROM runs. The smaller process geometry had better output drive characteristics (good, right?) but that lead to faster edges on output switching and that meant more emitted noise. The product ended up needing an extra shielding layer added to the design and that ate well into our already thin profit margin. From then on out, we were very careful to specify that no processes changes would be accpted after the initial acceptance runs. I don’t know how much we had to do to get Hitachi to admit the change as it was just before I worked for the company, but there was a lot of resentment in everyone over the outcome.

    2. I agree, and I’m a big 6502 fan for historical reasons.

      The 65C02S by Western Design Center is not even pin compatible with the original 6502: If you try to use it in a board designed for a normal 6502, you have to disconnect one pin which was ground on the original 6502 and is an output on the 65C02S, and you have to pull one or two formerly non-connected lines to VCC. Also the RDYB line (!RDY on the 6502) is an input as well as an output on the 65S02; many old circuits pull it straight to VCC with no pullup resistor but WDC says you shouldn’t do that.

      As far as I’m concerned, I would reluctantly call the Z80 the winner. It’s still for sale from the same company (Zilog), and it’s fully compatible with the original if I understand correctly. But the 6502 has a cooler story…

      1. The stories of Z80 and 6502 are kind of similar… Micropocessor design engineers left big companies – Intel and Motorola, that wasn’t giving a sh*t about the microprocessors at that time, formed their own companies – Zilog and MOS Technology, and designed micropocessors that were an improvement of their designs in their previous companies

        1. I’d dispute the idea that the Z80 is an “improvement”, other than from a marketing perspective. Its implementation of its new features is horrible, and it misses the big omission from the 8080 (which the 8085 would have fixed, had its LHLX and SHLX instructions been documented) – which the presence of IX and IY throws into stark relief; they were seemingly added to mirror the 6800, but of course, the 6800 can do LDX n,X whereas the Z80 can’t. (But then the 6800 has no straightforward way to push its X register, which is arguably worse.)

          Still, even by Jenny’s criteria, the Z80 should be the winner – the NMOS part may no longer be available, but the CMOS part is clearly a (99% bug-compatible!) straight conversion, rather than the reimplementation that is the 65C02.

          1. Z80 does support LD HL,(nn) and LD (nn),HL, which I think are equivalent to LHLX and SHLX. There are many more improvements in Z80 vs. 8080. On the hardware side: single 5V power supply. No need for external clock generator, easy to use (at least in simple systems) control signals – /MREQ, /IORQ, /RD, /WR, a bit weird, but yet efficient way to implement interrupts (IM2). On the instructions side: block instructions, bit manipulation, a set of shadow registers, and two additional 16-bit registers that you’ve mentioned

          2. The Z80 had some ram refresh built in. I think that was key. Just as DRAM was spreading in hobby circles. People got excited about the added instructions, but I gather much CPM software didn’t bother, in order to be compatible with all the systems using the 8080.

            And when there was a wave of dual CPU boards for the S-100 bus, I think they all used the 8085. Easier than the 8080, but nobody felt a need to go for the Z80.

          3. No, Sergey, it’s not the same. The instructions you mention are LHLD and SHLD, and documented in the 8080. The 8085 provided two additional 10-cycle instructions, unofficially dubbed LHLX and SHLX, which loaded and stored, respectively, the contents of HL via the address in DE; but for reasons best known to Intel, neither they nor the 8 other instructions the 8085 added were documented officially. (Only RIM and SIM were documented, and they were required for the serial port.) See this PDF for details: https://www.ee.iitb.ac.in/~sumantra/courses/up/undoc_8085.pdf

      2. The WDC65C02 doesn’t only have a different pin layout, it also doesn’t have the exact same behaviour (bugs were fixed) and it supports new instructions, addressing modes and other features. As one would expect from a processor that came out in 1983 and not 1975.

        I don’t really understand the criteria used by the author, but the WDC65C02 doesn’t qualify.

    3. The original NMOS 6502 seems to have been out of production for two or three decades, as the CMOS 65c02 is better in every way. (See my article at http://wilsonminesco.com/NMOS-CMOSdif/ .) So is it still the same processor? Well, yes, and no, depending on how you look at it. Some users are committed to the NMOS though because of vintage computers and a desire to continue to software that used the undocumented NMOS op codes.

    4. How many others reading these comments are still holding on to original documentation that influenced their outcome?

      On a curio shelf I display my Intel 8008-1 User’s Manual:
      * 12.5 microsecond instruction cycle time.
      * PMOS 18 pin DIP package.
      * Fixed seven deep internal stack (the 8080 changed it to a memory stack pointer register).

      * Who could have forseen the 8008 machine instruction set mnemonics to compile into machine code that still runs in every ‘WinTel’ PC being sold today?

  1. QFP 486 parts still seem to be available by the thousands NOS. Actually that seems to go for a lot of x86 in packages that don’t plug into desktop motherboards, there’s some TI 386/486 CPU and mobile Pentium MMX stocks also.

    Then there were some parts that were a bit stealth in that they were renamed, I think AMD were making 5×86-133 far into this century as part of the geode line, and higher models were Athlon XP, I believe they stopped both now though.

    As you say above though, many architectures survived but not in their original form, there’s 8042 and Z80 based things around still also.

    1. AMD SOCs with a 486 CPU core were widely used in thin clients and at least one “internet computer”. IIRC it was called MyPC or My PC. It had a custom GUI OS and had several programs built in.

      Sort of like a next generation NABU but used dialup TCP/IP and with the optional floppy drive and a printer it wasn’t completely useless without a network connection.

      ISTR there was a second generation of the device which never shipped because the company went out of business thanks to sub $1,000 PCs and free internet services like NetZero.

    1. The F8 design was kind of stolen.
      It’s ancestor was developed in Germany, originally.
      The original CPU was the CP3-F from Olympia-Werke.

      There’s an interesting story about it.
      Wikipedia does it partially tell, I think.

      Pictures: http://retrocpu-cp3f.net/

      Anyway, it’s kind of reassuring to know that “our” design helped to establish a computer industry.

      1. Interesting. I did not know that, even though the Fairchild/Mostek F8 Development Kit was my first computer, back in early 1978. It had a whopping 1K of RAM and 1K of ROM as the standard FAIRBUG monitor.
        I still have it but the peripherals we were hacking to use as its I/O (an IBM 3277 keyboard and an I/O Selectric) never got finished and we soon switched to building an S100 machine.

  2. CMOS versions of Z80 CPU and the corresponding peripheral chips are still being manufactured by Zilog and hugely popular among homebrew computing community.
    CMOS versions of 8088 and 8086 are still manufactured by Renesas (Harris/Intersil).
    Analog Devices / Innovasic still makes 80C186 and 80C188 compatible CPUs
    DM&P apparently still manufactures ALi M6117C/M6117D that integrate 386SX together with an AT-compatible chipset in one chip.

    1. The Z80 also had a few unofficial clones, I think.
      It was the most popular CPU in former East Germany (GDR), maybe the whole Soviet Union.

      https://en.wikipedia.org/wiki/U880

      In addition, the Z80 had fine support chips. Namely the Z80 SIO and the Z80 PIO.
      They were good serial/parallel port alternatives to the i8250 and i8255.

      The last 16-Bit CPU created in “the other Germany” (GDR) was an 80286 compatible, I believe.
      https://en.wikipedia.org/wiki/MME_U80601

      1. Iirc those Z80 IO chips were kind of odd ducks to program. Where the 6502/6800 family would map their registers to a range of address allowing direct access, the Z80 support chips were got at through a single I/O port (which makes sense given there were only 256 ports), meaning you had to hit the registers in prescribed sequences.

        1. Thank you for your reply!
          I think I understand your point of view.

          I suppose they were useful, though, for building a basic CP/M system out of few components.
          The SIO was fine for connecting a serial terminal, at least.

  3. > There are still PIC chips in production that share the original PIC architecture.

    If I ever have to deal with PIC16 or PIC18 again I’d rather kill myself by swallowing a dildo.

      1. Or better yet, use a high quality compiler. The Proton (now Positron) did an amazing job, and aside from its inbuilt functions created code far tighter than any C compiler for those chips.

          1. I suppose it could be said that in general the early processors did not make very good targets for compilers, and compilers for these turn out some terribly inefficient code.  OTOH, these early processors were much easier to write good assembly-language code for than modern processors are that have super-complex instructions, deep pipelines that must be kept full, multi-level caches (separate for instructions and data), out-of order execution, and other things that simply don’t apply to the 6502.  Add to that program-flow-control structure macros (which I have used extensively on the PIC16 and 6502) for nested structures like IF’s, looping, CASE structures, etc., and other good use of good macro assemblers we didn’t have in the early days, and assembly language can really sing, not just in performance but in programmer efficiency, reduced bugs, better maintainability, etc., with, in most cases, absolutely zero penalty in run speed or memory taken.

          2. Since nsfhb said, “If I ever have to deal with PIC16 or PIC18 again I’d rather kill myself,” I should have added to my post above (or will it appear below??), that part of my heavy use of macros is to hide the backward logic and other mickeymousities of the PIC16.  I still have full control, because I wrote the macros myself; but I don’t have to look at the ugly innards every time.  PIC18 is a step up, having added some things to get around deficiencies of the 16; but they’re still kluges.

  4. This does not count, since nothing that I started with back then has been in production for decades. But recalling one’s first experiences is tempting.

    I remember writing an octal debugger on an Intel 8008. It was done on an MDS8008 borrowed from a friendly manager at a local AT&T Bell Labs site (Columbus Ohio). They were no longer using the development system and could spare it for some Ohio State EE students trying to get in some practice. I believe that was in 1976. We ended up returning the MDS and Teletype ASR33 to AT&T after getting the debugger going, as we were getting into higher level studies and needed to concentrate on school.

    The basic UI was based on the Digital Equipment PDP-8 ODT and DDT debuggers that I used during my first few years at OSU.

    At that time, the extra-high-speed 8008 could be driven by a 200kHz clock, and the standard-issue 8008 was specified for 125 kHz.

    The company I work for today is still maintaining production data acquisition systems based on 32-bit X86, VME bus, Mototola 68302 and 68HC812 8/16 bit microcontrollers. As recently as 2017 I found and resolved a register-usage bug that had been present in the 68HC812 code for decades. The 68HC812 work is implemented in assembly language, which means that I was still coding in 68xx assembly language 5 years ago for chips that have been out of production for decades. In 1982 I thought my 8086 assembly work would be the last in my life, then in 1995 I thought some 80386/486 work would be the last assembly language project, again in 2006 with 8051-derivative processors and finally 2017 with the 68HC812. I promise that will be my last. :-)

          1. I’ve done extensive research on Intel 4004 history, and I’ve never encountered this myth, yea or nay. The 4004 was the brainchild of an understaffed side-contract for a single Japanese customer before Intel made it available to the wider market. The compatible 4002 RAM chips were truly bizarre (320 bits each) but very easy to wire up. Still, no way to sell lots of memory.

  5. Back in the early 1980’s OlI designed the front end of an ICE unit for the NSC800. Early 8 bit CMOS micro from National Semi. Had a Z80 type instruction set with an 8085 type handshake. For it’s day, the chip was pretty cool!

      1. The bulk of USB, PCIe, and SATA peripheral chips these days are managed by an 8051 core. I used to design external HD enclosures and in one product you can expect to find between one and several 8051s. Open any embedded USB, PCIe, or SATA device and look for 8 pin flash chips with a colored paint dot on them. Chances are exceptionally good these chips contain 8051 code.

    1. As I understand Many of the 87/8951 series, with internal ROM, from many manufacturers could be dropped-in in 8051 (without internal ROM) sockets without modification. The internal ROM would be completely disabled.

      I’ve found many other really weird microcontrollers like that, as well, NEC, some 4bitters, Hitachi, on old VCRs, etc.

      I always thought that was a brilliant move… Those pre/mask-programmed chips can be reused. Pull a pin high/low, add a ROM. Go!

      1. Ahhh, leave it to me to care about being mistaken about something no one seems to care about that I thought fit *perfectly* in the theme of the article so many folk seem to care about:

        FWIW: The 8×51 series *has* internal ROM, even the 8051. The 8031/8032 are allegedly “ROMless” versions of the 8×51/52.

        What that means to designers/hobbiests is, e.g.:

        You can design for an 8031, with an external ROM, then drop-in any old 8×51, even if its (OTP, mask, etc.) ROM was programmed with code from an entirely different product, or buggy, or whatever.
        Find some old 8×51 in some old piece of trash, think it’s worthless because it’s already been programmed…? Tie one pin to a voltage rail and use it as an 8031 in your own project.

        What that probably meant from Intel’s perspective:

        “Hey, a customer ordered a bunch of preprogrammed 8051’s, but discovered a bug before we shipped” and/or “We got a batch of 8051’s with flaky ROMs” and “we’re sitting on 1000’s of otherwise useless 8051’s. What should we do with them?” “Remarket them as ROMless 8031’s, and let the new customer supply their own ROM chip!”

        Forward-thinking, reusability, reduction of eWaste…

        And, in this new era of old paper datasheets now scanned and uploaded as pdfs (as opposed to my earlier experience where pdfs only existed for products designed in the pdf-era (THANK YOU to those who take the time and provide that effort!)), I’ve discovered that many previously zero-search-result ICs on old PCBs scavenged from VCRs, CD players, TVs, Stereo Components, and whatnot, now have full-on datasheets detailed down to instruction-sets and hex operands…

        And many of those, similar to the 8051, have a pin which can be tied to a voltage rail to disable the internal mask-ROM, enabling them, like the 8031, to run off an external ROM.

        Wow! SO MANY uC designers, of so many various architectures, even 4-bit, considered this worthwhile! And even if it was purely because they wanted to sell otherwise defective/wrongly-programmed uCs, it *still* benefits the likes of the customer and even the alleged us, that are hardware hackers.

        I dunno, I thought it was darn-relevant in the “chip shortage era.”

        1. I care! I didn’t know this, and am delighted to learn it! So, I suspect, would the collapseOS folk be, because this is exactly the kind of scenario they’re thinking of; especially given the chances that 8051-alikes outnumber Z80s and 6502s out in the scavengeable wild.

          Do any other microcontroller families act like this? (Obviously, they’d need to be those which expose an external memory bus, so that cuts out almost all the PICs and AVRs.)

      2. Why is this not posting?
        Ahhh, leave it to me to care about being mistaken about something no one seems to care about that I thought fit *perfectly* in the theme of the article so many folk seem to care about:

        FWIW: The 8×51 series *has* internal ROM, even the 8051. The 8031/8032 are allegedly “ROMless” versions of the 8×51/52.

        What that means to designers/hobbiests is, e.g.:

        You can design for an 8031, with an external ROM, then drop-in any old 8×51, even if its (OTP, mask, etc.) ROM was programmed with code from an entirely different product, or buggy, or whatever.
        Find some old 8×51 in some old piece of trash, think it’s worthless because it’s already been programmed…? Tie one pin to a voltage rail and use it as an 8031 in your own project.

        What that probably meant from Intel’s perspective:

        “Hey, a customer ordered a bunch of preprogrammed 8051’s, but discovered a bug before we shipped” and/or “We got a batch of 8051’s with flaky ROMs” and “we’re sitting on 1000’s of otherwise useless 8051’s. What should we do with them?” “Remarket them as ROMless 8031’s, and let the new customer supply their own ROM chip!”

        Forward-thinking, reusability, reduction of eWaste…

        And, in this new era of old paper datasheets now scanned and uploaded as pdfs (as opposed to my earlier experience where pdfs only existed for products designed in the pdf-era (THANK YOU to those who take the time and provide that effort!)), I’ve discovered that many previously zero-search-result ICs on old PCBs scavenged from VCRs, CD players, TVs, Stereo Components, and whatnot, now have full-on datasheets detailed down to instruction-sets and hex operands…

        And many of those, similar to the 8051, have a pin which can be tied to a voltage rail to disable the internal mask-ROM, enabling them, like the 8031, to run off an external ROM.

        Wow! SO MANY uC designers, of so many various architectures, even 4-bit, considered this worthwhile! And even if it was purely because they wanted to sell otherwise defective/wrongly-programmed uCs, it *still* benefits the likes of the customer and even the alleged us, that are hardware hackers.

        I dunno, I thought it was darn-relevant in the “chip shortage era.”

  6. My first project used the 6802. I was a member of the team that did the Data General D200 video terminal. I wrote the keyboard control code in assembler. After that, I did several personal projects with leftover parts, one of which was a controller for a ham radio repeater. Another was a decoder for the control stream on the GI Jerrold 450 addressable cable TV system. It was a very easy processor to learn, though it suffered from an index register that could only be incremented or decremented. Offset operations would have been mighty welcome (and appeared with the 6809).

    The 68000 was the next processor I used, but I didn’t do any coding on it. Its architecture would be very familiar to anyone who had used a PDP-11.

  7. The Intel 386 was the first real CPU for me, having gone from 8085 through Z-80, bypassing the 286. The reason was, it had an understandable MMU and virtual memory. This was the chip that Sun made billions on, porting Unix to it. Also came Minix and Linux.

    1. x86 and x86_64 are what drove Sun into near bankruptcy and into the arms of Oracle, they killed the SPARC business which is where All their profits came from.

      Sun executives made billions by buying Microsoft stock.

      Sun was always just an extension of Bill Joy’s masters thesis project.

        1. I worked on the 34082 floating point unit that complemented the sparkC chip as an engineer at Texas Instruments. We built their silicon for them. Later on I took the 34082 as an improved coprocessor to the Motorola 68O30 plugged into APC ISA port. What a nightmare that was. Big Indian ittle edian. Motorola instruction set, TI instruction set, C code on the PC to load all the instructions onto the board. Oh yes, add wire wrap. In engineers worst nightmare.

  8. I had always heard that the Z80 held this title, since it has been incorporated into countless TV remote controls. The rule “not a core included in a modern part” begs the question, what is a modern part? What about a core included in an old part? Where is the dividing line between “old” and “modern”? Riffing on previous commenter (Ramtop), no one makes NMOS parts anymore (but please, someone prove me wrong). None of the original microprocessors are still in production.

    1. Does Harris still make the PDP-11 on a chip? I thought they did but I haven’t checked lately. Kind of surprised the 180x (derived from 1802) isn’t still out there for space applications, although that market has opened up lately.

      1. Harris makes radiostations for PT-91 tanks (an upgrade of T-72), internally they’re based on PDP-11 with DSP extensions. If it’s working don’t break it. And to be honest FHSS comms is not rucket science nomore.

      2. The original NMOS CPUs (8008, 8080, 6800 , 6800, 6502 etc.) are gone, but Rochester Electronics *does* have the original RCA/Harris CDP1802 in stock. It was always CMOS, which gave it a longer life. Indeed, many are spec’d for military/aerospace applications. But there are plain commercial grade parts at reasonable prices as well.

        1. Rochester in fact has just completed recreation of the 6800 and 68030. 68020, 68040, 6802, 6809, 8080, 8085, 8086, 80186, 80C186, 80286, 80C286, cdp1802, 8051, 8751, 87C51, 80C51, 8031, 80C31, etc. are all available and more being looked at as well as 80486.

      3. My guess is that, when the Soviet Union became the former Soviet Union and tech embargoes dissolved, the motivation to continue making chips waned, and, while DEC crashed and burned, their motivation to home-grow PDP-11 clones was eclipsed by the x86 dynasty. It is heartening to hear that Ukraine still has a chip foundry or two. The demand for “old node” fab lines is high. May they live long and prosper.

  9. A couple of oddballs that date back to the beginning:

    1. 8×300 microcontroller which I got acquainted with in ’79.

    2. Bell system through Western Electric made their own microprocessor called a “MAC-8” around the same time.

    I wouldn’t be surprised if either of these are still available.

  10. I hand delivered the last production shipment of the 4004 to a customer in 1981 while working for Intel. Never thought it would be of interest 41 years later only that the single box was worth a little bit more than $1m at the time and that was my introduction to semiconductor economics.

  11. To be honest, for some reason I still prefer the Z80 over the 6502. I really wouldn’t be able to say why. But I think that it’s basically about the higher professional level of the Z80’s tools. But again, to be honest, the 6502’s tools might well have caught up in the past few years.

    But actually I think the easiest-to-program-for 8-bit CPU from that era is the 6809, or better: the Hitachi HD6309. I can’t find if it’s still being manufactured. A few years ago I bought a batch of 10, and have not run out of them yet. ;)

    If the HD6309 is still being produced, I think that should have won the title.

  12. My first processor ws the Fairchild F8 in 1975. It was on a board with the 3 other chips that make up the system, and I communicated with it via a terminal. I hand-assembled little programs and watched the results on the terminal monitor. The F8 was the largest-selling microprocessor system for a while, before single-chip versions of the 4-chip system were developed.

      1. It was given to me by my employer after we had a talk about micros. A while later I bought a 6809 add-on to the TRS-80 and did some programming on that, too. Now I use some of the advanced tiny boards. It’s a wonder how things have progressed.

  13. w at home was developing it in 8051 bored with all of the IO controllers from Intel. Programming was done in hexcode. It was stored on a cassette player where 1200 baud was a one and 2400 bod was a zero. All wire wrap. I still have that board and it still works.

Leave a Reply to Lou MCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.