Retro Computing

This article is a personal remembrance of the early days of digital electronics and computers. If you are of a certain age, this may be nostalgic for you too, but if not, you will be amused at the crude state of affairs of the computer field when I was growing up.

Since I was involved with the computer industry going back to the 1970s, I enjoy recollecting my experiences.

I have lately become fascinated by what seems to be a growing interest in the "retro" computing hobby, so there are some mentions on this page about projects recently initiated by others.

Early Logic Experiences

In middle school, we were subjected to the "new math" and we had textbooks from Yale University which looked like they were produced on typewriters, along with crudely drawn diagrams. The good part of this (and maybe the only good part) was an introduction to set theory and symbolic logic, which I found interesting. However, there were no practical ways for me to use that knowledge until much later. When I learned about Boolean algebra, later on, the symbolic stuff made sense.

Speaking of High School, I recently learned of the passing of a classmate, John Walker. John was a soft-spoken, private guy who became the founder of Autodesk and the creator of AutoCAD. John and I were not close, but I am amazed at his career and accomplishments. He stepped down as CEO and moved to Switzerland in the 1980s, continuing to develop software for the company and exploring his other extensive interests. Please take a look at his personal website here.

Programming

Computer Design

I later took an EE course in computer design. The textbook was based on the DEC PDP-11 computer architecture and went into extensive detail about instruction sets, arithmetic logic units, I/O structures, and memory mapping techniques. I loved this course, even though it was all theory with no hands-on. One day I was in the lab talking to the professor and noticed a box on a shelf that said "computer breadboard" or something similar. I asked about it and he said it had never been opened and no one had looked at it. I dug into it and discovered it was a huge set of breadboard modules with transistor logic that could be assembled on an easel and wired up with jumper wires. This was like deja vu for me, but on a much larger scale than my NORDAC. It turned out to be an 8-bit serial computer which the professor let me assemble in the lab as a special project. It had 8 instructions and functioned similarly to the DEC PDP-8S, which was a cost-reduced 8-bit serial minicomputer produced in 1967. It took me a couple of days to get it working (there were many, many jumper wires) and you could single-step the clock to watch the data march through the registers. This was my first hands-on experience with something you could call a real computer. 

At that point, I had no idea that I would end up working at Digital Equipment Corporation for most of my career, but that's another story.

Microprocessors

When microprocessors were a new thing, I was fascinated by the Intel 4004 chipset, originally developed for the programmable calculator market and introduced in 1971. I got the manuals for it and wanted to build a system. But things progressed rapidly, and Intel released the 8008, quickly followed by the 8080. There was an explosion of DIY computer kits such as the Altair 8800, introduced in 1974. This was soon followed by the IMSAI 8080 in 1975, pretty much a clone of the Altair. These were popularized by construction articles published in magazines like Popular Electronics and Radio Electronics and largely started the personal computer craze. These were considered hobbyist kits since they required a lot of hands-on building and low-level programming. Similarly, Apple released the Apple I in 1976, based on the low-cost MOS 6502 chip. It was not until the release of the Apple II in 1977 that a "finished" computer was available to the personal computer market at a reasonable cost. Apple was joined almost simultaneously by Radio Shack when they launched the TRS-80 the same year, based on the new Zilog Z80 chip. Both the 6502 and the Z80 were cheaper and simpler to implement at the systems level. Commodore joined in with the PET all-in-one computer, also based on the 6502, and followed up in 1980 with the lower-cost VIC-20. 

I followed all these developments pretty closely but did not become an early adopter until around 1975 for work purposes.

The Motorola 6800

In 1974, Motorola had introduced the 6800 microprocessor. To me, it had a much more straightforward architecture than the Intel chips, which seemed to have been heavily influenced by the earlier 4004 instruction set (to this day, the X86 architecture is something of a hodgepodge, but that no longer matters because the hardware is buried in layers of software). In fact, the 6800 was heavily based on the Digital Equipment Corporation's PDP-11 minicomputer architecture (which, as mentioned, I had studied in college), originally developed in the late 1960's. Since my business partner and I were looking to get into microprocessors for industrial control applications, we started looking at the 6800. Motorola published a huge applications manual that we obtained, and it was really helpful in learning how to use the microprocessor.

Sadly, our partnership ended before we could get into production with this series of computer parts, and I went off to work for Digital Equipment Corporation (DEC) as a support engineer. However, I had learned a huge amount about microprocessors in the process. I knew there would be a solid career for me in computers, so this was a good move for me.

The TI 99/4

The Commodore 64

DOS, Windows, Getting Online

Apple II computers were selling reasonably well and they had launched the Macintosh around 1984. However, these were closed, proprietary systems and priced above what many (including me) considered a reasonable investment for home computing. 

The original IBM PC (launched in 1981) architecture had an internal bus called the AT bus. Later IBM replaced it with the Micro Channel bus and peripherals were pricey due to IBM's licensing restrictions. Because the AT bus licensing was less restrictive, many clone motherboard makers adopted it and it was renamed the ISA (Industry Standard Architecture) bus, and this opened up the low-cost PC clone market. (Later, ISA was replaced by an enhanced version called EISA, standardized by a consortium of PC manufacturers). 

I could see that the PC clone market was taking off, fueled by the DOS operating system, which Microsoft was making available to anyone who wanted it. I built a home system from parts (a cheap Intel 286 motherboard with ISA bus, power supply, case, modem, CDROM, etc.) acquired from a local guy and built a system to run DOS and MS Windows in 1990. I can't recall if I installed Windows 3.0 on this computer, but I definitely ran Windows 3.1 when it launched in 1992 and then later 3.11, which had local networking capability. 

So the Commodore was relegated to the attic. Our kids grew up using Windows machines and since I worked for a computer company, newer hardware was not an issue (DEC had finally got around to making IBM compatibles, so I had access to them). Neither was software since I had an MSDN license courtesy of DEC and access to all the software Microsoft produced. Nowadays, we use MacBooks and iPhones at home because, well, they just work. 

But - Wait for it - I recently (2023) decided I needed to have a Windows computer back in my shop, so see my article on a cheap Windows computer.

The Retro Computing Hobby

The have been several attempts to resurrect or emulate old machines. I'll detail a few here, including the PiDP-11 that I have built. 

Recently, there seems to be a great deal of interest in simpler, Transistor-Transistor Logic (TTL) based designs. These are minimalistic designs intended to be educational and can be used as teaching tools.

Ironically, this design could have been implemented before Wozniak designed the Apple I around the 6502 chip. This design has even fewer parts! To be fair, the 32K static RAM chip used would have been expensive back then, but in reality, all RAM was pricey at that time. People were so enamored with using microprocessors, they overlooked simpler approaches like this. DEC had created (nearly) single-board TTL computers like the PDP11/05 in the early-mid-1970s but they did use bit-slice ALUs. This was followed by the PDP11/04 (a true single-board computer minus memory). One of my co-workers at DEC gave me the original wire-wrapped prototype of the PDP11/04 (he was directly involved with the project) on a hex-sized module which I hung on my office wall as a piece of techno-art. It probably had about 200 ICs, but bear in mind the PDP11 had a very sophisticated architecture and bus structures.

Marcel passed away but Walter still maintains the website. Unfortunately, he no longer sells kits, but you can still get parts (PCBs and a complete set of parts) from this supplier. Alternatively, you can get the PCB Gerber files from the Gigatron website and get your own boards made. Plus they have a complete BOM of parts. I think this would be ideal for a club or high school computer course. 

Here is a description from the Gigatron website: 

The Gigatron TTL microcomputer is a minimalistic retro computer. It’s special in its own oddball way, because it has absolutely no complex logic chips in it, not even a microprocessor! Its CPU is built out of a handful of classic 7400-series ICs, colloquially known as the TTL logic series. These chips combined form a powerful 8-bit processor. 

Besides running applications, this processor performs functions that traditionally require dedicated peripheral chips for video, sound and other I/O. By eliminating these the hardware remains small and understandable. Still the single-board system works as a full-blown microcomputer that you can play video games with. 

Now you can easily build one yourself. As you build, learn what happens inside a CPU by looking inside one. See the functional units, look inside the Arithmetic and Logic Unit, see its truth tables, and learn what makes up a ROM. Then go on to enjoy playing the built-in retro video games or write little programs in BASIC. You can also hack it in any way you like if you have a taste for that.

Ben Eater

Another experimenter who falls into the minimalist-retro computing category is Ben Eater. His website describes several designs and breadboards for "simple" projects such as an all-TTL 8-bit computer. This computer was based on a design by Albert Paul Malvino, who wrote a book called Digital Computer Electronics.* In it, he described a simple computer called the SAP1 (Simple As Possible), Eater also has a project to build a 6502 chip based design, but he recommends the TTL design for learning how computers work at a much lower level. He also covers the basics, like simple logic circuits and how adders work. He provides schematics and kits and has a great YouTube channel. Ben's explanations are crystal-clear and he is very humble in his teaching approach - it works for all audiances.

*This text has a lot more in it than just the SAP1 - it is a very complete survey of computer architectures where the author develops more advanced computer designs and has a comparison of early microprocessors as well as programming techniques at the machine code level. A worthwhile read for anyone interested in how computers really work.

DrMattRegan

Matt Regan is another figure in the minimalist computing field. He has made many YouTube videos explaining how Turing machines work and how to code them, as well as complete designs, one of which will emulate the Apple ][. You can learn a lot from his channel. He took the Ben Eater SAP1 design and made it even simpler, calling it the Turing SAP1. His approach is to use larger EPROMS to replace as much combinatorial logic as possible with firmware, including ALU functions, reducing the total chip TTL count to 17. This is basically a finite-state machine with registers used to hold data during state changes. He draws the schematics with KiCad in real-time (in itself a good learning experience) and wires up breadboards as he is speaking in the videos so you can follow along. His approach is not as straightforward as Eater's but is worth studying.

Regan has a second series of videos where he extends the functionality of the SAP1 into a (more complex) 6502 TTL computer capable of running Apple ][ software (games, etc.).

Hackaday Homebrew Projects

There is a lot of interest on Hackaday in creating small projects to build simple processors and systems. See this link for some of them. One person is even building a CPU with LED-Transistor logic gates. Other designers are using techniques like Field-Programmable Logic Arrays (FPLA's) to implement CPU functions.

Computer Museums and Festivals

There are a number of computer museums around the world dedicated to preserving the knowledge and history of the computing arts. A fairly complete list is here on Wikipedia. Not too long ago, very few existed, but DEC had one at its Marlboro, Mass facility. That was later moved to downtown Boston, and then relocated to Mountain View, CA, where it was combined with a much larger collections from various manufacturers. There is also one in Rhode Island and one in Maryland. Generally, these are staffed by volunteers who attempt to resurrect old machines and get them back to working condition.

David Lovett is a well known YouTuber, under the name of Usagi Electric, who restores ancient computers. He is even restoring an old Bendix (yes they made computers once upon a time) tube-based computer for one of the museums. He also has built a single bit (serial) computer entirely out of vacuum tubes.

The largest regular gathering of retro computing enthusiasts is called the Vintage Computer Festival (VCF). This draws hundreds of attendees, including David, and features demos of restored machines and software. Also - Oscar Vermeulen (see below) has been instrumental in reviving interest in old DEC machines with his Raspberry Pi based simulators and goes to these events. The most recent schedule is here.

My Recent Projects

The DEC PDP 11 Emulator - PiDP 11/70

The simulation software running in the PiDP-11 is called SimH (and available on GitHub here), a brilliant piece of code created by Bob Supnik, who worked at DEC and was famously responsible for a VAX CMOS chip project that obsoleted the enormously expensive VAX 9000 mainframe project less than a year after its introduction (that's a whole 'nother story). SimH is an open-source project that has been enhanced over the years by others to allow the emulation of all sorts of computers, mostly obsolete today (even the Altair).

In the case of the PiDP-11, there is a whole host of legacy operating systems (available in multiple versions) that can be booted up. These include:

• DEC RSTS

• DEC RSX11

• DEC RT11

• BSD UNIX

• SYSTEM V UNIX

• ULTRIX UNIX

• XXDP diagnostics

• and others

Each of these has programming languages, games, and utility programs so it is an endless playground for retro computing idiots like me.

The DEC PDP 10 Emulator

May 2024 Update: Oscar is shipping PiDP-10 kits now. There is a website describing the project here and also here. There is also a Google Group here for discussions about the project. The team's first YouTube video video is here.

YouTuber CuriousMarc has a nice video here where he interviews Oscar and gives some background on the PiDP-10 project.

The Guardian also recently published an article.

June 2024 Update: I received my kit and completed the build - see below:

Building the PiDP-10

I completed the build of my kit and added this page to document some tips I came up with in addition to Oscar's very good instructions.

Why is this interesting? A bit of history: 

The DEC PDP-10 was a very cool mainframe computer. It was developed just as the computer time-sharing industry was ramping up, and thousands were sold to companies such as Prodigy and CompuServe, as well as many universities. The DEC operating system, TOPS 10, was strictly aimed at timesharing. But other researchers had a slightly different idea. 

At MIT, there had been a long series of computing projects involving some of the early post-WWII computer architectures and software developments, most notably beginning with the Whirlwind tube-based computer. Next, the TX-0 and TX-2 computers were designed and built at MIT's Lincoln Labs. These were among the first all-transistor machines, replacing vacuum tubes. One of the engineers on the project, Ken Olsen (an MIT grad), saw an opportunity to start his own company and in 1957 Digital Equipment Corporation (DEC) was launched, Initially, the company built simple electronic modules for the scientific and research markets. But Olsen had a goal of building a computing machine consisting of many small modules and using the knowledge he had gained at MIT, so he directed the building of the DEC PDP-1. This was the very first "personal" computer. It had a graphics terminal and users could interact with the screen in real time. Olsen donated a PDP-1 to MIT in 1961 and it quickly became a favorite of the "hacker" culture. Some of the very first video games such as "Spacewar!" were developed on the PDP-1 there. 

MIT also created what was referred to as Project MAC (for Mathematics and Computation) which was looking at creating (among other things) a timesharing system running on an IBM 7094. This became known as the Compatible Time Sharing System (CTSS). Around the same time, the MIT Artificial Intelligence group, led by Marvin Minsky, was incorporated into the MAC project. CTSS evolved into MULTICS, the first highly available timesharing system. Both of these served as inspiration for Unix, a more modern implementation, developed around 1969 at Bell Labs using DEC machines.

Meanwhile, at DEC, the PDP-1 was quickly superseded by a series of newer DEC computers. The first large-scale 36-bit machine was the PDP-6, introduced in 1964. DEC only sold a handful of these but one ended up at the MIT AI lab. A small group of hackers was unhappy with CTSS and wanted a more "open" system. They developed a new OS, with the tongue-in-cheek name of Incompatible Timesharing System (ITS) which they wrote for the PDP-6 in machine assembly language and released it in 1967 for use by AI programmers at MIT. This eventually led to a split between the AI and MAC groups. The MAC group continued to focus on CTSS and MULTICS but was populated by people with ties to large corporations (like IBM and GE), which was repellent to the hacker-focused AI group. For a good history of the MIT groups and the evolution of these computer technologies, see this paper.

One of the most significant features of ITS was that it had zero security, by design. Users could look at and edit each others' files and projects and did not even have to log in, although it was considered polite. There was a "turist" login username for lab visitors. The group added all sorts of peripherals to the machine, notably graphics terminals such as vector and raster displays. Even the first laser printers were supported. The system was also one of the earliest connected to the ARPAnet. It even provided users with an instant messaging capability. All these features were way ahead of their time. 

In 1966, DEC released the 36-bit PDP-10, which was mostly compatible with the PDP-6 (but much more scaleable and reliable) and one was delivered to the AI lab. The ITS group ported their OS to the PDP-10 and even combined the PDP-6 in a shared memory configuration, so the installation became one of the first multi-processor computers! One of the developers, Tom Knight, created a terminal known as Knight TV, which was made available to multiple users. The TV system used a DEC PDP-11/20 (introduced in 1970) which had a Unibus backbone that was interfaced directly to the PDP-10, further extending the idea of a closely-coupled multi-processor system. Knight built a video switch and keyboard multiplexers to interface the terminals to the PDP-11 and Knight TV became the preferred method of interacting with ITS.

ITS spawned the creation of significant software projects, including the EMACS editor, the LISP programming language, the game Zork, Mac Hack (the leading graphical computer chess program for many years), and many others. More importantly, it became the foundation for open systems in general. 

The ITS project remained in use until about 1990, so it had a long life. Interest in the OS was continued at the Stacken Computer Club in Sweden until about 1995. Lars Brinkhoff has been instrumental in resurrecting many of the ITS projects and collaborating with Oscar Vermeulen on the PiDP-10 project, along with Rich Cornwell who is working on the SimH-based emulator for the KA-10 processor (the original PDP-10 CPU). 

So the history of the DEC PDP-10 and ITS is intertwined in a very interesting way. And now with new emulator projects and the resurrection of the original software projects, that history is being brought back.

The original (V1.5) ITS manual is located here. It has an overview and many details about operating the system. I suspect that the "hackers" felt obligated to document the system since a lot of the funding was provided by ARPA.

Other resources

Angelo Papenhoff has complied a nice website documenting some of the history of the PDP-6 and PDP-10, along with some info about the resurrection of Knight TV. He and Lars Brinkhoff collaborated on getting the Knight TV emulation package working with ITS.

A copy of the Digital PDP-10 System Reference Manual can be found here. This manual mainly covers the impressively large (for 1966) instruction set in assembly. A copy of the PDP-10 KA10 Central Processor Maintenance Manual - focused on the CPU hardware is here. 

Oscar's GitHub repository is here. Check here for the latest version of the PiDP-10 software. 

Oscar's KiCad files with schematics and PCB layouts are here.