What were the events and motivations that led to the design of the Mark-8?

The Mark-8 Minicomputer

Part 1: The Beginning

...being some memories of the machine from its designer, Jon Titus
Last revision: March 21, 1999, Copyright © 1999 Jon Titus

Jon Titus

This shrine to the Mark 8 Minicomputer discusses the machine with its designer, Jon Titus. Jon was gracious enough to write up most of the data you see here, and also provide some of the diagrams and pictures. Jon is happy to discuss the machine with you, but please look for your answer here, first!
Jon Titus
Milford, MA 01757-1362 USA
jontitus@cahners.com Work
harrowsmith@mediaone.net Home

The Beginning

I had been fascinated with computers and electronics for some time, but I'll tell you about the immediate events leading up to the Mark-8. (I still have some high-school-years "computer" projects in the basement from the early 60s.)

I was a graduate student at Virginia Polytechnic Institute and State University (Virginia Tech) in Blacksburg, VA, working toward a Ph.D. in chemistry. My research involved using minicomputers such as the Digital Equipment Corporation (Maynard, MA) PDP-8/L. The people in my research group were using minicomputers to control chemical instruments and experiments, and the computers also acquired the experimental information and processed it. At the time, acquiring 1 Msamples/sec (6- or 8-bits/sample) was considered high speed! My project involved designing a high-speed interface that would acquire 6-bit binary values, stack two values to form a 12-bit word and then jam these words into the PDP-8/L's memory using direct-memory access (DMA). The PDP-8 family operated on 12 bits at a time.

I enjoyed using the PDP-8/L computer for research and I had fun fooling around with it on my own. I taught myself assembly-language programming and I enjoyed making the computer do "neat" things. The neat things were controlling an X-Y display, controlling an X-Y plotter, dialling a phone line so we could get access to a mainframe, and so on. I often wondered how I could get my own computer. I realized that buying a PDP-8/L was out of the question. The computer alone cost about $5000 (US) in the mid 70s, and I would have needed a teletypewriter, too, for an extra $1000! At the time, a group called the Amateur Computer Society had quite a few members, and as I recall, some members were trying to clone a PDP-8. I can't recall whether or not they had any success. In those days, although some functions were available on small-scale integrated circuits, the biggest roadblock was the memory. The minicomputers of the time used core memory, which took quite a bit of external circuitry to properly drive the individual magnetic cores which actually stored information. Thus, memory was out of reach for almost all amateur computer enthusiasts, and so were computers themselves.

Intel


Small section of the original circuit diagram, showing the Intel 8008 processor. Select the image for a large, full view of the circuit.

In 1971, Intel introduced the 4004 microprocessor and some support chips for it. This 4-bit device seemed revolutionary, but with only four bits, it also seemed limited to calculator-type applications. I studied the 4004 architecture, the instruction set and the peripheral chips that provided I/O and memory functions. I quickly decided against using a 4004 CPU as the basis for my own computer. Intel's first 8-bit processor, the 8008 changed my mind. It provided a nice instruction set, an interrupt, a small internal stack, multiple internal registers, and it could address a whopping 16 kbytes of memory. To put this in context, a basic PDP-8/L minicomputer could directly address only 4 kwords. We were ecstatic when our basic PDP-8/L got an upgrade to 8 kwords, but the extra memory took up about the same space as the PDP-8/L computer itself.


Jon's Mark-8 prototype, as seen on the cover of July 1974 Radio Electronics.

Intel provided a nice databook for the 8008, and I devoured and lived with it for several months. While driving to Canada for a vacation in 1973, I resolved to adapt a demonstration circuit that Intel published in its book and use it as the basis of my own computer. I wrote to Intel and asked for some sample devices because the 8008 chips sold for $125 each from distributors who carried the chips. (Not many distributors carried them, because they didn't know what sort of market there was for them.)

Although I used the basic Intel circuit, a made some modifications. The Intel circuit provided about 1 Kbyte of read-write memory (Intel 2102 chips) and about 2 Kbytes of PROM (programmable read-only memory, in 1702 or 1702A devices, each of which held 256 bytes of data or instructions). I expected to add more memory, so I built in decoding for all 14 address bits. In effect, I could expand the memory up to the entire 16 Kbytes that the 8008 could directly address. At the time, that seemed like a tremendous amount of memory.

Because the PROMs sold for about $35 each and I had no way to either program them or erase them, I decided that my computer--like a PDP-8/L minicomputer--would provide a set of front-panel controls and indicators. By using the switches I could load in short programs in binary using the switches. The programs I had in mind would be short "loaders" or "monitors" that would then let me use a keyboard or a display device, such as Don Lancaster's popular TV Typewriter, to display information.

Success and Failure

In retrospect, the front-panel design was the most innovative contribution I made. The design was simple and elegant, and it used programmable counters to temporarily store an address during front-panel use. The counters served double duty as latches during normal operation. The design relies on simple debounced momentary-action switches to increment the counters, load the counters from the switch register, and so on.

If you look at the front-panel circuit on later computers such as the Altair, you can see what a kludge the designer made of the circuitry. The Altair used the next-generation 8-bit chip, the 8080, which provided for much simpler control of the device. The front panel circuit, though, hardly takes advantage of the chip's capabilities and instead relies on all sorts of timing circuits to accomplish front-panel control. I don't mean to be mean about the Altair's design, but it was a bit of a kludge.

I also redesigned the clock circuit. Intel's design relied on four monostables to create a two-phase clock needed by the 8008. I used a crystal oscillator, figuring that I might have difficulty finding a dual-trace scope to use during prototyping and debugging. The monostable clock required a user to adjust four trimmer resistors to get the timing right. I built a breadboard version of the monostable clock and found that it didn't always start when power was applied. So, this experiment provided another reason to scrap the monostable clock circuit.

The biggest failing during the design was the input-port circuit. I simply took what Intel provided and duplicated it. I could have used an three-state or open-collector bus for input, but it didn't cross my mind to design the input ports this way. The Signetics chips used to multiplex the input data proved a bit difficult to find, so I suspect some people who built a Mark-8 modified the input ports so they could have more of them and so they wouldn't use the Signetics chips.

Copyright ©Andrew Davie (site design) and Copyright ©1999 Jon Titus.
Contents may not be reproduced in any form without permission.
The Design