|
We were lucky enough to have a chance to interview Jonathan A. Titus. He was very helpful and was able to teach us a lot. Click the questions below to see his responses.
Q. What was your inspiration to make the Mark 8? A. Basically, I wanted my own computer to work with at home. I had gained experience with a Digital Equipment Corporation (DEC) PDP-8/L minicomputer in 1971 and 1972 while in graduate school at Virginia Polytechnic Institute and State University (Blacksburg, VA) and decided I could find a way to build my own computer. At about that time, Intel produced its 4004 microprocessor, a 4-bit CPU. Intel also provided specialized I/O and memory chips, but they had limited capabilities, and a 4-bit value would have made processing text information difficult. ASCII-encoded characters require eight bits each. As I recall, the instruction set for the 4004 had limitations that would prevent it from serving as a general-purpose processor. When Intel introduced the 8008 microprocessor, things looked brighter. The chip's instruction set seemed powerful and with an 8-bit processor, an engineer or student could do some useful work. So, I got manuals for the 8008 and started to think about a design. I developed an interest in electronics as a youngster and got a lot of support from parents and grandparents. While in high school, I designed digital circuits, such as a 4-bit binary adder. I didn't know much about digital logic and the techniques used to design circuits according to Boolean-algebra rules, so my designs were more brute-force efforts than anything elegant. Relays, lights and switches consumed a lot of my attention. While in college at Worcester Polytechnic Institute, I had an opportunity to operate an IBM 1620 computer with FORTRAN-language programs. We used FORTRAN II and did some rudimentary programming. I had a deep love for chemistry and was in the chemistry program at WPI, so I never seriously considered going into electrical engineering. At the time, the EE department at WPI had only a basic digital-logic course or two and nothing that dealt with computer architecture or computer engineering. I did enjoy programming the 1620 and had several opportunities to get my hands on the actual computer itself. After my sophomore year at WPI, though, students no longer had direct access to the computer, so I had to submit my punched cards--one line of code per card--through a "window" to a computer operator. Results and my card deck appeared in a small mailbox. Each student had a physical mailbox, much like those in a post office. So, the isolation from a real computer dampened my interest in hands-on work with the IBM 1620. Q. Did you ever think about expanding on the Mark 8? A. No, I never had plans to expand the Mark 8 beyond the original design. My design had a memory limit of 4 kbytes of program and data "space," although the 8008 could address as many as 16 kbytes. I cannot remember why I set the 4k limit. Some people changed the Mark-8 design slightly so they could address al 16 kbytes. Q. Have you ever thought about designing a successor to the Mark 8? A. Yes. In 1975, I designed the Mark-80, a computer based on the Intel 8080 microprocessor. That computer used dual-width dual-height Flip-Chip printed-circuit boards. Digital Equipment Corporation used that size board in some of its computers and the size and density of edge-connector contacts seemed good for the Mark-80 design. You can find an example of a single-width single height Flip-Chip board at: http://en.wikipedia.org/wiki/Flip_Chip_(trademark) The Flip-Chip boards measure about 8.5 in. by 5 in. You can contact Douglas Electronics (www.douglas.com) for a photo of the company's "33-DE-11" board that provides an example of the board size. I chose this board because it provided a nice set of edge connectors and a standard size board that easily fit into a rack-mount card cage. The Mark-80 boards included an 8080-CPU card, a memory card (3 kbytes of R/W memory and 1 kbyte of EPROM), a front-panel card with LEDs and switches, a serial I/O board, and perhaps another board or two, but my memory fails me right now. My brother, Dr. Chris Titus, designed the memory boards. I licensed the board designs to E&L Instruments, then of Derby, CT. My colleagues, David G. Larsen and Dr. Peter Rony and I wrote "BugBook III: Micro Computer Interfacing: Experiments Using the MARK 80 Microcomputer, an 8080 System" (paperback) and many schools and universities picked up the Mark-80 and the books as teaching tools. The Mark-80 was not aimed at hobbyists and experimenters and it cost more that most experimenters could afford. In a few cases, companies used the Mark-80 in industrial equipment. The serial I/O board, for example, went into some systems at one of the big-three auto manufacturers. The Mark-80 did not do as much for the personal-computer or hobby-computer movement as the Mark-8, but it was aimed at different types of users, mainly in educational institutions. Read More:  The Mini-Micro Designer (MMD-1) Brochure.pdf [1.28 MB]Read More: The Micro Designer Brochure I.pdf [1.85 MB]Read More: The Micro Designer Brochure II .pdf [735 KB]Q. How does it feel knowing that one of your creations is on display in the Smithsonian? A. It is a good feeling. I do wonder, though, whether people who see the Mark-8 in the Smithsonian's exhibit will understand the overall context in which I developed it. Unfortunately, the Smithsonian's exhibit does not provide much insight into the hobby-computer "movement of the late '60's and '70's. I'm glad the Smithsonian has a continuing interest in preserving artifacts from the age of computers. The Mark-8 and other computers, reside in the Smithsonian's Information Age exhibit in Washington, DC. Q. Was it difficult for you to design the working on the design of the Mark 8 and its circuitry? A. I based the Mark-8 design on a circuit that Intel published in its literature for the 8008 processor. So, much of the control logic in the Mark-8 circuit came from the Intel design. I modified the Intel design somewhat, but don't claim any originality for the basic Intel design. The Intel engineers deserve the credit for that circuit, which Intel used on its SIM8-01 board. The Intel design for the SIM8-01 board required the use of Intel 1702A 256-byte EPROMs into which Intel "burned" a boot-loader program. At the time (1973), those EPROMs cost $35 each, which seemed like a lot to spend for 256 bytes of storage. And, I would have needed an EPROM programmer to burn code into the devices and an ultraviolet lamp to erase the information if I made a programming error and needed to reprogram a 1702A with revised code. Intel sold a programmer board that connected to the SIM8-01 board, but that board cost several hundred dollars, and of course I would have needed preprogrammed 1702A EPROMs to use the programmer board. A classic chicken-and-egg problem. So, I opted to design the Mark 8 with only static random-access memory (SRAM). Intel offered 1101 memory chips that could store 256 bits of information, so using eight 1101 memory chips in parallel would provide 256 bytes of 8-bit storage. Several companies sold surplus 1101 memory chips for a few dollars each, so with a modest budget, I put together 768 bytes of storage in my original Mark-8 prototype. But, how would I get the information into those 1101 SRAM chips? My design used a set of switches on the computer circuit's input port and some logic that would let a user toggle the switches to drive data into a designated memory address. The switches would also drive 8-bit data into a low-address latch and into a high-address latch. (The low-address bits corresponded to address bits A7--A0 and the high-address bits corresponded to A13--A8. Recall the 8008 chip used the upper "address" bits, A15 and A14, for control functions, thus the 16 kbyte address limit.) The Intel design used simple latch circuits to hold the address information. The Mark-8 design used SN74193 programmable up/down counter chips instead. Thus, a user could load address information into a 6-bit high-address counter and into an 8-bit low-address counter. The use of the SN74193 counters preserved the larch function and also let a user automatically increment memory addresses while loading data and program code into the memory. If a user wanted to simply examine the contents of a block of memory, the counters would provide that function, too. I thought of that memory-loading and memory-examining approach and the use of the switches for data and address loading as my innovation and contribution to the design. Without a doubt, that part of the Mark-8 design let hobbyists bypass the need for EPROM chips because they could use the front-panel controls to store information directly in the SRAM chips. The PDP-8/L minicomputer provided the inspiration for the front-panel light-and-switch design. By the way, the name Mark-8 had nothing to do with any other computer. I simply came up with the name after a quick thought. I had approached Radio-Electronics and Popular Electronics magazines with the idea for a construction project and had to have a name for the computer. Only Larry Steckler, the chief editor at R-E magazine, responded and said he wanted to know more. So, I had to come up with a name quickly, and Mark-8 simply popped into mind. Radio-Electronics magazine published the Mark-8 Minicomputer project as the cover story in its July 1974 issue. Q. It took you x time to develop the Mark 8, how long do you think it would take you to develop a similar system using today’s technology? A. I remember studying the Intel 8008 manual on summer vacation in 1973 and having a working prototype that fall, so the design and building of the first version took about 3 months. I didn't have any help modifying the Intel SIM8-01 board circuits. I built the prototype on Veroboard boards that required me to cut traces of copper on the perforated boards to provide mounting positions for the 14-pin and 16-pin DIP ICs. Just making up the boards took a long time because I had to drill through long copper strips to isolate pads for each IC. As I recall, I left two contact points for each IC pin. Read More on Vero Board by Jon Titus I wired the prototype using point-to-point soldered wires on two pieces of Veroboard that each measured about 5 in. by 8 in. I had etched, assembled and tested a prototype memory board that provided 768 bytes of SRAM and connected that board to the two Veroboard-based circuits using wire connections. The three-board stack provided the prototype. Amazingly, the prototype worked properly as soon as I applied power. I could not believe it and stayed up until about 3:00 AM running simple programs. The prototype no longer exists, although the prototype memory board now belongs to the American Computer Museum in Bozeman, MT. I gave them the board during a visit in late 2002. Using today's well-documented microprocessor chips, development boards, free software tools, and overnight PCB-fabrication services, a student could put together a small, powerful computer in a few weeks. The builder could use free development tools that would include C compilers, assemblers for assembly-language programming and similar tools. I programmed the Mark-8 only in assembly language; writing code on pads and "assembling" a program by writing the appropriate 8-bit instruction on the pad. I got to the point where I had memorized many of the octal codes and could quickly assemble a program on the fly. People have asked me why I used octal in the Mark-8 and in subsequent computers based on the 8080 architecture. Most programmers seem to like hexadecimal. I had used octal code to program a PDP-8/L minicomputer and felt comfortable with 3-bit "digits." Also, the binary codes for 8008 operations broke down nicely into octal code groups. Here's an example: If you want to move the contents of one 8-bit register to another 8-bit register, the code breaks down like this... 11 DDD SSS Where the D bits designate the destination register's code and the S bits identify the source register's code. So, to transfer a byte from the B register to the A (accumulator) register, B-->A, you would use the code 11 000 001 or 301 in octal (hex C1) If you needed to reverse the process, A-->B, the code would be: 11 001 000 or 310 in octal (hex C8) Once you memorized the codes for the seven registers, you always knew an octal 300-type instruction involved a data-transfer operation and the register codes became obvious. The hex codes didn't readily provide that information. Most of the other 8008 and 8080 instructions broke down nicely into octal codes. Thus, my colleagues and I adopted the octal code. Q. You work with embedded systems currently, correct? So do you find it more difficult to develop an embedded system today then when you made the Mark 8? A. Easier today. If I had to design an embedded system now, I would buy PC/104 boards, PCIExpress boards, VMEbus boards, or other standard boards that met my hardware needs. Not much design needed at that level. Companies offer a plethora of development tools, drivers and technical support. So, compared to designing in the early '70's, today's designs look a lot simpler. True, the algorithms and class of peripherals are much more complex and they require more software effort, but companies sell and license signal-processing and machine-vision algorithms, for example, so in many cases, programmers and developers today face different hurdles than we did and I'm sure their jobs seem just as hard now as it seemed to us when we programmed 8008 chips in 1975. Q. When you set off to make the Mark 8, did you intend to sell it to the public or was it just a personal project turned global? A. I simply wanted my own computer and after I had a prototype running it occurred to me others might want to build a similar computer on their own. So, I decided to design printed-circuit boards and find someone to publish the do-it-yourself information. Luckily, Larry Steckler at Radio-Electronics magazine took the plunge and ran my article in July 1974. A company called Techniques, located in New Jersey, offered a set of the circuit boards for about $50. I did not expect to turn the Mark-8 into a commercial product and had no means to do so even it I had the desire to start a business. I just wanted to show other hobbyists and experimenters they could have their own computer. The Mark-8 minicomputer accomplished that mission. I made some money from the sale of the circuit boards and from the sale of a manual that Radio-Electronics sold for $5. The magazine article could not provide all the details, so Larry Steckler at Radio-Electronics magazine decided to offer everything in a booklet for serious experimenters. Some of the money I earned went to buy an IBM Selectric typewriter--the kind with the white self-correcting tape built in. That purchase helped me write more and got me more interested in writing and educating people about electronics and computers. Q. Do you have any advice for people interested in continuing education in computers? A. Jump in. Many companies offer simple development kits and software tools for under $100. The MSP430 from Texas Instruments, the 8051-class of computers from Ramtron and Silicon Laboratories and kits from Parallax provide excellent starting points. Many other companies offer similar hardware and software, but start small and simple. Just keep in mind that to do something well, you have to do it badly many times until you get some experience. These days, many books provide an introduction to microprocessors and microcontrollers. I favor microcontrollers as a starting point because these chips and modules provide all of the I/O and memory on one chip. That simplifies programming and development. Q. Any final words on the Mark-8 and its design? A. I regret not designing the circuit boards with plated-through holes and edge fingers that would plug into a motherboard. I wanted to save cost and cutting these corners met that objective, but it caused some hobbyists problems. Also, I could have put more circuits on each circuit board, so the design would use fewer boards. But, I designed and laid-out the boards by hand and wanted to get the Mark-8 boards into production as soon as possible. In retrospect, I should have abandoned the original Intel input-port design and used instead an open-collector or three-state bus for the input of data. That data could come from three sources, input ports, memory, or an interrupt vector. The front-panel design and board-to-board bus worked out nicely. The bus provided all the needed signals and the front-panel circuits provided a simple way to load memory and generate an interrupt. I looked at the Altair S-100 bus and thought, "What a mess." The same for the Altair front panel. It's a jumble of signals and circuits. I don't like to criticize people or designs, but although the Altair provided the more powerful 8080 CPU chip, the bus and front-panel were steps backwards in design techniques. Problems getting Altair computers to work vexed many experimenters. For the most part, the Mark-8 worked well right away. I give a lot of credit to Intel for the original SIM8-01 board design that I adopted. I consider myself "rich" because of the Mark-8 project. Not because I made a lot of money but because I made a lot of friends, helped people get a start with computers, and had a great deal of fun. Readers should keep in mind that other hobbyists and companies had computer developments underway. In the mid '60's, Stephen B. Gray founded the Amateur Computer Society and published a newsletter. Many members wanted to or tried to "clone" the Digital Equipment Corporation PDP-8 architecture. But at that time, getting reliable memory presented a major problem, so most efforts stalled. Steve and other members of the amateur-computer-enthusiast deserve a lot of credit for sparking interest in small computers for personal use. |