A look back at the development of computers as a real and understandable things, and a mourning of the days which are gone.
I find it difficult throwing things away. Even when I honestly answer myself that the chances of every using it are zero, and that I need the space - it can be so difficult to allow myself to part with a something familiar. It is all the more difficult because of those few times I have managed to throw something out and later regretted it. These departed objects haunt one with a nostalgia which only shines more brightly as time passes, unlike places we miss which can be revisited and brought back into reality.
What can one do to get out of such a predicament? We try to get over the final departure of a person with a wake, a celebration. Maybe this will work for some of the inanimate bits and pieces, trivial though they are. Maybe indulging in a short celebration and remembering will settled the score on the departed objects. It's worth a try. This is that celebration. It is a self-indulgent reflection.
For me, things I wish I had kept were the bits of computer. These are things which have been such fun in the building, and I miss the ones I have had to throw out over time. As I looked back at the things I had built over the years, growing up and gown up, I found I wanted to go into the whole thing, the whole process of discovering electronic and creation of things which compute. I found myself feeling unbelievably lucky that I had been around at the right time to ride the wave of the new things which ordinary people could do, from transistors to the Internet.
Mum and Dad were involved in the Manchester/Ferranti Mark I computer, and successive machines, Dad as a statistician analysing how things worked, and helping customers figure out how to use computers, and Mum as a programmer. The machines early on were programmed with 5 hole wide paper tape - like ticker tape, which is great to play with. You can push the middle out of the roll, you can unroll the outside, you can heap up masses of it as magic "meringue" for the family Chistmas pantomime.
Once I went in to Dad's office to see one, a Pegasus or Sirrius, I think. In any event it had a tall grey steel cabinet centered behind a steel desk. Built into onto the desk were the input/ouput devices: on the left, a paper tape reader, and on the right, a paper tape punch. Above the desk, stuck on the main cabinet, was a clock. That was my model of a computer, and I went home and copied it. I used my toy cupboard for the vertical cabinet, and my small table, with one cardbord box on each side of the table as paper tape reader and punch.
My brother Pete made a famous statement of one of is early creations, "It dosn't really give the right answer, it just pretends to". This was what the cardboard computer did, I suppose. Making things which really do something - however simple - is yet more exciting.
Our family had a mathematical bent. We also had fun with drama and outdoor adventure, though for me sports was not a strong point at elementary school. Rather than play soccer or rough games of tag I would walk around the edge of the playground with my best friend, Nick Barton. Nick was a very smart kid, read huge amounts at a very early age, and we would talk about dinosaurs and electrictity and such things. Nick, as a kid with the chickenpox, had informed his doctor of the details of the Bubonic plaugue in London in 1667 which he had leaned from his Junior World Encyclopaedia. He is now a professor of Genetics at Edinburgh University. Nick and I would pause by the tree stump with the stag beetle or under the big tree in the corner and quietly bring out of our pockets the pieces to wind an electromagnet. At home, we built circuits with batteries, bells, and lights, and made burglar alarms. We each had a secret panel in our room with burglar alarm controls, and Nick had a twelve core cable running out to his garage.
As we grew, electricity turned to electronics. What I mean by being lucky was that electricity was there when we were old enough to understand off and on, and later Volts and Amps, and as we were in a position to understand ratios, transistors magically appeared on the scene. Individual transistors were expensive but down the Tottenham Court Road the shops which seemed to have converted from war surplus were selling discarded ("untested") transistors for a few pence a packet of 50.
If you weren't into that sort of thing at that stage in your life, let me explain the difference. Things like switches either conduct electricity or not. Something like a lamp, as you put a greater voltage (electrical pressure) on them, respond by letting though more and more current. Resistors are passive - they just let through current in proportion to the voltage. Transistors are neat because the amount of current they take between two wires depends on the amount of control current flowing though a third wire. This means you can take a weak signal, feed it into the control wire and a strong signal will appear as a current flowing though the other two. There are whole lot of things you can't do with just resistors - you need transistors
You can put two transistors back to back so that when one is on the other is off. This circuit is called a "flip-flop". You can make it stable, so it remembers which state it is in. It is a memory cell, one bit of memory, a toy train running or stationary.
You can make it stable only one side, so that after flipping it will, after a while, flop back of its own accord. This is the timed delay of a burglar alarm, or a "sleep" button on an alarm clock.
You can make it unstable in either state, so that it will just flip back and forth flashing a light, or - running faster and fed into a loudspeaker - an audible note. Transistors allow you to make active circuits which do things. When I was 10, this seemed exciting to me. Just think, if we had been born earlier we would have had to use valves (vacuum tubes) which worked at much high voltages and would have been complicated, dangerous and expensive and we wouldn't have done it. If we had been born later, this would not have been new.
Emanuel school is a secondary school built in the triangle between the Brighton and Bournemouth lines where they diverge as they leave London. From 11 years, when I started at Emanuel, it was inevitable that I would get interested in trains, along with a whole load of other kids. Looking into the railway cuttings on either side of the school grounds we saw a mixture of local and express, and in 1967 we saw and mourned the last of the steam engines on British rail. We checked their numbers off in pocket books which comprehensively listed every locomotive ever built in the UK.
When my parents had the attic converted to living space to give each of the four kids a room, Pete and I got the new rooms. The design was open and the constraints on it were, firstly, that to be fair the area of the rooms should be exactly the same, and secondly that my room should be as long and thin as possible for a model train layout. The result lost a lot of headroom in the corners to the roof, but under the window in the center I built a four-track station in OO gauge. On each side, the tracks split in pairs to go into separate tunnels, looping back inside the hill. I remember the day I learned that plaster of paris, while great for soaking old sheets to make hills, should not be disposed of down the sink.
The trains, of course, needed controlling. By now I was old enough to build the power supply, and Nick and I attached to it all manner of flip-flops and power transistors to drive the trains. There were magnets on the bottom of the coaches, and magnetic "reed" switches in between the rails to trigger a pause at a station or a change of the points. (The points at the station could put a twist in the two loops in a Mobius way so that a single train would traverse each piece of track before it got back to where it started.).
Breadboards were our way of life then. Making circuits is time-consuming if you have to solder the wires. A breadboard such as the S-Dec and T-Dec was plastic box with a matrix of small holes which would take the lead of a transistor. The holes were connected electrically in one direction in groups of 5 or so. When we got a box of untested transistors, we wouldn't know how good they were - how little current though the controlling base wire would turn them on. We would set up our own test shop, plugging in different components to see what a transistor would so. We would sort them into boxes depending on their gain (hfe) value. We would prototype a circuit on a breadboard and then transfer it to a circuit board with solder.
Rather than make our own circuit boards by etching copper-plated laminate, we used hobbyists' Veroboard, which came predrilled in a matrix of holes every tenth of an inch, with copper strips in one direction. You could break the strips in appropriate places, and add components and a few extra wires to make the circuit. I think that the pinnacle of the train-oriented electronics was when we discovered the noise diode, a device which would generate a random hissing noise. We turned this into a steam train simulator complete with steam whistle.
I remember running out of the back door to go around to Nick's with pile of bits including a soldering iron. My mother stopped my in mid flight, demanding to know what sort of a mend there was in the iron's power cable. "Let me show you how to do that", she said. Expertly, she stripped off the insulating tape, which was sticky on both side in those days, rebinding it and flouring on the outside to take the stickiness off. I was impressed: I hadn't seen her in this mode. She explained about her work in a radar lab in Malvern during the war, and I left realizing that Nick and I didn't know everything yet. I am glad we survived that time of playing with electricity (and occasionally, under controlled circumstances, fire). Unfortunately our good friend Christopher was not so lucky and perished in trying to use model aircraft fuel as a new cure for insomnia when he was in his teens.
As time went on, we went though high school, Nick through the local one and I through Emanuel. Transistors gave way to the first logic chips.
The famous silicon chip is what a transistor is made out of, inside its can, but chips made news when they started to contain more than one transistor. You could then buy a little black block of plastic less than an inch long with two rows of centipede-like legs. [picture]. Transistor-transistor logic, or "TTL" came in at the stage when building more complex things out of individual transistors was starting to be tedious. One TTL chip could contain four flip-flops, and other interesting circuits such as shift registers and multiplexers and half-adders.
I knew about half-adders because I had seen my father build them out water or ping-pong balls. He used to write speeches for Ferranti's managers to give, explaining the new technology. He used pennies, halfpennies and defunct but still remembered farthings (quarter pennies) to explain how you could represent numbers in base 2. Each digit, 0 or 1, could be represented by a circuit high or low, on or off. To make it more graphic, you could use water flowing. When you add two numbers in decimal, you add "five and eight is 13, that's three carry one into the tens". In binary it is simpler. 0 and 0 is 0. 0 and 1 is 1. 1 and 1 is 0 with a carry into the twos. You can make that out of water jets crossing over. If either jet runs, it goes into a funnel as a "1". If they both run, they collide and make a mess (which you collect with a funnel in the middle as the carry digit) and and the output digit is zero because nothing goes into the original funnels. This is called a half-adder because in fact you need two, one two add a digit of one number with a digit of the other, and then a second to add in the carry from the previous column. Dad had fun getting wet and I think there was a Neptune computer demonstrated once. [where?]
[dig. of water half-adder]
But now, for Nick and me, we could get a half-adder in a TTL chip. Nick bought an old oscilloscope "as seen" in the Tottenhem Court Road. It worked (more or less) as a way for measuring signals as it had designed, but he converted it into a two-digit decimal display with TTL logic. He found he needed more wires between his bedroom and garage than he had, and so made TTL multiplexers to consecutively switch several signals over the same wire. I had previously wired up regular old telephones between the attic rooms and the kitchen. They worked very well, but they weren't smart enough, so I started on a version with digital dialing using TTL. These things were fun and worked more or less at one time or other, though my mother wished at times we had the old phones back. She pointed out resignedly that with a house full of gadgets the door bell still didn't work.
Dad had fun too. When he had been at high school, he hadn't been able to use transistors, but he had made an integrating machine using resistance wire and - what was it - a record player? He tried at home to make a self-steering lawnmower which cut the grass using a hot wire. I came upstairs to ask him whether he was ready for a cup of tea, to find him at a table on the landing surrounded by a cloud of smoke.
He had always wanted to make device (then called a mouse as there was no contender for the title which came with a keyboard) which would use a photocell (light-sensitive resistor) to follow a line painted on the floor. We'd seen one demonstrated at the Royal Institution and the implications for cleaning the house and mowing the garden were obvious. Nick and I took on the challenge and managed more or less once, though the device would waver across the line with ever increasing inaccuracy until it actually turned right around.
And so we had fun. There was no computing at school, but by the time Nick and I went to university respectively, we pretty much knew our way around digital electronics, being limited by our pockets and the time we had to solder wires.
The times, though never stand still.
Nick went to Cambridge to read natural sciences, and I to Oxford to read physics. Why physics? At school mathematics was my best subject, but at home I was fascinated by electronics. They were both fun. So, I took physics as sort of compromise between engineering and mathematics. It isn't a compromise, of course: its is something special in itself. And maybe that was best. The irony is to have done theoretical physics, it may have been better to have started with maths as a first degree, and to do practical physics as people did at CERN would involved huge amounts of electronics and computing. But at Oxford, a physics course was a Physics course, and people in the States view with justified horror the fact that it is undiluted with liberal arts or even other science. So I kept on the physics track, while Nick, at the more flexible Natural Sciences course at Cambridge, after an early chat with one Prof. Northcote started taking more cell biology, starting a trend toward his eventual field of genetics.
If you are going to build a computer, you need a device with which to communicate with it. Just as Nick has turned his oscilloscope into a two-digit decimal display, I set about making a computer display out of a television. A zany friend Pete Gilliard-Beer at Catz college was into anything electronic, and explained to me how all black and white TV sets were basically made alike. By then, they had evolved over time, so that all the shortcuts had been made to make them with the fewest components. For example, the same coils which generated the line scan also worked as a transformer to generate the extra high voltage for the electron gun. He showed me how to find a particular point in the circuit which would carry the video signal at a 1 volt level. The circuit before that was basically the radio receiver, and after that, just a video monitor.
Come the summer, I was back in London. Just on the boundary between the suburbs of East Sheen and Barnes in London is a an area of little row houses where Princes Road connects Queens Road and Kings Road. On the corner of King's and Prince's was then a small store, AJM Radio, which claimed on its business card offices in Queens Road, East Sheen, and retail outlet in Princes Road, Barnes. Its proprietor fixed electrical goods, and was closer than the Tottenham road when I needed something in a hurry. I asked him if had such a thing as a TV with a broken radio part but a working monitor. He had a number. I chose one, for which he charged me five pounds. I was started.
I took it home, and started fiddling to see what I could do. Sure enough, there was a 1 volt video point. I ran a video cable out of it and put the back on quickly. I didn't want to spend any more time than I needed to around the high voltage. I made an unstable circuit - a flip-flop which continuously flipped and flopped -, and fed the signal into the TV.
To paint a TV picture, spot on a television tube sweeps across the top line and then the next line and so on, to the bottom, then starts again out of sight at the top. The signal was on for a while, and the spot painted white lines, and the it would switch off and the spot would paint black. The result was horizonal blank and white stripes. Amazing. Wonderful. I could drive the TV with a TTL signal.
The rest of my family were interested to know how I was getting on. World had spread that I needed a TV and I was given a second one. I did the same radioectomy to that, and left it plugged in downstairs. I ran a cable to to my room so that the family could see the latest pattern I had succeeded in making. The challenge with the video signal I would need was that it would run a lot faster. The full frames in the UK come 25 times a second, so a signal which changes a few hundred times a second will make horizontal horizontal lines. The 625 scan lines come at a rate of 625x25=15,625 times a second. The dots for the pixels which make up the letters on a computer terminal, come much faster. To have 6 dots per character, 64 characters visible and some extra space on either side making a total of 6x88=528 dots on a line, you have to send the signal for a dot every 15625x528=8,250,000 times a second. 8.25MHz. That was frightening - I had never made a circuit which went anything like that fast.
I played with different ways of making an oscillator. As they got faster, the horizontal lines became thinner, and then turned to vertical lines. Eventually the vertical lines became thin enough. I then found I could buy a crystal which would vibrate at exactly the right frequency. Crystals are the very accurate components on which electronic watches are based. That then drove circuitry to count which dot number and character number the spot was on, which drove the memory which looked up what character should be displayed there, and then given the character code (a number between 0 and 255) and the scan line number within the character, what dots should be shown on the screen. Bit by bit I put the pieces together, with the instant gratification of new patterns on the display at each step.
I was using new chips which had only become recently practical for a hobbyist. The Random Access Memory (RAM) which stored the character data for each row and column needed for 16 lines of 64=1024 characters and seven bits for each chacter, and I could buy 1O24-bit RAM chips. To store the shape of the characters - which dots should be black and which white on each scan line of a character, I needed a Read Only memory (ROM), and such a character generator ROM I could now buy for around 13 pounds. This was the most expensive part of the VDU to date. In those days such devices carried dire warning about death (of the device) from static electricity, so, having none of the propoer tools, I just left an electric kettle turned on under my table to keep everything moist while I handled it.
One day in my final year at Oxford I took the terminal, basically completed, into the Clarendon undergraduate physics lab. The lab technician looked askance at it and made a few scathing comments. However he allowed me to borrow a TV, and even to connected it to the lab's small PDP-8 computer in place of the usual teletype. This was on the strict condition that I had optical isolators in the current loop interface (precursor of RS232 serial line) between my makeshift contraption and his precious machine. He had to withdraw his scepticism when it displayed output from the PDP-8 and actually worked as a terminal.
It was nearing the end of my time at Oxford. I had a computer terminal which had actually been shown to work on a real computer.
Two other relevant things happened during my time there. One was that in studying quantum mechanics and a little solid state physics, I felt that I could piece together some understanding of how a transistor worked. At undergraduate (or any) level there is a certain amount of hand waving, but I could at least see the shape a full understanding would have if one were to have it. This tied nicely on the front of a high-school-learned understanding of how logic circuits could be made from transistors, and an understanding of how the elements of a computer, like parts of the VDU, could be made from logic circuits.
The other thing which happened was that Julian Hammersley [sp?] and I were
thrown off the Nuclear Physics PDP-10 machine. We had accounts restricted to
using the BASIC language and we were forbidden from using the big
line-printer. However, it was common knowledge that the BASIC system had a bug
such that if one joined a four-letter and a one-letter string of characters,
(10 PRINT "ABCD" + "E") it would crash. One could thereafter do
more or less anything. We simply used the line-printer to print out mailings
for a student Rag Week [exp.] committee. We checked the machine room was empty
at lunch, set the job running, and came back to find pandemonium -- a crisis
in progress, our printout holding up the printer, and a very unhappy system
manager. Even if I had not been heading in that direction, being struck off
the PDP-10 user list would have convinced me I needed my own machine.
It is important to understand what a significant step up a real computer is. A transistor (unlike a passive component) is special as it allows you to build any sort of logic circuit, if you just have enough of them and connect them up right. Similarly, once you have a computer, you can program it to do virtually anything so long as it has enough memory and time and your program it right. Alan Turing proved that any computer worth its salt could be programmed to pretend to be any other one, so in a sense they were all equivalent. To have a computer is to have solved the hardware problem for good, and be left with only software -- that is, your imagination -- as the limit.
Now I could have gone about building the computer in the same way as I had built the VDU. I knew now how to connect the registers and the memory and it would have been fun to design the control unit which interprets the instructions into signals to operate the appropriate bits in the right order. However, it would have been boring. It would have been months of hard work building circuit boards out of TTL and wiring them together. But this is what I mean by being lucky: just as I got to that stage, the microprocessor was coming out. Whereas the 4-bit processors would have been really painful to use, the new 8 bit ones were really practical. In one chip, this provided all that circuitry, and made the task of doing by hand pointless.
I bought a Motorola M6800 evaluation kit. It had a much cleaner architecture and instruction set than the 8008 series from Intel. I bought a rack for small index-card sized circuit boards, and hand-wired a backplane bus to connect the slots. I put the processor (CPU) on one board, and the interface to the VDU on another. Another had memory - initially 256 bytes (later 4k) of RAM and 256 bytes (?) of ROM. I don't remember the first day it worked, but by that time I was working for Plessey Telecommunications on the south coast of England in my first real job.
I should emphasize that this did not look like the computer you use today. The card rack was 4.75" high and 19" wide but was mounted in 19" rack unit looking like a small fridge. When it came on it displayed a prompt character on the first column of the display - the television. Then you could enter commands to set or read out the contents of the memory as hexadecimal (0-9, A-F) numbers. You could make it execute a program starting at a given memory address. If the power was turned off, it would forget everything. Living in shared rented house with a coin-operated electricity meter made this extra dangerous. I remember losing a hand-keyed program to add up two numbers of any length, because someone took a bath without stacking the meter up beforehand.
This was of course rather inconvenient. The PDP-8 at Oxford, and the Interdata minicomputers at Plessey both used a fan of concertina-folded paper tape (now 1", 8 hole) to load the initial program. Fan-fold tape [diagram] is better than reeled tape as the stack of tape is immediately ready to reload without rewinding. I found an old paper tape reader head from an abandoned project, and built the 19" rack mount unit and the fan-fold arrangement out of wood, stained mahogany. Why does one throw away things like this? But I did, many years later, and the computer with all its hand-soldered boards.
With the years, the M6800 showed its age, and both at Plessey and later Image Computer Systems, the compters I was using at work were 8080/Z80 family. I met a computer-crazy school teacher, Mike Blandford, who had built himself a computer on a single large 7"x7"circuit board. It had a Z80, 32kB of RAM and a terminal interface. I designed a matching board to hold an extra 32kB (the Z80's limit was 64kB) , along with interfaces to a printer and the floppy disks which now existed. I actually laid out the design in black tape on plastic film just like the draftsmen I had watched at Plessey. The combination of the two boards was a fully-stocked Z80 computer which would run the operating system of the time, CP/M, or our own re-engineered version of the Mostek development system, hacked into a mutitasking system.
Mike and I went one level more professional than before - we took the design of the circuit cards to a local company and had a few professionally produced. I had made mine to have different edge connectors on the two sides of the board, one compatible with Mike's card and one with the commercial Nascom card. We even put in a very small add for the disk card and may have sold a handful. In California, Steve Wozniak around the same time was designing what his partner Steve Jobs called the Apple-II, and selling it in the local computer clubs. There was the same excitement, very similar design decisions, although "Woz" and Steve Jobs managed to make the product fly commercially.
I met the Woz at an event at the American Computer Museum in Boltzman Montana. We gave short talks to local students, and one middle school girl asked, "Can you just explain how exactly the computer sends itself a message telling it what to do next?" That was a very good question. We did not alas have time to attempt an answer. I did try to assure her that in fact it could be explained, and it could be understood, and it wasn't that hard. But it made me wonder what it would be like to be born surrounded by computers of huge speed and memory size. Would it be possible to feel how special was a transistor, a flip-flop, a computer, if one had not seen them move from the unknown, through the exotic unattainable, to the creatable? We were indeed a lucky generation to have surfed this wave as it came in, rather than just guess what it was like looking at the water all around us.
Why was it so exciting? Why is exciting to a skier to top a rise and see below a steep field of unbroken powder snow? Because the brain is built to become fundamentally happy when solving the real time dynamics problems which are skiing. Getting it right must generate a happiness drug. The same is true for the challenge of building systems. It is innately pleasurable to build something powerful out of smaller parts, to use what has gone before as a lever to move greater things. This (not cracking into computers) is called hacking. The ski terrain used to be hardware, before computers. Now, the computer has reduced a mass of hardware problems to building a computer (done that) and writing software. Maybe one day we will solve software problems by creating a form of intelligence which solves it for us, but in the mean time, when it comes to the joy of hacking, software is where its at.
Tim Berners-LeeDRAFT
Cambridge Massachussetts 2000
M6800 pin-out and instruction set
Jonathan Bowen, A Set of Microprocessor Programming Cards, Microprocessors and Microsystems, 9(6), pp 274-290, July/August 1985.
veroboard example
$Id: Hacking.html,v 1.27 2017/04/04 16:00:47 plehegar Exp $