Fullscreen HTML5Med Res (??MB)HTML4

It’s the modern-day equivalent of one of the world’s first “computer games” OSCAR: 2007 STYLE B y B R I A N H E A LY Noughts and crosses may rate quite poorly amongst young gamers of today but in the late 1960s, a machine, more often called an electronic brain than a computer, playing noughts and crosses against a human opponent, was quite a sensation. T HE FIRST COMPUTERS WERE built during World War II to attempt to decode German coded signals. From this early work sprung “EDSAC” (for Electronic Delay Storage Automatic Calculator), the first truly programmable computer. It was built at Cambridge University (UK) in 1949. This computer, shown in the background above, was used by mathematicians for research and learning. It 26  Silicon Chip contained 3000 valves and consumed some 12kW of power! History was made in 1952 by A.S. Douglas, a young PhD student, when he used it for another purpose: he programmed it to play noughts and crosses. The computer used a cathode ray tube to display its output, which means this was the very first video game in the world. In the mid-1960s, both Sydney and Melbourne technical museums at- tracted large crowds with a “computer’ which played noughts and crosses against a human opponent. In 1968, when the author was aged 24, he and a friend built a machine using 70-odd telephone relays and a uniselector to play the game. A uniselector, by the way, is a rotary, solenoid driven, 50-position switch. They were commonly used in automatic telephone exchanges at the time and were even found in some siliconchip.com.au This is a Uniselector, a 50-way solenoid-driven switch. Every automatic telephone exchange used these in their thousands – now they are virtually a museum piece. The Uniselector was used as the basis for the original “Oscar” built by the author back in the 1960s. older exchanges until quite recently, before first solid-state devices and then microcontrollers took over. In a busy telephone exchange, the noise of the uniselectors switching back and forth following the numbers dialled on a phone perhaps 10 kilometres or more away was quite deafening! Our machine was as large as a refrigerator and about twice as heavy! We called him “OSCAR” and he worked very well. When you pressed a button for your turn, the machine started whirring loudly (the sound coming from the uniselector), a row of lights flashed, relays clicked in and out and finally it all stopped as it brought up its reply. It was very impressive. We hired it out to retailer David Jones and a new shopping centre called “Westfield” in Wollongong (a large city south of Sydney) where it attracted large crowds. It must be very difficult for people who did not come through that era to understand such a reaction, so I will try to explain. It was akin to the crowds who gathered on footpaths outside electrical retailers a decade or so earlier to watch that new-fangled invention, television, playing in the windows. In 1968, very few people had ever seen a computer “in the flesh”. They may have seen one in a film where typically it would be in an almost sacred situation, a series of large metal cabinets, some with large tape spools rotating and all attended by well groomed, bespectacled technicians, wearing white lab coats, hovering over it like nurses over a new-born baby. What the computer actually did was a complete mystery and there was no way you were ever going to be able to get close enough to touch one. So, if suddenly you were now able to have contact with a computer, to challenge it at a game that you unD1 1N4004 K 4.7k 4.7k 14 Vdd 16 1 RA2 RA1 18 A1 OSC1 100pF A1 MCLR IC1 PIC16F84A RESET S1 K RB5 11 RB4 10 RB3 9 A1 LED1  3 OSCAR'S FIRST TURN S2 S3 Vss RA3 A1 A2 A1 A2     S5 K LED7 S7 S6  K  S4 A2 LED8   S8 S9 K S10 S11 2 5 2007 A1 A2 A2 LED4  K K K 8 RB2 RB1 7 6 RB0 RA4 A1 LED5   A2 LED9   K A2 LED2  A1 A2 LED6   K 17 RA0 RB7 13 RB6 12 4 A1 A2 LED3  SC 'OSCAR'  +6V 10 F 100nF 4.7k 10k A – 2007 version (RED) A1 K A2 (GRN) 0V LEDS1–9 1N4004 A K Fig.1: not a Uniselector in sight (or even hidden!). The PIC chip does all the work of the mechanical monster of 40+ years ago – and this OSCAR is much easier to build. siliconchip.com.au October 2007  27 new game and the red button is to allow OSCAR to have the first move in the game. It is polite and fair to give OSCAR the first turn at least sometimes. OSCAR is very clever here, as the square he chooses for his first move is truly random. Generating truly random numbers is difficult for a PC and very difficult for a PIC but OSCAR employs a trick here. At the end of a game, following a reset, OSCAR is not just sitting there doing nothing – not at all. He is repeatedly counting rapidly from one through to nine at high speed, until you press a key to start the game, at which time he stops counting. The number that he stops on will be the square he chooses if you give him first turn. How it works This clipping, believed to be from the “Wollongong Mercury” around 1968, shows the original fridge-sized OSCAR noughts and crosses machine. No, those aren’t LEDs in the display – they weren’t even a glint in a mad scientist’s eye back then . . . derstood well, this was very exciting. Lots of my older friends still remind me about OSCAR. We initially did not know even how to start to build such a machine. We eventually worked out that the machine had to go through a logical series of steps, in sequence. We never called it a program but of course, it was a program. I have never forgotten the sequence, so now that PICs have become available, I set out to put the exact same program from the 1968 OSCAR into a 2007 PIC16F84A. The new OSCAR The new OSCAR is a tiny fraction of the size of the original. And instead of requiring a lot of power to operate, it will run for months on a couple of batteries. It has nine green buttons in a 3 x 3 array and nine bicolour LEDs, also in a 3 x 3 array. 28  Silicon Chip When you press a button, to take your first turn, the corresponding LED illuminates green. OSCAR now “thinks” for a second or two and then has his move, illuminating his position red. It is then your chance to have your second move – and I think you know the rest. If OSCAR wins, which is pretty common, the winning row of three red LEDs flashes, calling attention to the player’s demise(!) and the game is halted. You cannot continue when you have been defeated. The first version of the software was wriiten so that OSCAR never lost a game. However, that quickly becomes boring and so the software was later reworked to give its opponent a chance to win approximately one in 10 games. If you do win, your three green LEDs will flash to indicate success. There are two more buttons. The white button is to reset OSCAR for a If you glance at the circuit you will see that it is quite simple and does not use any active components apart from the PIC. The circuit is powered from four AA batteries that sit in a plastic battery box available from Jaycar. The maximum voltage for the PIC is 6V, so don’t be tempted to install a 9V battery. This PIC can have various oscillators but in this case, we are using a resistor/capacitor circuit on pin 16 of the chip. With the values chosen, the circuit oscillates at around 700kHz. The RC oscillator is a little cheaper and somewhat slower than a crystal, making delay loops easier. You can see the 700kHz triangular waveform on pin 16 with a ’scope and high impedance probe. The PIC divides this by four to become the system clock and you can see the resulting 180kHz square-wave on pin 15. Because we are connecting so many devices to the PIC, we need to do some multiplexing. The PIC has only 13 input/output connections available, but we have nine position buttons, nine red LEDs, nine green LEDs and a couple more buttons. If you have a look at the circuit you will see that for each of the nine locations, the common cathode of the green/red LED and one side of the pushbutton are all connected together. So we have a common connection for the button, red LED and green LED, and of course there are nine separate common points. The PIC holds these lines high at 5V and then, one at a time, drops the line siliconchip.com.au + 4.7k OSCAR'S FIRST TURN RESET 10k 10 F 1 IC1 PIC16F84A S2 4.7k 4.7k 100nF 0V 100pF D1 +6V S1 17001180 LED7 LED4 LED1 LED8 LED5 LED2 LED9 LED6 LED3 1CIP RACSO S9 S6 S3 S10 S7 S4 S11 S8 S5 Fig.2: follow this parts layout diagram and the accompanying photo to assemble the unit. In the prototype, ordinary copper wire was used for the links but we suggest tinned copper wire to prevent oxidation. to 0V for around one millisecond, then puts it back up to 5V and drops the next one to zero for one millisecond and so on. So the PIC is scanning from one to nine, relentlessly, regardless of the state of the game. Let’s now look at pins 1 and 18 of the PIC. These pins are configured as outputs and are normally held low by the PIC. Pin 1 connects to all the green LED anodes and pin 18 connects to all the red LED anodes. If, during the scanning, the PIC needs to illuminate, let’s say, green LED number six, it waits until the scanning reaches position number six and then, just for one millisecond, while the cathode is held low, it raises the anode via pin 1 of the PIC to 5V. Only that LED will light because it is siliconchip.com.au the only one with power on one end and 0V on the other end. In this way, the PIC lights the LEDs one at a time at high speed, so you are unaware that they are actually flashing rapidly. It will never light both green and red for the one location, as that situation never occurs. The common sides of the pushbuttons are all connected to pin 2 of the PIC. This pin is configured as an input and is normally held high by a 4.7kW resistor. However, if you press a button, this pin will be pulled low when the scanning reaches that position. During an actual game, as the PIC is scanning each position, then for the one millisecond when the common point is pulled low, pin 1 will be switched high if the LED should be green. Similarly, if the LED should be red, pin 18 will be switched high. If the position is not occupied (no red, no green), then (and only then) the PIC looks at pin 2 to see if a button has been pressed. This means that if the player presses the button for a position already occupied, it is ignored. Software Let’s ignore the housekeeping software and just look at the game logic itself. When you press a button, the green LED is illuminated immediately, and then there is a deliberate delay of 1-2 seconds so that OSCAR appears to be “thinking”. Then the PIC very rapidly goes through four separate procedures, looking for a response. As October 2007  29 These two shots give a good idea of how OSCAR fits together. The PC board screws to the box lid via standoffs with the switches and LEDs poking through, while the battery is fixed to the bottom of the box. soon as a response is indicated, the PIC executes it and quits any further procedures until the next move. The first procedure, called “Win for Sure” is to test every position to see if in any row of three LEDs, there are two red LEDs lit and the third position blank. If it finds one, it of course puts a red LED there, stops the game, declares a win, etc. The second procedure is called “Prevent Win” and is similar to the first. Its job is to test every row to see if there are two green LEDs in a row and a third position blank. If it finds one it puts a red LED there to prevent defeat. The third procedure is the most difficult. It is called “Tactics”. It goes through quite a few algorithms and tries to do something intelligent. The fourth procedure, if the first three produce nothing, is simple: just find an empty position and go there. There is more software for responding when the player lets OSCAR have first turn and also to highlight the winning row of three red LEDs by making them flash. Assembly The whole circuit is built on one PC board which mounts inside the lid of a jiffy box. The most difficult part of the construction process is the precise drilling of the lid. Photocopy or cut out the front panel art and use it as a template. Tape it to the jiffy box lid and drill a small pilot hole for each marked spot. That done, increase the size of the drill, being careful to keep the drill perpendicular to the lid at all times. Check that the LEDs will fit into the holes easily and that the buttons have about 1mm clearance all around. If the switches get caught on the hole edges and jam on, the project won’t work! Assemble all the components onto the board except the LEDs. This is important – leave the LEDs until later. It’s best to use a socket for the PIC in case you need to remove it. The end of Parts List – OSCAR Noughts & Crosses Game 1 1 1 4 9 1 1 1 4 8 OSCAR PIC1 PC board, code 08110071, 145 x 86mm 158 x 95 x 53 mm (UB1) jiffy box (Jaycar HB6011) 4 x AA battery holder with switch (Jaycar PH-9282 AA batteries green PC-mount pushbuttons (Jaycar SP- 0724) red PC-mount pushbutton (Jaycar SP- 0720) white PC-mount pushbutton (Jaycar SP- 0723) 18-pin IC socket M3 x 10mm tapped metal stand-offs (Jaycar HP-0900) M3 x 6mm screws Semiconductors 1 PIC16F84A preprogrammed with OSCARv2.hex (IC1) 9 red/green 3-terminal LEDs (WES Components LED5GRY) 1 1N4004 silicon diode (D1) Capacitors 1 10mF 16V electrolytic capacitor. 1 100nF miniature polyester capacitor (code 104, 0.1 or 100n) 1 100pF ceramic capacitor (code 101 or 100p) Resistors (0.25W, 5%) 1 10kW (code: brown black orange gold) 3 4.7kW (code: yellow purple red gold) 30  Silicon Chip siliconchip.com.au 0v OSCAR’S FIRST TURN +6v R A C O OSCAR PIC1 S 5677s +6v RESET Figs.3 & 4: full-size artwork for the front panel (which can also be used as a drilling template) and the PC board, viewed from the copper side. Note the four cutouts required in the corners of the PC board so it can clear the pillars in the box. the socket with the notch in it is near the edge of the board. Leave the PIC itself out for the moment. Fit all 18 wire links on the PC board (we suggest using resistor and capacitor lead cut-offs as these are invariably tinned copper wire) and the four resistors. The small electrolytic capacitor is polarised. Install the four 10mm metal standoffs to the PC board. Test fit the board to see how well you have drilled the holes for the buttons; file or ream the holes a little if necessary. When you are happy with the fit of the buttons in the holes, fit all nine LEDs into the holes in the board, taking great with the polarity (flat side is on the anode 1 [red] leg) but don’t solder them just yet. Now mount the board on the lid using the standoffs and fit the screws siliconchip.com.au to both ends, so that the board is in its correct position. That done, push the LEDs one at a time hard into their corresponding holes in the lid. Make sure each LED is fully pressed into its hole, then solder its leads. Repeat until all the LEDs have been soldered in. If you have done this well, all the LEDs will be protruding through the lid by the same amount (around 3mm). The battery box can be attached to the bottom of the jiffy box with double sided tape. Finally, fit the PIC into its socket, install the batteries, switch on and give it a go. Faultfinding If you have any trouble with any of the functions you can check out the board as follows. First power off and remove the PIC, then make a short jumper out of a single strand of telephone wire. Put one end in pin 14 of the socket and the other end in pin 18. Put one end of a 220W resistor in pin 5 and the other into pin 6. The first LED should light red. Swap the resistor end from pin 6 to pin 7 and the second LED should light red. Keep going into pins 8, 9, 10, 11, 12, 13 & 17 and each LED should light red in turn. To check the green LEDs, change the jumper linking pins 14 & 8 to pins 14 & 1. Repeat as above with the 220W resistor and again, each LED should light green in turn. If you have a LED in backwards it will light green when it should have lit red and vice versa. If these checks are OK, then it is highly likely that you have a faulty PIC or a PIC that has not been programmed SC or is programmed incorrectly. October 2007  31