Make: Electronics

by Charles Platt

  Here’s a question for you: how many LEDs would you need to represent the decimal number 1024 in binary? And how many for 1023?

  Obviously binary code is ideally suited to a machine full of logic components that either have a high or a low state. So it is that all digital computers use binary arithmetic (which they convert to decimal, just to please us).

  Getting back to our project: I want to take the three binary outputs and make them create patterns like the spots on a die. How can I do this? Quite easily, as it turns out.

  I’m assuming that I’ll use seven LEDs to simulate the patterns of spots on a die. These patterns can be broken down into groups, which I have assigned to the three outputs from the counter in Figure 4-108. The first output (farthest to the right) can drive an LED representing the dot at the center of the die face. The second (middle) output can drive two more diagonal LEDs. The third output must switch on all four corner LEDs.

  This will work for patterns 1 through 5, but won’t display the die pattern for a 6. Suppose I tap into all three outputs from the counter with a three-input NOR gate. It has an output that goes high only when all three of its inputs are low, so it will only give a high output when the counter is beginning with all-low outputs. I can take advantage of this to make a 6 pattern.

  Note that it’s bad practice to mix the LS generation of TTL chips with the HC generation of CMOS chips, as their input and output ranges are different; so, the NOR chip has to be a 74LS27, not a 74HC27.

  We’re ready now for a simple schematic. In Figure 4-107 I’ve colored some of the wires just to make it easier for you to distinguish them. The colors have no other significance.

  Figure 4-107. A simplified schematic shows how outputs from the 74LS92 counter can be combined, with signal diodes and a single three-input NOR gate, to generate the spot patterns on a die. The wire colors have no special meaning and are used just to distinguish them from each other.

  Figure 4-108. Binary outputs from the 74LS92 counter can be used to power LEDs arrayed in groups to simulate the pattern of spots on a die.

  Each of the LEDs is grounded through a separate 4K7 load resistor. Unfortunately, this means that when they are displaying the pattern for a 6, all of them are running in parallel from the output of the NOR gate, which overloads it. As long as you don’t leave the display in this mode for very long periods, it shouldn’t cause a problem. You could compensate by increasing the load resistors, or by running pairs of the LEDs through one resistor, but this will make them so dim that they’ll be difficult to see, as they’re so close to their lower limit for current already.

  Notice how I have added four signal diodes, D1 through D4. When Output C goes high, it has to illuminate all four corner LEDs, and so its power goes into the brown wire as well as the gray wire. But we must never allow one output to feed back into another, so D4 is needed to protect Output B when Output C is high.

  Because there is now a connection between B and C, we need D2 to protect Output C when Output B is high. And because Output B must only feed two of the corner LEDs, we also need D3 to stop it from illuminating the other two. And, we have to protect the output from the NOR gate when either Output C or Output B is high. This requires D1.

  Figure 4-109 shows everything that I’ve described so far assembled in breadboard format, while Figure 4-110 shows the test version that I built. Note that the unused logical inputs on the 74LS27 chip are shorted together and connected to the positive side of the power supply. Here’s the rule:

  When using CMOS chips (such as the HC series), connect unused logical inputs to the negative side of the power supply.

  When using TTL chips (such as the LS series), connect unused logical inputs to the positive side of the power supply.

  Figure 4-109. With some extra components, the schematics from Figures 4-102 and 4-107 can be combined to make the working dice simulation.

  I assume that you have had enough fun watching the LEDs count slowly, so I’ve changed the capacitor and resistor values for the 555 to increase its speed from approximately 1 pulse per second to about 50,000 pulses per second. The counter could run much faster than this, but I just want it to cycle fast enough so that when the user presses and releases a button, the count will stop at an unforeseeable number.

  The button starts and stops the 555 timer by applying and releasing power to the timing circuit only. This is the equivalent of shaking and then throwing the die.

  While the counter is running fast, the LEDs are flashing so fast that all of them will seem to be on at once. At the same time, the circuit charges a new 68 µF capacitor, which I have added between the pushbutton and ground. When you release the button, this capacitor discharges itself through the 1K timing resistor. As the charge dissipates, the timing capacitor will take longer and longer to charge, and discharge, and the frequency of the 555 will gradually diminish. Consequently the LED display will also flash slower, like the reel on a Las Vegas slot machine gradually coming to a stop. This increases the tension as players can see the die display counting to the number that they’re hoping for—and maybe going one step beyond it.

  Note that to maximize this effect, the button has to be held down for a full second or more, so that the 68 µF capacitor becomes fully charged before the button is released.

  Figure 4-110. The electronic dice schematic applied to a breadboard, with a pushbutton at the top to start and stop the counter, and 7 LEDs at the bottom to display the output.

  So, this circuit now fulfills the original goal. But can it be better? Of course it can.

  Enhancements

  The main thing I want to improve is the brightness of the LEDs. I could add a transistor to amplify the current to each one, but there’s a simpler alternative: a TTL “open collector” inverter.

  I want to use an inverter because in the world of TTL, as I mentioned earlier, we can sink much more power into the output pin of a chip than we can source from it. So, I’m going to turn each LED the other way around and connect their load resistors to the positive side of the power supply. This way, they’ll sink their power into the outputs of the inverter.

  And the great advantage of an “open collector” version of the inverter chip is that it is designed to sink much more current than a normal TTL logic chip. It is rated for 40mA per pin. The only disadvantage is that it cannot source any current at all; instead of its output going high, it just behaves like an open switch. But that’s OK for this circuit.

  So the next and final schematic, in Figure 4-111, includes the 74LS06 inverter, which has also been added to the breadboarded version shown in Figure 4-112. I suggest that you put aside the little low-current LEDs and substitute some normal-size ones. Using Kingbright “standard” WP15031D 5mm LEDs, I find that each draws almost exactly 20mA with a voltage drop of about 2V with a 120 ohm series resistor. Because each output pin from the 74LS06 inverter powers no more than two LEDs at a time, this is exactly within its specification. I suggest that if you build this circuit, you check the consumption of your particular choice of LEDs and adjust the resistors if necessary.

  Remember: to measure the voltage drop across an LED, simply touch the probes of your meter across it while it is illuminated. To measure the current, disconnect one side of the LED and insert the meter, in milliamp mode, between the leg of the LED and the contact that it normally makes in the circuit.

  For a really dramatic display, you can get some 1 cm diameter LEDs (Figure 4-113). Check the specification, and you should find that many of these don’t use more power than the usual 5 mm type. But whatever kind you use, don’t forget to turn them around so that their negative sides face toward the inverter, and their positive sides face the resistors, which are connected to the positive side of the power supply.

  One last detail: I had to add two 10K resistors to this version of the circuit. Can you see why? Diodes D1 t
hrough D4 are designed to transmit positive voltage through to the inverter when appropriate, but they prevent the inputs of the inverter from “seeing” the negative side of the power supply when the counter outputs are low. These inverter inputs require pull-down resistors to prevent them from “floating” and producing erroneous results.

  Figure 4-111. If open-collector inverters are added to the dice schematic, it can drive full-size LEDs with up to 40mA, as long as the LEDs are turned around to sink current into the TTL output stage instead of trying to source current from it.

  Figure 4-112. The completed circuit using an open-collector inverter to drive full-size LEDs.

  The final enhancements are up to you. Most obviously, you can add a second die, as many games require two dice. The 74LS27 chip still has a couple of spare NOR gates in it, one of which you can make use of, but you will need an additional 555 timer, running at a significantly different speed to ensure randomness, and it will have to drive a second counter.

  After you get your dice up and running, you may want to test them for randomness. Because the pulses from a 555 timer are of equal length, every number has an equal chance of coming up; but the longer you hold down the Start button, the better your odds are of interrupting the counting process at a truly random moment. Anyone using your electronic dice should be told that “shaking” them for a full second is mandatory.

  Of course, I could have simulated dice more easily by writing a few lines of software to generate random numbers on a screen, but even a fancy screen image cannot have the same appeal as a well-made piece of hardware. Figure 4-113 shows white 1 cm LEDs mounted in a sanded polycarbonate enclosure for dramatic effect.

  Most of all, I derived satisfaction from using simple, dedicated chips that demonstrate the binary arithmetic that is fundamental in every computer.

  Figure 4-113. The open-collector inverter chip in the dice circuit is sufficiently powerful to drive 1-cm white LEDs that draw about 20mA each, using a potential of 2V. In this finished version, the LEDs were embedded in cavities drilled from the underside of half-inch polycarbonate, which has been treated with an orbital sander to create a translucent finish.

  Experiment 24: Intrusion Alarm Completed

  Now let me suggest how you can apply the knowledge from this chapter of the book to upgrade the burglar alarm project that was last modified in Experiment 15. You’ll probably need to check Chapters 2 and 3 to refamiliarize yourself with some features of the alarm.

  Upgrade 1: Delayed Activation

  The biggest flaw in the alarm was that as soon as it was activated, it would immediately respond to any signal from the door and window sensors. It needed a feature to delay activation to give you a chance to exit from the building before the alarm armed itself. A 555 timer can provide this functionality, probably in conjunction with a relay. The power to the alarm should pass through the contacts of the relay, which are normally closed. When you press a button on the timer, it sends a positive pulse to the relay lasting for around 30 seconds, holding the relay open for that period. You could mount the timer in its own little box with a button on it, which you press when you’re ready to leave the building. The 12-volt power supply to the burglar alarm passes through the box containing the delay circuit. For 30 seconds, the 555 interrupts power to the alarm, and then restores it, ready for action.

  Upgrade 2: Keypad Deactivation

  This is now really simple. You can substitute a latching relay instead of the switch, S1, on the alarm box (shown in Figure 3-110), and use a keypad to set and reset the relay in exactly the same way as in the combination lock in Experiment 20. You’ll have to run an additional three wires from the relay, out of the alarm box, to the keypad (one supplying power to the “on” relay coil, another supplying power to the “off” coil, the third being a common ground). You can either use a 9-volt battery to power the electronics in association with the keypad, or run an additional fourth wire from the alarm box, to carry positive power to the logic chips, bearing in mind that you have to insert a voltage regulator at some point, to drop the 12 volts that the alarm uses to the 5 volts that the logic gates require. As the gates consume so little power, the 12-to-5 drop should be OK for the voltage regulator; it won’t have to dissipate too much heat.

  With this additional feature, you can use the alarm like this:

  Press the pound key on the keypad to flip the latching relay into its “on” mode, so that it passes power to the alarm box, which is now armed.

  If you want to leave the house, push the button on the delay unit to give you 30 seconds in which to do so.

  If the alarm is triggered, enter your secret code on the keypad to deactivate it by flipping the latching relay to its “off” position and cutting power to the alarm box.

  These modifications are so simple that I think the block diagram in Figure 4-114 should be all you need. I don’t think I need to give you any schematics. The only change you have to make to the existing alarm is to substitute the latching relay for the on/off switch.

  But, there is still one obvious necessary enhancement needed: how can you get back into the house without instantly triggering the alarm?

  Figure 4-114. This block diagram shows the relative placement of the old and new components. The pushbutton power interrupt (which allows you to leave the house before the alarm switches itself back on) goes between the power supply and everything else.

  The latching relay substitutes for the DPDT switch on the previous version of the alarm. The transistor and self-locking relay, connected with the door and window magnetic switches, remain unchanged. The new delay circuit is inserted between the self-locking relay and the noisemaker. The test button is wired with the latching relay in the same way that it was wired previously with the DPDT switch.

  Upgrade 3: Delay Before Deactivation

  Typically, alarms include another delay feature. When you open a door on your way into the building and it triggers the alarm, you have 30 seconds to deactivate it, before it starts making a noise.

  How can we implement this delay feature? If I try to use another 555 timer to generate a pulse to inhibit the noise, that won’t work, because the output from either the transistor or the relay can continue indefinitely. The relay locks itself on, and the transistor can continue passing voltage for as long as someone leaves a door open. If either of these signals activates a timer in monostable mode, the pulse from the timer will never end, and it will suppress the alarm indefinitely.

  I think what I have to do is use a resistor and a capacitor to create a delay. I’ll power them through the existing relay, so that I can be sure that they’ll receive the full voltage of the power supply, after beginning from zero. Gradually the capacitor will acquire voltage—but I can’t connect this directly to the noisemaker, because the noisemaker will gradually get louder as the voltage increases.

  I have to insert a device that will be triggered to give full voltage when the input rises past a certain point. To do this, I’ll use a 555 timer that’s wired in bistable mode. This kind of modification is generally known as a “kludge,” because it’s not elegant, uses too many components, and does not use them appropriately. What I really need is a comparator, but I don’t have space to get into that topic. So, using the knowledge that you have so far, the schematic in Figure 4-115 shows how a delay can be added to the alarm—not elegantly, but reliably.

  The only problem is that if you power up a 555 timer in bistable mode, there’s a 50-50 chance that the timer starts itself with its output high or low. So I need to pull the voltage low on the reset pin (to start the timer with its output inhibited) and gradually let it become positive (to permit the output). At the same time, I want to start with the voltage high on the trigger pin and gradually lower it, until it falls below 1/3 of the power supply and triggers the output.

  So there are two timing circuits. The one for the reset pin
works faster than the one on the trigger pin, so that at the point when the timer is triggered, the reset won’t stop it.

  The schematic shows component values that will do this. The 10 µF capacitor starts low but is charged through the 10K resistor in a couple of seconds. The timer is now ready to be triggered. But the 68 µF capacitor starts high (being connected with the positive side of the power supply) and takes a full minute to be pulled down to 1/3 of supply voltage through the 1M resistor. At that point, its voltage is low enough to trigger the 555. The timer output goes high and supplies the noisemaker.

  You should be able to insert this little delay module in your alarm box, between the output from the relay and the input to the noisemaker without too much trouble. And if you want to adjust the delay, just use a higher or lower value resistor than 1M.

  Figure 4-115. This addition to the original alarm circuit imposes a one-minute delay before the alarm sounds. The 555 timer (wired in bistable mode) receives power through relay R1. The lower timing circuit initially applies negative voltage to the reset, ensuring that the 555 powers up with its output suppressed. This voltage quickly rises. Meanwhile the upper timing circuit applies a voltage to the trigger that gradually diminishes as the 68 µF capacitor equalizes its charge through the 1M resistor. When the voltage diminishes to 1/3 of the supply, the timer’s output goes high and starts the noisemaker. If the power to the circuit is interrupted at any time before this, the relay relaxes, the capacitors gradually discharge, and the alarm does not sound.

  The Wrap-Up

  If you add these three enhancements, your alarm will have all the features on my original wish list. Of course, if I were designing it from scratch, with all the information that has been added in this chapter of the book, it could be more elegant. But the modifications have not entailed making destructive changes to our original project, and all the design goals have been met.