Fullscreen HTML5Med Res (??MB)HTML4

Fast clocks running at six to eight times actual speed are desirable for model railways which run a 24-hour schedule in a compressed time of three to four hours. A fast clock for railway modellers Are you a keen railway modeller? Do you run your trains to a schedule? If so, you will want to build this fast clock which can be set to run at between 4.5 and 8.5 times faster than normal. By LEO SIMPSON Why would anyone ever want a fast clock? Surely time passes rapidly enough as it is, except, of course, in the afternoons at school or work. But there is a sound reason and it has to do with running trains to a schedule on a model railway layout. Because model railway layouts are always much smaller with respect to scale than the real thing, the times taken to run a train from one point to another are ridiculously small, in real time. 38  Silicon Chip For example, while the distances on real railways can be hundreds of kilometres and the trains can take many hours or even days to get to their destination, a typical large model railway would be unlikely to have more than 50 metres of track. For HO scale (1:87), this equates to 4.3 scale kilometres; for N scale (1:160) it equates to 8 scale kilometres. Even on these large layouts, the time for a train to make several circuits will be measured in minutes rather than hours. So to inject a little more realism into a model railway train schedule, it makes sense to use a “fast clock”. But there is another more practical reason which has nothing to do with scale factors and this has more to do with the number of spare hours in an evening. Typically, a model railway club will have a running session which lasts around three hours in an evening but a “realistic” operating session should last at least one day, or 24 hours. So the club needs to squeeze 24 hours of operation into a real time of three hours or so. The reason for the fast clock now becomes clear – it needs to run about 6 to 8 times faster than normal. The question is, how do you make a clock run this fast? Our approach was to take a typical crystal controlled clock movement which is more or less Fig.1: this is a typical circuit for a 1.5V crystal controlled clock movement. It uses a 32kHz crystal and drives the clock stepper motor coil with pulses at 1-second intervals. typical in zillions of battery operated clocks. Fig.1 shows a typical circuit using a Samsung chip. It uses a single CMOS IC operating from a 1.5V AA cell and con­trolled by a 32kHz crystal. The IC has an internal divider chain which produces complementary pulses to drive a stepping motor. Fig.2 shows the oscilloscope waveforms from the clock movement used in this article. As can be seen, there are two pulse trains, each with pulses about 30ms long and exactly two seconds apart. The pulse trains are staggered by one second. What actually happens is that the IC applies a pulse to the clock coil (the stepper motor) in one direction and then one second later, applies the same pulse in the opposite direction. This operates the escapement which makes the ticking sound and drives the clock hands. Our first approach was to see if we could make the chip operate six to eight times faster than normal. The simplest way to do this would be to Fig.2: this digital oscilloscope printout shows the waveforms from the circuit of Fig.1. In effect, there are two pulse trains with pulses two seconds apart. A pulse is applied to the coil in direction (upper trace) and then a pulse in the opposite direc­tion is applied to the coil (lower trace). The oscilloscope timebase for these waveforms is 500ms/div; the printout is five seconds long! replace the 32kHz crystal with one of 192kHz but such crystals are not readily available. Hence, we decided to remove the 32kHz crystal and to drive one of the oscillator pins of the chip with an external oscillator based on a 7555 CMOS timer. The first hurdle with this approach is that a 7555 will not operate at 1.5V. It will operate with a 3V supply so we cobbled together a suitable circuit with a voltage divider at the output, to make the signal compatible with the 1.5V clock chip. Fig.3 shows this approach. Did it work? Well, yes and no. It would work up to about 100kHz or so but higher than that and the clock mechanism itself refused to work. The reason appears to be the length of the Fig.3: our first attempt at a speed-up circuit involved using a 7555 CMOS chip driving one of the crystal input pins on the clock chip. The circuit conked out if we attempted a speed-up of more than four times. pulses applied to the clock stepper motor. In the standard clock, the pulses are typically 30ms or 46ms long and their length is a fixed relationship to the 32kHz crystal. At an oscillator frequency of, say, 128kHz, the clock pulses would only be one quarter as long (ie, 7.5 or 11.5ms) and this appears to be insufficient to operate the motor reliably. We tried a number of circuit variations, such as operating the clock chip from 3V which is in excess of the ratings but it still did not work. Final circuit The next approach was to scrap the crystal controlled cir­cuit and develop a new circuit to drive the clock stepper motor directly. This is shown in Fig.4. Again it is based on a 7555 CMOS timer, IC1. This operates at a frequency of between 4.5Hz and 8.5Hz, as set by the components at pins 2, 6 & 7. The fre­ quency is adjustable by trimpot VR1. The output of IC1 can be varied between 4.5Hz and 8.5Hz and thus the speed-up factor can be varied between 4.5 and 8.5 times by trimpot VR1. The output from pin 3 is inverted and buffered by NAND gate IC2a and then applied to IC3, a 74HC76 flipflop. This divides the output by two and produces complementary outputs at pins 14 and 15. These are gated toDecember 1996  39 PARTS LIST 1 1.5V crystal controlled clock movement 1 PC board, code 09112961, 67 x 38mm 2 1.5V AA cells 1 double-AA cell holder and battery snap connector 1 1MΩ trimpot (VR1) Fig.4: our final circuit for the Fast Clock Driver uses three ICs: a 7555 CMOS timer, a 74HC76 flipflop and a 74HC00 NAND gate chip. The circuit drives the clock coil directly, dispensing with the internal clock circuitry. gether with the pulses from pin 11 of IC2a to provide complementary pulses from pins 3 & 6 and of IC2. Fig.5 & Fig.6 shows the output waveforms at two different clock speeds, six times and eight times. Fig.5 shows the waveforms when the clock is operating at six times normal speed while Fig.6 shows it operating at eight times normal speed. Considering Fig.5, the upper trace (Ch1) is the waveform at pin 3 of IC2b while the lower trace (Ch2) is the waveform at pin 6 of IC2c. In effect, while the period of both waveforms in Fig.5 is 333ms, the clock coil receives stepping pulses 166.5ms apart which is six times faster than the normal stepping rate of one per second. A similar situation applies in Fig.6 except that the period of both waveforms is 250ms and the speed-up is eight times. Notice that the pulse width applied to the motor is between 15 and 16ms which is half that applied to the clock in normal operation and as shown in Fig.2. There are two reasons for this. First, the clock motor itself is designed to run from a circuit powered with a 1.5V cell whereas our circuit uses 3V. We have used 3V because the CMOS chips specified will not run reliably below 2V. This means that the pulses delivered from the modified circuit were twice the voltage they should be. Paradoxically, because the clock coil was being driven so hard, its operation became unreliable at the higher speeds. We could correct that problem by inserting a 330Ω resistor in series with the clock coil but then the effective battery life would be reduced; as the battery voltage dropped, the pulse drive was unduly reduced by the series resistor. Our final version, presented in Fig.4, Fig.5: waveforms from the circuit of Fig.4, taken at pins 3 & 6 of IC2. The speed-up factor is six times. The oscillo­ scope time­base is 50ms/div. 40  Silicon Chip Semiconductors 1 7555, LMC555 CMOS timer (IC1) 1 74HC00 quad 2-input NAND gate (IC2) 1 74HC76 dual JK flipflop (IC3) Capacitors 1 100µF 16VW electrolytic capacitor 3 0.1µF MKT polyester 1 .01µF MKT polyester Resistors (0.25W, 1%) 1 820kΩ 0.25W resistor 1 150kΩ 0.25W resistor compensates for the higher pulse amplitude by halving the pulse width and eliminating the series 330Ω resistor. This has the benefit of allowing the circuit to work reliably down to below 2V which means that the batteries last longer. PC board We designed a small PC board to take the circuit of Fig.4. It measures 67 x 38mm and is coded 09112961. Its component layout is shown in Fig.7. Fig.6: waveforms from the circuit of Fig.4, taken at pins 3 & 6 of IC2. The speed-up factor is eight times. The oscillo­ scope timebase is 50ms/div. Left: when you pull the back off the clock movement, it will look like this. Be careful not to scatter the parts. If you lift off the two top gears, you will be able to remove the PC board and coil assembly. The photo above shows how we made two cuts to the PC tracks and then connected two fine gauge enamelled copper wires direct to the clock coil terminals. Fig.7: follow this parts layout to build the Fast Clock Driver circuit of Fig.4. When assembling it, make sure that all three ICs are correctly oriented and that the 100µF electrolytic ca­pacitor is correctly polarised. You will need four PC stakes, two for the battery connec­tions and two for the clock coil connections. Assembling the PC board and get- Fig.8: this is the actual size artwork for the PC board. Check your board carefully before installing any of the parts. ting it going is the easy part. Pulling the clock apart and making the connections to the clock coil are a little trickier but it just takes a little care. Essentially what must be done is to remove the hands and time-setting knob, undo one screw and unclip the clock case. Then, while the clock is This photo shows the assembled Fast Clock Driver. Two wires connect it to the clock movement. face down, lift out two gears and then the internal PC board. In practice, you will find that the PC board actually sup­ports the coil so it cannot be removed and discarded. Instead, you must cut the PC tracks where they connect to the coil. Then you need two fine wire connections to the coil which can be brought out through the side of the clock case. You can then reassemble the clock and connect it to the new driver board. When power is applied the clock should immediately start running and the speed-up factor should be variable between about four and nine times, depending on the setting of trimpot VR1. We suggest that you leave the second-hand off the clock; it will go around so fast that the effect will be SC ludicrous. December 1996  41