Fullscreen

Fun with comparators This month, we’ll have a play with comparators. Most people regard comparators as pretty basic and boring devices but they should be regarded as important building blocks. You can do all sorts of functions apart from comparators. Examples are Schmitt triggers, oscillators, timers, pulse generators, AND and OR gates and even zero voltage crossing detectors. By LEO SIMPSON We will have a play with a “bog standard” device, the LM393 dual comparator but remember that virtu­ ally everything we do will be applica­ ble to other comparators such as the equally “bog standard” LM339 quad comparator or other higher perfor­ mance comparators. OK. So what is a comparator? It’s 74  Silicon Chip very similar to an op amp. It has a differential input stage, with noninverting (+) and inverting (-) inputs. The two inputs are used to “compare” two signals or two voltages and then the output indicates whether one input is higher (or lower) than the other. Typically, one input will be tied to a reference (REF) voltage while the other input is fed with the signal or voltage to be monitored. Let’s do a typical setup with the LM393, as shown in Fig.1. BELOW: this photo features all the parts shown in Fig.1 and Fig.2, except for the 47µF capacitor. One of the potentiometers is not used. Fig.1: this is a basic circuit for a non-inverting comparator. To turn it into an inverting comparator, swap the inputs, pins 5 & 6. Pins 2 & 3 on the unused comparator should be tied high. The 47µF is not used initially. Because we are wiring this up on a Protoboard and want to make it as straightforward as possible, I’ve hook­ ed up the second comparator in the LM393 dual package; ie, involving pins 5, 6 & 7. But remember that ex­ actly the same circuit can be hooked up with the first comparator, or any number of variations on the theme. Since we’re not using the first com­ parator, its inputs should be tied high (+12V) or low (0V). This is done to prevent it from producing any spuri­ ous oscillations which it could do if its inputs were left to “float”. Pin 6, the inverting input, is con­ nected to the junction of two 10kΩ resistors connected across the 12V supply. The voltage at the junction will be half the supply or +6V (nom­ inal) and since this voltage is fixed, we regard this as the REF (refer­ence) input. Pin 5, the non-inverting input, is connected to a variable voltage obtained from the 50kΩ pot (VR1) which is con­nected in series across the 12V supply. By rotating pot VR1, we can vary the voltage fed to pin 5 from +12V down to about +2V. Why not hook up just the 50kΩ pot and the 10kΩ resistor across the 12V supply to begin with and check with your multimet­er to see if this voltage range can be obtained? Check also that you get 6V (or half the DC plugpack supply voltage) at the junc­tion of the two 10kΩ resistors. Now hook up the rest of the components on the circuit of Fig.1, as shown in the photo and in the Protoboard wiring diagram of Fig.2. So that we can see what the compar­ ator does in response to the variable signal conditions, I have hooked up a LED (light emitting diode) in series with a 1kΩ resistor, between the +12V supply and the comparator’s output at pin 7. Now, if we set VR1 so that +12V is fed into pin 5, the LED will not light. If we then wind VR1 back the other way, reducing the voltage to pin 5, at some point the LED will light. If we then measure the voltage at pin 5 we should find that it is just below the voltage at pin 6. In our case, on the afternoon I was writing this, the voltage on pin 6 was +6.18V and as I wound VR1 to the point where the LED came on fully, pin 5 was +5.93V and pin 6 was +5.6V; ie, a smidgin below pin 5. This demonstrates a number of interesting points. The first question might be, “Why did the voltage at pin 5 change at all?” but we’ll get to that later. No, the main point is that when pin 5, the inverting input, is pulled low, the output at pin 7 also goes low. And Fig.2: use this diagram to wire up the circuit of Fig.1. Winding VR1 back and forth will turn the LED on and off. December 2000  75 Fig.3: using the simple comparator results in poor switching behaviour. The upper trace is the input sinewave at 1kHz while the lower trace is the output waveform. when 7 goes low, the LED will light because it is hooked up to +12V via the 1kΩ resistor. So what we have here is a non-in­ verting comparator. We can summa­ rise its operation by saying that when the non-inverting input goes above the inverting input, the output will go high; when the non-inverting input goes below the inverting input, the output will go low. Inverting comparator Say we wanted to change the sense of the comparator? Say, we wanted the output to go high when the input goes low – ie, below the reference input? Easy. Just swap the reference and signal inputs. Go ahead and do it: swap the connections to pins 5 & 6. Now what happens? What happens is that when pin 6 is above pin 5, the LED is alight. Conversely, when pin 6 is low, the output at pin 7 is high and so the LED is not alight. So that’s how you make an inverting comparator. The thing is, you can tell what the comparator will do just by look­ ing at which input is inverting and which is non-inverting. If we vary the non-inverting input, the output will essentially follow the input; ie, when it goes above the REF input, the output will go high as well. In other words, the output is the same as the non-inverting input or to put it anoth­er way, the output has not been inverted (non-inverting, get it?). Conversely, for an inverting com­ parator, the output signal will be 76  Silicon Chip Fig.4: this is the cleaner switching result when a 47µF bypass capacitor is connected to pin 6. This stops the voltage at pin 6 from varying while the switching action is taking place. inverted compared to the input. We can use these basic compara­ tor circuits in all sorts of ways. For example, if we replaced VR1 with a thermistor we could produce a temperature-sensitive switch. Or the potentiometer could be a throttle switch in a car or any one of a number of transducers. So comparators do an important job in sensing all sorts of circuit conditions and then switching an output in response. AC signals What else can a comparator do? So far we have only consid­ered the situa­ tion where a comparator is monitoring static or slowly varying signals. What about rapid signals? To demonstrate, let’s feed an audio oscillator into the comparator of Fig.1. We’ll feed the signal in via a 0.1µF (100nF) capaci­ tor to pin 5, set VR1 to give +6.5V at pin 5 and see what happens. With no signal from the oscillator, the LED is off. We apply a 1kHz sinewave signal, wind up the signal to about 400mV and the LED lights, although not as brightly as it was when we manually varied VR1. So what is happening? The scope waveforms of Fig.3 show the results. The upper trace is the input 1kHz sinewave and the lower trace is the voltage at pin 7 and the LED. Some­ thing is wrong here because instead of switching cleanly, the comparator is obviously dilly-dallying on the output transition between high and low. Hmm, what if the reference voltage at pin 6 was varying up and down with the switching action? We saw that this was actually happening on the static signal test previously. OK. So let’s hang a 47µF capacitor off pin 6 to the 0V rail. That will stop any short term signal variations on pin 6 and should clean up the output signal. The scope waveforms of Fig.4 show the result and the output waveform now switches much more cleanly. The LED also runs a little brighter as well. So in practice, we would not use a simple voltage divider for the REF voltage. We would use a well-filtered voltage, prob­ a bly derived from a zener diode or a more precise voltage source. However, even with a well-filtered REF source for one of the inputs, the switching action of a comparator may not be what we want. Say we were us­ ing a thermistor to drive a comparator in a temperature controller. If we had the simple circuit of Fig.1 (together with the 47µF capacitor at pin 5) it would certainly work but it would be far too sensitive and the circuit would hunt back and forth (ie, switch on and off) continuously with small temperature variations. This would be unsatisfactory if you were controlling a heater or cooling unit. Adding hysteresis The answer is to change the points at which the comparator switches from low to high and then from high back to low. To do this, we provide positive feedback from the output to Fig.5: to demonstrate hysteresis this inverting comparator ver­sion of the circuit has a 22kΩ positive feedback resistor con­nected between pins 5 & 7. the non-verting input. By positive feedback we mean applying some of the output signal back to the input, so that a portion of the output signal adds to the input signal. This is the opposite of negative feedback where the portion of the signal fed back from the output subtracts from the input. The circuit of Fig.5 is similar to Fig.1 but we have swapped the way the inputs are connected and we have added a 22kΩ resistor from pin 7 to pin 5. This positive feedback resistor shifts the switching threshold up and down as the output switches high and low. To set up the circuit of Fig.5, dis­ connect the oscillator, swap pins 5 & 6, pull out the 47µF capacitor and then check that the circuit works as before. As you wind VR1 back and forth you will find that the LED Subscribe & Get this FREE!* THAT’S RIGHT – buy a 1or 2-year subscription to SILICON CHIP magazine and we’ll mail you a free copy of “Computer Omnibus”. Includes articles on troubleshooting your PC, installing and setting up computer networks, hard disk drive upgrades, clean installing Windows 98, CPU upgrades, a basic introduction to Linux plus much more. switches on over a very narrow range. Pin 6 only needs to be raised or low­ ered by a small amount near +6V to turn the LED on or off. Now connect the 22kΩ resistor between pin 6 & 7. You will find that you now have to wind VR1 over a wider range to turn the LED on and off. In fact, you will now find that you have to vary VR1 so that it shifts pin 6 above +7V to turn the LED on and below +5V to turn it off. In fact, if you measure pin 5 as the LED turns on and off, you will find that it is moving up and down over a 2V range as the output goes high and low. This is a fairly crude way of adding hysteresis but it demonstrates the principle. You might also notice that the LED does not fully turn off. This is not be­ cause the comparator is not switching correctly but is due to the current flowing through the 22kΩ resistor. This small current is enough to keep the LED glowing feebly. Well, that’s enough for this month. Next month we’ll have a further play with the LM393 and make it work in SC a few more cir­cuits. APOLOGY ATTENTION KIT CONSTRUCTORS Some K3130 Temperature Control Switch kits were supplied with metal end panels, instead of the plastic panels originally specified. Depending on how the kit was assembled, this may compromise electrical safety. Constructors are advised to disconnect the kit from mains power and check the panels on their kit. If plastic panels have been supplied, no further action is required. If metal end panels have been supplied, customers should stop using the kit and obtain replacement panels. *Australia only. Offer valid only while stocks last. Subscribe now by using the handy order form in this issue or call (02) 9979 5644, 8.30-5.30 Mon-Fri with your credit card details. To obtain replacement panels contact: Dick Smith Electronic Kit Department "K-3130 End Panels" PO Box 321 North Ryde NSW 2113 or Phone: 1800 618 459 or e-mail: kits<at>dse.com.au Dick Smith apologises for any inconvenience caused December 2000  77