Fullscreen HTML5Med Res (??MB)HTML4

By RICK WALTERS Build this sine/square wave oscillator for your workbench Do you want a good quality audio oscillator that does not go “boingg” when you switch ranges? And has constant output amplitude as you sweep over each range and from range to range? If so, this could be the oscillator for you. It covers the fre­quency range from 2Hz to 20kHz and is suitable for a wide range of audio applications. 58  Silicon Chip N ORMALLY, THE FIRST choice of anyone contemplating building or buying an audio oscillator is a Wein bridge type. These have the advantage of low distortion (usually) but their output ampli­tude often bounces all over the place as you sweep over each frequency range and is even worse when you switch ranges. It is possible to avoid these problems with careful design but the circuit will end up being more com- Fig.1: block diagram of the oscillator. IC1c is a high frequency oscillator and its output is divided by 1, 10, 100 or 1000 by IC7b or IC8. It is then further divided by 2 and 5 before being applied to a divide-by-10 ring counter (IC2a, IC4 & IC5). This drives a resistor network which produces a stepped waveform which is fed to switched capacitor filter, IC6. plicated (see our low distor­tion design in the February & March 1999 issues). The second choice for an audio oscillator is typically a function generator but while these usually have good amplitude stability, their distortion content is usually fairly average. But now you have a third choice with this design which uses digitally generated sinewaves and employs a switched capacitor filter. While thumbing through the Jaycar Electronics catalog some time ago, I came across an “IC bargain”, an MF4CH-50 4th order switched capacitor Butter­worth low-pass filter, for the trivial sum of $1.50. This set me thinking (I do that occasionally) about what level of distortion we would get if we fed a pseudo sinewave (digitally generated) into that sort of filter. Many moons later, this low cost audio oscillator is the outcome of those profound thoughts. The oscillator is housed in a plastic zippy box measuring 157 x 95 x 50mm. It has three knobs on the front panel and these are the 4-position range switch, the frequency control and the sinewave output control. As well, there is a toggle power switch and three RCA sockets for the sinewave output and two square wave outputs. The circuit is battery operated but could be run from a plugpack if you wish; more on that later. Theory of operation Before we go too far, we need to explain just what is a 4th-order switched capacitor Butterworth lowpass filter. Let’s do the low-pass filter first because it’s easy: as its name sug­ gests, a low-pass filter is one that lets low frequencies through (passes) but progressively blocks (attenuates) the higher ones. The frequency at which the response is 3dB down is called the turno­ver frequency. Now what does 4th-order mean? A 1st order low-pass filter has an atten­uation slope of 6dB per octave above the turnover frequency and so a 4th-order has four times this or Performance • • • • • Sinewave output ...............................................2Hz - 20kHz, 0-2V RMS Square-wave output ................................2Hz - 20kHz, 5V peak-to-peak Square-wave x100 output ....................200Hz - 2MHz, 5V peak-to-peak Sinewave distortion ....................................................... less than 0.85% Current consumption............................. 15mA from +5V, 6mA from -5V 24dB per octave. This steep rolloff of high frequencies is used to get rid of the higher harmonics of our digital sinewave. The term Butterworth describes a filter response which is flat (0dB) until it begins to roll off. Other types of filters have a peak or ripples in the response before the rolloff begins. For audio work, the Butterworth response is usually the best and most suitable. Switched capacitor filters Any conventional filter circuit can be designed to roll off at any given frequency but this frequency can only be altered by changing the relevant resistor or capacitor values. For a 4th-order filter, this would mean changing the values of four resis­tors or four capacitors in precisely the same ratio. This makes things very complicated because an oscillator based on a variable 4th-order filter would then need a five ganged potentiometer, or a five ganged capacitor (if we include one for the actual frequency control). In practice, this approach would be just too expensive. This is where the MF4CH-50 switched-capacitor filter comes into the picture. It has four internal capacitors which are rapidly switched in and out of circuit to vary their values. Furthermore, the more rapidly they are switched, the less the effective capacitance. FEBRUARY 2000  59 60  Silicon Chip More specifically, the turnover filter frequency of the MF4CH-50 is 1/50th of its clock frequency, so if we clocked it at 50kHz it would begin to rolloff at 1kHz. Fig.1 shows the general concept of the oscillator and IC6 is the MF4CH-50 switched capacitor filter. IC1c is a high frequency oscillator and its output is divided by 1, 10, 100 or 1000 by IC7b or IC8. It is then further divided by 2 and 5 before being applied to a divide-by-10 ring counter (IC2a, IC4 & IC5). This drives a resistor network which produces a stepped waveform which is a very rough approximation of a sinewave which is 1/50th of the frequency output from IC2b. IC6, the MF4CH-50, is also clocked by the output of IC2b and so its turnover frequency exactly matches the output of the ring counter. It effectively removes the switching hash from the waveform, leaving a clean sinewave. Circuit description The circuit of Fig.2 is a little more complex but operates as we have just explained. While it may look to have a lot of circuit elements, it uses only nine low-cost ICs. Let’s start with IC1c, the master oscillator. It is a 74HC132 quad NAND gate with Schmitt trigger inputs configured as an oscillator, with the maximum and minimum frequencies adjusted using trimpots VR2 and VR3. These are set so that the sinewave frequency varies from just under 2kHz to just over 20kHz on the highest range and potentiometer VR1 then becomes the main fre­quency control. The frequencies for the other three ranges are generated by successively dividing this main frequency by 10 in IC8 and IC7b. These four frequencies are fed to range switch S1a which directs the selected frequency to IC2b (one section of a 4013 dual-D flipflop) and also Fig.2 (left): the master oscillator is IC1c and since it is a Schmitt device it requires trimpots VR2 & VR3 to set the maximum and minimum frequencies. Both IC1a & IC1b are unused but their inputs have been tied to related parts of the circuit. The circuit can be powered from a 9V plugpack, as shown in Fig.12. to IC1d which inverts and buffers the signal and feeds it to the front panel as SQUARE x 100. This signal is only an exact square wave on the lower three ranges which come from IC8 and IC7b. The top range comes direct from IC1c and its output is not a true square wave but has a high time of around 45% (the low time being 55%) of the oscillator’s frequency. As the oscillator output is inverted by gate IC1d, the high and low periods are also inverted (55:45). IC2b divides the selected frequency from S1a by two, giving an exact square wave which is required for the clock input of IC6, the switched capacitor filter. IC2b also drives IC3, a 4017 connected to divide by five. The output of IC3, pin 10, is fed to the clock inputs of flipflops IC2a, IC4b & IC4a and IC5b & IC5a. These five flipflops are connected as a twisted ring counter which divides the clock frequency by 10. The Q outputs of four of the flipflops are summed by the 10kΩ and 16kΩ resistors to pro­duce a stepped waveform and this is fed to the input of the switched capacitor filter, IC6. The stepped input waveform and the filtered output can be seen in the scope waveforms of Fig.3. Quite a dramatic improve­ment, eh? Twisted ring counter What’s a twisted ring counter we hear you asking? In a normal D-type flipflop (such as IC2b), the Qbar output (pin 12) is connected back to the D input (pin 9). This causes the Q output to change from high to low and back to high again on sequential low to high transitions of the clock signal at pin 11. In our twisted ring counter the Qbar output of IC5a is tied back to the D input of IC2a. Assuming the Qbar output of IC5a was low, the Q output of IC2a would be low and this low would then be propagated through the chain until the low level was applied to IC5a. This would cause the Q output to go low and the Qbar output to go high. Thus a high would be presented to pin 5 of IC2a and it would be propagated through the chain. It is “twisted” because a low level on IC5a’s Q output (the main output) propagates a high through the chain and vice versa. While it is hard to visualise, what happens is that a high is moved Parts List 1 PC board, code 04102001, 149 x 71mm 1 plastic box, 50mm x 90mm x 150mm with aluminium lid 1 front panel label, 150 x 85mm 1 2-pole 6-position PC mount rotary switch with two nuts (S1) 1 DPDT miniature toggle switch (S2) 3 panel-mount RCA sockets 2 25kΩ linear potentiometers (VR1, VR4) 3 knobs to suit 1 2kΩ horizontal mounting trimpot (VR2) 1 200kΩ horizontal mounting trimpot (VR3) 2 9V batteries 2 battery snap connectors Semiconductors 1 74HC132 quad NAND Schmitt trigger (IC1) 3 4013 dual D flipflop (IC2,IC4,IC5) 1 4017 decade divider (IC3) 1 MF4CH-50 switched capacitor filter (IC6) 2 74HC390 dual decade divider (IC7,IC8) 1 TL071 op amp (IC9) 1 78L05 5V regulator (REG1) 1 79L05 -5V regulator (REG2) Capacitors 1 100µF 16VW PC electrolytic 1 10µF 16VW PC electrolytic 4 0.1µF monolithic ceramic 1 .033µF MKT polyester 1 .0033µF MKT polyester 1 820pF 10% ceramic 1 330pF 10% ceramic 1 220pF 10% ceramic 1 33pF 10% ceramic Resistors (1%, 0.25W) 1 68kΩ 3 10kΩ 1 47kΩ 2 3.3kΩ 1 33kΩ 1 1kΩ 2 16kΩ For 9V AC plugpack operation Delete 9V batteries and snap connectors 1 panel mounting connector to suit 9V AC plugpack 2 1N4001, 1N4004 power diodes 2 100µF 25VW PC electrolytic capacitors 1 tag strip FEBRUARY 2000  61 The PC board is mounted on the back of rotary switch S1 which in turn is mounted on the front panel. However, you may prefer to further secure the board to the front panel by fitting a mounting pillar at each corner, particularly if the unit is going to be moved about. through the ring and this is followed by four more highs. As each Q output goes high, the stepped waveform of Fig.3 is produced by summing the Q outputs. Once the first high reaches pin 1 of IC5a (pin 2 will be low), a series of lows is shifted through the ring, causing the steps to fall towards 0V and this cycle repeats over and over. The clock frequency fed to the ring counter is also divided by 10 in IC7a, giving a true square wave output at the same frequency as the sinewave output. We found that the output amplitude from IC6 (MF4CH-50) increased on the highest range, starting from around 10kHz. The resistor/capacitor network between the output of IC6 and the sinewave level control VR4 help to flatten the output in this region although even with these components the response is still +1dB at 20kHz. Op amp IC9 is used as a sinewave output buffer with a gain of 3, to 62  Silicon Chip make up for the losses in IC6 and the two 3.3kΩ resistors in series with the output level control (VR4). It also sets the maximum output level to 2V RMS. While the switched capacitor filter does a good job of producing a clean sinewave, there is still some switching hash present and we do some more filtering in IC9. This is done by using S1b, the second pole of S1, to switch a capacitor across the feedback resistor of IC9 on each range. This helps to attenu­ate the high frequency switching spikes. This causes a rather interesting effect. The measured distortion actually decreases slightly as the frequency increases on each range, rather than the normal case where the distortion increases as the frequency increases. Mind you, since the hash is 50 times the fundamental, it is not the slightest bit audible until the fundamental frequency drops below about 200Hz. The sinewave output is symmetrical above and below the 0V line (ground) and is variable from 0V to 2V RMS which should be suffi­cient for any normal audio work. We fitted two voltage regulators on the PC board and these are fine for battery operation. If you plan to use a plugpack you will need to add two capacitors and two diodes which can be wired to a tag strip. This is explained in more detail later. Output & distortion waveforms As noted in the performance panel, the distortion content of the sinewave output is less than 0.85% but this depends on the frequency and the bandwidth of the measurement. The scope wave­ forms of Figs.4, 5, 6 & 7 demonstrate this. Fig.4 shows a 1.1kHz waveform on the top trace and the lower trace is the modulated distortion product which is mainly the 50kHz switching hash. This is equivalent to a harmonic distortion content of 0.83%, taken with a measurement bandwidth of 80kHz (ie, all Fig.3: these scope diagrams show the operation of the switched capacitor filter (IC6). The top trace is the stepped waveform and the lower trace is the sinewave output. Fig.4: the sinewave output at 1.1kHz (top) has a very slight “jagginess” due to 50kHz switching artefacts. The lower trace is the modulated distortion product – mainly the 50kHz switching hash (0.83% THD <at> 80kHz bandwidth). Fig.5: a 1kHz waveform is shown on the top trace, while the lower trace is the distor­tion waveform, measured with a bandwidth of 22kHz. (THD 0.26%). Fig.6: the top trace is a 10kHz sinewave while the lower trace is the residual harmonic content measured with an 80kHz bandwidth (THD 0.285%). Fig.7: the top trace is the sinewave output at 19.6kHz and the lower trace is the distortion which has a level of 0.76%, measured with a bandwidth of 80kHz. Fig.8: the 20kHz sinewave output (top) and the squarewave output. The lefthand cursor is not set correctly and so the frequency measure­ment of 20.7kHz is wrong. FEBRUARY 2000  63 Fig.9: this is the component layout for the PC board and it also shows the wiring to the front panel. harmonics and noise up to 80kHz are included in the measurement). Fig.5 shows a 1kHz waveform on the top trace but this time the distortion waveform on the lower trace has been measured with a bandwidth of 22kHz. This has removed most of the 50kHz hash from the measurement and results in a THD figure of 0.26%. The top waveform of Fig.6 is a 10kHz sinewave and the lower trace is 64  Silicon Chip the residual harmonic content measured with an 80kHz bandwidth. The result is a distortion measurement of 0.285%. Note that for an output at 10kHz, the switching hash would be at 500kHz and this would be well and truly eliminated by an 80kHz filter. Fig.7 shows the output waveform at 19.6kHz and its accompa­ nying residual distortion which has a level of 0.76%, measured with a bandwidth of 80kHz. In this case the switching hash would be at 980kHz. Finally, Fig.8 shows two waveforms at 20kHz. The top is the sinewave output and the lower trace is the accompanying square wave output. Construction All the circuit components, with the exception of the two potentiometers, are mounted on a PC board We used double-sided tape to secure the batteries but you might prefer to use battery holders fastened to the bottom of the case. measuring 149 x 71mm and coded 04102001. The component wiring diagram and the connec­tions inside the case are shown in Fig.9. While we have made provision for mounting pillars at each corner of the PC board, our method of mounting is somewhat simpler – we just supported it on the back of the rotary switch, S1. It is a good idea to check the PC board against the artwork of Fig.11 before beginning the assembly. Check for any undrilled holes or broken or open circuit tracks and fix any defects that you find. Capacitor Codes         Value IEC Code EIA Code 0.1µF  100n  104 .033µF   33n  333 .0033µF   3n3  332 820pF  820p  821 330pF  330p  331 220pF  220p  221 33pF   33p   33 Resistor Colour Codes         No. 1 1 1 2 3 2 1 Value 68kΩ 47kΩ 33kΩ 16kΩ 10kΩ 3.3kΩ 1kΩ 4-Band Code (1%) blue grey orange brown yellow violet orange brown orange orange orange brown brown blue orange brown brown black orange brown orange orange red brown brown black red brown 5-Band Code (1%) blue grey black red brown yellow violet black red brown orange orange black red brown brown blue black red brown brown black black red brown orange orange black brown brown brown black black brown brown FEBRUARY 2000  65 The connections between the PC board and the front panel hardware can be run using light-duty hookup wire. Keep the lead lengths reasonably short to maintain a neat appearance (you can use cable ties if you wish). tion correct (not upside down) before soldering the 12 outer lugs. The locking tab on the switch can now be set to position 4 (so that the switch has only four positions). This done, solder the battery leads to the switch and com­plete the wiring, as shown in Fig.9. By the way, we used a zippy box with an aluminium front panel as the frequency control is sensitive to hand capacitance. If you wish to use a plugpack instead of batteries, you will need a 9V AC plugpack and a rectifier circuit wired to provide positive and negative supplies, as shown in Fig.12. This circuit consists of positive and a negative half-wave rectifiers, each feeding a 100µF electrolytic capacitor. The extra components can be wired onto a length of tagstrip. Testing the oscillator This view shows how the PC board is supported on the back of the rotary switch. Note that this switch mounts on the copper side of the board. Begin by installing the PC pins, wire links and resistors, followed by the trimpots and IC sockets, which are optional. This done, insert the smaller capacitors, followed by the two electro­lytic capacitors which must be installed the right way around. Next, solder in the CMOS ICs. To do this, earth the barrel of your soldering iron to the 0V line on the PC board and solder the supply pins of each IC first, followed by the other pins. You can now install the two regulators. Make sure that you put each one in the correct position otherwise the circuit defi­nitely won’t work. must fit two wire links on the back of the switch as shown in Fig.10. This done, insert it in the PC board from the copper side. The lugs should be flush with the laminate side. Check that you have the orienta­ Calibrating the oscillator Rotary switch mounting The rotary switch is mounted on the copper side of the PC board (as shown in the photos) and this means that it is impossi­ble to solder the two centre pins of the switch to the PC board. Therefore, before you mount it, you 66  Silicon Chip You will need a multimeter and a frequency counter or an oscilloscope to calibrate the oscillator. Turn on the batteries or plugpack. Check for +5V at pin 7 of IC9 and -5V at pin 4. These voltages should be within 0.5V. If the voltages are correct turn off the power and insert the ICs if you used sockets. Power up again and check for +5V on pin 14 of IC1, IC2, IC4 & IC5, pin 16 of IC3, IC7 & IC8, and pin 7 of IC6. Also check for -5V on pin 4 of IC6. With the sine level control fully clockwise and the 200Hz - 2kHz range selected, you should measure about 5.6V peak-to-peak with your oscilloscope. If using your multimeter, you should be able to measure 2V RMS at the sinewave output. Using an oscilloscope or a frequency counter check that the X1 square wave frequency is the same as the sinewave frequency and that the x100 output is also correct. Fig.10: the two centre pins of the rotary switch must be wired as shown before it is installed on the copper side of the PC board. The last step is to calibrate the oscillator. Turn VR3 and VR1 fully clockwise and adjust VR2 until the sine­ wave frequency is 20.5kHz on the 2-20kHz range. Now turn VR1 fully anticlockwise and adjust VR3 until the frequency is 1.95kHz. There will be some interaction between the two presets, so you may have to make these adjustments a couple of times to get the frequencies just right. As the lower ranges are generated by digital division they will track exactly. Fig.11: here are the actual size artworks for the PC board and the front panel. If you cannot get the frequency adjustment right, set VR3 and VR1 fully clockwise and VR2 to centre. Check the oscillator frequency then alter the 220pF capacitor on pin 10 of IC1c until you are close to 20.5kHz. Then follow the calibration instruc­tions once again. If the frequency is too high, fit an extra capacitor in the holes adjacent to the 220pF capacitor. If the frequency is 20% high, add a 47pF capacitor. Conversely, if the frequency is low you will have to reduce the 220pF to 180pF or less, then perhaps fit a small SC value as described above. Fig.12: use this circuit if you wish to power the oscillator from a 9V AC plugpack. FEBRUARY 2000  67