Fullscreen HTML5Med Res (??MB)HTML4

High performance 10STEREO GRAPHIC EQU This stereo graphic equaliser is very compact and quite cheap to build. However, it has the performance to match full-blown commercial models which are far more expensive. As well, it can be used in a wide range of applications from AC or DC supplies. I t is a very long time since a graphic equaliser has been published in SILICON CHIP – way back in 1989, in fact. The Studio Series 32-Band mono equaliser appeared in March and April 1989 and the Studio Series 20-band stereo equaliser in August and September 1989. Both these designs have been unob18  Silicon Chip tainable for many years and we have not thought to revise them because of the high cost of the rack-mounting chassis and the multi-slotted screen printed and black anodised front panels which are really too expensive to make such a project economically viable. This new graphic equaliser was prompted by a reader’s suggestion to revise our 3-band Parametric Equaliser from the July 1996 issue, since the kit for that project has also now been discontinued. However, when we looked at updating the design we were also conscious that parametric equalisers can be quite confusing to use – you never quite know how to vary the controls to obtain a desired effect. siliconchip.com.au Performance of prototype Gain:............................................................Unity Input signal with no clipping at max boost:.....up to 2.3V RMS Maximum input signal with flat response: ......up to 9.25V RMS; 4.5V RMS                   with single 15V supply Frequency reponse (flat): ..............................+0.25,-0.75dB ...............................................................10Hz-60kHz (see Fig.1) Maximum boost: ..........................................±12dB (see Fig.1) Signal-to-noise ratio: ...................................-96dB unweighted                   with respect to 2V RMS Total harmonic distortion plus noise: .............<0.002%, 20Hz-20kHz,                   22kHz bandwidth;                   typically 0.0016% (see Fig.2) Channel separation: .....................................>-60dB 20Hz-20kHz,                   90dB <at> 1kHz (see Fig.3) Input impedance: .........................................100kΩ || 100pF Output impedance:........................................470Ω Supply current: ............................................55mA typical; 110mA maximum -Octave UALISER By JOHN CLARKE By comparison, graphic equalisers are much more intuitive – you can see which bands you are boosting or cutting and it is quite easy to repeat the settings after a particular listening or recording session. Used carefully, a graphic equaliser can make a considerable improvement to overall sound quality. siliconchip.com.au It is able to smooth out the frequency response of the reproduced sound, cure peaks, dips or lumps in a loudspeaker’s response or simply subtly change the program’s tonal quality to your liking. This 10-octave unit uses an individual slider potentiometer for each octave, giving you far more detailed control than is possible with simple bass and treble controls. And of course, the settings of the slider potentiometers provide a visual graph of the equaliser adjustments with the centre position providing a flat response in the respective octave, ie, no cut or boost. A slider adjusted above centre shows the level of boost and a slider below centre shows the level of cut. This is why it is called a “graphic” equaliser. Compact design Our new 10-Octave Graphic Equaliser is very compact and can be used as a stand-alone unit or incorporated into existing equipment. So having decided to produce a new design for a graphic equaliser, we had to concentrate on the problem of reducing the cost, particularly that of the metalwork, the large and complicated PCB with all those op amps and gyrator components, and finally all those expensive slider controls. Yesteryear’s approach was not going to work. The slider control was an easy choice, even though it is a bit of compromise. Compact ganged sliders with a 45mm travel and a centre detent are now readily available at low cost and their plastic actuators mean that multiple knobs are not needed. By using ganged sliders, we have been able to drastically reduce the cost and the size of the PCB. So what was the compromise? The sliders we have selected are linear types with a value of 10kΩ and a centre detent. However, for the best noise and distortion performance we would have preferred a value of 50kΩ. Further, we would have also preferred sliders with a 4BM taper instead of a linear resistance characteristic. The 4BM taper, as used in our 1989 designs (specially sourced by Jaycar Electronics at the time), has a log/antilog resistance taper; log in one direction, antilog in the other. If we had gone to the trouble of sourcing special 50kΩ 4BM slider pots, though, the final design would have been very expensive to build. Suffice to say that we have been able to get the performance up to or better than CD standard, so the compromise is quite satisfactory. Naturally, we are using a doublesided, plated-through PCB with the 10 ganged sliders on one side and all rest of the components on the other side (pretty closely packed). However, it is not a hard board to assemble. First, most of the resistors and some of the capacitors (all with a value of 100nF, used as supply bypass June 2017  19 capacitors) are reasonably sized (easy to solder!) surface-mount components. The rest of the components are easy to solder through-hole types. Furthermore, all the SMD resistors are clearly labelled with their values; OK, you will need keen eye-sight, a magnifying glass or spectacles! And the SMD capacitors all have the same 100nF capacitance so you don’t need to worry about identifying those. All the rest of the capacitors are normally-sized MKT polyesters. There are 13 low-noise LM833 op amps and again, to keep the PCB size in bounds, we have used surface-mount types. However, they have a pin spacing of 1.27mm so they are quite straightforward to mount in place. So the combination of 10 ganged sliders and a double-sided PCB with a mixture of surface-mount and throughhole components results in a compact assembly and avoids a large, expensive PCB. But what about the problem of the expensive metalwork and a precision machined, screen-printed front panel Fig.1: the green curve shows the frequency with all controls set to the centre position, giving a ruler flat response which is only 1dB down at 10Hz and 100kHz. The red and mauve curves show the response with all sliders in the maximum boost setting and all in the maximum cut setting. Finally, two blue curves show the sliders alternately set for maximum boost and cut and these show the effective octave width of each band. CON1 (CON3) LEFT INPUT (RIGHT INPUT) IC11a (IC12a) L1 (L2) 470nF FERRITE BEAD 1k 100k 3 2 100pF 100nF 8 2.7k 1 LM833 2 4 100pF V+ V+ (NOTE: SIGNAL CIRCUITRY SHOWN ONLY FOR LEFT CHANNEL; COMPONENTS FOR RIGHT CHANNEL SHOWN IN BRACKETS) 820nF 1 F 680 V+ 220nF 3(5) 2(6) 3(5) 2(6) SC IC1a (IC1b) 31.25Hz 1(7) 2(6) IC2a (IC2b) 62.5Hz 1(7) 2(6) IC3a (IC3b) 125Hz V+ 3(5) 8 1(7) LM833 2(6) 8 1(7) LM833 4 V– V– 82k 680 15nF 4 V– 91k V+ 3(5) 8 LM833 100nF 680 33nF 4 V– 100k V+ 3(5) 8 LM833 220nF 680 68nF 4 V– 20 1 7 V+ 1(7) 4 110k 680 100nF 8 LM833 390nF CUT CUT CUT CUT CUT 680nF VR5 10k VR4 10k VR3 10k VR2 10k VR1 10k 100nF BOOST BOOST BOOST V+ V+ 100nF 100nF 100nF BOOST BOOST 10 x 100nF CERAMIC CAPS (ONE BETWEEN PINS 8 & 4 OF IC1 – IC10) IC4a (IC4b) 250Hz 100k IC5a (IC5b) 500Hz 10-OCTAVE STEREO GRAPHIC EQUALISER 20  Silicon Chip siliconchip.com.au 1 Graphic Equaliser THD+N vs Frequency 02/05/17 14:57:15 0 0.5 Graphic Equaliser Channel Separation 02/05/17 15:14:25 -10 0.2 20Hz-22kHz bandwidth 20Hz-80kHz bandwidth Signal coupled from left to right Signal coupled from right to left -20 -30 0.05 Relative Amplitude (dbR) Total Harmonic Distortion (%) 0.1 0.02 0.01 .005 .002 -40 -50 -60 -70 .001 -80 .0005 -90 .0002 .0001 20 50 100 200 500 1k 2k 5k 10k -100 20 20k 50 100 200 500 Frequency (Hz) 1k 2k 5k 10k 20k Frequency (Hz) Fig.2: the harmonic distortion performance is limited by the residual noise “floor” of the crucial gain stage in the circuit. The actual harmonic distortion is much lower. with all those slots? Well, we have dispensed with metal-work altogether! The front panel is a black screenprinted PCB with precision milled slots – it looks great. And following Fig.3: the channel separation of the graphic equaliser and the two curves show that the separation between the channels is almost perfectly symmetrical. our recent practice with smaller projects, the case is made of black acrylic which slots together very easily. It looks neat and can be used as a freestanding unit or as part of a larger installation. If you decide to build the Graphic Equaliser into a larger piece of equipment such as an amplifier or recording console, you probably don’t need the acrylic case. You can simply mount the unit in a rectangular cut-out, with the V+ IC11b (IC12b) 1 F 5 6 LM833 2 CON2 (CON4) 470 7 LEFT OUTPUT (RIGHT OUTPUT) 1 F 1nF 1M 2.7k 8 10 4 1 100pF 10 V– 22nF 680 V+ 10nF 3(5) 2(6) 3(5) 8 1(7) LM833 2(6) 1(7) 2(6) IC6a (IC6b) 1kHz 91k IC7a (IC7b) 2kHz 3(5) 8 1(7) LM833 2(6) IC8a (IC8b) 4kHz V+ 3(5) 8 1(7) LM833 2(6) 8 LM833 1(7) 4 V– V– 82k 680 680pF 4 V– 110k V+ 1nF 4 V– V– V+ 3(5) 8 LM833 3.3nF 680 680 2.2nF 4 4 82k V+ 4.7nF 6.8nF 10nF 680 CUT CUT CUT CUT CUT 47nF VR10 10k VR9 10k VR8 10k VR7 10k VR6 10k BOOST BOOST BOOST BOOST BOOST IC9a (IC9b) 8kHz 62k IC10a (IC10b) 16kHz Fig.4: this circuit shows only the left channel – the right channel is identical apart from the IC numbers (shown in brackets). siliconchip.com.au June 2017  21 R2 680 Ic IC11b (IC12b) IN 2.7k C2 Iout Vin 5 7 From IC11a (IC12a) 6 OUT Vin 10k 2.7k CUT R1 Vout Ic BOOST C1 GYRATOR Fig.5: this is the circuit of a graphic equaliser reduced to its basic essentials – with just one op amp, one slider and one gyrator. But remember that there are 10 sliders and 10 gyrators. front panel PCB over the top. All the components are on the one PCB and there is no external wiring apart from the supply leads from the on-board connector. Even the RCA input and output sockets are directly soldered onto the PCB. What could be simpler? Typical applications Our new Graphic Equaliser can be connected to a stereo amplifier or receiver in several ways. First, it can be connected in the “Tape Monitor” loop that’s still provided on most amplifiers and receivers. Alternatively, the equaliser may be connected between the preamplifier and power amplifier. Some home theatre/stereo receivers include pre-out/in connectors for this purpose. If you only have a single sound source that has line level output level (anywhere between 500mV and 2V RMS) then the equaliser input can be connected to that source output and the equaliser output connected to the amplifier input. For sound reinforcement use, you can connect the equaliser between the sound mixer output and amplifier input. In that case, connectors other than the RCA types maybe required and you may need to add a balanced input and balanced output converter on each channel. We published a suitable project to do this in June 2008. See siliconchip.com.au/l/aacv Power supply options There are three supply options; you 22  Silicon Chip Fig.6: each gyrator in the circuit is essentially capacitor C2 and the op amp and the two together work as if they were an inductor. The accompanying waveforms at right shows how the current IOUT lags VIN, just like it would for an inductor. can use a DC supply of around 1820V, a 15-16VAC plugpack supply or a centre-tapped mains powered 30VAC transformer (or equivalent supply rails in a power amplifier, mixer desk etc). Performance The overall performance is summarised in a separate panel and a number of graphs. Fig.1 has a number of coloured response curves. The green curve shows the frequency with all controls set to the centre position, giving a ruler flat response which is only 1dB down at 10Hz and 100kHz. The red and mauve curves show the response with all sliders in the maximum boost setting and all in the maximum cut setting. Finally, two blue curves show the sliders alternately set for maximum boost and cut and these show the effective octave width of each band. Note that you would never use a graphic equaliser in these extreme settings – the sound quality would be just weird. Instead, you would normally use comparatively small boost and cut settings for the sliders. For example, if your loudspeakers are a touch too bright in the 4kHz region, you might apply a slight amount of cut to the respective slider. You could not do this with a normal treble tone control because it would drastically impact the higher frequencies. Or if you wanted to lift the bass response below 60Hz, you could apply a significant amount of boost on the 31Hz band and get a much more subtle effect than would be possible with Vout Iout a conventional bass control. We stated that the overall performance was effectively CD-standard and that is backed up by the figures for signal-to-noise ratio and harmonic distortion. Fig.2 demonstrates that the harmonic distortion performance is limited by the residual noise “floor” of the crucial gain stage in the circuit (that of IC11b & IC12b). In fact, the actual harmonic distortion is well below our quoted figure of around .0016% (typical) but is masked by the residual noise. Suffice to say that the harmonic distortion of this circuit is better than can be achieved by CD and DVD players, so it will not adversely affect the sound quality of signals from such sources. Finally, Fig.3 shows the channel separation of the graphic equaliser and the two curves show that the separation between the channels is almost perfectly symmetrical. Circuit details Fig.4 shows the full circuit of the left channel of the new 10-Octave Stereo Graphic Equaliser. The right channel is identical. The IC numbering and pin numbers for the right channel are shown in brackets. We have used dual low-noise/low-distortion LM833 op amps throughout for high performance. Before going into the detail of the circuit, let us discuss the operating principles of a typical graphic equaliser. The overall circuit is effectively an input buffer amplifier, op amp IC11a, siliconchip.com.au Another view of the completed 10 Octave Stereo Graphic Equaliser in its laser-cut black acrylic case. No knobs are used – the actuators on the slider pots are quite sufficient. followed by a non-inverting op amp stage, IC11b, with the 10 slider potentiometers connected in parallel inside its feedback network. Connected to the wiper of each 10kΩ slide potentiometer is a series-resonant LC circuit; one for each octave band. Inevitably the story is much more complicated than this because there are no inductors in the tuned LC resonant circuits. Close tolerance, low distortion inductors are very expensive and bulky, as well as being prone to hum pickup. Therefore all graphic equalisers designed over the last 50 years or thereabouts use gyrators which are an op circuit which performs just like an inductor and can be connected to a capacitor to provide a series resonant circuit. Series-resonant circuit So let’s break down the graphic equaliser circuit to show just one op amp and one 10kΩ slider and one series-resonant circuit, as shown in Fig.5. Remember that there are actually 10 resonant circuits but in order to simplify matters, we will only consider one. In the simplest case, the 10kΩ slider control is set to its centre setting. In this condition, the op amp stage has unity gain and a flat frequency response and the series resonant circuit hanging off the wiper has no effect, because whatever its impedance at a particular frequency, it affects the signals at the inverting and non-inverting inputs (pins 5 and 6 here) equally. siliconchip.com.au When the slide pot is set to the boost end, the negative feedback from the output pin tends to be shunted to ground by the low impedance of the series tuned circuit at frequencies that it is resonant. Since its impedance is high at all other frequencies, this means that the feedback is only reduced over the narrow band centred around the resonance of the series tuned network. So frequencies in that band will be boosted while others will be unaffected. When the slider is set to the other extreme, to “cut”, the negative feedback is at a maximum and the series tuned circuit actually tends to shunt input signals in its resonant band to ground. This results in a reduction of gain for the frequencies at or near the resonance of the series tuned network. As you would expect, the amount of boost or cut is proportional to the slider settings, so intermediate settings give an intermediate level of signal boost or cut. Note that the circuit of Fig.5 does not show an inductor in the series resonant circuit; it shows the equivalent component, a gyrator (mentioned above). Gyrators explained Fig.6 shows the circuit of a gyrator made with an op amp. It effectively transforms a capacitor into an inductor. In an inductor, the current lags the voltage (ie, the current is delayed in phase by 90°) while in a capacitor, the voltage lags the current (by 90°), as it charges or discharges. Another way to explain this is that if you apply a large voltage step across a capacitor, a very high current flows initially which tapers off as the capacitor charges up to the new voltage. By comparison, if you apply a large voltage step to an inductor, at first the current flow remains the same as it was before, while the inductor’s magnetic field charges but over time the current flow builds as the magnetic field density increases. To understand how the gyrator circuit behaves like an inductor, consider an AC signal source, VIN, connected to the input of Fig.6. This causes a current to flow through the capacitor and through the associated resistor R1. The voltage impressed across R1, as a result of the capacitor current IC, is fed to the op amp which is connected as a voltage follower (buffer), as its inverting input is connected directly to its output. The voltage at the output of the op amp thus tracks the voltage across R1. This then causes a current to flow through resistor R2. This current, IOUT, adds vectorially with the input current IC and the resultant current which flows from the source lags the input voltage. As far as the signal source is concerned, the gyrator “looks” like an inductor, not like an op amp with two resistors and a capacitor connected to it. The inductance is given by the formula: L = R1 x R2 x C2 where L is in Henries, R is in ohms and C is in Farads. If you’re having trouble understanding how this works, consider again the effect of a large voltage step at the input. Say the input rises suddenly by 1V. This is initially coupled through C2 directly to the op amp and so its output also rises by 1V, keeping the voltage across R2 the same. Thus the current flow from the input changes very little initially; it is just the current to charge C2 which is normally much smaller than that flowing through R2 (since it’s is normally a much lower value than R1). However, as C2 charges, the voltage across R1 drops and so does the op amp output voltage, causing the current flowing from the input, through R2, to increase up to 1.5mA (1V÷680Ω) higher than it was initially. June 2017  23 REG1 7815 POWER A FUSE S1 500mA? ~ CON5 15V 470 F V+ LK1 GND 10 F 25V CT E OUT IN BR1 W04 T1 47k + – LK2 A 15V 470 F ~ N IN POWER SUPPLY CONFIGURATION WITH A CENTRE-TAPPED TRANSFORMER 10 F GND 25V  LED1 LK3 K OUT REG2 7915 (OPEN) V– (IC13 NOT INSTALLED) REG1 7815 AC PLUGPACK ~ CON5 ~ ~ OUT IN BR1 W04 POWER S1 GND 470 F V+ LK1 10 F 25V 47k + – LK2 A 470 F ~ IN POWER SUPPLY CONFIGURATION WITH AN AC PLUGPACK 10 F GND 25V  LED1 LK3 K OUT REG2 7915 (OPEN) V– (IC13 NOT INSTALLED) REG1 7815 IN BR1 W04 POWER S1 ~ CON5 DC + SUPPLY IN – 470 F OUT (OPEN) GND 10 F 25V 22k 10k (OPEN) + – V+ LK1 LK2 A  LED1 ~ 10k LK3 K POWER SUPPLY CONFIGURATION WITH A DC SUPPLY V– 100nF W04 – + ~~ 78 1 5 LED K A GND IN GND 7 91 5 OUT IN GND IN OUT Fig.6: the three power supply variations, which allow you to operate from a mains transformer with centre-tapped secondary (top), a plugpack or similar mains transformer without a centre tap (centre) and a DC supply, such as might be available in existing equipment (bottom). Note that while a BR1 bridge rectifier is used (for convenience) in the two lower supplies only some of its internal diodes are used (unused diodes greyed out) – you could substitute 1N4004 diodes if you wish for those diodes used. As described above, this behaviour is very much the same as if an inductor was connected instead of the gyrator. Building a series resonant circuit To make the tuned LC circuit shown in Fig.5, all we need do is to connect a capacitor in series with the input to Fig.6. The result is a circuit with a dip in its impedance around a specific frequency. The “Q” of each gyrator is determined by ratio of R1 and R2. Note from the formula above that if you double the value of R1 and halve the value 24  Silicon Chip of R2, the simulated inductance does not change. The same is true for the opposite, ie, halving the value of R1 and doubling the value of R2. But the “Q” does change. If you think about the resonant circuit’s impedance like an inverted bell curve, the “Q” relates to the width of the bell. So if you were to increase the value of R2 and proportionally decrease the value of R1, you would reduce the “Q” and thus broaden the bandwidth of the filter. Note that there are limits to this. You don’t want to make the value of R1 too L CH GND 100 1 3 8 IC13a 2 IC13: LM833 100 R CH GND 5 7 IC13b 4 6 100 F low or else the error current through it could overwhelm the current through R2 and the gyrator would no longer be a very good simulation of an inductor. You don’t want to make the value of R2 too low either, as you will eventually reach a point where the op amp is no longer able to drive such a low load impedance and it will run into current limiting. And changing the value of R2 also affects the minimum impedance of the resonant circuit which may require changes to other circuit components to avoid reducing performance. siliconchip.com.au The value of series capacitor C1 also controls the “Q”; you can change the value of C1 without affecting the centre frequency as long as you change the value of the simulated inductor so that the product remains the same (by changing any of R1, R2 or C2). Higher values for C1 result in lower “Q” and vice versa. However, adjusting the “Q” with R1 and R2 is generally easier. The values in our circuit set the bandwidth of each slider to approximately one octave. You can see the degree of overlap provided from the red and mauve curves in Fig.1. We could have provided more overlap by increasing the values of R2 in our circuit, and reducing the R1 values (which differ for each band) proportionally, however this would also increase the interaction between adjacent bands. Back to the equaliser So remember that we have one op amp buffer stage IC11b, with 10 slider pots connected inside its feedback loop. The wiper of each slider is connected to one of the series-tuned circuit described above. Each is tuned to a frequency that is double that of the last, to provide octave bands. Refer to the main circuit diagram in Fig.4. This shows just the left channel of the stereo equaliser, with one gyrator circuit repeated 10 times, with different values for R1, R2 and C2. Looking at the top left-hand side of the circuit, the input signal is applied to CON1 and passes through a ferrite bead which acts like an inductance to attenuate any radio signals. A 470nF capacitor blocks any DC voltage while a 100kΩ resistor provides a charging path for the that capacitor and “grounds” the signal. An RC filter comprising a 1kΩ resistor and 100pF capacitor provides further high frequency filtering. Op amp IC11a buffers the input signal, giving it a low impedance, for the following equaliser circuitry comprising IC11b, the sliders (VR1-VR10), IC1IC10 plus associated components for the gyrators. The output signal of the graphic equaliser appears at pin 7 of IC11b and this is fed via a 470Ω resistor and a 2µF DC blocking capacitor (using two parallel 1µF capacitors) to the output at CON2. The 1MΩ resistor to ground sets the DC level for the output signal siliconchip.com.au Here’s a sneak peek at the laser-cut acrylic flatpack “case” mentioned in the text which significantly reduces the cost of building the Graphic Equaliser – and adds to the professional appearance. The pieces slot together to form a very smart-looking case in piano-finish black with white marking. We’ll show how this goes together – and how the PCB fits in place – in part two next month. while the 1nF capacitor shunts any out-of-band high frequency noise to ground. The 470Ω resistor sets the output impedance of the equaliser, while the 2µF output capacitor and 470nF input capacitor set the low frequency -3dB point of the entire circuit to about 4Hz. Potentiometer value doesn’t affect gain One thing to note about the equaliser circuit which may not be obvious is that if you changed the potentiometer resistances to another value, the output level and frequency response would not change but the noise performance might. Imagine that all the slider pots are centred for the moment and consider each tuned circuit as having a low impedance (since white noise exists over a wide range of frequencies). This means that half of each slide pot is effectively connected between pin 5 of IC11b and ground (with a 10Ω resistance in series). The impedance of ten 5kΩ resistances in parallel is 500Ω; add the 10Ω to get 510Ω. This 510Ω forms a divider with the 2.7kΩ resistor at the output of IC11a, providing a signal attenuation of 0.16 times (510Ω ÷ [2.7kΩ + 510Ω]). Now, IC11b has a 2.7kΩ feedback resistor and it also forms a divider with the other half of all the slide pots in parallel, again 2.7kΩ/510Ω. But because it’s in the feedback loop, it provides gain, not attenuation; 6.3 times in fact. Since 0.16 x 6.3 = 1.0, therefore, the gain from input to output of the equaliser is unity. If you change the potentiometer values to say 50kΩ, then you end up with an attenuation of 0.48 (2.5kΩ ÷ [2.7kΩ + 2.5kΩ]) and a gain of 2.08 times (2.7kΩ ÷ 2.5kΩ + 1), again giving 0.48 x 2.08 = 1.0. So the gain is still unity. So the lower the slide pot values, the more the input signal is attenuated and the more gain is applied later to compensate. Unfortunately, though, that gain also applies to any noise in the circuit. Thus, 10kΩ pots result in three times (6.3 ÷ 2.08) as much noise as if we were using 50kΩ pots, or a degradation in signal-to-noise ratio of around 9.5dB. But as we said earlier, 50kΩ slide pots with a centre detent are more expensive and harder to get. As the performance with 10kΩ pots is pretty good, we feel that this is a reasonable compromise. Power supply options As already noted, there are three power supply options and these are depicted in Fig.7. You can use a centre tapped 30V transformer, a 15-16VAC plugpack or a DC supply of up to 20V. There are two ground/earth connections shown on the circuit with different symbols for each. One is the ground for the power supply, signal inputs and signal outputs. The second is the ground reference signal for the op amp circuitry. The two are connected directly together when using a ±15V (AC-derived) supply. This is shown in the dual supply section of the circuit, where LK1 and LK2 connect the grounds together. The power supply ground is connected to the centre tap of the transformer and is also the ground for both REG1 and REG2. These regulators provide the +15V and -15V supply rails and receive voltage from the full wave rectifier (BR1) and the raw rectified DC June 2017  25 Parts list – Graphic Equaliser 1 PCB coded 01105171, 198 x 76mm (SILICON CHIP online shop Cat SC4279) 1 front panel PCB 198 x 76mm (SILICON CHIP online shop Cat SC4280) 1 Acrylic case and hardware to suit (optional) 10 dual ganged 45mm travel 10k linear slider pots (VR1-VR10) 2 vertical PCB mount white RCA sockets (Altronics P0131) (CON1,CON2) 2 vertical PCB mount red RCA sockets (Altronics P0132) (CON3,CON4) 1 3-way PCB mount screw terminals with 5.08mm spacing (CON5) 2 5mm long ferrite RF suppression beads (L1,L2) Semiconductors 12 LM833D SOIC-8 op amps (IC1-IC12) 1 5mm high brightness blue LED (LED1) Capacitors (through hole 5.08mm pitch, all 5% tolerance except for surface mount types) 6 1µF MKT polyester 2 820nF MKT polyester (Rockby Electronics #32693) 2 680nF MKT polyester 2 470nF MKT polyester Acrylic case parts 2 390nF MKT polyester 1 Acrylic case 211 x 89 x 40mm 4 220nF MKT polyester 1 SPST rocker switch (Altronics S3210, 4 100nF MKT polyester Jaycar SK0984) (S1) 12 100nF X7R ceramic ^ 1 panel mount 2.1 or 2.5mm DC socket 2 68nF MKT polyester to suit supply plug 2 47nF MKT polyester 1 15mm length of 5mm heatshrink tubing 2 33nF MKT polyester 1 20mm length of 10mm heatshrink tubing 2 22nF MKT polyester 4 6.3mm long M3 tapped spacers 2 15nF MKT polyester 4 25mm long M3 tapped spacers 4 10nF MKT polyester 4 3mm nylon washers 2 6.8nF MKT polyester 4 15mm long M3 screws 2 4.7nF MKT polyester 6 10mm long M3 screws 2 3.3nF MKT polyester 2 M3 nuts 2 2.2nF MKT polyester 4 1nF MKT polyester 2 680pF MKT polyester 6 100pF ceramic Resistors (0.25W 1%; # = metal film; ^ = 1206 thin-film surface mount) 2 1MΩ# 2 100kΩ# 4 2.7kΩ# 2 1kΩ# 2 470Ω# 4 110kΩ^ 4 100kΩ^ 4 91kΩ^ 6 82kΩ^ 2 62kΩ^ 20 680Ω^ 4 10Ω# AC supply 2 2-way pin headers with 2.54mm spacings (LK1, LK2) 2 shorting blocks 1 W04 1.2A bridge rectifier (BR1) 1 7815 positive 15V regulator (REG1) 1 7915 negative 15V regulator (REG2) 2 470µF 25V PC electrolytic 2 10µF 16V PC electrolytic 1 47kΩ resistor^ DC supply 1 2-way pin header with 2.54mm spacing (LK3) 1 shorting block 1 LM833D SOIC-8 op amp (IC13) 1 W04 1.2A bridge rectifier (BR1); 1N4004 diodes may be substituted (see text) 1 7815 positive 15V regulator (REG1) or 7812 12V; or no regulator (see text) 1 470µF 25V PC electrolytic 1 100µF 16V PC electrolytic 1 10µF 16V PC electrolytic (not required if REG1 not used) 1 100nF X7R ceramic ^ 1 22kΩ resistor^ 2 10kΩ resistor^ 2 100Ω resistor^ 26  Silicon Chip is filtered using 470µF capacitors. One capacitor is for the positive supply and the other for the negative supply. Power LED (LED1) lights with voltage applied between the +15V and -15V supplies and is supplied current via a 47kΩ resistor. You can use a 15-16VAC plugpack instead of a centre-tapped transformer. This connects to CON5 between the 0V and an AC terminal of CON5. The bridge rectifier then half-wave rectifies the input AC voltage. Two of its internal diodes are thus unused, and are shown shaded. The resulting ±15V supply rails then run the circuit. For a DC supply, the positive voltage is applied to one of the (normally) AC inputs and the negative connection to the 0V terminal of CON5. Bridge rectifier BR1 then operates as if it were a diode, providing reverse polarity protection (the other three internal diodes are unused and thus are shaded in Fig.6). For input voltage above about 18V, you can use a 15V regulator for REG1, as with the AC supply options. If the input DC supply is less than this, use a 12V regulator (7812). With a supply voltage below 15V, REG1 should be left out, and its input and output terminals shorted, so that the external supply runs the circuit directly (but via BR1). When using a DC supply, there is no negative rail available and so REG2 is left off. LK3 is fitted to connect the V- supply rail to the negative side of the external supply (ie, 0V). LK1 and LK2 are left open. As there is no negative rail, all signals to the op amps now must be biased at half supply so that there will be a symmetrical signal swing. The half supply voltage rail becomes the op amp signal grounds. This is provided by additional op amps IC13a and IC13b. A half supply rail is derived from two series 10kΩ resistors across V+ and V- that are bypassed with a 100µF capacitor, to remove supply ripple. Op amps IC13a and IC13b buffer this half supply for the two channels. The signal grounds are separate to minimise crosstalk between channels. IC13 can be left off when using an AC supply. Construction That’s it for this month. Next month we will go over the details for assembling the PCB and case, putting it all together and getting it up and running. SC siliconchip.com.au