Fullscreen HTML5Med Res (??MB)HTML4

By NICHOLAS VINEN While this crossover network PCB was specifically developed for the Majestic 2-way loudspeaker system featured earlier in this issue, it can be used anywhere a 2-way loudspeaker crossover network is required. It has an optional treble peaking circuit which can be switched in and out to compensate for tweeter roll-off at high frequencies and also incorporates a high power attenuator for the tweeter. 2-Way Crossover Network With High-Power Attenuator T HIS CROSSOVER works very well in the Majestic loudspeaker system, giving it a commendably flat frequency response with low distortion and excellent power handling. But it’s also suitable for other 2-way speakers systems, such as tower or bookshelf units. The component values just need to be changed to suit the driver and cabinet properties. As with the crossover network built into just about all hifi loudspeaker systems, this circuit is passive, ie, it has no ‘active’ electronic circuitry to provide the required attenuation of the drivers above and below their respective crossover frequencies. We published an active crossover network in the January 2003 issue and 32  Silicon Chip this could be configured as a 2-way or 3-way crossover. However, while active crossovers do have some advantages, they are a much more complicated approach because separate power amplifiers are required to drive the woofers, tweeters and midrange speaker (the latter being required for a 3-way system). Passive crossovers Why is a crossover network required? First, because woofers do not reproduce high frequencies and tweeters do not produce low frequencies. Second, because woofers may produce a distorted output of high frequencies and tweeters can be damaged by too much low-frequency signal. So, for a 2-way system, involving just a woofer and tweeter, we separate the audio signal from the amplifier into two frequency bands: low and high. For a 3-way system with woofer, midrange and tweeter, we separate the audio signal into three bands: low, midrange and high. This ensures that each driver (ie, woofer, midrange and tweeter) is fed only with a frequency band it can effectively reproduce. The crossover network must also set the signal levels of the two (or three) frequency bands to achieve an overall flat frequency response. Typically, the woofer is less sensitive than the midrange and tweeter, so the signals to the latter drivers must be reduced so that the siliconchip.com.au 3.3 µF output levels from all the drivers are well matched. Hence, our 2-way crossover incorporates an attenuator for the tweeter. OK, our 2-way system has a lowpass filter to driver the woofer and a high-pass filter to drive the tweeter. Such filters can be first-order, secondorder, third-order, etc. A first-order high-pass filter rolls off the signal above the corner frequency at 6dB/ octave; quite a gentle slope. A secondorder filter rolls off at 12dB/octave; a steeper slope. We are using the simplest filters, ie, first-order. When the low-pass and high-pass filters which comprise the crossover have the same corner frequency, it theoretically gives flat amplitude and power responses (ie, no peaks or dips in the output), which no other type of crossover can achieve. Another unique property of first-order cross­ overs is their ability to retain the input waveform’s shape once the tweeter and woofer outputs are combined, ‘in the air’. These ideal properties assume that the drivers have a perfectly flat frequency response, that they have perfect time alignment and that the listener’s ear is on-axis equidistant from them. That’s an unlikely set of circumstances but if you refer to the Majestic Speaker design you will see that despite this, the overall result using the first-order crossover is very good. However, first-order crossovers provide a signal roll-off that is not very steep and this means that each driver receives substantial signal content beyond the crossover point, at reduced but still audible levels. So this type of crossover is best used with drivers with a significant overlap in frequency-handling capability. For example, if you have a woofer rated for 30Hz-1.5kHz and a tweeter which will handle 750Hz-20kHz then you have one octave of overlap (750Hz-1.5kHz) and so they should work quite well with a first-order crossover. Actually, it isn’t strictly necessary that they operate over a wide range of common frequencies; what really matters is that they do not misbehave when driven with a signal somewhat outside their design range. This means that tweeters must be robust enough to accept some low-frequency signal without damage and their resonant frequency should be high enough that it is outside this overlap zone. For woofers, siliconchip.com.au 1 Ω 5W HF PROFILE S1 12 Ω 10W CON3 (R1) 12 Ω 10W (C1) + (R2) 5.6Ω 5W 5.6Ω 5W TWEETER CON4 4.7 µF CON1 L1 2.7mH CON5 INPUT – SC 20 1 4 + – + WOOFER CON6 CON2 – First-ORDER Loudspeaker CROSSOVER Fig.1: the crossover circuit is quite simple, consisting primarily of inductor L1 to act as a low-pass filter for the woofer and a 4.7µF capacitor as the highpass filter for the tweeter. Resistor pairs R1 and R2 attenuate the tweeter signal so that its output level is matched to the woofer. The remaining components form a switchable treble boost circuit. this means that they should not generate excessive distortion when driven with low-level signals above their normal upper operating frequency. Tweeter attenuator Because the tweeter is usually more efficient than the woofer, we also use a resistive divider to attenuate its signal. This can be omitted if not required. The horn-loaded tweeter used in the Majestic Speaker has an efficiency of around 109dB/W<at>1m while the woofer is 97dB/W<at>1m – and this is a very efficient woofer. You might think that we could attenuate the signal using a single resistor, ie, put an 8Ω resistor in series with an 8-ohm driver to halve the voltage level and thus provide 6dB of attenuation. However, this also increases the source impedance “seen” by the driver by 8Ω (from the very low figure provided by the power amplifier) and performance will be significantly impacted due to poor damping. By using a divider, we place a low resistance across the driver and thus keep its source impedance low. In fact the source impedance is the parallel value of the two legs of the resistive divider, typically around 2Ω. The driver is thus better damped, keeping distortion low. The resistive divider also provides more precise attenuation as it swamps the effect of the tweeter’s inductance. Power dissipation in this resistive divider is a significant issue. In the Majestic Speaker we are attenuating the signal to the tweeter by 12dB and that means 75% of the treble power delivered by the amplifier is turned into heat by the resistors. At a peak program power of 300W, that’s a lot of power to be dissipated! But there are a couple of reasons why we can get away with much lower-rated resistors. Even when driving the speaker at a peak of 300W, the average program level will typically be only a small fraction of this; maybe 10W or 20W, at most. Secondly, a good deal of that program power will be going to the woofer. With a typical recording, the energy in each octave is about half that of the octave below. So even though we using resistors with a total power rating of 30W, for home (hifi) use, these should be more than sufficient. For PA use, it would be a good idea to mount higherrated resistors on a heatsink and wire these up to the board instead, via the provided spade-lug mounting pads. Circuit details The circuit of the 2-way crossover network is shown in Fig.1. The sole component of the low-pass filter for the woofer is a series inductor, which for the Majestic Speaker and its 8-ohm driver is 2.7mH. This is a standard air-cored choke, used because air is a perfectly linear core material. Its resistance is a little over 1Ω. The rising impedance of this inductor, coupled with the (more or June 2014  33 +30 “Majestic” Speaker Crossover Response Simulation +20 +10 Relative Amplitude (dBr) 0 -10 -20 -30 -40 Tweeter response Tweeter with treble boost Woofer response Woofer with impedance equalisation Woofer with parallel capacitor Woofer with both -50 -60 -70 20 50 100 200 500 1k 2k 5k 10k 20k Frequency (Hz) Fig.2: simulated response for the Majestic Speaker crossover. The drivers are simulated as 10.2Ω/1.8mH (tweeter) and 9.3Ω/1mH (woofer). The woofer response doesn’t drop much below -10dB due to the voltage divider formed by its own inductance and the 2.7mH filter inductor. The tweeter plot has been raised by 12dB to allow for the difference in driver efficiency. 90 “Majestic” Speaker Crossover Phase Shift Simulation 60 30 Phase Shift (Degrees) 0 -30 -60 -90 -120 Tweeter phase shift Tweeter with treble boost Woofer phase shift Woofer with impedance equalisation Woofer with parallel capacitor Woofer with both -150 -180 -210 20 50 100 200 500 Attenuator design 1k 2k 5k 10k 20k Frequency (Hz) Fig.3: phase diagram for the same set-ups as in Fig.2. With the crossover as designed, the phase shift is around 90° across most of the frequency range. Circuits with faster roll-off have more phase shift. less fixed) impedance of the driver, rolls off the signal as the frequency increases. 34  Silicon Chip leading to a shelving effect, as seen in the simulated response of Fig.2. This could have been tamed using an impedance equalisation network (involving an extra capacitor and resistor) however with the Majestic Speaker the woofer’s natural roll-off combines with the crossover to provide sufficient attenuation at higher frequencies. Similarly, the high-pass filter for the tweeter is basically just a series capacitor, which is 4.7µF for the Majestic Speaker. As with the woofer, the tweeter is also an 8-ohm driver. The corner frequency (-3dB) points can be calculated as follows. For the woofer it’s F = R ÷ 2πL which gives us 472Hz with a 2.7mH inductor and 8-ohm woofer. For the tweeter it’s F = 1 ÷ 2πRC. For a 4.7µF capacitor and 8-ohm tweeter, that gives us a figure of about 4kHz. As you can see from Fig.2, the roll-off points are quite far apart but remember that the drivers themselves have some roll-off which is not shown here (as this is an electrical simulation) and these values have been chosen imperically to give the flattest response (see Majestic Speaker article for details). Given these formulae, you can adjust the components used in the crossover board as desired. We’re assuming that if an attenuator is used for the tweeter, its impedance is the same as the tweeter’s nominal impedance. The calculations below show how this is achieved. Table 1 gives some example values that could be used. Normally, you would start with similar turnover frequencies for bass and treble but experimentation may show, as with the Majestic Speaker, that changing one or both slightly can give a flatter response when the driver and enclosure characteristics are taken into account. Actually, since the voice coil is also an inductor, at higher frequencies the driver’s impedance also begins to rise, The following calculations allow you to select attenuator resistors based on the difference in driver efficiency from the manufacturer’s data. However, note that due to cabinet design etc, you may need to tweak it from there. The upper resistor in the divider is R1 and the lower resistor (to ground) is R2. The calculations are: R1 = Z x [10(A÷20) - 1] ÷ 10(A÷20) R2 = Z ÷ [10(A÷20) - 1] Where A is the required attenuation siliconchip.com.au Woofer Reactance & Filter Compensation + L1 2.7mH 8Ω INPUT – + L1 2.7mH INPUT (A) LR FILTER – 8Ω 10Ω 8.2 µF (B) LR FILTER WITH IMPEDANCE EQUALISATION + INPUT – Parts List L1 2.7mH 8Ω 22 µF (C) LC FILTER Fig.4: the basic crossover configuration is shown in (A) with two options to increase the low-pass filter roll-off for the woofer shown in (B) and (C). As mentioned in the text, the interaction between the inductor used to provide the low-pass filtering for the woofer and the woofer’s voice coil inductance leads to a shelving effect where the level applied to the woofer drops to about -12dB and then stays flat for high frequencies. We then rely on the woofer’s insensitivity to high-frequency signals to continue the roll-off for us. However, this isn’t always desirable. Some bass drivers will reproduce higher frequencies but add significant distortion. In this case, there are some ways to defeat this effect and cause the response to continue to roll off. One way is to add a so-called ‘impedance equalisation’ network consisting of a series resistor and capacitor across the woofer – see Fig.4(B). While the voice coil’s impedance rises with increasing frequency, the impedance of this network drops with increasing frequency and thus the overall impedance remains relatively stable. This prevents the shelving effect from occurring and allows the roll-off to continue, as can be seen in Fig.2 (light mauve trace). Another possibility is to change the LR filter [with the ‘resistor’ being the driver; Fig.4(A)] to a second-order LC filter, by placing a capacitor across the driver – see Fig.4(C). This requires a larger capacitor value but provides a much steeper 12dB/ octave roll-off compared to the 6dB/octave of the first-order filter. It does, however, dramatically increase the phase shift of the signal reaching the woofer and thus the phase difference between the tweeter and woofer (see Fig.3). This can cause ‘lobing’ and ‘beaming’ due to constructive and destructive interference between the audio coming from the tweeter and woofer, which adversely affects the speaker’s directivity and frequency response. That is why why we have avoided doing this. Finally, it’s possible to combine these two approaches, with a capacitor across the woofer as well as an impedance equalisation network. This gives a similarly steep roll-off to the LC filter but with more attenuation around the corner frequency and with slightly less phase shifting of the signal (light green traces). Overall, the configuration we have used has the most benign phase shift for the bass signals, with a maximum of about -45°, but it does rely on the bass driver being well-behaved at higher signal frequencies. If using this board with a different speaker design and different drivers, you may wish to experiment by adding an impedance equalisation network. in dB and Z is the driver impedance. If we plug in the figures for the Majestic Speaker of 12dB attenuation and 8Ω tweeter impedance, we get R1 = 5.99Ω and R2 = 2.68Ω. To save time, you can use this online calculator: www.sengpielaudio.com/calculator-Lpad.htm We’re paralleling pairs of resistors for reasons of power handling, so this means we chose two 12Ω 10W resistors for R1 and two 5.6Ω 5W resistors for R2. The latter gives 2.8Ω, resulting in an inconsequential error of -0.2dB. These calculations give an overall nominal impedance that’s almost siliconchip.com.au identical to that of the driver itself, in this case 8Ω. Treble peaking circuit All that’s left to describe is the treble boost step circuit. Its effect is shown in Fig.2. Essentially, it just reduces the attenuation of the resistive divider slightly, starting at about 7kHz and ultimately providing about 4dB of boost. This is designed to correct a roll-off in the response of the tweeter used in the Majestic Speaker above 10kHz. We determined by experimentation that this capacitor value is close to 1 PCB, code 01205141, 107 x 120mm 1 2.7mH air-cored inductor (Jaycar LF1330) 1 M4 x 10mm machine screw and nut 1 300mm length 0.7mm diameter tinned copper wire 6 PCB-mount 6.3mm spade connectors, 5mm pitch (Altronics H2094) (CON1CON6) OR 6 chassis-mount 6.3mm spade lugs plus M4 machine screws, shakeproof washers and nuts 1 3-pin header, 2.54mm pitch (CON7)* 1 jumper shunt* 1 SPST or SPDT toggle switch* 1 2-way cable terminated with female header plug* 4 No.4 x 12mm self-tapping wood screws 1 20 x 20 x 5mm section highdensity foam rubber or synthetic rubber material * optional component for treble peaking network – see text Capacitors 1 4.7µF polypropylene crossover capacitor (Jaycar RY6954) 1 3.3µF polypropylene crossover capacitor (Jaycar RY6953) (optional, for treble boost) Resistors 2 12Ω 10W 5% 2 5.6Ω 5W 5% 1 1Ω* 5W 5% Note: values listed are for the Majestic Speaker & may need changing for other designs – see Table 1. Additional Parts For Connecting To Speaker 1 pair long binding posts, red & black (Altronics P2004/P2005) 8 yellow 6.3mm female crimp spade “quick” connectors (Jaycar PT4707, Altronics H1842) 1 2m length heavy duty figure-8 speaker cable (eg, Jaycar WB1732, Altronics W2130) optimal and that the resistor value is not critical but it works best when it’s reasonably low, so we settled on 1Ω. Because some recordings may have excessive sibilance, thereby making June 2014  35 S1 S1 3.3 µF K 250V 5W 1 Ω J L1 2.7mH TO TWEETER – 2-Way Crossover 5W 5R6 J 10W 12 Ω J 5W 5R6 J + 10W 12 Ω J 4.7 µF K 250V Fig.5: follow this PCB layout diagram to assemble the crossover. It includes provision to connect off-board attenuation resistors via spade terminals if required for very high continuous power applications (eg, PA). Note that extra pads are provided for wire supports for four of the wirewound resistors, to help take the stress off their lead solder joints (see text). + + FROM INPUT TERMINALS TO WOOFER – – Table 1: Inductor/Capacitor Values & Associated Turnover Frequencies 8-Ohm Woofer Inductance (L1) 0.47mH 0.56mH 0.82mH 1.0mH 2.7mH 3.0mH 5.6mH 9.0mH 12.0mH Turnover Freq. 2.71kHz 2.27kHz 1.55kHz 1.27kHz 471Hz 424Hz 227Hz 141Hz 106Hz 4-Ohm Woofer 8-Ohm Tweeter Capacitance (C1) Turnover Freq. 1.5µF 2.2µF 3.3µF 4.7µF 5.6µF 6.8µF 8.2µF 10µF 13.3kHz 9kHz 6kHz 4.2kHz 3.6kHz 3kHz 2.4kHz 2kHz 4-Ohm Tweeter Inductance (L1) Turnover Freq. Capacitance (C1) Turnover Freq. 0.47mH 1.35kHz 0.56mH 1.14kHz 0.82mH 776Hz 1.0mH 637Hz 3.3µF 12kHz 2.7mH 235Hz 4.7µF 8.5kHz 3.0mH 424Hz 5.6µF 7.1kHz 5.6mH 114Hz 6.8µF 5.9kHz 9.0mH 70Hz 8.2µF 4.9kHz 12.0mH 53Hz 10µF 4kHz Table1: this table shows the inductance (L1) and capacitance (C1) values to use for various turnover frequencies. L1 ensures that low-frequencies are fed to the woofer, while C1 ensures that high frequencies are fed to the tweeter. 36  Silicon Chip high-frequency treble boost undesirable, there is provision for this network to be switched in and out. You can of course link out the switch header if you want it to be permanently in, or leave the components off if this feature isn’t necessary for the speaker you are building. We feel that with the Majestic Speaker, its sound is improved with these extra components included. Construction Fig.5 shows the PCB layout. Start with the spade lugs; we used the PCB-mounting type however chassismounting spade lugs can also be pressed into service. For the PCBmount type, there are various ways they can be fitted as there are four holes per position but we aligned them with the board edges and placed them as close to the edge as possible. Solder them in place with a highpower iron. Start with the pins on the bottom side but it’s also a good idea to ensure that there are solder fillets from the top side pad to the sides of the spade connectors too. siliconchip.com.au The 1Ω resistor has no provision for support wires and can be pushed all the way down onto the PCB if desired, as it handles relatively little power. The next step is to fit a pin header to connect S1, if you are using it. Once it’s in, install inductor L1. First, scrape the enamel off its two leads; they are supplied pre-tinned, however the tinned sections are too far from the bobbin to allow it to be soldered to the PCB. You will have to scrape them back to the point where they exit from the bobbin, then tin those sections. It’s then just a matter of mounting the inductor in place and securing it using an M4 machine screw and nut before soldering and trimming the leads. Mounting & connecting it If using chassis-mounting spade lugs, use either the single-lug type or cut off one lug from a double-lug connector. Install each one by first feeding an M4 x 6mm machine screw up through the hole in the bottom of the board, then fit a shakeproof washer, then the connector, then another shakeproof washer and finally the nut. Tighten the nut with the lug projecting out from the edge of the PCB. The capacitor(s) go in next. Bend the leads to fit the pads and push them down so they sit flat on the PCB before soldering them in place. Note that we have provided multiple pads in case you prefer to use radial types (eg, X2-style polypropylene capacitors). Polyester capacitors are not ideal as they are less linear but would probably work OK. The capacitor next to L1 must be fitted. The other is optional depending on whether you want the treble boost feature. Solder the capacitor leads on both sides of the board, assuming you’re using the specified axial capacitors. We’ve provided pads so that the wirewound resistors can be supported by sections of stiff tinned copper wire, so that if they are exposed to shock or vibration, their primary solder joints are not the only means of support. You don’t have to fit these support “trusses” but it’s probably a good idea siliconchip.com.au to do so (see photo above). They are made as follows. First, bend the resistor leads so that they fit through the holes in the PCB, then cut a length of tinned copper wire at least 100mm long, straighten it and bend it through 90° about 20mm from one end. Place this end of the wire parallel with the resistor leads, with the longer section resting across the bottom of the resistor body and with the shorter section aligned with the edge of the resistor, then wrap the longer section of wire tightly around the resistor body, going over the top and then across the bottom again. Finally, bend this end through 90° so that the remainder of the wire is parallel to the initial short section and lined up with the other side of the resistor body. You will need two support wires for each 10W resistor and one each for two of the 5W resistors. These support wires are then fed through the appropriate holes on the PCB at the same time as the resistor leads. These supported wirewound resistors should be spaced off the PCB by about 6mm. That’s done by pushing each resistor down onto a 6mm-thick spacer. You then turn the PCB over and solder the leads, along with the support wires. The resistor leads should be soldered on both sides of the PCB (not necessary for the support wires). We mounted the crossover in the Majestic Speaker as follows. First, we marked out the four mounting hole positions in the bottom of the speaker (on the opposite side of the divider from the port) and drilled these to a depth of about 10mm with a 2mm bit. We then cut a 20 x 20 x 5mm piece of high-density foam into four sections and drilled 3mm holes through the middle of each section. We then fed a self-tapping screw through each of the four PCB mounting holes and slipped the foam sections over the screws. The assembly was then lined up with the pilot holes and the screws tightened progressively until the four pieces of foam were well compressed. This provides a shockabsorbing mount for the board and also helps prevent the screws from vibrating loose. It was then just a matter of crimping 6.3mm yellow female spade connectors onto the ends of the wires from the woofer and tweeter and plugging these into the appropriate connectors on the PCB. We also made some 150mm-long spade-lug to spade-lug cables using spare speaker wire off-cuts to connect the input terminals on the PCB to the binding posts mounted on the rear panel of the speaker. If using the treble peaking switch, drill a hole through the rear panel and wire the switch up across one of the pairs of terminals marked on the PCB (ie, the middle pin and one of the upper pins). Alternatively, use a jumper shunt instead, shorting out the indicated pins to enable the treble peaking or placing it across the lower SC pins to disable peaking. June 2014  37