Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman PE Mini-monitor crossover for Wavecor drivers – Part 1 Fig.2. The Wavecor TW022WA04 tweeter is small thanks to its rare-earth ring magnet. Fig.1. The PE Mini-monitor can be used for near-field listening. (Yes, the ‘speaker stands’ are house bricks!) T he LS3/5A is arguably the best mini-monitor speaker, but it is very expensive and getting the parts can be tricky. Having spent years trying to get a sound quality as close as possible to the LS3/5A, we present the PE Mini-monitor. All speakers are quite imperfect and it has been said you have to fiddle with the crossover for many iterations until the required ‘timbrel soup’ is obtained – however, the basic technical parameters have to be right first. Getting parts for speakers is always a problem, so the drivers, crossover PCBs, inductors, capacitors and even enclosure kits will all be available from the PE shop. This small speaker is designed for maximum accuracy for its size, so maximum volume is necessarily limited. Its low efficiency of around 80dB/W at 1m will mean an amplifier of minimum 25W RMS into 8Ω is required. The speaker will overload with amplifiers above 60W. The MX50 amplifier described in EPE May 2017 is ideal, but I will be offering even better solutions! Using the Universal Passive Crossover PCB The PE Mini-Monitor crossover is built on the Universal Passive Crossover PCB 62 described last month. Component numbers here refer to those on the general circuit diagram (Fig.9, PE, January 2019). This means some component numbers which are not being used will be omitted; for example, C1. Finding the drivers The crossover design cannot be started until the drivers have been selected, which means perusing many data-sheets. The B110A is thought by many to be the best small bass driver, even after 40 years, since speaker technology has moved slowly. There have been several contenders on the market, but all too often they have been deleted just as I have completed a design. After getting many samples, I found the Chinese manufacturer Wavecor to be nearest in bass quality with some modern technological improvements along with proven continuity of supply. The tweeters are also good, using small neodymium magnets rather than the large ferrite ring used in the T27, so one of these was also selected. Selection of crossover frequency Many PE constructors will be wedded to their computers, so it was thought the ideal mini-monitor should be useable either side of a video display (near-field) on a desk, as well as the optimum free-standing arrangement. Most two-way speakers are ‘phasey’ sounding (disjointed – the tweeter and woofer sound as though they are separate sources) when listened to close-up; often a minimum distance of say 1.5m is needed to get the outputs of the woofer and tweeter to integrate properly. With this design, the drivers are small and mounted as close together as possible. If a relatively low crossover point is used the acoustic centres of the drive units can be less than a wavelength apart. This prevents the lack of integration close-up. A typical computer desk set-up is shown in Fig.1. Normally, with a 120mm woofer and a 19mm tweeter, a crossover from 3 to 4kHz would be used. In this design a crossover around 2.5kHz was the target. Since the frame of the tweeter is only 65mm diameter, the vertical distance between the drivers centres is just 94mm if their frames are almost touching. The wavelength of 2.5kHz is 138mm (speed of sound (343m/s) divided by frequency) is significantly longer, so a coherent wavefront should be obtained. Resonant frequency Most 19mm tweeters have a high resonant frequency of around 1.2 to 1.8kHz, which means they cannot handle a 2.5kHz crossover. A 25mm tweeter would be fine, but then the dispersion at high frequencies would be inferior to the T27. I found Wavecor had the ideal compromise tweeter Practical Electronics | February | 2020 n 40Ω 40Ω 10Ω 10Ω 20 Hz 50 100 200 500 1k 2k 5k 10k 20k Fig.3. Impedance curve of the Wavecor TW022WA04 tweeter. The ‘hump’ shows the resonant frequency. Note that the vertical scale is linear and measures ohms (2Ω /div). 20 Hz 50 100 Bass unit Fig.5. PE Mini-monitor’s Wavecor TW022WA04 Tweeter, flush mounted. – a 22mm unit with a resonant frequency of 800Hz, designated the ‘TW022WA04’, as shown in Fig.2. It is always worthwhile plotting the impedance curves of any drivers of interest, since they show the resonant frequency clearly as a hump. They can also reveal break-up glitches, damping and inductance. The tweeter curve is shown in Fig.3, revealing a resonance of 850Hz. Note that this tweeter is only available in 4Ω, possibly to keep the coil light, which could cause crossover design problems. Luckily, it’s more sensitive than the woofer, so a series resistance can be used to prevent the overall system impedance going below 8Ω. Fig.4 shows the woofer curve with a 5-litre box, giving an 82Hz resonance with high Q. Note that resonant frequencies generally drop 5% as the speaker is used, akin to ‘running-in’. The hardest thing to replicate about the B110A is its high compliance of 3mm/N and its low 37Hz resonant frequency. The Wavecor WF120BD06 4.75-inch unit was higher in both these respects (1.5mm/N and 50Hz). Its high total-driver-system ‘quality’ or ‘peakiness’ of resonance, Qts of 0.4, and its smaller diaphragm diameter (100mm as opposed to the B110A’s 115mm) enabled a bass performance only slightly inferior to that obtained in the same box volume as the LS3/5A. The smaller cone area means the free-air resonance is raised less by the ‘air-spring’ from the confined air in the enclosure. This unit can work well in anything from about 4 litres to 9 litres internal volume. The resonant frequency will vary from 100 to 71Hz (total system Q of the of the box = Qtc = 1 to 0.6) so the LS3/5A standard and PE 9-inch-deep LS3/5A boxes are ideal. A little bass lift from the amplifier tone controls also helps at low volumes. With the large box it is possible to install a 4-inch by 1.4-inch reflex duct to give an option with more (but worse quality) bass, for those who like electronic dance music. I initially worked out the enclosure figures from a free on-line box-calculator program: http://bit.ly/pe-feb20-ao1 I did look at the Wavecor 5.5-inch and 5.75-inch drivers, but their higher 57Hz resonance and larger cone areas would mean a bigger box would be needed. The bass drivers are available with paper or paper/glass fibre-mix cones. I prefer the glass-fibre one because its slightly higher Fig.6. (left) PE Mini-monitor’s Wavecor WF120BD06 woofer; (right) the Wavecor woofer front mounted in the box (note the doped diaphragm). Practical Electronics | February | 2020 200 500 1k 2k 5k 10k 20k Fig.4. Impedance curve of the Wavecor WF120BD06 woofer. The resonant frequency has been raised by the air-cushion effect of a five litre sealed-box (shown in Fig.1). The inductive rise at high frequencies is much less than normal. moving mass gives a lower resonant frequency. Wavecor provide PDF data-sheets for both drivers, available on the Wavecor website (http://bit.ly/pe-feb20-ao2 and http://bit.ly/pe-feb20-ao3) and also for download on the February 2020 page of the PE website. Fig.5, Fig.6 and Fig.7 show the drivers in question. Pumping iron The motor system of the Wavecor unit is better than the B110A, which helps compensate for the smaller and stiffer diaphragm assembly. The voice coil has a bigger diameter of 32mm, as opposed to 25mm, although the length is the same at 12mm. The linear excursion capability is 1mm higher than the B110A because of a patented specially shaped extended pole-piece called ‘Balanced Drive’, which gives a more symmetrical current-vs-displacement characteristic, resulting in lower second-harmonic distortion. The bass performance remains very clean, until it limits abruptly. There is a technical paper on the Wavecor website (http://bit.ly/pefeb20-ao4) which covers the concept in detail, also available for download from the February 2020 page of the PE website. The B110 has a more gradual overload characteristic. At medium levels, the Wavecor has lower distortion. The B110 has a higher usable maximum output, but it costs more than twice the price. Flux modulation reduction measures, consisting of an aluminium pole-piece ring and a copper cap, both missing from the B110A, also reduce Fig.7. Rear view of Wavecor WF120BD06 woofer; note extensive venting. 63 0dB 0dB –5 –5 –10 –10 –15 –15 –20 20 Hz 50 100 200 500 1k 2k 5k 10k 20k Fig.8. Frequency response of the WF120BD06 woofer in a 5-litre box. The 10kHz peak is characteristic of a stiff cone with low damping. The crossover is designed to stop this resonance being excited. distortion. These features are reflected in the impedance curve as a reduced rise at high frequencies. Another plus point is the use of a die-cast chassis as opposed to pressed steel. This avoids ringing, I’ve often had to support the magnet of the B110 with a damped strut to avoid this effect. A rather strange feature of the Wavecor bass unit is the vented spider which reduces wind noises and aids voice-coil cooling. It is possible to see the voice-coil through the gaps (see Fig.7). Unfortunately, it is also possible for iron filings to get in. My workshop floor is littered with steel wire off-cuts from soldering resistors, which are a real hazard for speaker magnets. Overall, the suspension of the Wavecor is very low-loss (high Qm), a useful characteristic for a driver in a sealed box. Desirable curves For a successful crossover, the frequency responses of the drivers should overlap by about two octaves when mounted in the cabinet. It can be seen from Fig.8 that the bass unit goes up to around 10kHz and the tweeter goes down to 1kHz shown in Fig.9. The tweeter has the flattest response curve I have ever measured. The woofer is quite odd, with a massive dust-cap resonance at 10kHz and the usual mid-range peak at around 1.2kHz. The high-frequency peak is not as damaging as first appears, since it falls away sharply off axis. Crossover filters 20 Hz 50 100 200 500 1k 2k –20 20 Hz 50 100 Doping scandal The cone and dust cap are made of a hard stiff paper/glass-fibre mix, which ensures pistonic operation up to a higher-than-normal frequency. Unfortunately, when break-up does occur, it does so with great severity, giving rise to the 15dB 10kHz peak. I found this could be reduced to around 10dB by doping the cone with plasticised PVA coating, such as Scola Master Medium (from art suppliers). This also smoothed the response generally and lowered the fundamental resonance, giving better bass as shown in Fig.10. Unfortunately, efficiency is also reduced by a couple of dB. This doping modification is not essential, but it does make the crossover equalisation more effective. I can supply doped drivers if anyone is worried about painting. Note, only the front cone and dust cap should be painted; dope must not get on the flexible rubber surround; if it does, wipe off with a damp cloth immediately. Crossover circuit design The editor has expressly asked me not to write a thesis on crossover design, so I’ll just go through my basic procedure. Most designers use simulation programs such as LTSpice or LEAP to get started. (A precise 10k 500 1k 2k 5k 10k 20k Fig.9. Frequency response of the Wavecor TW022WA04 tweeter flush mounted in box. (No, I can’t believe it is that flat!) do both the job of crossing over and also equalisation of anomalies, so the resulting woofer filter is quite complex. 5k 200 20k 20 Hz 50 100 voice coil electrical model is even available from Wavecor). I’m an experienced, old-fashioned, analogue designer, so I dive straight in with my ‘junk-box’ of standard value parts and circuit topologies in my head that I’ve used before. However, this intuitive/iterative/empirical technique only works if you have frequency and impedance-curve plotting equipment, along with a fairly anechoic test room. First attempt, low-pass section I like to keep things simple, but not so simple that they don’t work. Because of the 10kHz peak, a first-order filter consisting of just a series coil, wouldn’t provide sufficient out-of-band attenuation. A basic second-order low-pass filter consisting of a 2.6mH inductor and 15µF capacitor was initially tried, giving the electroacoustic curve shown in Fig.11. This gave a cutoff frequency of 1.2kHz; too low. I raised the frequency by reducing the inductor to 1.2mH, giving the curve in Fig.12. There was now a hump at 1.2kHz, so I went to one of my old tricks of putting in another inductor in series with the woofer to equalise the hump. I added a resistor in parallel, which I could tune to adjust the equalisation ‘shelf’ required. I reduced the capacitor to 6.8µF to up the roll-off, giving a flat response to 2.5kHz, as shown in Fig.13 near the required crossover frequency. The resulting electrical response 200 500 1k 2k 5k 10k 20k 0dB 0dB –5 –5 –10 –10 –15 –15 –20 –20 Fig.10. Wavecor WF120BD06 woofer frequency response after doping. Dust cap resonance has dropped by 6dB. Fig.11. First attempt at second-order low-pass filter using a 2.6mH coil and 15µF capacitor. Selected frequency too low at 1.2kHz. 20 Hz 50 100 200 500 1k 2k 5k 10k 20k 20 Hz 50 100 200 500 1k 2k 5k 10k 20k 0dB 0dB –5 –5 –10 –10 –15 –15 –20 –20 Fig.12. Reducing inductor size to 1.2mH pushed frequency up, but a ‘hump’ at 1.2kHz appears. Fig.13. An extra coil with parallel resistor, plus reducing the capacitor to 6.8µF bends the electro-acoustic curve to the required shape. 64 Practical Electronics | February | 2020 L 1 L C lo w - p a s filte r R 2 E q u a l i sa t i o n n e tw o rk Ω 0dB 1 .2 m H –5 In p u t fro m a m p lifie r + R 9 D a m p in g n e tw o rk 2 W 2 m H C 2 a 6 .8 µ F C 2 b 1 0 µ F 0 V –10 L 2 Ω W o o fe r W F 1 2 0 B D 0 6 –15 + –20 20 Hz 50 100 High-pass section filte r Now we have a working low-pass C 6 R 4 C 5 section, which is always the most dif7 .5 µ F 10Ω 5 .6 µ F ficult bit, so next we have to design + the high-pass. Since the crossover + frequency is quite low for a 22mm In p u t fro m L 3 T w e e te r a m p lifie r 0 .2 2 m H tweeter, a minimum of third-order T W 0 2 2 W A 0 4 A i r co r e will have to be used, because too 0 V much low frequency input could cause distortion and possibly damage the tweeter. The tweeter has a Fig.15. Initial high-pass circuit, standard third4Ω coil, so inductor L3 in the centre order filter, but feeding a 4Ω tweeter. of the T-section is smaller than for a conventional 8Ω , and the caand circuit are illustrated in Fig.14. This pacitor is bigger. Because the crossover approach is ‘backwards’ – most designers is a bit lower than the standard 3kHz design the electrical curve first. I don’t, beused in most speakers, all the values cause it is the acoustic output that matters. are increased again. See Fig.15 for the It sounds like quick work, but to get to this high-pass circuit. The input resistor R4 point took around five hours. R 2 L 1 R 4 10Ω + In p u t fro m a m p lifie r 2 W 2 m H F e r r i t e co C 2 a 6 .8 µ F C 2 b 1 0 µ F ( b ip o la r ) 0 V C 5 5 .6 µ F 2k 5k 10k 20k sets the tweeter attenuation and its relative level to the woofer. This is a very sensitive adjustment and is dependant on room acoustics and positioning. The value can range from 12 to 5.6Ω. I found the optimum in my living room was 10Ω and in the workshop 6.8Ω. In the end, I compromised with 7.5Ω (I used a 27Ω wired across 10Ω, giving 7.3Ω). If the woofer is undoped, the resistor will have to be reduced by around 1 to 2Ω. Incidentally, 10cc’s song, I’m Not in Love, is a good test for tweeter level. If the strummed tizzy miced-up electric guitar can be heard clearly on the left and the vocal isn’t too sibilant and spitty, the resistor value is correct. Since a relatively high-value resistor is in series with the tweeter input, it means thin cheap wire can be used to feed the tweeter circuit if bi-wiring is used. The circuit in Fig.16 is not the final crossover design, but the basic curves shown in 16b to 16e are now satisfactory. Next month, we are going to do the vital impedance check to make sure it doesn’t go below 6.7Ω necessary for it to qualify as a 8Ω speaker. I’m also going to do some minor crossover additions to improve the subjective frequency response, called ‘voicing’. C 6 7 .5 µ F L 2 Ω 1k Basic circuit Ω re R 9 500 Fig.14. (left) Wavecor crossover low-pass circuit and (right) its electrical response. T h ir d - o r d e r h ig h - p a s 1 .2 m H F e r r i t e co 200 re W o o fe r W F 1 2 0 B D 0 6 + L 3 T w e e te r T W 0 2 2 W A 0 4 0 .2 2 m H A i r co r e + Fig.16a. Complete provisional crossover circuit obtained by combining the low-pass section in Fig.14 with the high-pass in Fig.15. 0dB 0dB –5 –5 –10 –10 –15 –15 –20 20 Hz 50 100 200 500 1k 2k 5k 10k 20k Fig.16b (top left) Low-pass electrical response curve displaying the effect of adding the damping network shown dotted in Fig.14 Fig.16c (bottom left) resulting electro-acoustic curves of woofer 20 Hz 50 100 200 500 1k 2k 5k 10k –20 20 Hz 100 200 500 1k 2k 5k 10k 20k and tweeter overlaid to show crossover at 2.5kHz; Fig.16d (top right) electrical response of the high-pass filter; Fig.16e (bottom right) The overall system response with damping resistor. 20k 20 Hz 0dB 0dB –5 –5 –10 –10 –15 –15 –20 –20 Practical Electronics | February | 2020 50 50 100 200 500 1k 2k 5k 10k 20k 65