Fullscreen HTML5Med Res (??MB)HTML4

A 2021 variation on an old theme . . . The Bass Block If you’re building a home theatre system, or want to listen to music with small monitor or tower speakers (because you don’t have room for huge ones, perhaps), then this subwoofer is for you. It’s compact and easy to build, but it pumps out plenty of bass to fill in the gaps left by smaller speaker systems. Virtually all music and movies can benefit from a healthy dollop of low-end sound! S ubwoofers have become collisions. So ideally, you want a commonplace in recent Features & specifications sound system which doesn’t just Frequency response: 40-100Hz, ±3dB; decades. There are sevdie off below 50Hz. 25-150Hz, ±5dB eral reasons for this: One is the Impedance: Traditional speaker designs nominally 4Ω Ω popularity of home theatre sys- Dimensions: (whether sealed or bass reflex 240 x 272 x 396mm tems with 5.1 surround sound Material: boxes, or more exotic designs 16mm thick MDF (where the .1 refers to the sublike horn-loaded or transmission woofer). Another is the increaslines) all have similar difficulty ing trend towards small speakers which are less obtrusive in reproducing this bottom octave without large drivers in a home setting. and enclosures. This is especially true where high sound Compact speaker systems (and many larger ones) tend pressure levels (SPL) are needed. to have a bottom end roll-off in the region of 50Hz. While Speaker manufacturers have responded by developing much of the satisfying bass components of music is in the drivers with extremely long excursions to “move more 50-60Hz range, there is still plenty below this level. For air”. Unfortunately, these drivers still tend to be large and example, the bottom A on an 88-key piano with modern expensive. tuning has a fundamental frequency of 27.5Hz. In recent decades, mathematicians and audio/acoustic For speakers with a -3dB point of 50Hz, at least half an engineers have developed new speaker enclosure configentire octave will be severely diminished, and the funda- urations which enable these bass frequencies to be repromental of the bottom note perhaps not heard at all; only duced in much smaller physical volumes. the overtones and harmonics. One such design, implemented by Julian Edgar, was the It isn’t just classical or piano music either; other types of “Bass Barrel”, presented in the August 1997 issue of SILImusic which have a lot of content in the 20-50Hz range in- CON CHIP – (see siliconchip.com.au/Article/4846). clude reggae, hip-hop, rap, rock and pop. And action movIt used a “Compound Isobaric 6th Order (A) Bandpass ies make good use of the lowest octave, Double Vent” enclosure. This type of cabBy Nicholas Dunand inet employs two drivers mounted faceto reproduce sounds like explosions and 68 Silicon Chip Australia’s electronics magazine siliconchip.com.au Behold the Bass Block, in all of its blocky magnificence! It is made from MDF, which you could just leave “natural”, but if you rout the edges and corners and paint it like this one, it looks a whole lot better. You could also glue speaker carpet to the outside (as was done with ye olde Bass Barrel). Read on for more details on how I achieved the finish shown. to-face in a ‘push/pull’ (out of phase) configuration, with each driver working into separate volumes with different vented tunings. The net effect of the chambers interacting is an acoustic bandpass response, where the upper roll-off, lower rolloff and passband frequency response can be manipulated. This is particularly useful for subwoofers. The Bass Barrel design has some advantages; chiefly, it is small and cheap to build. It used a novel construction technique that made building it much easier for people with limited facilities. I built a couple of these subwoofers (as conventional rectangular MDF boxes) for two small sound systems, and they were very effective. Fig.1: the predicted response of the subwoofer design, produced by “Speaker Box Lite”. The goal was to design a small sub with a useful response up to at least 100Hz, and down to as close to 20Hz as possible. With a free version of the iPad speaker design app “Speaker Box Lite”, and using the original design as a starting point, I set about investigating the substitution of these new C3055 drivers. The design goals were: 1. Keep the boxes as small and unobtrusive as practical, with the smallest footprint. 2. Obtain the lowest possible bass extension. 3. Cross the subs over at around 90-100Hz to relieve the Tannoys of some of the bass demands. 4. Be able to take advantage of “room gain”, managed with equalisation and via the crossover. Initially, I plugged the Thiele-Small parameters of the new drivers and the original enclosure dimensions into the software. The predicted response was not a beautiful thing, so I started experimenting with different chamber volumes and ports. After many iterations using common sizes of PVC pipe for the ports, I settled on the following design. The total internal volume is 20L in two chambers: one of 15L, with a 210mm length of 32mm inner diameter electrical conduit for the port, and one of 5L, using a 180mm length of 63mm internal diameter plumber’s pipe for the port. The box is made of 16mm MDF with both ports facing forward. In the course of testing, I ran simulations on larger box sizes. One design produced predicted bottom-end extension flat within ±1dB down to the mid-20Hz region. I built a test box, and the measured response proved that it was delivering well down in the predicted region. The internal volume of this design was 36 litres, but in the end, I rejected it as simply being too large for me. The predicted response from the software for the configuration I selected is shown in Fig.1. After building a test box and measuring in a ‘free-air’ environment, the measured response is shown in Fig.2. It isn’t precisely as predicted, but close. Note that this measured response has 1/6 Fig.2: the actual ‘free-air’ response of the test sub built to the specifications used to produce Fig.1. While not an exact match, it’s pretty close and certainly meets the design goals. The response changes somewhat when the sub is placed within a room. Fig.3: here is the room response, and by comparing it to Fig.2, you can see the standing waves created at certain frequencies by sound waves reflecting off hard surfaces. This results in a faster high-end roll-off but also a useful amount of low-end boost. Updated version Having recently acquired a pair of Tannoy ‘bookshelf’ studio monitors for another system, I decided to make another pair of stereo subs to go with them, to fill out the missing bottom octave. Referring to Altronics catalogue for the original drivers used in the Bass Barrel (“Redback 6.5-inch woofers”, Cat C3086), I found they were no longer available. There is, however, a ‘replacement’ driver, the 165mm (6.5”) 30W Woofer / Midrange Polypropylene Speaker (Cat C3055). This driver has advantages and disadvantages compared to the original. It has a reduced power handling capacity, so the maximum possible SPL is lower. If you want to build a subwoofer for a large home theatre set up, and have the plasterboard crack whenever a Star Destroyer rumbles overhead, this may not be for you. On the other hand, the driver parameters are more suited to this application, allowing a deeper bottom end extension than the original design. So it’s not that this design is bad for home theatre use; in fact, it is very well suited, just at more moderate levels. (Your neighbours can thank us later!) Design process siliconchip.com.au Australia’s electronics magazine January 2021  69 Fig.4: here, the response from the two ports is shown, along with the overall response of the subwoofer. This gives you an idea of how the two separate cavities and ports contribute to the extended flat response of the subwoofer. Fig.5: the response that’s possible from this subwoofer with equalisation applied. It is now mostly flat from 24Hz up to just over 100Hz; an excellent result for a sub this small! octave smoothing applied. This is a nice, clean response. On the face of it, the response is not ideal due to the gradual and increasing roll-off at the bottom end. However, it is only down by 6dB at 30Hz and about 9dB at 25Hz. This is less of a problem than it appears. the order of +6dB of ‘room gain’ at 25Hz, which effectively enhances the raw bass performance of speakers. Unfortunately, the same reflections which can enhance the bass also interact with the direct sound coming from the speakers, producing what are commonly (probably incorrectly) referred to as “nodes”, where the amplitude of the sound waves add, and “nulls” where they cancel out. The actual result is entirely dependent on the speaker, its placement, the room size and shape and the types of surface treatments (eg, carpet, timber or tiles). The result is that it is often difficult to predict and control the room nodes. I placed the stereo subs in my room and measured the response at the listening position, which is shown in Fig.3. Again, this measurement has 1/6 octave smoothing applied, and all the good and bad results of room effects are plain to see. While room nodes at 40Hz and 60Hz are a problem, the worst peak is only +6dB. Moving the speakers would likely change the response considerably. Depending on the phase relationships at these points, it may be possible to cancel out the nodes. Room gain At mid-to-high frequencies, the propagation of sound from speakers is increasingly directional. This is commonly referred to as a “two pi” response. However, at lower frequencies, the sound propagation becomes more omnidirectional or spherical, referred to as a “four pi” response. There are several consequences of this characteristic. The first is that it becomes less apparent where the sound is coming from, and the speaker placement becomes less critical for the stereo image. The second is that the very long wavelengths at these frequencies pressurise the room to a certain extent, and interact with the room boundaries (especially the floor, where subs are typically placed). The net effect of this is bass boost, which increases as the frequency drops. It’s fairly typical to get something in 396 396 272 240 272 240 240 240 240 396 240 396 240 900 x 9 00 SHEET FOR A SINGLE UNIT 240 Fig.6: the easiest way to cut the 16mm panels for the subwoofer(s) is to cut three strips from a 900x900mm (or 900x1800mm for two subs) sheet of MDF, then cut the strips into the lengths shown. 70 Silicon Chip Fig.7: the basic layout and dimensions of the subwoofer. The hole which is used for mounting both drivers is 148mm in diameter and comes within 18mm of the edge. The upper port hole is approximately 68mm in diameter while the lower port hole is 40mm in diameter (if using the recommended pipes). Australia’s electronics magazine siliconchip.com.au Fig.8: start by glueing and screwing (or nailing) the first three panels together like this. Fig.9: next, add the inner baffle and base panel. On the plus side, the room gain has raised the measured response so that it is now only about -4dB at 25Hz. Subjective listening tests bear this out. Observing a Spectrum Analyser while listening to music will reveal that not a lot of recorded music actually has much full-level content in the range from 20Hz to 25Hz. However, deep bass from 25-50Hz is often present in rock, dance and reggae music. Where it is, the boxes produce a satisfying level of tight and clean deep bass at any volume level that I listen to, and when integrated with the main speakers, the overall full range response is rich and smooth. When the boxes were driven with higher levels of pure low-frequency sinewaves, there was some “chuffing” or port noise coming from the small diameter port, but in real-life use (eg, listening to music), it was inaudible to me. More for interest than anything else, I measured the output from each port separately. This is shown in Fig.4. As expected, the low-frequency response from the larger chamber with the small port, while the higher frequencies come from the smaller chamber with the large port. Crossover and equalisation Although acoustic bandpass designs like this have an inherent high-frequency roll-off, this is not good enough to use as the crossover. These drivers have a rated response up to 4kHz. Without a crossover, these higher frequencies are audible from the finished sub. This would lead to undesirable interaction with the main speakers, so signals at these frequencies need to be removed. My Tannoys have a response down to around 50Hz, but I wanted to relieve them of some of the bottom end effort, so I aimed to cross them over at about 90Hz. So the subs had to reproduce up to at least that frequency. As mentioned earlier, the directionality of low-frequency sounds is less apparent than higher frequencies, but at 90Hz, this effect is certainly not absolute. Directional information in the music content is audible at 90Hz. So for hifi use, I needed a stereo pair of subs. The cost of these drivers is so modest that it hardly broke the bank, and the upside is a doubling in the sound output enables higher ultimate SPLs without overdriving the subs. I am using miniDSP signal processors to cross over the subs to the main speakers, and also to equalise the speaksiliconchip.com.au ers and the room. There are two versions of this device, the standard miniDSP (siliconchip.com.au/link/ab4c) and the HD miniDSP (siliconchip.com.au/link/ab4d). You could also use our DSP Active Crossover and Equaliser (May-July 2019; siliconchip.com.au/Series/335) or the 3-way Active Crossover for Speakers (September & October 2017; siliconchip.com.au/Series/318). The miniDSP units provide many options for crossing over and parametric equalisation of both its inputs and outputs, to help manage speaker and room idiosyncrasies. Applying a modest amount of correction with these units can easily yield a corrected free air response like that shown in Fig.5. This is an advantage of a design with a long shallow roll-off rather than one which is initially deeper, but drops off steeply. Construction Refer to the parts list to gather the required supplies. Fig.6 shows a cutting diagram to help you cut the panels required. The 240 x 240mm sheets are for the top and bottom of the enclosure, plus the internal baffle. The sides are 240 x 396mm while the front and back pieces are 272 x 396mm. If you haven’t already, cut the conduit and pipe to length. Fig.7 shows what we are aiming to build. The small port dimensions I chose were optimised to Parts list (to make one subwoofer) 1 900 x 900mm sheet of 16mm thick MDF 1 210mm length of 32mm internal diameter electrical conduit (40mm outer diameter) 1 180mm length of 63mm internal diameter PVC (plumber’s) pipe 2 165mm (6.5in) 30W polypropylene woofers [Altronics Cat C3055] 1 pair of panel-mounting speaker terminals 1 1m length of twin conductor speaker wire 4 spade crimp terminals, to suit speaker wire thickness 1 roll of acrylic speaker dampening material [Jaycar Cat AX3694] Nails, wood screws, construction adhesive, paint as required Australia’s electronics magazine January 2021  71 Fig.10: then add the side panel and glue in the port pipes (if you haven’t already). Make sure the joints are well sealed. provide the long, shallow roll-off I was pursuing. I made the ports from thick-wall 40mm outer diameter electrical conduit. Although it is quite cheap, it is typically sold in 4m lengths (for around $9), leaving a lot left over. An alternative is to use 40mm plumber’s PVC pipe with an inner diameter of 38mm. This has the advantage of being available in short lengths from hardware stores, and the larger diameter would likely reduce the risk of port noise. However, this small difference in diameter produces a notable difference in response, with a flatter initial (higher frequency) curve, but a steeper roll-off. Taking into account room gain, this would likely result in a peak at around 30-40Hz, which is not so good for HiFi use, but may well suit a home theatre application. Another option is to use PVC pressure pipe with an inner diameter of 30mm. This produces a predicted response closer to my chosen solution, but the smaller diameter of the pipe risks increasing port noise under higher SPLs. I chose the dimensions of the box to provide both a small footprint and to simplify cutting. The pieces come from three MDF strips. After cutting the strips, you can then cut the individual pieces to length. If you have limited facilities for cutting straight lines, cabinetmakers and even timber suppliers will sometimes cut pieces to order, or perhaps just the strips if you have a drop or slide saw to cut the lengths. The sheet sizes specified are commonly available at hardware stores, and there will be a little left over, but not much. The general construction procedure is: 1. Cut the individual rectangular pieces. 2. Use a jigsaw to cut the 148mm driver hole offset in the baffle, the port holes (note: not portholes!) in the front panel and (if appropriate), a hole for the speaker terminal in the back panel. 3. Drill the holes for the speaker mounting bolts in the baffle. 4. Cut the port tubes to length, glue them into the front panel with construction adhesive and put it aside to cure. 5. Starting at one end, glue and screw the first three pieces together flush, as shown in Fig.8. If you have a small 72 Silicon Chip nail gun, putting a couple of tacks in first will hold everything in place until the screws go in. 6. Fix the baffle in place, then the other end piece (see Fig.9). Remember to mount the baffle with the offset driver cutout closest to the still-open side 7. Mount the drivers and wire them up to each other (out of phase) and to the speaker terminals. I recommend applying a small amount of sealant to the rims of the speaker and around the mounting holes, as well as the hole where the lead passes through the baffle. You can wait until the facepiece is mounted to do this, but it’s easier now. 8. Fix the face panel (see Fig.10). 9. Place some polyester wadding around the inside surfaces of the two chambers. 10. Fix the last side in place or, if you want to make it removable as with my test boxes, apply some thin foam as a gasket and screw the side on without glue. Aesthetics There are various options for finishing the boxes. Automotive type carpet was particularly practical for the original Bass Barrel because the cylindrical shape was relatively This test speaker was built with a thick piece of acrylic in place of one of the MDF side panels. I did this so that I could observe the driver excursion, to make sure it was not excessive. I don’t recommend that you do this, but you could if you really want to. Australia’s electronics magazine siliconchip.com.au Panel: Making the measurements The software I used for measuring the subs’ actual response is Room Eq Wizard (REW). This excellent, comprehensive software produces a sinewave swept from 15Hz up to 20kHz and samples the response picked up by a microphone (or sound level meter). It can then apply many analytic processes to the measured result, not just frequency-domain measurements. The process involves first measuring the ‘native’ response of the speakers, then measuring the whole system response in a real-world room setting. The native response shows what the speakers would produce in a completely neutral environment, but in real life, of course, this never exists. Acoustic engineers make these measurements in an anechoic chamber where all reflections and external interference can be effectively eliminated. Without an anechoic chamber, the unwanted influences can be reduced in a couple of ways. One is to make the measurements in the most open environment possible. Making the measurements outside in the middle of a sports field would go a long way to eliminating the effects of room interference, but is hardly practical. Many of the response graphs in this article were made in a very large empty workshop, with the speakers about 2m above the ground. These are the measurements I have referred to (perhaps erroneously) as ‘free air’. Although it is certainly not equivalent to an anechoic chamber, it is as near as I can come for practical purposes. Another way to reduce unwanted interference is to make measurements ‘nearfield’. This involves placing the microphone quite close to the speakers and making the measurements at modest SPLs. In a nearfield position, the relative SPL coming from the speaker is much higher than that of the reflections coming from the environment. Consequently, impingement of the interference on the measurement is largely reduced. For these measurements, I used a calibrated microphone from miniDSP (siliconchip.com.au/link/ab4e). Unlike professional microphones costing many hundreds or thousands of dollars, these USB microphones are cheap! The microphones don’t need to be fancy (or accurate for that matter), they just need to be able to sample the full spectrum of audible sound, and be themselves measured. Each microphone is supplied with a siliconchip.com.au matching calibration file, which is then loaded into the measurement software, to adjust the incoming measurements accordingly. Note that in all the response plots, you can disregard the varying absolute amplitude measurements on the left Y-axis of the graph. These simply reflect different measurement volumes at various locations and points in the room. What we are really interested in is the relative flatness and smoothness of the response. The small-scale variations in the curve are dependent on the measured frequency response, of course, but also the ‘smoothing’ applied to the response graph. Speakers never reproduce all frequencies equally, and room effects produce responses similar to comb filtering, where nodes and nulls cancel or enhance particular narrow frequency bands. These can easily be heard if you put a sine wave generator with uniform amplitude into the system and very slowly sweep through the frequencies. Many dips and peaks can easily be heard as changes in volume as the frequency changes. In many cases, though, these narrow bands are never heard in real-world listening to music. REW can take up to one million samples per sweep of the audio spectrum (although I settle on 512,000). This means that it can resolve tiny frequency band variations which might not be at all audible. For practical use, the response plots can be ‘smoothed’ for different purposes. The software offers smoothing options from one octave (which produces a curve that manufacturers might like to present to customers) through to 1/48th of an octave, which reveals many artefacts which might not be audible. There are also specialised options like “psychoacoustic smoothing”. Plots of the sub’s response with various smoothing options are shown in Figs.a-e. Throughout the article, I’ve used 1/6 octave smoothing, which reveals plenty of detail without showing extraneous information which is probably not relevant. The 1/6 octave smoothing comes out looking much like the psychoacoustic option. Note that the psychoacoustic smoothing reduces some of the low-frequency peaks and troughs visible with 1/6th octave smoothing, and accentuates some in the higher frequencies. Without knowing the algorithm used to make this smoothing, it’s probably fair to say this is intended to provide a more accurate representation of what the human ear would perceive. Australia’s electronics magazine Fig.a: one-octave smoothing gives an almost useless result – it’s just too smooth! Fig.b: 1/6th-octave smoothing is about right. You can see the details of the response, including standing wave peaks and troughs, and accurately gauge the turnover points and roll-off steepness. Fig.c: 1/48th-octave smoothing also gives a reasonably good result, although it’s questionable whether the extra detail is helpful. In some cases, such as when optimising edge diffraction, it could be. Fig.d: without smoothing, the result is similar to 1/48th octave smoothing over most of the portion of interest, but gets very noisy above 200Hz, mainly because the sub isn’t producing much (if any) output at those frequencies. Fig.e: psychoacoustic smoothing is an interesting option as it appears to give a useful curve that supposedly compensates for the properties of human hearing. January 2021  73 easy to wrap, and trim out with edging. The block shape here would make the trimming a bit more of a fiddle to get a neat finish. An alternative is polish over iron-on timber veneer, or a laminate finish or, as I did, a paint finish. I began by rounding over all the edges with the router (making sure all the nail and screw heads were well down so the router didn’t hit them). I also rounded over the port openings. Theoretically this smoothes the passage of the air as it pumps in and out of the ports and reduces the likelihood of chuffing. I couldn’t hear any difference, but I liked the appearance better. After routing, I filled and sanded all the holes and applied a general primer to seal the MDF. It takes quite a lot of work of filling and sanding to completely hide the joins in MDF boxes – they can stubbornly show even after several coats of automotive spray putty. I used a pressure pack can of “Granite Effect” paint to create the texture. This paint comes out as splatters of different greys to simulate granite. I didn’t want the light/ medium grey colour of the paint, but I used it to create the base texture surface. The top coat was a satin dark “charcoal” colour. But this material is expensive, difficult to work with, and certainly not needed for the functioning of the box. You can also see that the boxes are “empty”. It is common practice to put damping material inside speakers, which can provide various benefits. I tested the boxes with differing amounts of stuffing, but the frequency response didn’t change at all. That does not mean that it serves no function. I didn’t test impulse response, for example, and damping material may well help in this respect. In the end, I left some in on the surface opposite the drivers. SC What about a barrel? The design presented here is not particularly suited to the PVC pipe construction used in the original “Bass Barrel” article because of the port lengths. But it is possible to tweak the design so you can build it that way – see Fig.11. The material for the baffle and the ends is 16mm MDF again. In this case, the 63mm inner diameter pipe is 200mm long instead of 180mm. The 32mm inner diameter pipe is still 210mm long. This results in a predicted response as shown in Fig.12. The response is similar to my original design with a 38mm diameter small port: the overall response is flatted, but it has a steeper roll-off, which when room gain is taken into consideration, might produce a less ‘manageable’ result. Using this design with the 38mm port accentuates this characteristic, with a further raising and straightening of the initial curve and a steeper low-end roll-off. I was not personally interested in this construction method, so I did not build and measure a test speaker. Most likely, further tweaking could produce alternative (possibly enhanced) variations for this construction method. Speaker Box Lite (and similar) software enables many different configurations of drivers and enclosures to be investigated easily. TUBE 68 DIAMETER 168 ~300mm (eg STORM WATER PIPE) Fig.11: if you want to make a “Bass Barrel” as per the August 1997 article, but with currently available drivers, here are the dimensions. If you don’t have an anechoic chamber but want to characterise speaker response accurately, you either need to do it in a wide-open space, or else perform ‘nearfield’ measurements, as shown here. This involves placing the microphone very close to the speaker, so that reflected sound waves are at very low levels compared to the direct sound being measured, and thus do not unduly affect the results. This test was done prior to the final box coating. 74 Silicon Chip Fig.12: the predicted response of the barrel version of the subwoofer is very close to the rectangular version. Australia’s electronics magazine siliconchip.com.au