|Elliott Sound Products||Speaker Box Project - Part 1|
The speaker system shown here is part of my own system, and is intended for tri-amping, so there will be no passive crossover details. I am using the Linkwitz-Riley crossover (Project 09), my existing amps for the bass and midrange, and (probably) Project 19 LM3876 amp for the tweeters (this is still to be decided at the time of writing).
The enclosures are designed as mirror image pairs for the best imaging. Midrange and tweeter drivers have been carefully located so that diffraction effects are minimised, by ensuring that the distances are different from the centre to each edge of the baffle.
The drivers are as described below, with each having been selected for response, linearity and power handling. Each is excellent in it own right - not necessarily the most expensive, but all have very good performance.
|Make / Model||Visaton / GF-250||Focal / 5K4411||Audax / TW025M0 *|
|Z (Ohms)||8 (2 x 4)||8||8|
|Xmax (mm)||32||1.55||0.30 mm|
The Focal 5K4411 is described as a 5¼" Polykevlar Midrange with phase plug, and although I am normally not a great fan of Kevlar, the cone is well treated to prevent the normally nasty breakup effects at high frequencies.
* The Audax tweeter was replaced by a ribbon tweeter in 2010. Otherwise, the boxes are as originally built and are still just as satisfying as when they were first made. The stands were also replaced by ones that are a little more elegant than the originals shown further below.
The cabinets are made from 18mm Medium Density Fibreboard (MDF), with a laminated construction for the baffle. Figure 1 shows the laminating in progress - I used every clamp I could find to keep them together.
Figures 1 and 2 - Laminating the Baffles ... Enclosure, Half Assembled
The boxes are very solidly braced, and the various stages of the construction are shown below. The midrange is in its own sealed enclosure of approximately 14 litres, and the total volume for the woofer is 35 litres. The tweeter is in its own tiny enclosure, fully isolated from any back pressure from the midrange.
Midrange and tweeter are to be surrounded by a felt-filled cutout, and the front edges of the enclosure will be rounded to reduce refraction. All drivers are mounted as close to each other as possible. To this end, the bottom of the tweeter has been cut off so it sits closer to the midrange than would otherwise be possible. This necessitated the separate tweeter enclosure, since the mounting flange no longer seals the tweeter properly. No grille cloth will be used, as the frame would cause refractions I would much rather be without.
All drivers are secured with metal thread screws and tee nuts. The latter are very firmly attached and glued in position to ensure that they don't come off while I am assembling the system. Those for the woofer are visible in the photo. The brace above the woofer and the base and side of the midrange enclosure are also seen. The small section beside the midrange enclosure forms part of the main cabinet, and will be stuffed full of fibreglass before final assembly.
The enclosures are quite intricate, and represent a significant amount of work (not to mention sawdust!). As always, the initial preparation takes the longest in terms of assembly. Once all the panels are cut, holes drilled and rebates rebated, the actual assembly is fairly fast, and the boxes were brought from basic bits of MDF to their current status in less than one day. The preparation took the best part of a full weekend.
Figures 3 and 4 - Rear View of Cabinets ... Front / Rear Views
In Figure 3, you can see the rear view of the two boxes (upside down). The block of plywood below the midrange cutout forms the tweeter enclosure, and again, you can see the tee nuts glued in position. At this stage, no corner reinforcements or 'minor' internal braces have been added - these are being prepared from various offcuts, and will be added next.
The front view of the left speaker shows the cutout for the felt around the midrange and tweeter. The right box is a mirror image. As you may be able to tell from the photos, these shots were taken shortly after assembly - the glue is not dry yet. All panels are glued and screwed or nailed - the selection of screws or nails was based on the stage of assembly. A nail gun is very fast, and does a great job, but is fairly useless in the first stages of the assembly where initial alignment is critical.
Now that the boxes are basically assembled, the corner braces, panel braces and rear panel support trim can be added. Figure 5 shows one box from the front, with the drivers mounted to check that everything lines up. Once the backs go on, there is nothing that can be done about a misaligned brace or inadequate clearance, so everything has to be right before this happens.
The colour of the Focal driver's cone is a little unfortunate, but the system is being put together for the sound, with appearance taking second place. Despite the foregoing, the screw heads securing the drivers will be painted black before the boxes are finished. The recess for the felt can be seen in Figure 5, as can the rounded edges (only at the front - the others are 'conventional', and just lightly sanded to take off the sharp edge that would be too easily damaged).
Figure 5, 6 and 7 - Front and Rear View of Box Showing Drivers and Braces
The internal braces were installed using wood glue, but this was reinforced after the photos were taken, using automotive body filler. This is very strong, and sticks to wood extremely well, providing a very well braced and extremely sturdy box. Don't underestimate the usefulness of 'bog' for cabinet making - it is a better adhesive and better gap filler than almost anything I have ever used. Highly recommended . Unfortunately, it is no good with solid timber that is to be stained and varnished, but it can be veneered over perfectly well.
Note that the braces are all at an angle to help break up any internal standing waves. Considering the amount of fibreglass that will be used, this is probably unnecessary, but it is worth the very minor effort. Note also the carpet that is glued and stapled to the bottom of the midrange enclosure. There is also some carpet wrapped around the tweeter enclosure 'block' to prevent diffraction within the box itself, although this cannot be seen in the photos.
At the bottom of the box, the tee nuts for mounting the stand can be seen. As with all tee nuts in this project, they are glued into position so they cannot come loose.
Not a great deal here - the boxes were designed pretty much 'on the fly', but knowing exactly what I wanted. The drawing came last - basically after the boxes were completed to the degree shown above. This is not a complete drawing - braces and the side of the midrange are not shown, but these are not overly critical (perhaps surprisingly, or perhaps not). All dimensions shown are in millimetres, and the drawing is not to scale.
Figure 8 - Side and Front Elevations
As noted above, the midrange housing is about 14 litres, so it is quite easy to work out where the side panel goes - simply determine the distance based on the volume of the enclosure.
The actual volume becomes slightly less than the calculated value, because of the braces and the volume of the speakers themselves. The response will be plotted (then measured) in the next episode. In the meantime, the photos below show the boxes with all the fibreglass in place, both with and without the back installed. The enclosures are now ready for final sanding and finishing (after a listening test, of course .
Figures 9 and 10 - All Fibreglass Installed ... One Back In Place
As can be seen, the packing is extensive. The panels are all very acoustically dead, and a full test for airtight sections will be performed before the drivers are installed. I am not entirely sure how to do this properly yet, but I'm sure I will think of something before next weekend .
The cutout for the connector panel can be seen in Figure 10, and again, this is held in place by tee nuts that are glued in place. It is a source of considerable annoyance to many reviewers that so many speaker manufacturers use particle board screws to attach the drivers to the baffle - and this includes many 'high end' models. While my speakers will not get 'reviewed' (other than by friends in the industry), I do not think that anything less than a metal thread screw is adequate for fastening drivers - or anything else for that matter. The combination of the metal thread screws and tee nuts is almost indestructible - the box will rip apart before the screws fall out! The drawing of a tee nut may help those who don't know about these wondrous little fasteners. Drill a hole and hammer it in (and use a little glue to ensure it never falls out).
As some readers may be aware, the amp I am using at present has an output impedance of about 2 ohms. This has the effect of increasing the Qts (total Q) of the drivers, which was useful in my current speakers, but is equally valuable with the new drivers. I have not determined the exact impedance I will need yet, but the basic results that I will need are shown below.
Visaton GS-250 Woofer
Qts - 0.28 Standard, 0.34 Desired
Figure 11 - Desired Woofer Response (Qts = 0.34)
This is about as good as it gets, the response is much more desirable than it would be if driven from a 'conventional' amplifier with close to zero ohms output impedance. Resonance is at 53Hz, and F3 is 50Hz.
Figure 12 - Existing Woofer Response (Qts = 0.28)
Resonance remains at 53Hz and F3 is 61Hz - combined with the slow droop, this is not as good as it should (or could) be. The suggested size for the box is 22 litres, giving F3 of 63Hz - somewhat higher than I want.
The slow droop in response is quite typical of any speaker in a box that's too big, but the downside of making the box smaller is that the low frequency -3dB frequency increases (as it must). A Linkwitz transform circuit could be used to achieve the same result, but with added electronics and a lot more work than simply increasing the amp's output impedance.
Focal 5K4411 Midrange
Qts - 0.32 Standard, 0.60 Desired
Figure 13 - Desired Midrange Response (Qts = 0.60)
Resonance is 88Hz, and F3 is 83Hz, which is just fine, since it will be crossed over at 300Hz. This is an excellent result - all I have to do now is determine the exact output impedance I need to obtain this result.
Figure 14 - Existing Midrange Response (Qts = 0.32)
Using a conventional drive amplifier gives resonance is still 88Hz (as it should be), but F3 is 187 Hz - way too high. Again, equalisation could be used, but adds additional electronics to the equation. The slow droop extends all the way to the crossover frequency, a less than desirable outcome. Again, using a higher than 'normal' amplifier impedance can correct for parameters that are not exactly what you want.
The first listen has shown that these loudspeakers are exceptional. Hooked up to my triamped test bench amp, all I had available for the inaugural listen was an FM tuner, but the results sounded very impressive. Bass, mid and treble are well balanced, there is virtually no panel resonance, and the sound quality remains extremely good above, below and each side of the axes. Listening from a separate section of my workshop showed that the overall balance is very good indeed.
It never sounded like a speaker around the corner, but with speech I was almost tempted to rush back to see who was there - very lifelike indeed. Bass is actually much better than expected, and with an amp output impedance of 4 ohms, there was very good extension to 40Hz, and even 20Hz was reproduced (but somewhat subdued, as is to be expected).
The resonance of the Visaton driver in the cabinet is 52Hz - quite close to the calculated value. The Focal was a different matter, with resonance at 69Hz - considerably lower than the specifications indicate. This is of little consequence for a triamped system, but shows the importance of verifying the important driver parameters which will have a profound (and undesirable) effect on any passive crossover.
A near field scan of frequency response of each driver (bass and midrange) indicates that the response is almost dead flat across the designed frequency ranges. There are the usual minor peaks and dips, but absolutely no major 'suck-outs' or resonant peaks were to be seen - or heard. The response across the crossover frequency is harder to measure near field (actually it is almost impossible), but there was no audible variation - so much so, that I had to check the frequency and feel the drivers to find out which one was reproducing at the time.
The system is very well behaved with any crossover frequency between 100 Hz and 400 Hz (bass to midrange), and is also extremely tolerant of anything between 2 kHz to 4 kHz between midrange and tweeter. I will be using 250 Hz and 3 kHz crossover frequencies, and a power analysis is yet to be performed to verify the relative power needs of each driver. Bass and midrange are expected to be about equal because of a slightly lower than my 'normal' crossover frequency, and different driver sensitivities. I don't expect that more than about 10W will be needed for the tweeter to balance the 70W available for bass and mid.
The drivers have now been removed from the enclosure, and final finishing is in progress. This is destined to be time consuming, but I used four coats of black spray enamel, followed by four coats of clear. The final finish will be to rub the cabinets down with Scandinavian finishing oil and fine steel wool. This imparts a lovely smooth satin finish, and is similar to the final treatment of the subwoofer project described in the Projects Pages.
The final finish is rather nice, but cannot really be shown to any great advantage on a web page. You will get some idea from the photos that follow. Figures 15 and 16 show front and rear views respectively. The felt surrounding the midrange and tweeter can be seen, and the connection panel is visible in Figure 16.
Figures 15 / 16 - Front and Rear of Completed Boxes
Figure 17 shows a close up of the midrange and tweeter - not exactly exciting in hindsight, but it does give you a better idea of how the drivers are close-coupled, the felt surround itself, etc.
Figure 17 - Mid and Tweeter
The full length photo in Figure 18 lets you see the finish (to some degree at least), and what the final enclosure looks like. The stand is mounted using metal thread screws and a felt spacer to prevent vibration. This is shown in a little more detail later.
Figure 18 - Finished Enclosure
The photo in Figure 19 shows what had to be done to get the height right. The only stands I could get were 300mm high, and this was too much. I cut off the top section, then re-attached it to the shortened stand with screws and nuts. Welding was out of the question, since it would have damaged the finish badly.
Figures 19 / 20 - Speaker Stand and Connection Panel
The connection panel shows how the binding posts were glued in position to prevent them from coming loose. The colour coding of the connectors should have had a violet or blue binding post for treble, but my supplier didn't have any .
Figure 21 - A 1.5V Cell ??
Figure 21 seems somewhat incongruous at first sight, but I used it to ensure that all speakers were properly phased. When a positive potential is applied to the plus (+) terminal of a speaker, it is meant to produce a positive pressure wave, so the cone should move outwards. This is readily seen even with 1.5V on any speaker, and the current is low enough that it will not damage the most sensitive tweeter. However, it is definitely not recommended (and doesn't work anyway) with ribbon tweeters, so don't even think about it!
The speakers are in operation, and my old faithfuls have been relegated to a spare room (and have since been discarded after the woofer foam surrounds disintegrated). Since I am by nature somewhat impatient (OK, very impatient), I wanted the speakers operational before I'd built the electronic crossover and tweeter amplifier. Since all speakers simply terminate on the rear panel, it was a simple matter to use an external passive crossover network between midrange and tweeter as a temporary measure.
As regular readers well know, I am not a fan of passive crossovers, however, listening to my new speakers was more important than waiting until I could triamp them properly. I used a commercial 12dB/octave crossover, which would normally mean that it would be Butterworth alignment (See Passive Crossover Design for more info). This I did not want, so with a bit of an educated guess I decided that 12 Ohm resistors in parallel with each driver would do quite nicely, converting the filter alignment to a sub-Bessel (Linkwitz-Riley) having a Q of 0.5. A quick bench test confirmed that this was almost perfect, but in reality, there is a small discrepancy because I didn't equalise the midrange driver's inductance. This was acceptable, since it is temporary only.
The sound is wonderful! Even before careful equalisation and the full electronic crossover (including baffle step compensation), I am extremely satisfied with the performance. There is (was!) a hint of over-brightness, but otherwise colouration is extremely low, and imaging is superb. The definition is extremely good, and voices sound so natural that you'd almost swear that the person was in the room.
The second installment covers the electronics - See Part 2
|Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is Copyright © 2001. Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro- mechanical, is strictly prohibited under International Copyright laws. The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference. Commercial use is prohibited without express written authorisation from Rod Elliott.|