Humble Homemade Hifi

Progress – concept and ideas

Cabinet size and shape

A definite trend can be seen in high-end loudspeaker designs recently and that is of the tall slim cabinet. Designs like the Wilson Audio Grand Slamm, Avalon Sentinel, etc are making place for new reference models like the Dynaudio Evidence and Confidence range or the Mission Pilastro. Also taking into account that many people have said “I love the Andromeda but its way too big for me” I decided to go for a tall and slim cabinet. This doesn’t automatically mean woofers no bigger than 7” or 8” even if I don’t believe in side-firing woofers. These side-firing woofers require steep and low crossovers so that they don’t cause weird phase cancellations in the upper-bass / low midrange. I prefer a simple first or second order crossover for smoother integration. The height and depth of the cabinet have no restrictions other than the optimum position for the midrange unit and tweeter and looks.

A separate mid-range unit

As with all types of loudspeaker designs, even a high-end project like this will contain certain compromises – if you choose for one thing then you automatically can’t choose for something else. There are a few conditions that must apply for all drivers in this project:

● Wide frequency response with at least two octaves of cone break-up free output above and below the predicted crossover frequency.

This gives the possibility of low order and simple crossover networks. Considering I will be using a series network in which all components interact this condition is fairly high on my priority list. Each driver will also work well with-in its limits, which should contribute to less distortion and greater power handling.

A separate mid-range unit has several advantages:

● Small lightweight cone for “fast” and detailed midrange.

In this case 6,5 grams for the Scanspeak 12M4631G00 compared to 20 grams of the Scanspeak 18W8545 often used in 2 or 2,5-way systems as a bass-midrange unit. It sort of bridges the “speed-gap” between a mid-woofer and a tweeter. The magnet and therefore the BL force factor is lower with the 12M4631G00 but more important is the combination of BL and Mms, also known as the acceleration factor, is much better. Here is the calculation and for comparison also that of a well known tweeter:

Scanspeak 18W8545 Scanspeak 12M4631G00 Scanspeak D2905-9700

BL=8,0 Tm BL=5,3 Tm BL=3,5 Tm

Mms= 20 gr. Mms= 6,5 gr. Mms= 0,45 gr.

Г= 400 Г= 815 Г= 7777

The acceleration factor is basically the ratio between the driving force and the moving mass: BL/Mms

The same idea as with a car. A lightweight car with a powerful engine means great acceleration! Because the calculation for BL contains Re of the driver, an 8-ohm driver will have twice the acceleration factor as a 4-ohm unit (considering all else is the same). This is logical as the 8-ohm unit has twice the amount of wire in the air gap on which the force of the magnet can be transferred (considering all else is the same, i.e. voice-coil wire thickness, etc). A pity the 12M4631G00 isn’t available as an 8 or 16-ohm unit!

● Small cone for better off-axis response.

If we take the crossover point fx at 3000Hz for the moment (it may be something else when I start filtering) the 60˚ off-axis output level of the Scanspeak 12M4631G00 is 3,5dB’s down compared to the on-axis response. The Scanspeak 18W8545 is 13,5dB’s down at that same point. For a lower fx of 2000Hz the Scanspeak 12M4631G00 is only 1dB down compared to the on-axis response. The Scanspeak 18W8545 is 4dB’s down at that point. I believe good smooth off-axis response to be as important as the on-axis response for a stable and wide stereo image (I hate speakers that beam!) and it is also as important for good dynamics. If there is less energy being produced in the frequency range around fx from say 1500-4500Hz compared to the frequencies above and below that (even if it is off-axis energy) this will have an effect on the overall perceived dynamics of the system. I don’t have any scientific measurements or so to prove this but I just get this impression from the many speakers I have designed in the past.

● A separate midrange means the mid-woofer can be less “mid” and more “woofer”.

This opens up the possibility for a woofer with low fs, slightly heavier cone and therefore better bass response.

‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗‗

30-01-2003

I am a bit late in posting information, been rather busy lately. For the last few months I have been making sketches and calculations to determine the rough size of the cabinet, for looks I want to keep it to about 220x400x1200mm (WxDxH). The main thing that will determine the size of the cabinet is the size and amount of bass drivers. Also I am still investigating the various possibilities of bass loading. Reflex, closed, transmission line or combinations of two types are all open options at this moment.

Here are a few commercial designs which I am looking at for inspiration: The new Dynaudio Contour S5.4; Aerial Model 3 and 5; Sonus Faber Cremona; TAG McLaren F1; leather!

06-02-2003

To determine the position of the drivers on the baffle I simultaneously use a CAD programme and crossover design software. By moving the drivers up and down the baffle and backwards or forwards relative to the listening position in the CAD programme I can get a fairly accurate measurement of the distance from each driver to the listening seat. The relative offset is then feed into the crossover design software to account for the acoustic phase shift due to this offset. I want to try and get away with first-order crossovers if possible; the idea being if it will work with 1^st order crossovers, it will also work with higher order crossovers, not the other way around.

But what ever I tried with the existing preliminary design with its straight baffle (moving the drivers up and down, TMWW, WMTW or other configurations) the only way to get all the drivers into phase was to offset the tweeter and midrange unit several centimetres backwards. This lead me a totally different design: the sloped baffle. Commercial designs that immediately jump to mind are the Thiel CS7, Sonus Faber Grand Piano, Kharma Exquisite, etc. – (By the way, I used to have double-bass lessons from her bass-player and boy-friend Peter Bjørnild)! - Sonus Faber just tilts the cabinet backwards by using longer spikes at the front.

Okay next step into cabinet construction. No matter what the speakers will look like, I want a nice strong and dead cabinet that will incorporate the reduction of standing waves inside the enclosure. I could use thick panels, internal bracing, angled internal partitions, etc but I want to take a totally different approach than usual (that’s the designer in me wanting to be satisfied). If you look at a vertical cross-section of the cabinet there is a very easy way to combine a sloped baffle and irregular internal shape: the laminated panel technique. This concept has been around for quite a while and can be found in many DIY and commercial designs on the Internet. One of the first companies to exploit it commercially was Sonus Faber with the Extrema launched 12 years ago back in 1991.

Other designs are the previously mentioned Kharma Exquisite. The Kharma Grand Ceramique uses a similar construction for it upper tweeter/midwoofer section.

Then there is the Gershman Acoustics Opera Sauvage and the Vandersteen Model 5.

All these ideas lead to the following preliminary design:

14-02-2003

Still working on different cabinet concepts. The external design should be original (so no Kharma or Thiel type cabinets) and I want to get it down to the bare minimum needed, that means a cabinet (closed and/or reflex) for the woofers; the tweeter only needs something to hold it in place. The midrange unit could be an open-baffle type – if I do this then it will also only need something to hold it in place. This concept opens up new possibilities in cabinet design. The visual aspect of the design must be very pleasing so at the moment I am willing to make minor compromises on the technical side of things if they will lead to significant improvements on the esthetical side. Another interesting consequence is the open baffle midrange, I have read many positive things about this concept and am very curious to try it out. Commercial designs are for example those by Alón. Seeing as the cabinet is becoming more and more final I am also focussing in on the woofers, they will either be 7” or 8” and there will be two per cabinet, so 4 woofers in total.

27-02-2003

The open-baffle-di-pole concept. When the driver is mounted in a rectangular baffle, the wavelength at which bass roll-off commences is equal to the shortest distance between the front and the back of the driver. This is when the distance reaches half a wavelength between front and back. Thus the cone must be at least a quarter wavelength from the edge of the baffle. When L =l/4. For a 15cm wide baffle this would give about 570Hz as the “bending point” of the output curve. Below this point the bass roll-off with 6dB per octave down to fs and then drops with 18dB per octave. This automatically means the woofer must be able to reach well into the midrange without break-up or non-linearity distortion. The advantage of an open baffle is that the many resonances and colourations produced by even a well-designed enclosure are eliminated. The polar propagation of a baffle speaker is that of a doublet, which is a figure eight. Sound pressure at an angle from the front or rear axis is the product of that obtained on-axis and the cosine of the angle. At high frequencies the angle of propagation narrows (as with any speaker) and the rear lobe is distorted by reflection and diffraction caused by the speaker frame and magnet (as with any speaker) – luckily the 12M has a well-designed frame and compact magnet. As frequency increases and the driver becomes more directional of its own, the polar pattern still has an approximately a figure eight shape, because the driver radiates front and rear and little to the sides of the open baffle. The front and rear output will differ because of an acoustic filter formed by the basket openings and trapped air between cone and basket. This filter is the reason for the differences in high frequency response between front and rear. A flat baffle should be operated well clear of any wall otherwise the rear wave is reflected to reinforce and cancel the front wave at quarter and half wavelength spacings and their multiples, respectively. Adding damping material on the rear of the baffle can minimalist this effect.

As you can see the latest design has been taken back to its minimum, I have used the “Golden Ratio” for determining the relation between height, depth and width of both the baffle structure and the woofer cabinet. The Scanspeak R2904-7000 tweeter and the Scanspeak 12M4631G00 midrange unit are mounted on a sort of L-shaped baffle (seen from the side). The original idea was to make the baffle from solid anodised aluminium but that turned out to be too expensive and the aluminium would taper at the bend (ugly). I am now thinking of laminating thin mdf to a total thickness of about 20-25mm. This baffle is fixed to a completely form-follows-function bass cabinet. The front woofer will be running far into the midrange so I want to have it as near to the mid as possible, thats why the tweeter is on top with the midrange unit below. Also phase issues in combination with the shape call for a top mounted tweeter. With the mid at the top the baffle (as seen in some other oopen baffel designs) the baffle would have to slope forwards. For looks the second (sub)woofer is placed in the side of the cabinet, I will worry about phase cancellation later. You may notice that the woofers have grown to 2x 8-inches so that there will be enough cone area for a decent and low bass response. The bass cabinets will be mirror images of each other. At the moment I am leaning towards the new Seas Excel wood-pulp/sisal cone drivers to be used for the woofers – nice looks and a smooth top-end roll-off. Can’t wait until Seas publishes the datasheets!

01-04-2003

Okay, I think I have finalised the concept stage! The second (sub)woofer and reflex ports have moved to the bottom of the cabinet. Some commercial speaker manufacturers like Avalon with the Opus also do this down-firing woofer principle for example. They also use a 3-way design with a built-in subwoofer of the same size as the woofer placed in the bottom. This decision has several advantages:

● Less drivers in sight.

A want this speaker not only to sound superb but also to look great. The fewer the amount of visible drivers the better, following the less-is-more approach. Furthermore “hiding” the subwoofer will strengthen the impression of amazement about the amount of deep bass coming from such a “small” cabinet with only “one” woofer. Driving the second woofer as a subwoofer means it doesn’t need to be directly near to the other woofer.

● Greater flexibility in driver choice.

Driving the second woofer as a subwoofer means it doesn’t need to have an ultra smooth response in the (upper) midrange because of the simple reason that it won’t be working that far up. Even if you filter the peaks out with notch filters there are still sharp phase shifts in that range due to the steep top end cut-off. The opens up the possibilities for metal cone drivers which give great bass detailing but do break-up at the top end of their range.

● Coupling of the air mass to the floor.

Because the subwoofer will only be a few centimetres above the floor, the air “trapped” between the cabinet and the floor on which it is spiked to will be exited by the woofer. This means a larger acoustic source to produce the lowest octaves. Also the boundary reinforcement of the floor will lift the output level of the bass. It should work like a half space radiation more than a semi-free space radiation when a woofer is placed higher up in an enclosure free standing in the listening room.

The ideal woofer for this concept is the new Seas L22RN4X/P. The L22RN4X/P is an 8" cone driver with an extremely stiff and stable injection moulded metal basket. Large windows in the basket both above and below the spider reduce sound reflection, air flow noise and cavity resonance to a minimum. The extremely stiff aluminium cone should give tremendous bass precision. The cone and the low loss rubber surround show no sign of the familiar cone edge resonance and distortion associated with soft cones. On the other hand, the cone break up modes at higher frequencies call for special attention in the crossover design work. A 4-layer, extremely long throw, high temperature voice coil wound on an aluminium voice coil former gives a high power handling capacity. The phase plug reduces compression due to temperature variations in the voice coil, eliminates resonance’s that would occur in the volume between the dust cap and the pole piece and increases the power handling capacity. The combination of low fs – 23Hz! – low Vas and a handy Qts give the possibility of deep bass from a small cabinet. Simulations so far predict a –3dB point of about 28Hz from a 35 litre cabinet! Efficiency is correspondingly low but that isn’t that much of a problem in this case because they will only be supporting an existing 8” woofer in a 3,5-way set-up.

So to sum things up so far: Scanspeak R2904-7000 ringradiator tweeter; Scanspeak 12M4631G00 sliced non-coated paper midrange, both mounted on an open baffle; Seas Excel W22EX-002 coated sisal wood/paper pulp conemid-woofer in a closed enclosure (still got to audition these yet); Seas L22RN4XP aluminium cone subwoofer, in a down-firing reflex enclosure.

03-04-2003

Okay, I think I have finalised the concept stage again! – my motto: never be satisfied. The down-firing subwoofer has grown in size to a Seas L26RFX/P and the mid-woofer has shrunk to a Scanspeak 18W8531G00. The reason I changed the subwoofer is because (to quote myself in the Andromeda article): “I believe that you can only produce “real” bass by moving a large area of air gently and not by moving a small area of air violently. This means that anything smaller than a 10-inch woofer would simply not do”. This statement had been nagging me for a while, so why should I compromise on such a project. Furthermore the Seas L26RFX/P works very well in a closed cabinet. The reason I changed the mid-woofer is two-fold: because the subwoofer has grown in size I would still have enough cone area even if the mid-woofer was slightly smaller. Secondly I wasn’t completely happy with the idea of using two different cone materials in the midrange. Seeing as the di-pole midrange unit will have low efficiency in the lower midrange due to cancelation of the front and back waves, the mid-woofer will need to reach relativly far into the midrange to compensate this (there will be a large overlap between the two drivers). Seas Excel coated sisal wood/paper pulp cone was a good compromise, but if I will be using sliced non-coated paper for the mid, why not just stick to sliced non-coated paper for the lower midrange as well?

So to sum things up again: Scanspeak R2904-7000 ringradiator tweeter; Scanspeak 12M4631G00 sliced non-coated paper midrange, both mounted on an open time-aligned baffle; Scanspeak 18W8531G00 sliced non-coated paper conemid-woofer in a closed enclosure; Seas L26RFX/P aluminium cone subwoofer in a down-firing closed enclosure.

NOTE: This design is strictly for the home DIY enthusiast and not to be used professionally without my permission!

Tony Gee

The Netherlands