Enclosure Part I

This page is still WIP. The pic above is the latest available snapshot of wip status.
- latest update -

Design

As mentioned elsewhere, this is a WMTMW configuration speaker. The MTM cluster is group in a triangular layout with the T at the apex and the 2M forming the base (refer diag. below) versus the more traditional horizontal straight line. This from a 'point source'#1 when reference at a distance as well as reduce the overall width. Routing may be a tough challenge with the 3 drivers rims touching one another. Hopefully with 2M drivers on a horizontal plane help expand the imaging as well as compensate for off axis response.

The port vent is design as slots instead of conventional round ports to further reduce the width. I find this design resulted in better aesthetic too. Naturally they are placed on the front baffle in order to facilitate close to back wall placement. Enclosure volume sizing was simulated using manufacturer provided TS parameters rather than actual measurement. Though there will likely be some variation to the actual drivers, I assume these to be negligibly small and of little significance. Furthermore, I've little confidence in my own measurements since there's a severe lack of proper measurement equipment and setup.

The MTM will be self-contained in a air tight sub-enclosure. The Unibox simulation modeling produces the following bass response.

MID-RANGE MODELING


Modeled parameters:
Volume :: 3.5 litre
F3/Imp Peak :: 177Hz

WOOFER MODELING



Modeled parameters:
Volume :: 29 litre
F3 :: 47Hz
Imp Peak :: 28Hz, 65Hz
Port 1st resonance :: 828Hz

From the response charts, it will be necessary to select a WM xo point to avoid the impedence peaks and resonance frequencies btw 200-800Hz. A xo point at 400Hz looks promising. The huge rise in M imp also dictate proper notch filter to be applied (being only 1octave from the suggested xo).

Taking profile cues from B&W and KEF, I ended with the following design;

not shown are extensive joint bracings and supports

Other key design elements;

  1. Front baffle thickness of 36mm. Minimum 24mm wall thickness throughout.
  2. Woofers are 'coupled' to the bracing.
  3. Slight time alignment by positioning the drivers on different layers of the 3 composite MDF (t=12mm) that form the front baffle. Woofers are flushed-mounted on top most layer, midrange mounted without flushing and tweeter flushed-mounted on the 2nd layer.
  4. Drivers to be secured by t-nut set on the back of top & middle baffle layers.
  5. All internal corners to be rounded.
  6. Cross bracing btw all profile panels to increase stiffness.
  7. Total of 8 layers of 3mm mdf panels, each layer of panels laid with staggered seams to the one on top and bottom.

Overall dimensions:

Width :: 723 mm
Height :: 280 mm
Depth :: 380 mm
Weight :: ~25kg without drivers & accessories (estimated)

Preparation

Key to the successful execution of this design is ACCURACY. I hope to limit the tolerance of all assembled dimensions to 1mm. Having completed the CAD drawings, it was time to send them in to the lumber supplier for cutting. Concurrently, the necessary tools and accessories were purchased. The following is the tools list (not in any order of importance);

Commercial speaker cabinet cost is mostly due to manufacturing cost rather than material. Yet, even for speakers costing up to $1k, the cheaper (and acoustically inferior) particle board is used rather than MDF! Example - the previously owned JmLab Electra CC900.

SAW DUST

A couple of days after placing my order, the MDF boards are ready for collection. Total cost is only SGD$76 for the lot. Before making any cut on the raw MDF, all dimensions are checked against the CAD drawing. The first challenge was to make the profile template. A good practise in wood working is marking the cutout outline. This was done for all the major pieces in as detail as possible. This greatly help in visualization of cutting and assembly sequence. The process throw up potentially difficult routing procedure for the front and back panel.

Routing impression using an industrial grade router vs the Dremel rotary tool is like heaven and earth. Use the proper tool for the proper job as they say.

To route the curve profile, the circle jig that came with the router had to be stretched to the maximum. As a result it was rather wobbly during operation. Consequently, the 1st attempt ended in failure when the pilot pin slipped out of the pilot hole. Improvising with free weights holding down the pilot pin, I managed to get 2 nice curve cuts in the 2nd attempt (with a very slight slippage error at the edge joining the front baffle). The two curved sides are not exactly mirror image of each other as supposed on the drawing#2. These are minor issues and within my target 1mm tolerance. With this profile template and a straight piloted route bit, routing the rest of the profile shapes was a breeze, albiet with lots of sweat from sawing each mdf plank to the rough shape b4 routing.


Router & templates; Completed profile template; Internal sub-enclosure; DIY-bending jig to shape the curve panels;

Once ready, the base structure was assembled. 1st, to assemble the 3 pieces forming the internal enclosure. It is very important to perform a dry assembly and to constantly check to ensure the pieces are at 90 degrees to each other. Having satisfy myself, glue was applied, assembled, re-check angles and finally secured in place with clamps for drying. Every glued pieces are left to dry for a minimum of 12hrs. Critical pieces for 24hrs.

Concurrently, the 3mm panels are pre-conditioned to the curved profile using self-made bending jigs. The panels are spray with water to 'soften' and ease the panel into taking the curve profile shape. This was done several days before assembly/fixing. The clamping difficulty was grossly underestimated (it is rather difficult to describe). The 3mm mdf panel when bent over the curve profile does not exactly follow the template curve!! Instead, gaps exist along the profiles with the panel only in contact at pressure points. This will be inadequate to couple the curve panel to the main structure. In the end, I need to make special clamping support jigs combine with wood screws to ensure full contact. Another unexpected problem was undulating surface across the length rather than a straight horizontal line. It seems that the internal structural stress raises the free ends that are not connected to the curve profile braces. Hopefully, the undulating surface will even out once the subsequent layers are applied over it.

[2006, April 22] With each panel needing overnight drying, it is estimate to take slightly more than a month to build up both sides. At this stage there's little to add until the panel layering completes.

[2006, May 31] I'm at layer 6 on one side and layer 7 on the other. At this thickness it was possible to begin routing out the recess for the back panel. I spent over a week conceptualizing how I would go about doing it, instead of long winded boring words, the following pictures should save me the thousands words.


setup-shimmed to height & clamp to workbench; note weights for added stability; dry fit


routed surface; uneven thickness due to inconsistent bending

[ 25 Jul 2006 ]

Having paused for a long while due to taking up a new job and the Fifa WorldCup 2006, work finally began on the front baffle. The cut-out diameter is smaller than what the bundled circle jig could accomodate. Thus a customized jig was made. The result is not too bad...


All lined up, bottom piece yet to be routed; Dry fit of top 2 pieces

Work also started on the binding post plate. A 6mm thick Al plate was sawn, filed, drilled and finally polished to give;


worked Al plate; WBT binding post installed

$$ money invested in QUALITY parts;


WBT Binding posts ; model WBT-0730.12 Topline (Signature)

Took me quite a while before finally splurging on these babies. The build quality is certainly examplary. It is hefty and huge - the pics are rather misleading.

Click here for part II ......

#1 This is only logical assumption on my part, unsupported by any scientific study.
#2 On hindsight, it is possible to produce a mirror image curved side. Instead of routing both curve using the circle jig, complete one curved side, use it as a template to make a copy profile. Use the copy profile as the template to route the mirror curved side. This is a much more accurate method than trying to route a mirror curve using the circle jig as the later depends on drilling accuracy of the pilot holes in addition to slippage error.

Last update :: Sunday, October 29, 2006 12:06 PM