Vertical Array Cabinet
if you have a popup blocker you may not have seen the disclaimer about the design described on this page please view it before going any further. Thank You.
A joint project on Harmony Central is developing a Line Array loudspeaker system. This web page runs through some preliminary design work to come up with my own design based on their brief. There was some debate whether the speaker should be two or three way; this opts for three way. The only other criterion was that the cabinet should be 30" wide. Presumably they don't want something 60" deep, and other truck pack dimensions included 22.5" which I will take as the maximum allowed depth. The height of the cabinet will be determined by the bass drive unit size hopefully 10" but maybe 12" if the output from a 10" proves insufficient. Obviously the cheaper the better.
|
To get a three way system in a 30" wide cabinet there are a couple of options. First you could go with an asymmetric layout with perhaps a 12" bass, a couple of 6" mids and again a couple of horns . This is very miniature 1980s and I know that I am not the only one who used 2 x 15", 4 x 8" and a couple of horns in similar style enclosures back then (see the diagram). The alternative is a symmetrical layout similar to my existing design, which again will require moving the drive units from off the front baffle. By staying with a similar layout but going with the bass reflex option the design that I already tested would be reduced to 32" x 24" x 14". It shouldn't take too much ingenuity to 30" x 22" x 12". |
 |
Although placing the bass drive units on the end of the cabinet was something that I considered previously but never did, it has recently been adopted by EAW. This gives the technique credibility in the sound reinforcement world, but the method is hardly new. Many car subwoofers have fitted loudspeakers at each end of the enclosure as have a number of ghetto blaster CD/radio/cassettes. In the pro-sound world the technique is also used in Servodrive's SPL-C^3. The low frequencies are omni directional so the direction that the drive units face is not critical. When the width of the cabinet becomes half a wavelength the drive units act as a dipole and no sound is radiated sideways to the cabinet. At this frequency if you stood at the side of the cabinet, looking directly at the drive unit you wouldn't hear much. With a 30" width the half wavelength would be 223Hz. This fixes the bass/mid crossover at approximately 200Hz.
My main concern with the previous mid-horn was the roughness at the higher frequency range. This was something not apparent when listening, but non the less I considered worth experimenting with if I developed the design. Thanks to the great data published by EAW this has become less of a concern. If you look at the raw frequency plot of the KF730 there is a huge dip in the response between 1KHz and 2KHz. This is compensated for by boosts of up to 12dB in the processor settings. My instinct would have led me to believe that the mid horn design that is used in the KF730 would have been smoother than the horn design that I used, but the upper response is severely limited and the high frequency drive units presumably will not operate down low enough to prevent the chunk missing from the response.
|
|
Click on the thumbnail to view the frequency response graph of the KF730. This is taken from the manufactures data sheet freely available from their web site. It shows the falling response of the mid horn above 1KHz |
Given the measured performance of my previous mid-horn the low frequency performance was such that even using a single cabinet the mouth size could be reduced slightly with no detriment to performance if crossed over at 200Hz. A plot of the predicted response for the new horn is shown later.
|
|
This is an early measurement of the midrange horn of my previous cabinet. I lost later plots where the software compensated for the downward slope. With the red line as a datum point the two grey lines show 3dB either side. The notes are some thoughts about the peaks and dips in the response. The measurement was made with the horn 1 metre off the floor and the measuring microphone 0.5m in front of the horn mouth. |
Having chosen a specific cabinet layout the focus can now be on the details and finding out if there are suitable drive units. I haven't mentioned the dispersion angle, because that has already been decided. With the cabinet width and depth chosen the flare angle is set.
|
|
The diagram to the left shows the layout of the throat section for the mid and high frequency horns. The centre slot for the high frequency horn has an internal exponential flare and the mouth of the slot works down to about 2KHz providing good loading for the compression drivers. By using a small diameter compression driver the mid/hi throat assembly can be made more compact. By moving the mid drive units back the alignment with the compression diver becomes smaller too. |
To try and maximise the space I've imposed a few restraints on the compression driver. With a 1" exit and using 0.5" ply, the width to the mid baffles is 4". If a suitable compression driver can be found with this diameter, the mid drive units could be moved back enabling a longer horn. See the drawing. The RCF N350 is 102mm diameter so slightly moving the baffles apart would accommodate this driver. Although the cross over frequency would have to be raised another alternative would be the Eminence ASD 1001.
Looking at a few 6" drive units for the midrange gives the distance to the next baffle.
|
Eminence |
71mm |
|
RCF |
85mm |
|
P-Audio |
71mm |
|
18sound |
58mm |
Based on a 75mm clearance the approximate volume for the bass enclosure would be 24 litres.
One of the production difficulties with my previous design was fixing the midrange drive units. An alternative option instead of an additional baffle is to fit the drive unit in a small enclosure and then attach that to the baffle. This would also increase the volume for the bass driver.
Looking at the bass section in more detail we first need to choose a drive
unit. Searching on the internet bought up a cheap unit, the BX10 that seemed
ideal. The specifications in pdf form are
here. Searching further found very little information and what there was
proved conflicting as far as the specifications were concerned. If anyone knows
more about this drive unit it might be worth reconsidering. Eventually I settled
on the Eminence HL-10A.
which has a predicted response similar to the BX10. This is a new unit I haven't
been able to find any prices here in the UK yet.
|
|
The diagram shows the predicted maximum SPL output from 1, 2 and 4 cabinets using the HL-10A(green, red, white plots). The blue plot shows the predicted output from 2 x JBL 2226H speakers and the yellow that from 2 x P-Audio C12 300MB. All responses were calculated on winisd. The enclosure size for the HLs was guessed at 25ltr per cabinet as a rough size that might be expected in the proposed cabinet. |
I'm not sure if using JBL units as a comparison has the same impact else where
in the world as in the UK, but five of the BX10 units units at £50 could be
bought for approximately the same price as one JBL 2226 at £260 (prices at Jan
2004).
Moving on to the horn section, this was modelled using two Eminence Alpha 6 drive units using David Mcbean's horn response program. The horn was split into three sections each with a conical profile. The first section is the short slot section. The second uses a very short horn to model the step at the end of the slots and the third models the final flare with the front of the cabinet used as the mouth size. The areas used are given below.
|
|
This is the predicted response of the horn using two Alpha 6 drive units. The parameters are as below:
Throat area combined 40 Cm^2 Click on diagram for more info on the horn design |
At this point a more detailed drawing can be done to see whether our guessing and estimates stand any chance of success. The following diagram shows a first attempt. Not shown on the drawing is the internal height which is taken as 28cm.
The internal horn dimension is 15cm at the throat and has a straight flare to the mouth which is the full height of the cabinet.
The first thing that becomes apparent from the above drawing is that the volume of the bass enclosure is less than guestimated. It is not much but maybe enough to have a rethink on the drive unit used or increasing the size of the cabinet. The first thing to do is compare the response in a smaller cabinet to that done previously.
|
|
This diagram shows that even quite a substantial reduction in cabinet size doesn't affect the response too much. You should note that the box volume is for two drive units, which is why the figures are double that shown above. 44ltrs (22 per drive unit) was chosen assuming that the drive unit and port would take up approximately 1.5ltrs. The blue plot shows that a reduction to 35 litres for both cabinets still gives a good bass response. Giving the cabinet a trapezoidal profile could be an option. |
With the bass sorted we can give the mid horn a quick once over again with the new length entered into the parameters.
|
|
As can be seen from this graph, the only effect of shortening the main horn is to reduce the resonant peak at 200Hz. Although the response looks uneven, it is within +/- 2.5 dB from 200Hz to 2KHz. |
|
|
This graph was shows the effect of stacking cabinets. As expected the sensitivity goes up and the bass response is slightly extended and smoother. The effect of multiple drivers was simulated by reducing the radiating space to 1π |
Now the mid response has been checked it is time to look at the hi frequency horn. There seems a lot of debate and discussion about the high frequency sections of line arrays, you can read some more ideas about line arrays by following the link (lots of pics might be slow download). The design for this cabinet is a simple slot horn with an exponential (sort of) expansion rate, the mouth of which exits into the mouth of the main horn. This horn should operate down to a frequency of 2KHz given the size of the mouth. The idea of using a slot is that the diffraction effect would normally endow the horn with an extremely wide dispersion pattern but in this instance it is limited to that of the mid horn. With the throat, mouth and length fixed the cut off frequency is also determined. If two drive units are used, both are trying to load the same horn. To make things easy I just selected an arbitrary cut off frequency below the actual cross over frequency and let the slot expand out until the full height of the slot is taken up. This has worked well in the past for me but the actual loading on the drive units could be measured after building if it is felt there could be a problem i.e. if the drive unit is run to a lower cross over frequency than might be healthy. With a slot width of 2.5cm and a cut off frequency of 600Hz the slot should expand as given below.
|
Distance from throat (cm) |
Total Height of slot (cm) |
Height each section (cm) |
|
0 |
4.5 |
2.25 |
|
1 |
5.7 |
2.85 |
|
2 |
7 |
3.5 |
|
3 |
8.8 |
4.4 |
|
4 |
10.9 |
5.45 |
|
5 |
13.6 |
6.8 |
|
6 |
16.9 |
8.45 |
|
7 |
21.1 |
10.55 |
The alternative to calculating is drawing an appropriate looking flare on the plan, especially when the cut off frequency will be well below the cross over frequency.
The high frequency horn will look a bit like this from the side.
A quick redraft of the plans looks something like this. Click on the image for a PDF version
That is about it. The next stage is to build a cabinet and measure the results. For a prototype I would make the throat of the mid horn variable by making the opening in the baffle larger than needed and then making a thin sliding partition to reduce the size down. The throat section could then be tuned to give the best response.
To mount the mid drivers I would make an additional baffle to which the drive unit is flush mounted to the front. Using T-nuts that could then be fixed by bolts passing through from the front of the throat section and from the compression driver section. The compression drivers would be fitted using an intermediary plate. The plate would be screwed to the driver in the normal way and in turn the plate, which would stick out past the edge of the drive units would be fixed to the baffle. The diagrams should make it clear.
Eminence only publish the weight of the magnet (if I've overlooked something please let me know) so making an accurate estimation of the total weight becomes a guess. The RCFs are 2.1Kg, I would assume something similar for the Alpha 6s and maybe 6.5Kg for HL-10s. As a rough guide to the cabinet weight I approximate 10Kg per 5mm thickness of 8 x 4 sheet (how's that for mixing metric and imperial units). So with approx 20Kg for the drive units and 25Kg for the box, the total weight is around the 45Kg region.
Update number 1.
An alternative to the Eminence bass drive units is a Peerless 10" unit. Although it wasn't designed as a sound reinforcement drive unit the following graph from winisd shows that it could be used as a direct replacement. The graph shows the maximum SPL compared to the Eminence unit. Apart from its physical size the Paudio TM10 is another possibility.
 |
For an even smaller line array cabinet follow this link to a Very small line array part (concept) or V-slapc. |