6th order bandpass

This project is a result of swapping some six-inch flared ports for an IXL-18 driver.
Given the number of sealed and EBS ported boxes out there using this driver, it was time to come up with something a bit different...

 

Concept

The 6th order alignment consists of a ported chamber either side of the driver, resulting in a bandpass response.
The small 60 litre chamber has a single 6-inch port tuned to 53hz.
The larger 200 litre chamber has a pair of 6-inch ports tuned to 25hz

 

Design

WinISD

Output

Shown compared to an LLT Sonosub (365 litres, 15hz tune light red)
Both designs have 1000w applied
The bandpass box has less extension, but more slam (6dB more at 30hz), and has a slightly smaller box

Cone excursion

A high pass filter is recommended to control excursion below the tuning point of the large chamber.
A Butterworth high-pass filter tuned to 16hz is sufficient
Another characteristic of this design is that port loading decreases cone excursion at the tuning frequencies, which can lead to a decrease in the cooling of the voice coil. Care needs to be taken not to exceed the rating of the driver

Ports for large chamber

Peak velocity is 21 m/sec at around 25hz

A pair of 6-inch ports with 30mm exit flares will be noise free

Port for small chamber

Peak velocity is 31 m/sec at around 60hz

Whilst Flare-it has only been experimentally verified for frequencies up to 35hz, in this case the results can be extrapolated to get an idea of port performance. A single 6-inch port with a 30mm exit flare will be noise free

 

Dimensions and Cutting list

Dimensions

Volumes

Cutting list

• Six full sheets of 1800mm by 600mm by 19mm MDF are required.
• The approach is the same as outlined in the Speaker Building page
• Where an exact edge is required, it is initially cut 5mm oversize using a sawboard and then trimmed back to the green dotted line with a router

 

The Build

The long ports even manage to make an 18inch driver look small......

 

After several days of marking up and cutting, routering etc, you get.....

 

After after a few more days....

• The rear wall of the inner chamber extends the full height of the box to provide bracing
• A pair of shelf braces sit between the inner chamber and the rear wall of the box and also support the ports
• Shelf bracing is rebated into the surrounding panels
• All holes in the shelf bracing have edges removed with a rollover bit
• The outer shell of the box has chamfered edge bracing around all seams
• An access window allows the driver to be installed after the box is built
• The driver baffle is double thickness, for strength and to provide depth for mounting screws
• One-piece ports can be used

 

 

The 18inch Mach5 driver will be quite snug in there....

 

Hot-melt glue was used to secure the ports to the front baffle and the lower brace

Hot-melt glue is great for getting a quick and strong bond. It is however, difficult to ensure a complete seal. Going over the joint with silicone sealant, which can be worked into any voids solves the problem. The next picture shows the hot-melt only on the left versus the dual-glue solution on the right

 

Edge bracing and door support panel glued into place.

Detail of small port flanged into driver chamber.

 

I was in a hurry and accidentally glued the bottom on at an angle.

Another detour whist I worked out how to use a router as a milling machine

 

Carcass just before front panel is glued on

Carcass just before back panel is glued on

 

Less interesting with it's clothes on......

Fitting the driver took the total weight to 68kg, which is 150lbs in the old money...

 

There's a certain presence ....

 

Flush mounting of ports turned out well

 

Performance

Everything was dragged outside for measuring with REW

The response was measured at 1m and 2m with the microphone in line with the ports and at a lower height. This done was to identify any ground bounce effects or port coupling effects. The graphs were essentially the same, so only the 1m inline graphs are shown.

Response over the working range

Blue: Response at 1m measured with REW
Green: Predicted response from WinISD

Tuning frequencies turned out to be a bit higher than the design frequencies and the peaks are stronger than expected by around 7dB
The red graph explores the WinISD prediction when the volumes of the two chambers are adjusted so that the frequencies of the response peaks line up with what was measured. The port lengths were left unchanged as these are a known length.

This shows that even if the build volumes are off, the higher peaks are still not predicted by WinISD. Perhaps WinISD is a bit woolly when it comes to this style of box

Sweep extended to 500hz to examine resonances

The longer ports have a first resonance of 227hz and a second resonance of 454hz
Sources of the peak around 330hz could include: Box side-to-side resonance of 349hz Driver chamber top-to-bottom resonance of 338hz
There's a +/- 6dB dip/peak centred around 95hz - I'm not sure what that is!

 

Dialing it in

BFD Filters

An REW sweep was taken in-room and then some filters added using a Behringer Feedback Destroyer (BFD)
Since the high-pass filter has not been built as yet, an extra cut was added at 20hz for some excursion protection

Unfiltered response shown in green. Response with BFD filters applied shown in brown
The dip at 50hz is a room null

 

Measuring the output voltage of the EP2500 allows the power to be determined

Power = RMS voltage squared divided by DC resistance of driver

With the AV amp pushing enough signal to light the yellow warning LED on the BFD on the sensitive setting (-10dB), the subwoofer amp is only able to manage about 100w into 4ohms. Bridging the EP2500 lifts this to about 400w - still short of what's needed.

Switching the BFD to the less sensitive input level (+4dB) and maxing out the LFE level of the AV amp gets us up to 900w, which is the design level.
This measures 122dB @1m and 117dB at the seat, corrected for RS meter at 30hz - pretty much in line with what WinISD predicted.

After careful listening, it seems that 900w is pushing things a bit, with some excessive noises evident. I've been slack and haven't built the high-pass filter yet, so it may be an excursion issue. For the moment, it seems prudent to back the power off to 600w, dropping the peak output to 120dB @1m.

A cleanbox has been aquired and modified. Along with winding back the gain controls on the EP2500, this allows the signal to be trimmed to just light the red LED on the BFD at maximum output, which will make it easier to adjust the LFE levels for different movies.

This design is known as a parallel tuned 6th-order bandpass. There is also such an animal as a series tuned 6th-order bandpass. Dan McGrath from the UK has built an isobaric version of one, and has a pretty good write-up of his experience.