The answer to that involves hours and hours of intense mathematical argument with John. The proof is in the pudding, though. John's sub is incredibly loud and very musical for its size - I haven't heard the other John's sub yet but apparantly it is much louder and has an even wider dynamic range, so I'm confident I'll get something good.

Well, the figures speak for themselves:

6th order bandpass (2 chamber, driver in middle, chambers individually ported).
91 litre box (split 75/16).
Single Eminence Kappa Pro 12" driver (400w, only cost £100!)
Dimensions approx 60cm x 37cm x 42cm. Weight 20kg.
Materials: 1/2" plywood, 1" on top panel (weak point), 8x1/2" cross bracings, 3x2.5" (top) and 1x4" (bottom) drainpipes for ports.

And it manages: -3dB 32-120Hz, -20dB 23Hz.
Free air efficiency: 92dB/w/m (cars give you about 10-20dB more)
Free air max SPL: 118dB.
Free air response: 35-110Hz +/- 1dB.
In-room max SPL: >126dB at 36Hz (actually off the scale of my SPL meter)
In-room response: +/- 10dB :) (bloody rooms)
In-car max SPL: untested, my guess is 20dB in-car gain at 35Hz, around 135dB.

It's not hugely efficient, but I did design it for very low frequency response. This is a trade-off with any subwoofer. The lower frequency bass you want, the less efficiency you'll get, unless you make it much bigger. The other John (who's joining empeg soon btw) designed his sub around a higher -3dB (about 45Hz) and a larger enclosure of 120 litres, giving an efficiency of around 96-98dB/w/m in free air.

The enclosure cavity sizes, driver parameters, and port sizes are all tightly linked in bandpass designs. I used a combination of a few utilities to find the right design ("Dosbox", "WinISD" and one of two of my own). The most important and most often ignored parameter when tweaking is excursion. In my design, excursion is within 1% distortion xMax for pass-band at 250 watts, by which time it's reached about 115dB. A similarly designed sealed enclosure would probably be hitting end stops at <110dB at around 40Hz.

Before sealing the box completely, I wrote a little utility to measure the impedence magnitude and phase response. In vented enclosures, the impedence phase crosses 0 degrees at the port frequency. This gives a good measure of how far off your port lengths are compared to ideal. I started off with long ports and gradually sawed them down to the right size after several measurements. After sticking the box together, I measured the ports as being within 1Hz of the intended tunings.

As for that "tight" sound... what this actually means is having a low group delay, or at least a smooth, similar group delay. It is true that almost any bandpass or vented design will have a higher group delay than a sealed enclosure. However, this only holds true from around 70-100Hz in my design, and even then it's only around 3-5ms. This happens to be faster than the ear can measure at such frequencies. Below 70Hz, it rises to around 10ms, and a peak of 20ms at 35Hz. At such low frequencies however, the human ear is very poor at recognising group delay. Also consider that the sealed enclosure would have stopped outputting appreciable amounts of audio below 40Hz.

There's only two things I would have done differently had I designed it now:
1) Make it bigger, up to 120 litres. The parameters were fairly critical at 90 litres for the intended response curve.
2) Flare the ports, or make them wider. It "puffs" when it's near max wattage. This wouldn't be audible if my treble system could match the SPL of the sub :) (but I'd mess my ears up)

If anyone else is considering building a subwoofer, try out various designs in an appropriate program, and see for yourself the advantages of a vented or bandpass enclosure.


- John (from empeg)

(The above may not represent the views of empeg :)