SubZero

Peerless 12" XLS based Subwoofer

A subwoofer project

The project started out actually some years ago, and was intended for my home theater. It ended out as a sub for my stereo speakers.

It took some time to construct both the preamp and the power amplifier, and in the meantime I got around to building my home theater, which also took considerable time.

When I say it was meant for my HT, and ended as a sub for my stereo, it is because I got my hands on some THX certified JBL subwoofers (HT1), and despite what a lot of people will tell you, it's really hard to build a good very big vented subwoofer, especially for a HT, as you'll be ok with it only going down to something like 40 Hz. On the other hand the JBL subs are realy big. 2 times 120 liter!!!!

If you don't have space for such big beasts, then a closed box design with a big amp is the way to go. This is what this project is about.

The construction is based on the famous 12 inch XLS woofer from Peerless. This is a really good subwoofer driver, with some very good thile-small parameters, a reasonable long through, and a good power handling. Made especially for this kind of construction.

Without a passive driver, which would have been another good option, it needs some equalisation to make it play real bass. To make it flat to 25 Hz you'll need 11 db, in the chosen box.

With the 400 W the class-d amp will deliver, you'll have around 99 db at 25 Hz, which I think is respectable.

In contrast my two JBLs will produce around 108 db! ...... that's why I have chosen these for HT.

On the other hand, there is no doubt that the XLS will go deeper and will sound better. Therefore it is better suited for my stereo setup.

The result, the sound

The main speakers are not bass shy, and will go quite deep. They use two ported 8 inch woofers each, also from Peerless, and goes down to about 30 Hz.

Not that much music material  goes below 30 Hz, but the few that does really makes the experience more physical, with weight and a better feel of room.

Some examples:

		Yello - The eye Track 5 "Junior B": First of all this is a track that is very hard to get right. The bass is really heavy and will sound boomy in most rooms. It has high dynamics, and need to be played loud. With the sub on, you feel the movement of the bass, it's a bit like the feeling you sometimes get from the air pressure of a big door closing. It's felt more than heard.
		Tjajkovskij - 1812 (Telarc) First of all it a great piece of music, but of course this particular recording by Telarc, is known for the real canons. About 12:30 and 14:40 into the overture, real canons are pounding. With the sub playing you feel the air moving, and its like the canons have more breath. It simply sounds more real. This is a recording that will really test any bass system, and also made the XLS move close to its limit.
		Hans Zimmer - The house of the spirits I always liked this sound track. Didn't like the movie much, but this is one of Hans Zimmer's best tracks. It's a bit melancholic and dark, but the tunes are great. 8:10 into track one "The House of the Spirits" what sounds to be really really big drums are played. With the sub the deep rumble a bit like thunder is head and felt. The sub gives it weight, and room, an it feels like more physical and present.
		Karunesh - Zen Breakfast This is one of my favourite CDs. Simply relaxing and stunning music. Track 8 "Returning to now": Has an electronic organ that goes really deep. With the sub it's like the foundation of the music is more solid, there is the feel of vibration and depth. Brilliant.
		Jennifer Warnes - The Hunter Track 8 "Way Down Deep", has always been a favourite for showing off bass. But with the sub on, I'm hard pressed to tell the difference ..... conclusion this track simply don't go that deep. Anyway the entire CD is just brilliant, the voice, the warmth and tenderness, together with some really good songs is just great.
		Tom Danley's recordings of fireworks and freight trains (and others download here) With these two tracs, the difference between having the sub or not, is more than obvious. The fireworks, has a lot of energy in the 20-30 Hz region (and probably below), and this is rally felt. A big difference to the system without the sub.

The Woofer

Peerless XLS Series 830500

Technical Specifications:

Simulation of the woofer in a 23 l sealed enclosure (volume needs to be added for the driver it self).

The freeware program WinISD 0.5a7 was used for all the simulations.

As can be seen the output does not resemble that of a subwoofer much. The - 3db point is around 60 Hz and to lower this to 30 Hz will require around 9,5 db of gain. That is defenatly also within range.

With 400 W input we will get around 99 db output at 25 Hz, which is quite respectable. Placed in a corner the output will be around 6 - 9 db higher. This will not meet the THX standard which says 105 db at 20 Hz room responce, but i think it will be quite sufficient for normal use, and it is not far off!

Checking the cone excurtion we see that for 400 W we get 24 mm peak-peak at 25 Hz, which is near the linear limit, so that should be ok.

Group delay is very low ... which is good!

The circuit used for the XLS subwoofer preamplifier

Block diagram of the pre-amp (click here for pdf version)

Input

The input consists of a inverted summing amplifier, which also incorporates a 1. order lowpass filter cutting off at 1 kHz.

The input is designed to accept both high level signals form one or tw speaker terminals, og from one or two line level terminals. The most common use will be in a 5.1 system, where it is fead from the dedicated LFE chanel, but it could easily be used to help in a dedicated stereo system.

Highpass Filter

Next stage is a 4. order highpass filter, which sole purpose is to protect the speaker from to low frequencies. Using a large amplifier could easily exceed the mecanical excursion limit, and damageing the driver.

Bass Boost

The boost section is made by a Linkwitz Transform. This i an almost magical circuit, lowering the resonance frequency to any desiable level, and determining an arbitrary system Q. The section includes two of thise circuits, as it can be desiable to be able to change the lower frequency limit, to adjust the overall room response. At the lowest frequencies, the room will actually amplify the bass, as the wavelength is getting larger than the room size. The rise will normally begin at around 30 Hz and rise with 12 db/octave (2. order). This is the same rolloff charateristics as the Linkwitz Transform, just with opposite sign, and it would therefore make sense to target your Fs for 30 Hz, which should give a teoretic flat in room resounse.

After some trials I have opted for a fs of 25 Hz and a Q=0.5 meaning critically damped. I have also decided to use this sub for stereo only (as I have got myself two JBL 12" units in large ported boxes for movie ....... they don't go as low, probably have much more distortion, but man can they move some air :)

Simulation of two Bessel corrections at 15 and 25 Hz (-6 db points). This gives a - 3 db point at 20 and 32 Hz.

The group delay for the same simulations, both showing about 12 ms at 20 Hz.

Same simulation now just with Speakersim1 being 15 Hz Bessel and Speakersim2 being 15Hz Butterworth. The 3 db point is a bit lower but the group delay is almost the same!

A simulation of the step response for the two systems also reveals very little difference, which is also a bit surprising:

New Alternative (Bagend)

An alternative boost section was shortly tried. It compressed two integrating circuits in series, which will therefore give a 2. order lowpass filter at 0 Hz ..... this should in theori give a flat response under the resonance frequency for eny closed box system. The system is commercially used by the ronowend firm Bagend which they call INFRA^{TM, but has also been described by ESP Audio.}

The topology was simulated, but turned out to have a very high, and rising, group delay, which I think will be very unwanted, resulting in a very muddy bass sound.

Speakersimout1 is the integrator circuit, and Speakersimout2 is the Linkwitz circuit.

Integrator: f3 = 10 and 55 Hz (determined by DC filters)

Linkwitz: f3 = 20 Hz (chosen)

At 20 Hz the group delay is actually the same for both circuit, but a bit longer for higher frequencies, and steeply rising at lower frequencies. Actually bypassing all capacitors in the model resulted in lover group delay for the integrator circuit abowe 10 Hz, than for the Linkwitz circuit!!! Hmmmm, not realy how I thought it would be. Maybe we should try it out anyway :)

Phase Correction

To get a flat resounce around the crossover frequency, you'll need to adjust the phase. The circuit will be able to adjust the fase between 0 and 180 degree at 80 Hz.

80 Hz is the normal crossover frequency for a surround system, but you might want to use a different crossover to get a more semeless transition to the main speakers. This is no problem, but in ths case it will not be able to adjust the phase a full 180 degree.

In practice you should play a 80 Hz tone and adjust the phase until you get the maximum output.

Lowpass Filter

The lowpass filter is the to crossover to the main speakers or satelites, and is adjustable between 30 and 160 Hz.

I have chosen a second order filter, as this will match a sealed box roll off characteristic. If you are using the circuit together with a vented system then just close the vent (roll a couple of socks and pluck them in the port/vent), this will give you a sealed box instead.

The filter is the well known Sallen Key filter, for which the components can be bound by:

Where R1 = R2 = R.

For a Butterworth response Q = 0,707

We would like to be able to shift fc from lets say 30 to 160 Hz with 80 Hz as the center frequency.

With fc = 80 and R1 = R2 = 10k we get C1 = 280 nF and C2 = 140 nF.

Now by making R variable we can shift Fc. We'll use a 5 k resistor in series with a 22 k lin potentiometer for R

P2 is used for adjusting the cut-off frequency. By using a 5 kohm resistor in series with the 22 kohm linear potentiometer the cut-off frequency can be adjusted  between 30 and 160 Hz. By using a linear potentiometer the cut-off scale is almost logarithmic. On the figure bellow the cut-off frequency has been calculated as a function of the potentiometer position (blue curve). The purple line equal the logarithm of the cut-off frequency, for the scale to be logarithmic the line must be linear (straight). As can be seen this is not far from being true.

If you are using other components you can use the excel sheet for calculations, just press here to get it.

Parametric Equilizer

In all smaller rooms you will have problems with room modes (standing waves) at low frequencies. This to my experience is the highest contibuter to boomy bas sound. Most people blame their speakers, especially big vented speakers, that can actually play some real bass, and will therefore excite these modes. People then finds this sound unacceptable and instead byes some small 2 way closed speakers, that have limited output in the low end, and therefore less boomy sound. I wanted to do some thing about this, by lowering the output at the strongest mode. This can be done by a parametric equilizer, that can target a narrow dip or cut at a certain frequency. I theory you would need 2 or 3 equalizers to combat the 2 or 3 strongest modes, but that would make it a lot more complex (Ravel uses 3 in their Performa B-15).

It is not that easy to make a parametric equalizer and I have worked on this for a long time, constructing and simulating.

Take a look at the design here.

See references: Infinity [3], Rane [4][5], Revel Speakers (Performa B-15), Ethan Winer [6]

Simulations

Setting: min freq, max boost, max Q. Result:  15 Hz, + 20 db

Setting: min freq, max cut, max Q. Result:  15 Hz, - 20 db

Setting: max freq, max boost, max Q. Result:  159 Hz, - 20 db

Setting: freq position = 0.75, max boost, Q position = 0.25. Result:  45 Hz, - 20 db, Q=2

(green line is output before the simulation of the speaker for)

The scale of the different adjustments are not linear, and must be handled by adjusting the silkscreen for the knobs.

Scale for the tuning frequency

Scale for Q

Cut and boost scale

The silkscreen for the panel

Result

Schematic and Printed Circuit Board

The schematic is divided into two:

Input summing amp, Thile Transform and phase correction

Variable Low Pass filter, Parametric Equalizer and output

The PCB is a two layer construction using non SMD components

The Enclosure

The enclosure is made of 44 mm MDF (2 x 22 mm), with a separate compartment for the amplifier. It's very heavy and and very dead. I think you could easily make due with only 22 mm MDF. Nothing fancy, but it works quite well. The size turns out 23.1 litre after subtracting the woofer it self.

Measurements

In room measurements using Arta (measured in the middle of the room at the seating position)

Measurement of main speakers + subwoofer. Orange = Right and Green = Left

Left channel + subwoofer, magnitude and distortion .... not bad .... (Using Arta Steps program)

In room measurements made using Room EQ Wizard (measured in the middle of the room at the seating position):

Average measurement of Right chanel, Left Chanel and the Subwoofer. Lower port in main speakers blocked

Right channel alone, lower port blocked.

Right channel alone, lower port blocked.

SubZero alone

Pictures inside out

The final sub

Back panel

Stuffing and wire. The wire has been wrapped not to rattle!

A look into the amp compartment. 500 VA transformer and 22 mF RIFA!

Bitumen and polyester

The plate amp. My 400W amp, 12V PSU for the gates, relay circuit, +-6V PSU for the pre amp, and the preamp.

References