I like my Quad ESL speakers a lot. They've a very accurate and natural sound, but they've one limitation. Bass stops at about 50 Hz so the fundamental frequencies of the lowest bass notes and of bass drums isn't reproduced. This is the reason why I've experimented with a subwoofer to provide the missing bottom octave.
The driver and box
My subwoofer is on its second iteration. The box remains the same, the driver remains the same, but the electronics have changed.
The driver is not exotic. It's a 10 inch Kenwood KFC-W2500 sold as a car audio driver back in the early 1990s. I originally built the subwoofer as a vented box tuned to reinforce the output of the driver itself with a resonance around 30 Hz and electronic equalization at about the same frequency which resulted in a reasonably flat response in a 42 litre box. However, I was never entirely happy with the sound of the ported box, which could be a bit "boomy".
The purpose of the recent redesign was to achieve better performance at the lowest possible cost. That's why it uses the same driver in the same box, a convenient size as a coffee table. However, I have now closed the port as it seemed likely that a closed box or "infinite baffle" design would achieve a flatter in room response. This of course meant that the electronic equalization had to change to suit the modified box because there is no longer any reinforcement from the port resonance. It also means a compromise in that the driver will have to be driven harder at the lowest frequencies. A single ten inch driver may not sound like much, but some commercial "subwoofers" use smaller drivers in smaller boxes. And in any case I don't listen to music at a ridiculous level and my intention all along was to place the subwoofer in the corner of the room so that it benefits as much as possible from room gain. For me this isn't a big deal.
On its own, the 42 litre box is not optimal for the driver working as a closed box. This combination results in a resonance of around 46 Hz though the -3 dB is at 64 Hz. Both of these are rather high for a "subwoofer" and the speaker couldn't be expected to reproduce much low bass unless assisted in some way.
Luckily others have been here before and help is at hand. the "Linkwitz transform circuit design calculator" spreadsheet calculates values to plug into a simple op-amp based equalizer circuit for any given combination of box and driver. I used it to design a shelving filter for flat response to 20 Hz with my box and driver.
Kenwood quotes an Fs of 28 Hz, Qts of 0.327 and Vas of 74.25 l for my driver. Plugging these figures into the spreadsheet together with my box volume of 42 litres results in the graph shown. The blue line is the frequency response for the subwoofer without equalization, the red is the response of the equalization and the green is the expected result of the equalizer and subwoofer working together.
The maximum boost of 15 dB is applied only at the very lowest frequencies of interest, where there is little recorded information in music. As such, this won't stress the driver as much as you might imagine.
The rest of the circuitry
My equalizer is built around two 25 year old LM833 dual op-amps which I happened to have hanging around. However, almost any op-amps would do. The frequencies of interest are very low, after all, so you don't need something exotic which can cope with high slew rates. The NE5532 is easily overkill and actually you could build it using one LM324 quad op-amp.
I've sketched out the circuitry below using my usual "CAD" tool. Each part connects through to the next:
The power supply for this circuit is very simple in design. A plug in switched mode supply rated at 12 V and 200 mA was the starting point. It provides enough power and enough voltage for this circuit. Following it is the circuitry shown on the right, using two capacitors in series, paralleled with resistors and LEDs to create an artificial "ground" between two rails at +6 V and -6 V. This only works if the plug in power supply is completely isolated. If the "0 V" rail from the power supply is connected to the real earth or to the 0 V of your circuitry then the -6 V supply is shorted out.
The LEDs are optional and provided here only to provide a power on light. While testing, having two of them makes it obvious if you've shorted out one rail of the power supply. The circuitry above and below the artifical ground should be identical as otherwise a DC offset will be created.
Note that both of the op-amp chips also have local capacitors between their +ve and -ve supply rails.
The stereo input from the pre-amp connects through two 4.7 uF capacitors (to block possible DC offsets) through 66 k ohm resistors (a simple mixer) and is tied to ground by a 150 k ohm resistor to avoid DC offset. The values are not very critical, but they should be relatively high in order that the high pass filter that they create is at a very low frequency which doesn't affect the subwoofer response. It may seem un-necessary to have these capacitors, but without them there is a risk of DC voltages being propagated through the circuitry.
The two 10 k ohm resistors result in this op-amp having a gain of 2. This means that the following circuitry can operate on higher voltages, leading to lower noise.
Not shown is a very low value capacitor (0.01uF) connected between the -ve and output pins of the op-amp to remove the chance of high frequency resonance.
Adjustable low-pass filter
The adjustable low pass filter is built around a stereo 1 k ohm potentiometer which is connected to a knob on the front panel. This allows adjustment of the low pass filter frequency.
The initial circuit came from an online tool for 3rd order Sallen-Key filter design.
I selected the Bessel filter (for maximally flat group delay and hopefully better integration with my main speakers), a low pass frequency of 50 Hz and gave values of 4.7 uF for the two capacitors. The resulting circuit had two 300 ohm resistors where I have put the potentiometers. Adjusted to 300 ohm I should therefore have the correct Bessel response -3 dB at 50 Hz. However, it's obvious that we're only adjusting two of the three stages of the filter and at other settings of the control the filter won't still behave as a Bessel filter. Adjusted all the way to the right so that the potentiometer has zero resistance, the filter reduces itself to a first order filter, which turned out to be quite useful when testing as at least there was some sound to hear.
At this point we insert the Linkwitz transform which adjusts the speaker to be flat within our desired pass-band.
The output stage is similar to the input. Again it is based around a single op-amp with a gain of 2.
The input to this stage includes a level control. This is connected to a second knob on the front panel which is used to match the level of the subwoofer with the main speakers.
The output features a resistor to protect the op-amps internal circuitry should it be shorted to ground and a capacitor to block any DC offset. Again, the capacitor value is large to keep its effect out of the way of the desired response of the subwoofer.
Not shown is another very low value capacitor (0.01uF) connected between the -ve and output pins of the op-amp to remove the chance of high frequency resonance.
Connecting to the hifi
Input to the equalizer was taken from the wire which runs between my pre-amp and power amp. The high impedence of the input of this circuit means that it has no effect on the signal so it's perfectly OK to split one output and take it to two inputs.
The output goes to an additional power amplifier working in bridged mono mode which drives the subwoofer itself.
The previous incarnation of my equalizer (designed for the ported box) was in a much more visible box, but my box making skills aren't up to much and it was a bit ugly. For that reason, this equalizer was built into a very compact black plastic box. This is almost invisible on the shelf between the pre and power amplifiers.
The spreadsheet used to create this graph can be viewed here.
You can't make useful measurements with music so while sitting where I usually sit to listen to music I played a copy of the test tone CD from Arve Bersvendsen and used my trusty Radio Shack SPL meter to measure the level at individual frequencies. Plotting this for the subwoofer alone, my ESLs alone and the two combined resulted in the three different frequency responses. The result can be seen in the graph to the right. I'm quite happy with this result which shows that the subwoofer fills in underneath the main speakers exactly as it was intended to.
While listening to the tones it was obvious that the output from the subwoofer rapidly became less audible as the frequency rose. This was the intention of course. Quite soon it was obvious that there was no chance of audibility with the main speakers turned on, but from 360 Hz upwards I couldn't hear anything at all from the subwoofer in a quiet room with the main speakers turned off. Similarly, in the opposite directly, I couldn't hear the 35 Hz tone or anything below it from the ESLs.
It's difficult to make a direct comparison between the subwoofer now and its previous incarnation with a port and the earlier equalizer because it would take a lot of time to adjust between one and the other. I made an attempt at measuring the response of the old subwoofer and equalizer a few years ago and didn't think it was worth making public. It was definitely not so flat as this. As such, I'm very pleased with the result using the new equalizer with the Linkwitz transform - it makes what is a very basic and low cost subwoofer perform to a much better standard than might be expected.
How does it sound ?
Subjectively, the subwoofer doesn't really have any sound. When the subwoofer is switched on, it sounds as if the ESLs have developed a new ability to produce extremely low bass, without any overblown "boom". That was my aim, and I'm happy to say that I think I succeeded.
The steep low pass filter means that there is very little sound from the subwoofer except sub-bass. This is good because you don't want to hear where a subwoofer is, especially when it is in the corner of the room where I have placed this one to maximise the room gain that comes from being next to the floor and two walls. Even if you sit right next to the subwoofer, and completely off axis to the main speakers, everything still sounds as if is coming from the main speakers.
Because there is just one 10 inch driver, there is a limit to how much sound it can make at low frequencies without distortion or damage. However, I'm not trying to make my eardrums bleed or annoy the neighbours. Music doesn't have very much content at such low frequencies and so it is unlikely that the subwoofer will be driven into gross distortion in practice and therefore I doubt that this will be a limitation in use.
LinksSource for the Linkwitz transform spreadsheet
Siegfried Linkwitz's own website
Okawa Electric Design's online filter design tools
Arve Bervendsen's Test tone CD including good suggestions for using the test tones to improve the sound of your hifi at zero cost.
My pre and power amps
My main speakers
The subwoofer power amp
I've more hifi component reviews on my hifi page.
Link back to my home page