From Hi-Fi World - June 2009 issue

READ ONLINE

BUY THE MAGAZINE (back issues subject to availability)

Designing Loudspeakers

WD25T EX – The next step, by Peter Comeau

I am often asked ‘can you send me a crossover for these drive units’ and the correspondent goes on to name some mishmash of drivers that they’ve picked up from eBay and plonked in a cabinet. Of course I have to point out that crossover design is far, far more complex than just calculating theoretical component values for a standard second or third order crossover.

Clearly the role of the crossover is still not fully understood amongst many DIY speaker builders who put more emphasis on the drive units than anything else. In fact I’ve spent half my life (and am about to spend another third of it the same way) making even fairly non-descript drivers sound stunning simply by making sure the crossover gets the best out of them.

Careful crossover design can make a bass-midrange unit with characteristics that make it sound ‘ard and ‘orrible into something highly revealing and natural sounding. Good choice of crossover slope can make a treble unit which sounds ear piercingly jarring into something that has a silky presence that is a delight to the ear.

That is not to say that crossovers can solve all ills. It certainly helps to have good sounding drive units to start with. In fact the better the drivers the easier it is to design a good crossover that creates a stunningly revealing and natural sounding speaker.

Which is where the concept for WD25 EX came from. The past two articles in this series have concentrated on the stand mount WD25A EX and now it’s time to meld the crossover work into the WD25T EX. In short, take a 26cm bass midrange unit that produces better midrange qualities than it has any right to, drop it into an enclosure that yields extended, tight and clean bass performance in any room, and marry it to a treble unit par excellence, hence the ‘EX’ moniker.

The goal – to make a speaker which practically disappears when you are listening to it – a speaker which is so revealing that you can hear right through to the performance with, apparently, nothing in the way.

It’s a tough goal. In fact each version of WD25T has achieved it to a remarkably high level. In the EX I wanted to take the goal one step closer to perfection.

CROSSOVER WIZARDRY

So, how is it done? Is there some sort of crossover wizardry required? Does there need to be some unique circuit that magically processes the audio signal before it reaches the drive units?

Thankfully, no.  We just have to aim for certain technical parameters and then start the process of listening and refining. The engineering requirements are straight forward. We require an optimally flat frequency response AND power response in room, perfect phase integration between the drive units through the crossover region, and an off-axis character that matches the on-axis performance.

The latter is often overlooked. Too many speakers are designed solely to give a ruler flat on-axis response. These types of speakers become both room and seating position sensitive. They work providing the listener is seated on the measurement axis and the speakers are far from walls, floor and ceiling so the off-axis reflections are considerably delayed.

These types of speakers don’t work for Mr Average. Usually the room is too small to place the speakers well away from the side walls, and the rest of the family wants to enjoy the music as well as the listener in the ‘stereo seat’.

In addition these types of speakers are often genre dependant. They may suit classical music but not jazz or rock. They may sound great on ‘audiophile’ recordings but terrible on ‘ordinary’ discs. It doesn’t take much to show up the imbalances in the real-world performance. However if a speaker is designed with careful analysis of the off-axis, both horizontal and vertical, responses and with consideration of the total power response in-room, then the speaker should sound natural and well balanced wherever it is placed. Also listeners are free to sit where they like – they may not receive a holographic stereo image but they should still enjoy a lifelike image even when sitting well off-axis.

Similarly the speaker is not music dependant. Certainly it may be revealing and clean and detailed to the point where poor recordings sound only half decent and good recordings sound great, but in both cases the music should shine through. And there has to be something inherently right with a speaker that is capable of playing all genres of music without favour.

One more technical aspect is worth consideration. The impedance of the final speaker should not present a difficult load to the amplifier. To have an impedance graph which oscillates wildly between peaks and troughs throughout the frequency range usually means that the speaker becomes amplifier dependant. In other words it may work brilliantly with some amplifiers but not with others.

There’s no clear cut differentiation between good and bad amplifiers here. Not many amplifier designers understand that the speaker is a voltage generator and current distortion device. Some do have cognisance that a speaker load is far more complex than an 8 Ohm or 4 Ohm resistance. In any case you stand a better chance of the design suiting a wide range of amplifiers if the speaker doesn’t give the amplifier a hard time.

NEUTRALITY AND REALISM

There’s another school of thought that claims that a tonally neutral speaker, one with a flat response, will make the performance sound real. That ‘ain’t necessarily so, as the song goes. Our measurements are all based on sinusoidal analysis, even when the test signal is a pulse. It’s the only way to see, graphically, what is going on.

However music is rarely based around steady-state waveforms. Yes, you can analyse the musical waveform into a series of sine waves at any instant, but the human appreciation of music is based upon what happens over a period of time. In addition our brain is pre-programmed to respond to transient information. That’s why the most interesting and arresting music has sudden starts and stops, silences broken by loud interjections.

Any musician will tell you that timing is the essence of making music come alive, that the silence between the notes is as important as the striking of the note. And, of course, the ‘note’ is highly complex in its beginning, its decay and its harmonic structure.

These are all aspects which are nigh on impossible to resolve by pure technical analysis. As design engineers what we can do is find a design system which works, usually by trial and error, and apply that system to any new design. But to reach the goal outlined above has to be done by musical analysis, in other words a lot of listening and fine tuning.

Having just finished the WD25T EX crossover I was demonstrating it to my colleagues at World Designs towers. Our chief engineer was sitting in the preferred ‘stereo seat’. I was sitting at the side of the room, very close to one of the speakers, around 60 degrees off axis. I put on a simple recording by the Wailing Jennys that starts off with acoustic percussion and female vocal. At the point where the vocalist starts to sing we looked at each other – she was palpably there, in the room, a vocal presence that almost seemed to have a physical body. You couldn’t see it, but you could hear it. The point is that she was there, in the room, even to me sitting way, way off axis. That’s what I call real.

THERE'S MORE THAN ONE WAY...

The first step was to apply what I’d learnt with the WD25A EX to the WD25T with the addition of the SEAS Millennium treble unit. You might think that plugging the same crossover in would yield similar results. But it doesn’t. The reason is that the WD25T has a larger baffle area and a baffle that reaches right down to the ground.

This physical change alters the way power is radiated from the drive units into the room. In addition the larger, sealed, enclosure of the WD25T has a very different bass performance to the smaller, aperiodic ported, WD25A. So the overall balance of the speaker is changed.

There’s more than one way to configure a crossover to achieve similar technical performance but with a very different acoustic result. The crossover frequency remains at 2kHz, this being dictated by the acoustic roll-off of the 26cm bass midrange unit, but more energy is now imparted to the treble output by raising the coil in the electrical 2nd order crossover to 0.42mH.

I have shown before that the best acoustic performance is given where there is a mild 1 or 2dB depression through the crossover region for the on-axis response. This doesn’t necessarily translate to a loss of acoustic power in this area, however, due to the wider dispersion, and therefore greater radiating power, of the treble unit.

The level of this depression is set by the resistor in series with the 0.42mH coil.  Increase this resistor and you flatten the transfer curve of the treble crossover, decrease the resistance value and you sharpen the transfer curve as the crossover turns over. Feel free to play with this value; reducing it produces a slightly brighter performance, increasing it produces a warmer performance. You only need a change of, say, 0.5 Ohm to appreciate the difference.

You can see the overall response of the system in Fig. 1.  together with the crossover slopes overlaid. Following the typical Linkwitz-Riley responses the crossover  point is -6dB at 2kHz. The larger 0.42mH coil chosen, with its series resistance, gives a small shelf in the treble response below crossover, but this is the compromise chosen on the basis of listening tests!

The phase coherence of the drive units and crossover is easily indicated just by reversing the phase of one of the units. If the acoustic crossover is phase perfect then there would be just a sharp notch at 2kHz. I’m very satisfied by this result (Fig. 2) which marries up with the listening result where it’s not easy to discern the individual drive units and the stereo image is rock solid at all frequencies.

Figure 1.  Response on axis and the individual drive units showing the crossover slopes.

These approximate to acoustic 3rd order with Linkwitz-Riley characteristics.

Figure 2.  Response on axis and crossover (red) and with one driver reversed in phase (blue).

The deep notch confirms excellent phase integration between the drive units.

You can also see, in Fig. 3 and Fig. 4, how the WD25T EX measures off-axis. For the speakers to sound tonally balanced in a room the character of the response off-axis must look very similar to the on-axis response, both horizontally and vertically. This is a tough trick to pull off but, I believe, necessary if the speakers are to present a realistic sounding, natural musical performance.

Figure 3. shows the responses on-axis, 15 degrees horizontal and 30 degrees horizontal. You can see, now, why the WD25T EX sounds so realistic even when sitting to the side as the character of the response is maintained right up to 15kHz.  Similarly your seating height makes little difference to the response you’ll hear either. Figure 4. shows the response at 5 degrees vertical above and below the listening axis.

Figure 3.  Frequency response on axis (blue) with response 15 degrees and 30

degrees horizontal off-axis.  The character of the speaker is maintained well

off-axis so room reflections will sound tonally identical to the direct sound.

Figure 4. Frequency response on axis (blue) with response 5 degrees above and 5

degrees below. The character of the speaker is largely unaffected by seating height.

Figure 5. In room response, taken as an average of multiple mic positions,

showing excellent integration into room.

Figure 6. WD25t EX impedance. Apart from the droop to 4.2 Ohms at 5.5kHz,

which won’t cause any amplifier a problem, this is a nominal 8 Ohm load.

Overall response, in room, is indicated in Figure 5. which is the averaged output across a number of microphone positions with the speakers placed near a rear wall. From the on-axis semi-anechoic responses you’ll note that the bass output shelves down below 200Hz. Placement near a rear wall brings that output back up again and the room gain maximises the bass extension offered by the closed cabinet loading.

Again this response character is very similar to the on-axis response through the midrange and treble which augers well for a natural and realistic sounding musical performance in any room.

Finally, the impedance plot shows that no amplifier should have trouble driving these speakers. There are no sharp peaks and troughs in the impedance through the frequency range and the marginal fall to 4.2 Ohms between 5 – 6kHz won’t bother any amplifier – to all intents and purposes this is an 8 Ohm speaker.

Figure 7. Actual crossover circuit and values for the WD25T EX.  Bass coil, L1,

is a ferrite core inductor with internal resistance of 0.2 Ohms, treble coil, L2, is

an air core with internal resistance of 0.9 Ohms.

Overall, then, the design and technical goals have been met and we can confidently claim that this is the highest performance version of the WD25T that you can get. There are a few constructional details, such as housing the crossovers external to the speaker cabinets, that will be the icing on the cake and, I hope, next month we’ll see an independent review in this very magazine.