Double Driver Dynamic Loading for Low Q Drivers

Or

"A different way to get excellent sound from Lowther or similar Drivers"

First of all it must be asserted that very topic of this article, namely the use of conventional Reflex boxes with Lowther Drivers has long been a controversial issue. Not only that even Lowthers themselves have long be a "love or hate" topic. Indeed, many people will declare anyone crazy who even bothers with such drivers and worse uses them in such enclosures.

Well, it's Friday afternoon, the inconvenient jackets are off, we are allowed out, the wardens and nurses don't look too closely - nothing anyone has to be ashamed of. So we shall indulge fully in all the lunacy we can and enjoy ourselves. So here then the looniest, balooniest and totally out of it idea so far with Lowthers and similar Drivers. We will even have a simple suggestion (untested) for a DIY Speaker using this system.

One quick warning. If you have any serious background in conventional Loudspeaker design, this will make you cringe BIG TIME.... Better stop reading now. The first time I advanced the original idea any number of professional speaker designers jumped up and told me flat out: "This WILL NEVER WORK".

Only Servodrive Inc's Tom Danley and Jean-Michel Le Cleac'h encouraged me and offered some assistance, so most what is written here is attributable more to those gentlemen than to myself. Errors, omissions and the rest are mine of course, solely....

The Problem:

Low Q wideband Drivers (like Lowthers, Goodmans Axiom 80, Fostex FE208 Sigma and similar) tend to have a weak bass response in normal, easy vented enclosures. This is usually combined with a rising midrange response that levels out to a (usually quite ragged) Plateau in the 2kHz region (ballpark) up to 8....15kHz depending upon the Drivers.

Even with full corner loading and huge reflex boxes these kinds of Drivers just don't give a very even frequency response.

As a result often Horns are employed. Now physics demand that for a Horn to play low it needs to be huge. Even then they are often still not able to offer a well-balanced response. Even worse is that most horns for domestic applications are heavily folded, causing many unwanted resonances to appear. These can severely colour the sound in the lower midrange and bass. Some people maintain that Horns sound bad and propose that Reflex Boxes are the only solution. Let's have it then, shall we?

Looking at an in room response simulation of one of a Lowther DX3 in one of the commercially available reflex cabinets USING optimised corner reinforcement suggests about 95db @ 40Hz. Then we see a broad dip down 6db centred around 100Hz, then climbing response reaching back to 95db @ 400Hz and climbing further from there, reaching around 98dB by 1khz. Adding in the known additional about 6db Lift above 1kHz (not shown on the Graph) on top of the nominal 98db Sensitivity gives us overall a maximum differential of 89 to 104db in room. That is a tonal imbalance of around 15db, with a weak lower midrange fierce and excessive treble.

That is where the point comes in that Lowthers are suitable ONLY for horns. frontloaded, rearloaded, TQWT. Any type of horn, really any will give a better overall balance than a Reflex Box. But of course, horns, especially of the kind with many sharp folds have their own problems, many of which are equally unpleasant but in a different way.

What is worse, the same holds true for ANY low Q Driver (Qt < 0.3).

And that is that. Or is it?

What a bloody quandary....

Marc Wauters has a spot on his web pages for "alternative" loading for Lowthers. It's been there for ages, nothing ever turned up. So, is using a Lowther in anything not Horn(y) equal to squaring the Circle? I almost thought so for a long time....

Even with plenty of help from the source Resistance of a SET Amplifier we still cannot get the Qt of a Lowther above 0.3 (if we add even more resistance we loose efficiency at a frightning rate). Horns we want to avoid, especially folded horns.... Reflex boxes are also out of the picture, if we do not want to apply an EQ and loose efficiency, as well as LF power handling.... And so on and on in a circle....

Mommy! Why did the lights suddenly go out? Why is my head spinning?

At that point I resigned myself to believe that Lowthers as a matter of fact need horns. And that was that. Until I acquired several old Goodman's Axiom Drivers that is. Now I had a pair of the best fullrange drivers in the world (according to Jean Hiraga), the Axiom 80's and a pair of excellent 12" fullrange drivers, the Axiom 201....

The Axiom 80 has a very low Qt, below 0.3.... The Axiom 201 on the other hand where fairly high Qt Drivers (> 0.5). Both have a high 96 to 97 db/2.83V/m Sensitivity and 15-Ohm Voicecoils. Mounted into a 75 Litre, 35Hz Reflex Box predictably, the Axiom 80 sounded thin, weedy but with some of the best, clearest and most detailed Midrange and (for a Dual Cone Design) excellent Treble. In the same box (they where experimental boxes) the Axiom 201 sounded almost too bassy, very strong, warm and gutsy, but not very delicate. Interesting was that the high Q driver totally lacked that "toppy" response of the Low Q Driver, instead sounding subjectively rolled off in the upper midrange. So what if we use both Drivers.... The High Q Driver will carry most of the bass as it peaks up there anyway. In the lower midrange both drivers will combine their Output to achieve about 6db more SPL output. This will likely match the overall SPL of the raised plateau of the Low Q driver, which will take over and carry pretty much the whole sound above a bit over 1.5kHz (as it is 6db louder), or the idea goes at least.

Said, tried and for a while I listened to the system with Axiom 80 on a small open baffle seated on top of the Reflex Boxes with the Axiom 201.... This still had glaring response problems, but sounded better, a lot better than either driver alone in the box. It gave most the strength of each Driver and few of their weaknesses, so basically both Drivers can be seen as truly "complementary".... One is the complement to the other in ranges where the other does not work too well without much "infighting".

A Possible Solution

So I thought - hey - what about mounting these Drivers into one large box together, with a subdivision, and run the Drivers parallel. Said, not done.... Why?

I dimly remembered a Pair of Speakers I had owned in the early 80's. A Coral 5-Way speaker with more drive units than you can shake a stick at.... But this thing had a 12" Woofer and an 8" Woofer operating in the SAME box. No Subdivisions. Vented. It sounded surprisingly good too, was around 100 Litre in Volume and managed 100db/W/m above 40Hz.... When I let the Idea of the "One Box, two very dissimilar Drivers, no Subdivisions" Box fly on the Joenet, it was the old AC-DC Song all over again:

"Shot down in Flames, shot down in flames!"

Boy, do I hate it when that happens.

Even more do I hate it, when it happens with everyone who is knowledgeable on the subject telling me IT CANNOT WORK, but not bloody telling me any good reason WHY it cannot work. All that came after asking "But why can it not work?" was some blubbering about the dissimilar parameters and such, but that was the Idea in the FIRST place.... Basically, do that to me and it's the safest way to make go ahead and do it anyway.

Even better, one of my most admired Speaker Designers (Tom Danley) said "hmmm...., Very interesting question..." and Jean Michel dug out some stuff on double reflex systems, so at least there was a documented precedent on systems using multiple resonant combos in one enclosure.... I worked (with help from Tom and Jean Michel) ways out to model at least at a good first approximation such systems and then simply build such a box. The resulting speaker system is nothing short of stunning, with around 102dB/W/m sensitivity and in room response down to 25Hz. Here is the picture, for those interested.

I have since found a number of Examples of commercial Speakers that also employ Drivers with different Parameters (and even sizes) in one Box. This includes Wilson Audio (MAXX, X-1), Nestorovic and the VPMS Subwoofers. I'm sure more exist. Even if not, at least I'm in good company.... ;-)

But how does one model such a system using existing software? Eventually it crystallised that one could use two approaches to "approximately" simulate such a design using conventional Speaker Simulation software (like my "XLBOX" Spreadsheet) by combining the results from two methods:

We all agreed that if one drew out the electrical equivalence models of the drivers behaviour, connected the drivers in parallel and attempted to simplify the model to a single equivalence circuit with new values this would not really be doable. However, we had a consensus that modelling the two drivers as a new single driver with combined parameters would give at the VERY LEAST some reasonable indication of the possible behaviour of the combo. It was clear that the correlation with reality would be sketchy, but it would get at least a starting point.

Adding the Equivalence Volume (Vas) of both Driver and adding the Cone Surfaces (Sd) is part of mathematically assembling the new "compound" Driver. The new Resonance Frequency, Qes/Qms/Qts becomes the geometric mean of the two drivers values, so basically square the actual two values, add them up, divide by two and than take the square root. The Voice Coil DCR (Re) simply becomes the result of parallel resistors.

If for arguments sake we take the Lowther DX3 Driver and the Eminence Alpha 8 Driver as an example we get:

Vas = 54 Liter

Sd = 200 cm^2

Fs = 56 Hz

Qes = 0.22

Qms = 6.33

Re = 8 Ohm

SPL = 98db/2.83V/m

Vas = 20 Litre

Sd = 200 cm^2

Fs = 72 Hz

Qes = 0.58

Qms = 13.0

Re = 5.34 Ohm

SPL = 95db/2.83V/m

(Both SPL Values are T/S Parameter derived, not from manufacturer datasheets).

From this we make the compound Driver and get:

Vas = 74 Litre

Sd = 400 cm^2

Fs = 64 Hz

Qes = 0.44

Qms = 10

Re = 3.2 Ohm

SPL = 102db/2.83V/m

This is interesting. The new Driver looks perfect for a simple Reflex Box using the "EBS" Principle. There we set the Box Volume to the Driver Vas and tune somewhat lower than the resonance frequency. Say we tune the Box to 50 Hz and make it 74 Litre's large and mount our new, purely theoretical driver in it, we get what?

Well, we get something that looks in room good for about 100db/2.83V/m average SPL, with only a minimal rise towards the midrange (about 1db) and a in room -3db Point of 43Hz. The largest Dip is 3db at 100hz.

If we look at the pure predicted 2-pi response without room influence, the new compound Driver in its Box clocks in at 102db/2.83V/m with a -3db point of 70Hz and a very gentle rolloff. This rolloff is easily offset by proper application of room gain, absolutely classical EBS basically.

Admit it. That looks a LOT MORE like it than the previous same size box with only the DX3 in it.... While there will still be a slightly emphasised Midrange region (the DX3 swings itself up to almost 104db at the highest peaks) we now have something that overall looks quite okay....

Average REAL 100db/2.83V/m with (unfortunately) a 4 Ohm Impedance and in room (excluding narrow room modes) 43Hz to about 10kHz wit no more than 3 - 4db Peaks and no dips of much more than 3db until we hit the cone/whizzer resonance's around and above 4kHz....

To me at least that looks eminently listenable.

However, as said, this thing has a high "fudge" factor, but it seems that at least such an approach could get us into the ballpark.

This one is really only applicable after our fudged compound driver has given us box-size and tuning and is more of a reality check.

Here the Box Volume gets divided between two "separate" boxes according to the ratio between the Vas of the Drivers. As we set (for convenience sake) the Box Volume to Vas in the example, simply enter the Vas as Vb and the tuning as the 50hz we determined earlier.

With that we can see the individual Drive Units Responses, overlay them and do a little eyeball adding and apply some pen damping. Surprisingly the Curve from that is close to what the compound driver gave.

This then does it and I must say that having done the same for my "Magnificat" Speaker using the Axiom Drivers, the measured reality is quite well in step with the predictions.... It is noteworthy that still no electrical crossover is used, though it might be expedient to try a 0.5mH Coil or thereabouts in front of the second woofer if the midrange is too much lifted. I got away without such.

Further Consideration

So far we have concentrated ONLY on the on Axis LF frequency response. This of course being the reson d'entrée for this article is obviously normal. But not only does the use of two LF Drivers allow us a more even Bass response, it also offers a few further benefits, but also slight challenges.

The Drivers should be arranged in a vertical array, spaced very closely together. Timealignment between the Drivers is best ensured by placing the drivers on the front baffle in such a way that the imaginary line from ear level (seated) is at a height between the two drivers. The Lowther should be placed above the other driver.

Then, using (for example) interstation noise from a FM Tuner as source (any white or even pink noise will do) have someone tilt the Speaker backwards slightly backwards in small increments until you hear the sound "lock in". This is a quite distinct behaviour though often with placement of both drivers at a suitable high the time alignment is often more than "good enough". If you perceive a notable improvement with the Box slightly tilted back please fit some material to elevate the front of the speaker suitably or adjust the spikes (if fitted - it is a good idea to so) so that slight backwards tilt is retained. The tilt required should be no more than a few degrees.

It is actually also quite useful to have this two Driver Array in the lower Midrange for the overall sound dispersion. Depending upon the size of the Cone and the mounting any driver will start to have a very wide dispersion up to frequencies related to the cone Diameter where the radiated sound "cone" extending from the centre of the Driver narrows with rising frequency. So at high Frequencies the Driver will radiate in a fairly narrow beam but the lower we get the wider the angle of sound becomes.

Most modest size drivers (including Lowthers) start to "beam" above about 2 - 3kHz. This means that the sound at frequencies lower than that is "splashed" quite liberally around the room. A vertical 2-Driver array as used here will have a radiation pattern of about 50 degrees vertically and 80 degrees horizontally down to a few 100 Hz. This a lot more controlled and allows an overall more precise stereo imaging by reducing the sound reflected back from room walls, floor and ceiling.

The rising response in the upper midrange from the Lowther will also mean that by the time we have passed the 1kHz or thereabouts mark ALL the sound will seem subjectively to emanate from the Lowther as it contributes most of the sound.

Moreover, because of the now quite flat on-axis response of the overall Speaker listening almost on axis is now a good option. I personally found repeatedly and consistently with a wide range of speakers that the best imaging is obtained from a very WIDE placement of the speakers. There the imaginary axis's trough the drivers� centres should cross each other notably in front of the listener, who should be at least as far from the speakers as they are spaced apart.

The last benefit accrues in a very unexpected manner. As both Drivers have substantially different resonance frequencies AND parameters, placing them in parallel with each will significantly reduce the reactive nature of the Systems impedance at low frequencies. True, there will be now even more bumps than before for a reflex tuned box, but the magnitude is significantly lower�.

Finally, in all simulations the Lowthers hit's a 5 Watt Power Input Limit at around 35 - 40Hz. This means that below this frequency permissible input power for linear operation falls. How bad this is subjectively must be left to you assessment, but it is advisable to keep the Amplifier at or below 300B Output Valves (< 10 Watt into 4 Ohm). So even as it stands with improved LF behaviour these speakers will never make room shaking bass. But that is not the Idea anyway.

If you want that sort of thing, buy a Cervin Vega Earthquake Basshorn with suitable Cervin Vega LOUDspeakers....

Notes on Tweaking and implementation

Finally some notes for the selection and modification of the "fill in" woofer. It should be a fairly high Q unit of broadly similar cone size (usually slightly larger is a good idea). So with a Lowther my take would be a musician/Pro Audio 10" unit in a much larger enclosure than discussed so far.

Such drivers, but also the discussed Eminence Drivers often suffer from a number of resonances. Usually it is advisable to cut/burn a fairly large hole in the centre of the dustcap. This can be used as an acoustical lowpass filter, try aiming for a 1.5kHz cut-off when the other Driver is a Lowther. The Drivers used in the examples have cheap stamped frames. Damping these with dynamat or bitumen mats for sound deadening cars and self-adhesive lead strips does oodles of good.

Placing some absorbent Foam on the inside of the Basket spokes (if there is enough space) can cut reflections through the cone. In the Box ideally the wall behind the driver is fitted with some form of strong diffuser, operational down to lowish frequencies. Deflex is NOT good enough in that respect. I had good success with closed cell foam "eavesfiller" intended for use with corrugated tin sheeting. If no diffuser material can be found use felt.

In addition both cone and remaining dustup should be lacquered with some kind of damping/tuning Lacquer. I personally use these days the famous C37 Lacquer from Dieter Ennemoser. This is not cheap however and for those who do not have the prerequisite green stuff for C37 many other choices from the classic Dammar Varnish to the "boomjuice" offered at www.boomjuice.com and shellac based hairspray. Take your pick.

I would also still recommend to apply Marc Wauters $0.98 Lowther Tweak to the Lowther (it is reversible and will not affect the resale value of the Driver) AND the use of the new DX4 style Phase Equaliser (the "Holey Doorknob" one) which is available from Lowther.

A little practical design example

Now if you already have a suitable Box and Drivers, well cut another hole and fit the "extra" fill in unit. If this is not the case and you have so far shied away from Lowthers and similar systems, how about a Box using a readymade "Woodstyle" Tower Enclosure from Madisound with a Fostex FE208 Sigma and a Eminence Delta 12 LF per side?

These will not cost overly much and will be quite easy to assemble and tweak.

So, we need:

From Madisound (www.madisound.com)

"Tower" - Oak, Black Oak 73.3 Litres, 14.5" W x 38.75" H x 12"D, grill w/ Fastex Fasteners, aprox. 60 lb,/each cut for CB/GB-cup $ 300.00/pair

From Fostex (http://store.yahoo.com/fostex/)

FE208Z $ 168 each, so $ 336 the Pair

From Parts Express (http://www.partsexpress.com/)

Eminence Delta 12 LF $ 105 each, so $ 210 the pair

The 12" Woofer is being placed on the front of the enclosure. This will be a tight fit, the hole needed for that woofer needs to be routed out very precisely, centred and with exactly 11.187" Diameter. The woofer will most likely have a VERY SLIGHT overhang on both sides of the 12" wide enclosure, as it is exactly 12.187" outer Diameter. The woofer should be placed below the Fullrange driver but as close to the Fullrange driver as possible. It is possible to fit a connection Cup provided for Bi Wiring and Bi Amping. I do not consider it necessary though.

The Sides and Rear/Front walls should be covered by what Madisound calls "Polyester 1" Batting". this comes in 3' wide and 8' long pieces. One probably need about two for these speakers.

http://www.madisound.com/accessories.html

The rear of the Magnet of both Drivers should also be covered with this stuff, perhaps best applied and held in place with some strong string.... The Box will require tuning to about 45Hz. My Software and experience suggests using a single 3" Diameter Hole as starting point. Should that be found non ideal, in terms of a too one note bass, use instead three 3" inner Diameter Reflex Tubes, around 6 - 8" each in length. In any case the ports belong onto the rear of the Speaker and near the floor, their centres about 6" above the lower edge of the Box, or higher if required.

Some notes at the predictions from my own Simulation Software for this compound system. Excluding all room effects the Speaker will manage a -3db point of 60Hz (anaechonic) combined to a mean sensitivity in the 101db/W/2.83V region. In room, placing the Speaker 1.6' from sidewalls toed in by about 30 - 45 degrees and with a distance to the rear wall 4' the speaker will have -6db point of about 38Hz, as shown below....

In order to see how a modified Beta 12 LF and a Fostex FE208 Sigma would integrate I simulated the combination in Calsod. This is a very precise program that takes into account drver positioning and many other factors. The Calsod model of the Fostex was made based on Data measured by the German Magazine "Klang + Ton" based on being 15 degrees off axis, the Eminence Driver is based on the manufacturers data.

Shown below is a screenshot of the combined response, please note the db scale showing a 20db range rather than the more usual 40db range often found with plots in magazines.

In the region above 85Hz to 10kHz the predicted response at around 15 degrees off axis horizontally and on axis of the Fostex vertically fits into a +/-2db window!!!! For a 3m Listening Distance the SPL Level at 2.83V is 93 - 94db, suggesting 102 - 103db sensitivity for 2.83V or 99db to 100db per electrical Watt.

In the impedance plot the reduction of the reactive nature of the LF is visible.

I need to note that I have NOT build this particular Speaker. I'm pretty convinced that it will provide quite substantially decent sound, but (obviously) cannot guarantee it.

Also, both chassis will need a lot of run-in.... Best done at 20Hz sinewave with about 1mm movement from the Fullrange Driver for a few days, followed by at least one day of pink noise, about as loud as you can bear standing next to the speakers (I know this is loud).... As the sensitivity of the Drivers is so high and the normal operating power is so low, I'd say with normal listening levels these Speakers would literally take YEARS to break in....