Copyright 2010 Troels Gravesen


This is a speaker I've wanted to do for a long time. A 3-way design from 10" bass drivers supplemented by relatively small midrange drivers. I liked my 3W-Classic and I still use the L26-3W from time to time and I wanted to try this concept from the best possible drivers, thus the SEAS W26 nextel bass drivers, Crescendo tweeters and not least the 5 inch M15CH002 middriver. T29CF002 was chosen due to excellent performance in the CNO. A second choice will be the slightly cheaper T29CF001.
It's also an experimental speaker and I never finalised the cabinet work as it evolved into other speakers. So, I suggest you go with the Focal inspired design seen at bottom of page allowing you to tilt all drivers for optimal listening point.

The construction was started in February '08 and has been on hold for several months due to other projects and despite not having made the final cabinet design, I needed to get it off my chest because I told a number of people about it and they keep asking how it goes. So, here it is:

Commercial speakers I had in mind for this construction were Spendor SP100, Naim DBL and Harbeth M40. All classic 3-way speakers being around for many years and apparently still - I guess - holding a small audience. The Naim DBL is a huge 65 cm wide speaker, featuring a 15" bass driver surprisingly married to a small 5" mid and a ScanSpeak 8513 tweeter, where Spendor and Naim feature 12 inch bass drivers.
WAF is low for classic 3-ways in rather large rectangular boxes. High-WAF speakers these day have super-model properties and I'm not sure only acoustics alone call for this approach. So, this is a speakers for singles or those blessed with a tolerant spouse.

What I also wanted to do here was to deviate from all my other constructions in using an either closed or aperiodic box for the bass driver. If we want at least some bass from small two-ways, the vented box comes in handy but with large bass driver not having a too low Qt, other options are available and may provide a more dry and firm bass response with better transient response. As discussed later, you may use this construction vented too. Taste cannot be argued so in all three options are available according to your taste or belief.

It's always exciting to set up a new driver for test for the first time. I used my 34 litres test cab for the W26. Frontpanel is 30 x 50 cm. Microphone is placed 50 cm from frontpanel and input is normalised for 2.8V/1 metres distance. From the SEAS data file my expectations were so-and-so. Some dip at 650 Hz and a bit peaky around 1600 Hz.
I have three emotional responses to the first time a feed a new driver the CLIO MLS signal: 1) Bollocks!, 2) Hmm.... and 3) YES!
Having made MLS measurements thousands of times you know how you want the reproduced MLS signal to sound. There are drivers that will shred your ears and you know it's major trouble and there are drivers that will caress your ears and you know it's a potential winner. And then there are a lot of drivers in between....

The W26 is a YES!
Next I play some music without any crossover attached. I play it fullrange and despite a sloped response you may already enjoy what you're hearing. Some drivers just sound awful due to serious cone break-up or whatever. The W26 is so good I think it might even do in a two-way system with a 1-1.5 kHz point of crossover. I'm not saying this approach would make a great speaker, but it seems possible. (I later tried and it didn't work).
Secondly I play some bass lines (Fender) from a test CD and finally a kick-drum. The latter I compare to the good old
Goodmans 12" Axiom 150 mkII driver, which is the king in my speaker collection of making you feel the impact of a kick-drum. Now, the W26 doesn't pull the Goodmans off the throne, but it does deliver a punchy kick and I was surprised to also hear speedy transient attack from snare drums.

Left: SPL from SEAS data sheet measured at metre distance in 28 litres test box. Front panel dimensions not given.
Right: My measurement of the same driver in a 33.6 litres test box with 30 x 50 cm front panel dimensions. Measurement done at metre distance normalised for 2.8V/1m and no smoothing!

The drivers
W26FX002 + M15CH002 + T29CF002
Download specs from SEAS by clicking headings.

After numerous "6-inch-projects", it's nice to have a chunky 10" bass driver in your hand. It carries a massive 134 mm magnet delivering the magnetic force to a 2 inch voice coil gap. The M15CH002 is the exact opposite; very light-weight due to the small neo magnets. The T29CF002 is a heavy beast compared to most other domes due to the thick alu frontplate and extraordinary thick-walled rear housing. The nextel coating looks gorgeous but is fragile to nicks and scratches, so take care in handling the drivers. The T29CF001 seen above was finally replaced by -002, the Crescendo.

TS data, bass driver

It's always interesting to try a cross-calculation of claimed data and see if they fit with the math. To the left the most important TS data from the SEAS data sheet. To the right I've inserted selected data to the LspCAD box calculation software and indeed we see a firm correllation between the SEAS data and LspCAD prediction. Efficiency seems spot on at 89-90 dB/1W/1m. SEAS says 89 dB. So far, so good.

- my measured TS data:

TS data on both drivers, not fully burned-in, but they probably won't change much from here. Good correlation, I think. A driver well suited for vented, closed or aperiodic systems.

The mid box

Why a section on the midcab alone? The M15CH002 is a tricky fellow with a potential "problem" in the 1-2 kHz range, that is when the baffle isn't right. We have a dip here and it adds a significant fingerprint to vocal presentation if you have flat upper midrange up to 1 kHz followed by a ~3 dB dip in the 1-2 kHz area. Using the M15CH on a wide baffle didn't work - I've tried the PMS cabs and had to abandon the PMS-NEXT project - it didn't sound right. Too much presence - in the wrong way. Read below.

Left: M15CH002 center mounted on a 400 x 1000 mm wide baffle, 340 mm from the top. I won't go into details with the four graphs, only that these measurements were made from flat to various degrees of curving the panel.

Right: Mounting the driver in a 200 x 300 mm baffle (my 10 litre test cabs) produces the red graph. The driver center is 95 mm from top. Now, this looks a lot better. Adding 130 mm to the baffle height, makes the blue graph. Less baffle step loss but also a dip at ~1.5 kHz. The beginning of the PMS trouble.

I had great hopes for the smoothness of the M15CH driver from Dickason's measurements in Voice Coil; reading above. This almost looks as good as the MCA15RCY on the PMS baffle. Vance Dickason uses a 203 x 430 mm (8" x 17") baffle and unfortunately does not provide info on where on the baffle the driver is mounted. So, let's see how the M15 performs on a similar baffle in my lab.

Left: M15CH002 on a 200 x 430 mm baffle with driver placed in the middle. Readings done at 0, 10, 20 and 30 deg. Not quite as smooth as the VD readings. VD uses a gated sine wave technique, apparently at little more gently on FR presentations compared to the mercyless MLS readings.

Right: M15CH002 on a 200 x 300 mm baffle with driver placed 95 mm from top. Readings done at 0, 10, 20 and 30 deg. Now we have an even better reponse in the 1-2 kHz region. Nowhere near as bad compared to the PMS baffle.

Overall we have an even power response up to 4000 Hz +/- 30 deg., allowing a high point of crossover to the tweeter.

Left: Four samples of M15CH002 on the 200 x 300 mm baffle. Right: Impedance of four samples of M15CH002 impedance in free air. Fs = ~80 Hz. The minimum spread in performance seen here is one reason we pay more for these drivers.

Back to "the mid box" intro picture. Leaving the M15 in the test cab, various bits and pieces of MDF and cardboard can be used to simulate the upper part of this W26-Classic. The cab is 36 cm wide and adding a 45 deg. chamfering to the sides of the mid cab + minor chamfering of mid cab top (the cardboard) makes the above seen frequency response. This is clearly manageable in terms of making an overall smooth response in the so important upper midrange/lower treble (600-2400 Hz).

Why this odd interior for the middriver? Well, when SEAS has done all that possibly can be done to eliminate reflections from the driver's rear structure, the least we can do is ensure that our cabinet doesn't reflect any energy back through the cone. So, I've tried to design a box where the rear energy is absorbed as good as possible and due to the slots in the rear, we have an aperiodic system, reducing impedance peak at resonance frequency. Same as for the bass, only here managed by adding compressed damping material to the cavity.

The Woofer Box

Below 3 possible scenarios for vented systems. Actually it is my intention not to use a vented box, rather two ScanSpeak Variovents in an aperiodic system, but to get an idea of performance I had LspCAD suggest QB3 and SBB4 boxes.

Above the QB3 box at 75 litres and a vent tuning of 32 Hz. Here we get an almost flat response down to 40 Hz. Not bad at all, but 75 litres is quite a lot. I was looking for something around 65 litres.

The SBB4 tuning predicts a 60 litres cab and a vent tuning of 32 Hz like the SBB4. The response at 40 Hz is down 1-2 dB compared to the SBB4.

My intended 65 litres cab suggest an F3 around 35 Hz and this volume is what I'll be going for. We can tune the port a little higher and gain a decibel or two in the 50-100 Hz range. But hardly necessary here. F3 = 35 Hz is deeeeep if you want to go this route.

Placing the W26FX002 in a 65 litres closed box would produce the above seen response. We are some 8 dB down at 35 Hz compared to the vented system.

How the performance from W26FX002 in a 65 litres aperiodic box will be is to be seen. The rear panel will have a small separate section for either vents or Variovents.

- why vented boxes (for small speakers) ?

I've often had the question why I always use vented boxes and the answer is simple: To get more bass from small designs. An additional 3-4 dB at 40 Hz is clearly audible and the bass may be more precise from a closed box, but who cares if you can't hear it. My speakers are always 3/4-1 metres from the rear wall and some 1-2 metres from side walls, so I can't rely much on room gain. If you don't like the sound from the vented systems, stuff the vent with a pair of socks! Dynaudio sometimes supply a polyester foam plug for those who suffer excessive roomgain.

I cannot simulate the response from the driver in the same 65 litres enclosure fitted with acoustic vents. We'll have to wait for the measurements to see how it performs.
The Variovent is device sometimes causing much debate. What is an acoustic vent? An acoustic vent allows some ventilation at low frequences and virtually none at higher frequences. What's that supposed to be good for? Neville Thiele studied the Variovent in detail and concluded it didn't offer any advantage over a properly designed vented systems. I don't subscribe to that statement. It depends a lot on the actual driver used.
The general advice on the use of acoustic vents is simple: Try it! Which doesn't offer much comfort to those addicted to math. But box calculation based on Thiele/Small data really only apply for speakers with a Qt = 0.35, which only very few drivers have. Based on experience we use speakers with high Qt in cabs too small and drivers with low Qt in cabs too large compared to what math may suggest.

Comparing the impedance peaks of a driver in free air and a perfectly closed and empty box should reveal the same peak height, so we have no loss in the closed box. Obviously the resonance frequency is higher. Adding damping material to the closed box we see a decline in peak height; now we have some loss. Adding an acoustic vent may produce a significant reduction in peak height depending on how open the vent is. So, the system Q-value is lowered (= more loss) from applying an acoustic vent although we really can't speak of a Q-value of an aperiodic system.
Basically a closed box has 2nd order roll-off, a vented system a 4th order roll-off and the aperiodic system a 3rd order roll-off.
In this system I'll try all 3 options and see how my 20 wpc SET amp responds to the various impedance profiles generated - and obviously try to evaluate the quality of the bass performance. It'll be interesting to A-B test the various options.

Cabinet Construction

For the final set-up the bass cabs must be tilted some 6-7 deg. and the mid cab placed vertically.

Pics from cab construction
(Identical to
JBL L100 Reconstruction)

Making the mid cabs

Cabinet Damping

Cut the following sheets of felt material: 4 pcs 32 x 21 cm (top and bottom), 4 pcs 24 x 13 cm (sides) and 2 pcs 30 x 13 cm (rear reflector).
The slots to the rear must be damped with some folded acoustilux/MDM3. More details to come.

Simulated Crossover

LspCAD Predicted Response:

Final measurements of drivers, no crossover

Left: impedance and phase of all drivers in final cab. Red/black = bass, green/yellow = mid, red/purple = tweeter.
Right: SPL of bass driver i final cab. SPL bass driver merged at 250 Hz. Blue = minimum phase. An overall system sensitivity of 91 dB/2.8V seems likely.

Left: SPL and minimum phase of mid-driver in final cab. What really pleased me here was an even response from 500-4500 Hz making overall crossover construction very easy.
Right: SPL from tweeter in final cab. No problems here either.

Mid cab and test crossover on top.

Crossover fine-tuning, version 2 = final crossover.

Crossover version 2
. As can be seen, a lot of small changes to version 1.
This is what happens during actual measurements and listening tests.
R2021 is the actual resistance of 4.00 mH coil, so no true resistor here.

Complete kit only available from Jantzen Audio,

Kit can be bought with/without drivers.

Please state where you live for full quotation incl. shipping.

All kit and component prices may be subject to change and are always to be confirmed by Jantzen Audio Denmark.

Download Kit Sales presentations here  (pdf file)

All technical questions to me at

Please specify tweeter option on order.

Measurements, version 2

Left: Actual response at tweeter height, 1 meter distance, 2.8V input.
Right: Point of crossover between mid and tweeter is 3 kHz.

Left: Horizontal dispersion at 0, 10, 20, 30 and 40 deg. (tweeter height)
Right: Vertical dispersion from bass height (middle) to 10 cm above tweeter height. Red and purple are the extremes.

Left: SPL 1m/2.8V, merged at 300 Hz with woofer nearfield response.
Right: Impedance of final system, an easy load. Blue = electrical phase.

Left: CSD, 40 dB. Right: CSD, 20 dB.

System performance

Left: 8 wpc 300B SET. Middle: 20 wpc Audio Mirror 6AS7 PSE. Right: 100+ wpc, modded Rotel RB-981.

First of all: Do not use any capacitors (for mid and treble) of lesser quality than the Superior Z-caps used here. I'm not saying these are the worlds best, but you will never get the best from these magnificent drivers if not driven from a high-quality crossover. I do not recommend spending mega bucks on foil coils and the like, use your money on the caps. Buy perfectly wound and baked coils made from round copper - now available from a number of suppliers.

Next: If you think this speaker sounds bad, take a look at your CD-player, your phono cartridge and phono stage, your line-amp and your power amp and see if there are bottlenecks causing sonic degradation. Or your room may be totally inadequate for hifi systems, i.e. forget about modern homes with concrete walls, tile floors and no bookshelves or curtains! Sell the leather sofa and buy one with some heavy fabrics, and bring in some floor carpets, etc.

If you still think it sounds bad: Well, we have very different taste in loudspeaker performance....and I suggest you read these comments from Art Dudley/Stereophile. My point in referring to this again is that we're into the best of traditional high-end hifi here. I won't even try to describe the sound in detail because it's meaningless.

I will however comment on the bass performance because of the use of Variovents. If you want deeper and a more voluminous bass, add vents (find dimensions above). As can be seen from the impedance plot, this is an overall very easy load on your amplifier and due to the Variovents we have a dry and "short" bass performance, free of any bass boom overhang on transients.
And one more thing: I am particularly surprised by these speakers ability to reveal depth and space around the performers. Three-dimensionality is some of the best I've ever had and this follows up to even very high listening levels, a clear sign of very good drivers displaying very low distortion. The M15CH002 is indeed an extraordinary fine driver.

The 300B amp was on the stand when I brought the speakers to our living room. I didn't expect 8 watts to do much here, but it could actually drive the speakers well beyond my wife's usual listening level. And the mid and treble sounded really good.
The Audio Mirror brought some more punch to the bass and overall suitable for most situations, that is if we never throw a party or want to impress out hifi friends with some butt-kicking bass.
Give these speakers some 35 wpc PP minimum - or one of the very best 50+ wpc solid state amps you can find and it will deliver. With the modded Rotel RB981 on the stand I'd not even consider a vented version. The 10" bass driver in its aperiodic 65 litre cabinet is doing just great; solid, tight bass - but you will get more and - subjectively - deeper bass from a vented design.


This speaker did not turn out to be a classic design as it was planned. Sloping the cabs is a must due to size and due to the bass driver handling a good part of the midrange. The construction shown above will work excellent with a 5-7 deg. tilt of bass cabinet and vertical mid cab.
Thinking of possible cab designs, the Focal approach seems the best way to make the best of the current construction. Something like this:

Thanks, Rodolphe, for the pics!

This will allow the drivers to be individually positioned to suit any listening distance.

Suggested cabinet design la Focal.


Feedback from Cyclop builders