Coaxial ‘Mildness’


Onken ‘charged’ Selenium 15″ coaxial high efficiency loudspeaker

As my previous speakers did not satisfied me, and, moving towards the single ended tube amplifier topology, I started to dream about a loudspeaker system that has to meet the following needs:

  • to be SET friendly (high sensitivity and easy to be driven by a low damping factor single ended topology amplifier)
  • to produce live-like slam (in order to enjoy not only hard &heavy music, but also to have a good ‘base’ for jazz and large scale classical recordings reproduction)
  • to be easy to implement and tweak, regarding the box construction as well as the crossover

The solutions considered at the time varied from open baffle – O.B. by using PHY HP drivers  – until intricate multiple way horn projects. The journey was a short and lucky one: after almost an year of dreaming and researching, I ended up with the ONKEN style loudspeaker idea: just decided to try a coaxial 15″ –  with paper suspension for the woofer – in a Onken style enclosure build in a variable volume fashion.

Why coaxial? Because it should be the most coherent source given the dimensions of the woofer and the enclosure, and also easy to implement for the lazy diy carpenter 🙂

Why Onken? I didn’t know at the time, that’s why I was eager to experiment with it. Just seemed somewhere in between OB and reflex enclosure, easy to implement and simply to tweak.

The drivers


After a small research, I ended up by choosing the driver – SELENIUM 15CO2P 15″ coaxial driver,with a paper cone 15″ woofer and 2″ medium-high driver, with an efficiency of  97dB-SPL (1Watt-1m), proudly made in Brazil. The real data sheets out of the box are a bit different compared to the official pdf  file, but not by much. These babies were build by SELENIUM company for professional Public Address applications, not for home made hi-fi freaking projects 🙂 Still, given the price of one unit at the time (approx eur 170 for a coaxial unit) I decided to gave them a try.


A ribbon driver ended up to be used as super-tweeter, after using a dynamic Selenium driver first. P Audio and Eminence tweeters were also tested, but the beast was the ribbon one. A Fountek Neo CD2.0 ribbon tweeter was finally put into position, with marvelous results! Here there are the characteristics of  NEO CD2.0-v2.

The enclosure

Calculating the Onken enclosure was possible by using Cyr-Marc Debien excell sheet made public by the

onken enclosure digaramAlso, I was using the upper drawing (sorry, I don’t remember the source) as inspiration for crafting my own enclosure. The dimensions form this illustrative diagram are not the same with those of my own enclosure!!!

The Onken style enclosure was created from two boxes, each of them missing one panel, the small one being introduced in the big one. The two boxes fits perfectly on height, but on the width the outside box is larger than the inside box; if the inside box is centered, two lateral slits remain, those being the Onken reflex ports!  Now, if one wants to vary the total internal volume of the enclosure, he just slides the small box in or out from the big one, in small increments. Thus, the internal volume could be varied and tests at different volumes could be performed.


The outside box was made from 2.5cm (1″) thick high density MDF, as the small box was crafted from 1.8cm thick raw MDF, and stiffened with an internal cross bracing ( 3D cross inside made bay 4cm X 4cm section hardwood sticks).

To preserve the lateral symmetrical reflex ducts, soft wood (fir) sticks were used, three on each side of the small boxes. This way, the whole structure is becoming stiffer, and the small box is perfectly aligned relative to the big box. After auditions and measuring tests were performed, inside box was fixed forever into the proper position by using glue and few wood screws (don’t be afraid to use wood screws, the myth you are ‘infected’ with it’s not true).

Finishing the enclosure

The enclosure was covered with real wood veneer, installed by hand by using soft glue. White pine essence was used, then, the veneer was sanded and covered with two layers of wood lacquer. The enclosure is supported by three big  iron spikes (4.5cm in base diameter)

Stuffing the enclosure

Inside the enclosure, behind the coaxial driver, was stuffed some synthetic wool, long fiber type, about 500 grams for each enclosure, as well as some deflex foam pads mounted exactly behind the back of the main driver. Also, small acoustic foam  pads were mounted on each second inside walls (in ‘dead end – live end’ style). All these treatments were done by experimenting few weeks, and not by calculations, all the changes being done with patience and in small steps 🙂


The coaxial driver was delivered with special  PA style connectors. If you will use connectors for the drivers, they will oxidate in time. I just remove the connectors and I soldered the wires directly to the driver lugs. For the bass driver I was using 2.5mm solid copper wire, cotton isolated.. For the medium and super-tweeter drivers, silvered copper multistranded Ecosse CS2.2 cable was used. A pair of decent quality binding posts were installed on the back wall of the enclosure, supporting both spades and banana connections.



Well, creating and tuning the crossover it was a long an painful process. I have tried several topologies of X-overs, and the best one prove to be no crossover at all! Joking … The best musical response was obtained with minimal crossover: juts a capacitor in series with the driver, as well as a L-pad attenuator for equalizing the different sensitivities of the medium-high driver and super-tweeter driver, related to the bass driver. Don’t try this at home, you may hurt your drivers! I was using this way because I have PA drivers that resists up 5oo watt rms, as exciting them with less than 10 watts ! I am joining my system at quite low levels, so my drivers are not working in a dangerous regime.

Burnin’ in!

Please be aware that a paper cone woofer, with paper/cloth suspension, needs months of burning-in until it gets to the final sound. Mine was reaching the smoothest sound after about six months!!! It took a long time as I was not using special burning-in device or techniques, nor high levels of signal to accomplish that in a short time. Take this into consideration before to throw your drivers away for sounding too shy or not heaving deep enough bass 🙂


I will be cruel: the sound of my coaxial Onkens is far from perfection! And ist is not so neutral… Heresy!

But man, musically speaking, they produce a coherent, powerful yet mild sound. You could spend a full night, playing CD-s or vinyl records, and you will never get  ear fatigue. The whole spectrum is reproduced even at lower intensity levels. The voices and acoustical instruments are naturally reproduced, acoustical recordings being a pleasure to listen to. The highs are accurate and pretty fast, being delicate and non obtrusive.  The soundstage is large, but the depth is maybe not so deep like multi K loudspeakers.  The lower register is deep enough even at lower levels, having a pleasant energy in upper bass register which makes things rounded when listening classical recordings.

The only think, hard for me to mention, is that these speakers are not so linear, and if one wants high efficiency loudspeaker driven by a single ended tube amplifier, something must to be sacrificed ! It’s not possible to make this loudspeaker quite linear without sacrificing the sensitivity. Also, the lowest bass notes are not perfectly 100 % controlled, considering the small damping factor of a single ended amplifier. Despite these two imperfections, the Mildness produces a ‘mild’ sound indeed, very fast and accurate, portraying quite precise acoustic music scenes, having that big diameter woofer slam even at the lowest audition sound pressure.

Here there is a spectrum analyzer graph of the speaker during the early tests. A Selenium super-tweeter was used, and the crossover, third order type, was a two ways P.A. commercial design modified just for testing. The mike was set at 1.7 meters from the coaxial unit, on the speaker’s axis, so the measurement was affected by the room boundaries.

Also, here there is pictured the spectrum figure for a measurement made at 5 cm from the coaxial unit, the measurement microphone being mounted at 30 degrees off-axis

The appearance of these big babies into my old system (DHT tubes pre and mosfet classA finals) is showed here. During the tests/breaking in period I was so exited, that I wasn’t using super-tweeters in order to properly observe all the changes.

A detail on coaxial driver:

This is how the Mildness integrate into my current system:

If someone needs some info regarding this project, for non-commercial purposes,  don’t hesitate to write me an e-mail.

Happy listening  after building !

© attitube 2011


find me:
grid (at ) attitube (dot com)
wordpress counter