Last revision, 7 Januari 2006

Modular 0.1 Januari 2002

Bert Doppenberg visited me to hear the results that I obtained with my system. He pointed out to me that there were large resonances in the bass and the midrange was to laid back due to the correction network of the Jericho. So, again, time for a change. That same day I heard the bass horn of Peter van Vegchel, which sounded very impressive. Nevertheless I decided that I would not go for that idea, one of the reasons being that I didn't want to spend very much time on it. After a thorough search I decided to go for a closed bass box which could be realized easily in the Jericho horn. That would give it a volume of 100 ltr. A unit that should work well is the Peerless XLS-10. This speaker has an Fs of 19 Hz so it can go very low and it is fast because it is only 10 inch diameter. Its sensitivity is rather low, 88 dB/1 W, so subwoofer amplifiers are required. Through Internet I found the place with the best price was Antwerpen, only one hour by car from Eindhoven (www.diyparadiso.com). Their built-in filters and volume controls make it easy to integrate them with the system. I changed my Jericho's by closing them in two places: at the throat and high at the back. See the red parts in the drawing of version 0.1. The plates are fixed airtight with a lot of glue. I called this version 0 because it was a try out version and the plates could easily be removed when a return to the horn was desired. The total rebuilt of both boxes in April 2002 took less than two days. The result is also shown on the photo (the decorations are made by my son Toon).


When I switched them on the first time I heard a lot of bass coming when I turned up the volume at the woofer modules. At the listening position I heard almost no bass. I experienced the same effect with the full range horns but not so strong. After a while I realized that it sounded quite like speakers being out of phase. So I changed the phase on one module and indeed the bass response at the listening position became much better and after some fine tuning of the cross over frequency and the volume the overall sound became really acceptable. The bass itself was low and more heavy than from the horn its speed was, however, far below my expectations. This has improved a lot in the two weeks that the speakers have been used (40 hours in total?). I am now very happy with this change, which has brought me two active subwoofers at a price of one third of one commercial subwoofer that I owned before and was inferior in frequency response, speed and distortion. I have turned the volume of the system to maximum for only a few seconds and I was amazed on how clean it still sounded. And it was really loud because at 20 dB lower volume the floor is already strongly vibrating. In the evening I listen at much lower levels and I now hear lots of low frequency information that I have never heard before and it does give a positive thrill to the music. A final change I made is to leave out the correction network. This improves the clarity of the mid frequencies and the bass volume can be increased on the woofer modules. This does make the lack of very high frequencies more clear but overall I prefer this change. 

Modular 1.2 Summer 2002

So this was the end of my horn period. I still think that the idea is very good but it is very difficult to obtain a well balanced output over the entire audio spectrum. One problem I noticed with version 0.1 was a strong resonance at 39 Hz, much higher than the lower resonance frequency of my room. This frequency turned out to match exactly the distance of the speaker units straight to the opposite walls. The easiest solution I could think of was to make a separate box for the bass. This way the bass box can be positioned for optimal frequency response and a separate box for the FE208S can be placed for optimum stereo imaging.  This makes it a real modular design, on the photo version 1.2 is shown.


The resonance is now less pronounced and spread over the range of 32-40 Hz. This makes the lower bass somewhat stronger so it is also heard at low listening levels. Again I was able to re-use the wood of the horn enclosures. Version 1.1 made in June 2002 had a very short life. I made them 30 ltr. because I wanted to obtain a low cross over frequency with the bass. On Internet I found information from Joost Riphagen that 15 ltr. would make the FE208S sound much better (www.melhuish.org). With some left over pieces of would I was able to try this. I concluded that Joost was right and it almost did not influence the bass output. According to Joost it should go down to 110 Hz, but I have put my filter to 140 Hz to obtain (by ear) a balanced sound. The last change is that the output signal of the Kit One does no longer go directly to the low input of the amplifier but I made a resistive divider which goes to the line input. I use a 60 kOhm Caddock and a 100 kOhm Noble potmeter in series with the 13 kOhm input. This gives an adjustable bass level which be tuned over a large range. The advantage is that the bass level can be adapted if the bass level on a CD is too high or too low.

Modular 1.3 Oktober 2002

It is quite common to add a tweeter to a Fostex wideband speaker. I heard this also at Peter van Veghel, and I read that Joost Riphagen uses one. The question is, of course, which tweeter matches the FE208S. My first guess was the Fostex FT17 horn tweeter which has the right sensitivity (97 dB/W) and a very extended frequency response (35 kHz). An alternative could be the Visaton DHT9, at the same price of around 35 Euro. I contacted Benny Glas about this topic and he advised me the Beyma CP22. I found that its specifications and response curves looked very promising, but I could not find any review of its sound. Benny assured me that it does not sound like a metal dome tweeter which he and I don't like. So I decided to try and I went again to Antwerpen because Benny was by far the cheapest with 73 Euro. Others ask prices up to 112 Euro. The unit is built in a new front of the top box and it is connected with a 10 Ohm Caddock resistor, a 1.5 uF capacitor and a few centimeters of silver wire in parallel with the Fostex. I selected the capacitor from left over stuff by measuring the loss factor using a Fluke RCL-meter. The Hovland Musicaps had the lowest loss factor of all but I only had four pieces of 0.22 uF and I combined them with small foil capacitors of 1.0 uF which were second best at the loss factor. This change is one of the differences that are recognized as a clear improvement upon the first second of hearing them. A lot of clarity is added to almost all instruments and voices, but the sound is never sharp or overdone, so I am very pleased that I finally made this change and I recommend it to all FE208S users.


Modular 1.4 Summer 2003

After listening to the 1.3 for longer it became clear that they sounded worse at higher volumes. This was not very important because I almost never go higher than ~80 dBA. But one starts to think about what is going on and one day I incidentally touched the sidewall of the Fostex enclosure. Although this was 4 cm thick it was quite strongly vibrating. I remembered that a friend who studied architectural acoustics once told me that when you feel vibrations you will certainly hear them. So why was this sidewall vibrating so much and how to make something better. The first question is related to a more fundamental one in my opinion: why does everybody use wood to construct speaker boxes. I guess that it is because it is easy to use and it is cheap. But, is wood from acoustical point of view a good material? Answer: no! Wood is light weighted and stiff but to damp sound and/or vibrations the material must be heavy and flexible. Try to make a loud sound by hitting a piece of lead rubber with a hammer and compare it with a piece of wood and a sheet of metal. Then you will know the difference. Unfortunately, as an amateur constructor I cannot obtain and machine these kinds of exotic materials. Another possibility would be to change the shape. Rectangular boxes have well defined resonance frequencies but, again, other shapes are not so easy for a hobbyist. I tried unsuccessfully to find a thick walled plastic tube and I have been thinking on how to use sand or concrete. Finally I decided to try a octagonal tube of wood covered with one of concrete and a layer of sand in between. When the was finished I had found out that it was impossible for me to make this shape regular so I did not even start to make one of concrete. I did make three internal braces from sidewall to sidewall and filled up the V-shaped open spaces with the concrete that I already bought.


This box sounded already dull and dead. So I decided to finish it by wrapping around one layer of rubber and two layers of dense and heavy carpet in a color that matches the wood of the woofer box. This gives a total wall thickness of 4 cm, which is the same as with version 1.3. The internal volume is 15 liters which is adequate. I filled it up completely with wool and hard foam (made to damp wooden floor resonances). The tweeter is mounted on top without any enclosure and the filter is at both sides so that the wire length is minimized.


On the photo you can just see the Caddock series resistor to adapt the level, on the other side is the capacitor. I first changed one old box for a new one for comparison. The difference in sound was larger than I expected. The new box sounds more clean and clear by such an extend that it took me only a few seconds to realize it. The second test was to listen how much sound comes through the sidewall. With my left ear pressed against the sidewall I could easily hear music coming throuhg with the old box. With the new box I simply heard nothing coming directly from the box, but only the sound that come through the air from the speaker cone. With two new boxes in place same the improvement is also evident at my listening position. This is really a worthwhile upgrade and at very low cost, less than 50 Euro. I found an interesting article in the Journal of the Audio Engineering Society (volume 51, page 234) that gives scientific support to my experiences with box resonances. Below is the abstract.

 

On the Acoustic Radiation from a Loudspeaker's Cabinet
Kevin J. Bastyr and Dean E. Capone   234
Although it is well known that the walls of a loudspeaker cabinet vibrate at low frequencies, the authors determined the actual sound energy being radiated. Initially, a vibrometer was used to measure the surface velocity along the surface, and then a boundary-element method was used to model the acoustic radiation. By applying this method to production loudspeakers, changes in the internal bracing allow the designers to control these unwanted surface resonances, which are true radiating sources. The effects of enclosure vibration affect the overall radiation pattern of the loudspeaker and must be included in the design.

Modular 1.5 November/December 2003



Differences between 1.4 and 1.5

After listening for more than one year to the XLS-10 I became aware that its sound is deep but not fast or detailed. So I started looking for alternatives. Two options with quite different specifications were the final candidates. The BD-15 designed by Bert Doppenberg to match the dynamics of high efficiency wide-band units and the Eton 581/11-HEX. I finally chose the latter one because it goes lower, it has excellent impulse response and distortion figures and it fits well in the same enclosure as the Peerless. Of course it would have been nice to build new ones for the BD-15s but that would have required new material and, much more prohibitive, a lot of time. The Etons were installed in less then 2 hours. I started with a closed box as with the XLS-10, although both Benny Glas and Bert Doppenberg strongly recommended a reflex enclosure. Already with the first seconds of music I heard that the Etons were much better than the Peerless. After one month of waiting I decided to mount the reflex pipes that I obtained from Benny. To my surprise there was hardly a difference in the sound. Bert gave the answer to this: a reflex box is similar to a closed one if too much damping is applied. Luckily it was quite easy to remove most of it and indeed now the speed of the bass increased. Nevertheless the overall sound was still quite dark. The final trick was to adapt the low pass filter. I ended up with an extremely simple filter of one capacitor (0.22 uF Hovland Musicap) and two potmeters. One tunes the cross-over frequency and the other the volume. The estimated -3 dB point is now at 200 Hz and the slope only 6 dB/octave. Now the overall sound is very good, the bass is very deep, but it also has speed.

Modular 1.6 Summer 2005



New in Modular6:

In May 2005 I finally decided to compare the performance of my Eton's with Bert Doppenberg's BD-15. After getting used to the deep bass I noticed that the sound of bass drums and guitars did not have the attack of the real instruments. Bert claimed on www.audioasylum.com that his units are able to produce a life-like drumsound. Further I was interested in comparing my Theta CD-tranport and -DAC with Bert's Twindac plus laptop. So I listened to the Swing for about one hour and went back home with two BD-15 units in my car. It was quite easy to sell the Eton's via Internet, but I did not want to throw away my old enclosures. So their volume had to be increades from 110 to 160 liters. This has been done by putting a very large wooden cap over the front. I found a nice plate of 26 mm solid beech that was just big enough for two caps. I finished them with sandpaper until they were very smooth and then put wax on them until they shining. The disk that was taken out for the BD-15 was just big enough to mount the Fostex and the Beyma in, so the total speaker now looks much better. The sound is again significantly improved, the bass attack is much better and closer to the acoustical instruments.