Signboard of Tonochi's Audio Room       Site Map日本語
Home Gaudi NOBODY About Me Misc Info Contact
Home > NOBODY > SS-309 2011/04/03 created
2015/10/01 updated

SS-309: Three-way Loudspeaker System

 SS-309A
Description:
Three-way floorstanding loudspeaker system. Drive-units: Fostex T925A (horn-loaded tweeter), Fostex D1405 (midrange driver), Fostex H400 (wood horn), Fostex FW305 (12" woofer). Enclosure type: Vented-box.
Dimensions: Woofer module: 402mm(W) x 752mm(H) x 352mm(D). Weight: ?kg.
Network: Not included.
Frequency range: 25Hz〜40kHz
Sensitivity: Tweeter: 108dB/W, Midrange: 104dB/W, Woofer: 95dB/W (@1m)
Cost: approx. 70,000 yen (enclosures). Woofers: approx. 44,000 yen. Midrange driver+horn: approx. 180,000 yen. (2015/10/01added){Tweeters: 55,600 yen. Total: approx. 350,000 yen}
History: Made in 2001. The cables were upgraded in 2009. The woofers were renewed in 2009. The midrange module was upgraded in 2012 (Rev. A). The tweeters were renewed in 2013. Now used in Gaudi.


Concept

The concept of SS-309 is an improved version of SS-307. Of course, the sound quality should be better than SS-307, though, the better appearance was top priority of this project.
The enclosures of SS-307 were made of ordinary plywood, not painted, only waxed, and looked shabby. Moreover, after many upgrades and customizes its appearance had become almost like a junk. My wife often reproached me, "why did you install those junks in the living room? Throw them away as soon as possible." It was urgent to build new loudspeakers that matches the interior of the living room.

I considered that the basic design of SS-307 was so good. As explained in the page of Gaudi, a three-way speaker system based on a 12" woofer is my ideal type. Though there are so many types of speakers in the world, I single out the three-way speaker system as the best. My way of thinking is simple; for deep bass, a large woofer is the most suitable; for super high treble, a super tweeter is needed; for the middle range they both can't cover, a midrange driver should be used.
Of course, I would select a full-range drive-unit if it could produce from deep bass (25Hz) to super high treble (40kHz) without distortion. But, in practice, there is no such a drive-unit, and I don't think it will become a reality in the future, unless a great invention worth Novel Prize is achieved.

The basic design of SS-309 is the same as that of SS-307: the three-way multi-enclosure system. The woofer module is placed on the floor, the midrange module is placed on the woofer module and the tweeter module is placed on the midrange module. Some might call it only a piled-up speaker boxes rather than a speaker system. It does not have a network in it, since it is specialized for a tri-amplified system. The speaker cables are connected directly to each drive-unit.
As for drive-units, I decided to reuse the ones used on SS-307, because they were all excellent and I couldn't afford new and higher grade drive-units. However, the midrange driver, which was originally a full-range driver, was supposed to be replaced with a horn-loaded midrange in the future.
The midrange module can be switched to the full-range module in Gaudi (and vice versa). This feature is effective until the midrange module is replaced with the horn-loaded midrange.


[Contents]


Design

Selection of key parts

As mentioned above, the Drive-units used on SS-307 were reused.

I believe FW305 is an excellent woofer. It produces comfortably light and sharp bass. In addition, it can also reproduce deep base like an earth tremor and dynamic bass that has a strong impact and is felt like a wind pressure. Its distortion is so low even in the low midrange that it can be used with rather high fc (the crossover frequency). In my opinion, it might be superior to W300A, a more expensive woofer produced by Fostex.
6N-FE88ES is one of the best full-range driver I've ever heard. Its sound is so clear. Most full-range units are not satisfactory because of poor quality and declining response in high treble region. But 6N-FE88ES produces so good high treble that it could be used as a tweeter.
T925 is an old tweeter I've been using since 1982. I decided to reuse it because I didn't detect aged deterioration.

As for the midrange module, as mentioned above, I already planned at that time to replace it with a horn-loaded driver in the future. I considered Fostex H400 (horn) and D1405 (driver) as the strongest candidates.
The reason why I singled out H400 was that it was one of few wood horns available. A metal horn will resonate and produce noise, but a wood horn not. It is rather hard to eliminate the noise. And, as for the appearance, a wood horn is apparently better than a metal horn. I wanted a wood horn at any cost. Only few manufacturers produce horn speakers, and only Fostex produces wood horns. Among the Fostex wood horns, I chose the smallest one, H400. Other wood horns were too large for SS-309.
H400 does not need a baffle. To install it, all you have to do is put it on the woofer module. This easiness is another reason why I chose it.
As for the driver, I have more options, because any drivers with 1" throat can be used, though, I chose a Fostex again. I'm a fan of Fostex, because quality of its products is always good, production tolerance is minimal and reliability is so high. I had two candidates: D1400 and D1405. D1400 is the top-of-the-line driver using an alnico magnet. I was afraid that the magnet was so strong that its leakage flux could affect the cartridge of the record player. And it is often true that the next highest grade product is more excellent than the highest. Consequently, I selected D1405, the cheaper one.

I also reused the cables that were used on SS-307, but I don't remember clearly what they were. According to my vague memories, they were:

I chose MDF boards for the material of the enclosures. MDF absorbs vibration. It has been spread because of this property. For example, it is often used as floor material of condominiums. MDF boards under a wooden floor prevents noise like a footstep from being heard on the floor below. Though MDF is made from wooden fibers, it is not wood itself. It does not have grain, so there's no limitation for cutting plan. For wood-based material, MDF is not so hard that it is easy to cut and plane it. These properties are helpful for me, who is so clumsy. I decided to create the enclosures with MDF only and not use sliced veneer for finish.

Design of the woofer box

I chose a vented-box system. Though I should have chosen a closed-box system if I had followed my motto, "The best is simple," I didn't because it's difficult to adjust the base response of the closed-box speaker system after it is built. When I designed SS-309, I lived in a condo. I already had a plan to move to a single-family house at that time. So, no choice but I selected the vented-box, which is easy to adjust its bass response.

The targeted lower -3dB cutoff frequency was 30Hz. I decided not to boost bass sound in particular on the assumption that I play music at pretty high volume. Because the preamplifier of Gaudi does not have the loudness and tone controls, it is necessary to turn up the volume high enough to make the frequency response flat.

I determined the dimension of the box by reference to the recommended dimension in Fostex's application note. Exceptionally, I fixed the external width of the box to 402mm so as to fit it to the size of Fostex H400, since I was planning to obtain H400 in the future. The thickness of the walls was 21mm. Not too thick nor too thin. The internal dimension turned to be 360(W) x 710(H) x 310(D)mm, and the internal volume 80l. Though it was slightly larger than the recommended box of 75l, the practical volume is almost same due to volume of bracing.

The joints are not processed in particular. The boards are simply glued by wood bond. Though wood screws are also used, they are only a supplementary means. The purpose of them is to keep holding the joints till the bond dries.
One of the side boards is not glued, and fixed with wood screws only. For most speaker enclosures, the backboard is detachable, though, the side board is removed when SS-309 is maintained.

The bass-reflex port (vent) is installed in the rear wall (backboard), in order to prevent the radiation from the back of the diaphragm from interfering the sound from the front as much as possible.
A variable-length tube, Tritec BR-70, is employed as the part of the port assy. The port assy is composed of a small wooden board and the BR-70 fixed in it. The port assy is screwed to the rear wall. It is easy to adjust the length of the port, as the port assy is detachable.

[Dimensional drawing of the woofer box]

I utilized an Excel spreadsheet for the design of the port. I had made the spreadsheet by quoting the procedure and the formulas for designing a port from a book I regarded as a guidebook for building loudspeakers. It took me only five minutes to design it.
I began designing the port with the target fb (the Helmholtz Frequency) of 30Hz, but it turned out to be 28Hz after calculation. The length of the port was calculated to be 130mm (the diameter is 68mm), which was the same as the shortest length of BR-70. BR-70 is composed of two tubes: the inner and the outer tubes. The outer tube slides on the inner tube to adjust the length of the port. I discarded the outer tube and used the inner tube only.

[The spreadsheet for designing the port of the woofer box]

There are chiefly three methods to mount a drive-unit in a baffle as shown in the figure below.
(a) is the simplest method. The drive-unit is mounted on the baffle from the front. So its advantage is that the drive-unit can be replaced without disassembling the enclosure. Its shortcoming is that the frame of the drive-unit protrudes. Unevenness of the baffle surface causes acoustical distortion. The frame and the baffle should be flush as much as possible.
(b) is a method to mount the drive-unit from behind. This method also has disadvantage about acoustical distortion. If the baffle is thick, it forms a circular cylinder attached to the drive-unit. The cylinder adversely affects the frequency response by making unwanted peaks and dips. This is called cavity effect.
(c) is a method to reduce the protrusion of the frame by making counter boring on the edge of the cutout.

How to mount a drive-unit

The method (c) seems to be the best, but I am so clumsy that counter boring is too difficult for me. Then, I came up with a good idea. In addition to the method (a), a sub-baffle is added to make the frame of the unit and the baffle surface flush. The thickness of the sub-baffle has to be the same as the protrusion of the frame. It was 15mm for FW305. The sub-baffle cutout for the unit is slightly larger in diameter than the frame.

Sub-baffle

This method has a couple of more advantages rather than making the baffle surface flush. The thickness of the baffle becomes as thick as 36mm in total, so the mass of the baffle increases too. Holding the drive-unit at a point firmly is another important function of the enclosure other than keeping the drive-unit's backwave in the box. The heavier baffle has the greater inertia. Though it is not great enough to be called the mechanical earth, it surely keeps the drive-unit's frame from vibrating. And, the edge of the cutout is stronger than the counter-bored edge.
The sub-baffle is screwed on the baffle. It's not bonded to make it removable. In case of a fault of the drive-unit, there is a possibility that the same model is not available at that time. The removable sub-baffle can be replaced with one that fits another drive-unit.

Three kinds of materials are used for bracing of the enclosure. 18mm square spruce pine bars are used as bracing in each corner between all internal walls so that it forms a rectangular frame inside the box. A 40mm diameter cypress bar is used to connect the baffle and the backboard. The walls except the top and the bottom walls are reinforced with 30mm square lauan bars. The dimension and mounting position of each bracing part is shown in the dimensional drawing of the woofer box in blue ink.

I purchased two pieces of 21x1820x910mm MDF and one pieces of 15x1820x910mm MDF. The parts of both the woofer and midrange boxes are cut out of those three pieces of MDF. In the cutting plan, each part is marked with a symbol (one alphabet) shown in the table bellow. There are different sizes of parts with the same symbol; the larger one is for the woofer box and the smaller one is for midrange box.

 F Baffle 
 S Side board 
 K Backboard 
 T Top board 
 B Bottom board

[Cutting plan]

Sound absorbent, Micron Wool
I selected to Micron Wool manufactured by Nihon Muki as filling. Because its fiber is more minute than that of glass wool by far, its degree of sound absorption is so high that even a small amount of Micron Wool works effectively.

It's generally believed that an acoustical distortion occurs due to diffraction phenomena if the edges of the baffle are not rounded. In practice, the distortion is so small that it doesn't affect sound quality. Actually, there are many good loudspeakers with the baffle not rounded.
However, I decided to round the edges of the baffle (sub-baffle, more precisely), because the rounded edge has decorative value too.
For this purpose, I was going to make a special tool. I described it later (see "Building").

Dimension errors of the parts may cause unwanted gaps and/or protrusion when the box is built. I came up with an idea to bring all the errors to the back wall side.
(1) Put one side board (hereinafter referred to as the fixed side board) on a level workbench. (2) Assemble the parts except the other side board (hereinafter referred to as the detachable side board) on it using wood bond. (3) Slide each part to make the baffle side flush. Since the parts are not mortised, they can be slid a bit while the bond is wet. (4) Make flush the top board side. (5) Screw all the parts with each other.
The kerfs that joint the detachable side board are made flush with a sander, and rubber sheets are affixed to retain the box air-tight.
The sub-baffle is screwed on the baffle after the box is completed.

How to assemble

I wouldn't like to do something special to finish the enclosure. Anyway, I don't have a proper place to paint it, so I couldn't do anything special. However, the most important goal of SS-309 was excellent appearance, so I considered what I could do for it.
When I looked at a real MDF, I knew the surface of MDF is so smooth that it could be painted after a little sanding. I chose spray colored varnish, because it seemed to be the easiest mean. The color I chose was maple. After the varnish was sprayed and dried, instant lettering was affixed on it, and finally clear varnish was sprayed.

The grille are indispensable item too. Though I don't care if the drive-units are showing (rather, I like it), my wife hates that not only the drive-units but also any other things mechanical are showing.
The frame is made of 18mm square spruce bar, which is also used for bracing of the enclosures. The net is stapled on the frame, after it was cut to the appropriate size. In order to make the grille detachable, commercially available grille holders are used. The male parts are screwed on the grilles, and the female parts are hammered in the baffles.
For your information, I bought all those accessary parts: the net, the grille holders, the Micron Wool, and Tritec BR-70, at Kimura Musen located on the fourth floor of the former Radio Kaikan at Akihabara.

[Drawing of the woofer grille]

The cable is soldered directly to the driver's terminal, and drawn out of the enclosure without any terminal. Although it needs a lot of labor to replace the cable in the future, I chose this direct cabling method, because I learned through long experience that a contact in a low impedance path through which a large current flows will impair tonal quality.

Cable connection

Design of the midrange box

The midrange box is also a vented-box system so that it can be used as the full-range module. The port (vent) is installed in the back wall like the woofer box.
6N-FE88ES is specially designed for a backload horn system, and the maker's recommended enclosures are backload horn systems. Because its vibration system is overdamped, it was anticipated that the volume of bass sounds would not be enough. But, I didn't consider particular countermeasures like raising Q to make a peak in the low frequency band. And I aimed to make the frequency response as flat as possible at frequencies higher than the lower cutoff frequency, in return for setting it relatively high. The response at the frequencies lower than the cutoff frequency would be sharply cut off like a typical vented-box speaker system.
The f0 of 6N-FE88ES is 110Hz, so I set the target fb to 100Hz, and designed the midrange box by the same way as the woofer box.

[Spreadsheet for designing the midrange box]

As the result of the calculation, the box volume was determined 6.9l. As for the dimension, I made the width the top priority. The width has to be the same as the woofer box, 360mm internally and 402mm externally, so that the midrange box can be integrated with the woofer box. The next priority was given to the height. The height needs to be low so that the tweeter is not too apart from the midrange. I decided the internal height is 120mm and the external one is 150mm. Automatically, the depth was calculated to be 160mm internally, and 190mm externally.

The parts of midrange boxes are cut out of the same pieces of MDF for the woofer boxes. Only the side boards are 21mm thick. The other parts are 15mm thick, the same thickness as the sub-baffles of the woofer boxes. Fostex P49 is used as the port (vent) tube after it is cut to the designed length.

The construction is the same as the woofer box. The widths of the baffle, backboard, and top and bottom boards are all 360mm that equals to the internal width. In a way, those parts look sandwiched by the side boards. The way of assembling is the same as the woofer box too. The fixed side board is put on a workbench, then the baffle, backboard, and top and bottom boards are assembled using bond on the fixed side board. After the adjustment, all the part are fixed with screws.

[Dimensional drawing of the midrange box]

The midrange box is so small that I didn't consider installing bracing.

The baffle is two-ply like the woofer box. This is not aimed to make the frame of the drive-unit and baffle surface flush, but to double the mass of the baffle. The frame of 6N-FE88ES is designed so that it is simply mounted on the baffle from the front.
Like the woofer box, the sub-baffle is screwed on the baffle, not glued.

The filling is Micron Wool, the same as the woofer box.

Many uneven wood bars are affixed to the inside of the top and bottom walls, in order for prevention of standing waves. The uneven surface diffuses acoustic waves and results in preventing standing waves from occurring. Later, I realized this method is not sufficient and an additional countermeasure is needed.

For prevention of diffraction phenomena, the edges of sub-baffle are rounded like the woofer box.

The box is assembled so that dimension errors of the parts do not appear in the front and the top of the box, just like the woofer box.
Due to the cutting plan and the order of cutting, the baffle, backboard, and top and bottom boards, which are sandwiched by the side boards, could be cut out in the exactly the same width. Consequently, the kerfs that joint the detachable side board should be flush.
To make sure, I show the cutting plan once again (the same as one in 'Design of the woofer box').

[Cutting Plan]

The design of the grille of the midrange box is common to that of the woofer box. The basic construction is the same, though the size is smaller.

[Drawing of the midrange grille]

The cable is soldered directly to the terminal of the driver and drawn out of the box without any terminal just like the woofer box.


Return to Contents


Building

Woodworking

Parts for the woofer boxI used to cut a wooden board with a saw or a power saw I rented. But, this time, I decided to have the MDF boards cut by a professional for the first time, because I didn't have enough space for woodwork in my condo and the number of parts was too large for me to cut out with hand tools.
It is commonly believed that the dimension error of the parts must be 0.1mm or less to make an exactly rectangular and air-tight box. I thought I couldn't afford if I ordered the cut to a workshop that guarantees that precision, so I ordered it to a do-it-yourself store in my neighborhood (the store is closed now), where I bought the pieces of MDF.
A craftsman of the store said the tolerance was 1mm. I asked him, "Please make the precision as high as possible, because these materials are going to be the parts of speaker enclosures." He had craftsmanship and accepted my request. In spite of my little worry, the dimension of the parts cut out by him were so precise. I was relieved.
The store usually didn't accept an order of a large circular cutout like one for woofer, but the craftsman did it. The photo on right-hand side shows the parts for the woofer boxes.

I cut out the parts for bracing and the grilles for myself.

I cut out the sub-baffle hole for the driver for myself, because the hole for 6N-FEE88ES is not a simple circular form. Since I am so clumsy, I had to concentrate keenly on the work. I managed to cut out the hole as specified in the drawing.

 Cutout for 6N-FE88ES Sub-baffle for the midrage box Part for the midrange box
 6N-FE88ES baffle hole dimension The processed sub-baffle for the midrange box The parts for the midrange box

I rounded the edge of the sub-baffles with a molding plane and a special tool I made.
The molding plane can plane the edge of the board at the angles of 30, 45 or 60 degrees.
My self-made tool is a special file to round the work with radius 15mm. I made it by cutting a 30mm internal diameter PVC pipe for water pipe into a handy length, cutting it vertically into two halves, and affixing sandpaper on the inner surface of the half pipe with double-stick tape.

 Molding plane Planed sub-baffle   Hand-made curved file
 Molding plane  The sub-baffle edge was planed with the molding plane. Self-made curved file
The planed sub-baffle edge was filed with this special tool, and the edge got smoothly rounded.

The rounded edge of the sub-baffle


The sub-baffle with the rounded edges









I drilled only prepared holes for wood screws that pass through the boards beforehand. I decided to drill prepared holes in kerfs just before screwing. I was afraid that a pair of prepared holes may be misaligned if the one in the kerf is drilled beforehand and the joint may be slid when the wood screw is tightened.

Prepared holes
Before the assembly, I screwed a wood screw in a kerf of a leftover piece of MDF as a trial. I found out MDF is unexpectedly vulnerable and splits easily, although I had guessed MDF is a solid material like plastic. In fact, it is layered like many thin cardboards are stuck. So, it is likely to split if a thick wood screw is used or many wood screws are screwed in the kerf.
For SS-309, wood screws are not used to fix the joints permanently, but to fix them temporarily till the bond dries instead of cramps. So I decided to use the minimum number of thin wood screws required to fix the joints temporarily.
Piece of MDF split by a wood screw

Assembly

I assembled the boxes following the procedure specified in the design phase. Both the woofer and midrange boxes were assembled in the same procedure.

Assembly of the woofer boxThe fixed side board was put horizontally, the baffle, backboard, and top and bottom boards were put vertically on it, and the parts were glued each other with wood bond. Bond was applied plenty so that, in case that a slit occurs at a joint, the bond will fill it. Bracing was glued in each corner between all walls of the woofer box. By the way, the bond I used was a commonly-used type, not a quick-drying type.
While the bond was still wet, each part was adjusted in position to make the front and top surfaces flush.
After the positions of all parts were fixed, screw holes were drilled in the kerfs by inserting the bit of an electric drill into the prepared holes that had been drilled beforehand. This work needed an extra care not to move the parts.
Each wood screw was not tightened individually. First, all the screws were screwed halfway. Then, in order to keep the joints from sliding, one screw was tightened a little with a screwdriver (not an electric screwdriver), then another was tightened in the same way, and so on, till all the joints were secured. Finally, all the screws were tightened firmly with an electric screwdriver.

The above process must be done quickly before the bond dries. When the work was done, a lot of bond leaked out of the joints. I left it as it was.

Sound deadeners of the woofer boxWhen the bond became adhesive, I turned over the box with care not to slide the fixed side board. Then I screwed it on the box. Now the box was rigid.

Next, the sound deadeners were fixed with wood bond and wood screws.

Then, the detachable side board was going to be screwed. Because the dimensional errors were so small, the kerfs were flush. I decided to use insulating tape instead of rubber sheets to keep the box air-tight. The tape is cheaper and easier to use.
The tape was affixed on the kerfs and small holes were cut out at the positions of the prepared holes for the wood screws. The detachable board was screwed, and the box had all six walls at last.

Finally, the box was completed by fixing the sub-baffle on the baffle. The sub-baffle is fixed with only wood screws. Bond was not used.

The woofer box just before the detachable side board is fixed    The woofer boxes just before the sub-baffles are fixed    The midrage boxes just before the sub-baffles are fixed 
 The woofer box just before the detachable side board was fixed    The woofer boxes just before the sub-baffles was fixed    The midrange boxes just before the sub-baffle was fixed
         
 Assembly accuracy of the woofer box   Assembly accuracy of the midrange box    The inside of the midrange box 
 The back of the woofer box; the assembly accuracy was not bad.
(This photo was taken in June 2013)
  The back of the midrange box; the assembly accuracy was not bad.
(This photo was taken in June 2013) 
  The inside of the midrange box with the driver temporarily installed 


The frames of the grilles were made of the 18mm square spruce bars, which was the same material as the bracing of the woofer boxes. Fast-drying type of wood bond is used to assemble them, and neither wood screws nor nails are not used.
The net was stapled on the frame after it was cut into a piece of an appropriate size.
Finally, the grille holders (male part) were screwed on the frame. The grille was completed.
Grille for the woofer boxGrille for midrange boxGrille holder (male part)

Because the female part of the grille holders were going to be attached to the sub-baffles, the prepared holes were drilled in the sub-baffles. Besides, the prepared holes for the insert nuts for a Fostex genuine grille were drilled on the sub-baffles of the woofer boxes too.

Finish

First of all, I cut off the bond sticking out of the joints with a cutter.
I applied putty on the heads of the wood screws that would never be removed. After the putty got dry, I sanded a little the outer surfaces of the boxes with an electric drill with a circular sandpaper attached.

Now, it was time to paint the boxes. I made a temporary workroom by using cardboards on the balcony of my condo, where I painted them.
Soon I realized that it is very difficult to spray varnish evenly. Even though I did my best, the sprayed varnish became uneven. But, this uneven varnish looked terrific like oil stain paint. So I continued to spray the varnish. The color of the varnish was maple, and if it's painted on MDF, the color became rather reddish. I was attracted to this color too. I sprayed the baffle four times, and its appearance became so cool (the other surfaces were painted three times).

I affixed instant lettering to read my brand name (NOBODY) and the model number (SS-309), then sprayed clear varnish three times. This finished the paint work.

The heads of the wood screws that would possibly be removed (not covered with putty) were painted metallic dark gray for better appearance.

Mounting the drive-units and other small parts

The female parts of the grille holders were hammered into the prepared holes on the baffles after varnish was completely dried. On the baffles of the woofer boxes, the insert nuts for the Fostex genuine grilles were hammered in too.

Because the port tube for the woofer box, Tritec BR-70, is a simple plastic tube, it needed to be deadened. It was rapped with a 0.1t brass sheet and cloth tape on it.
The tube was mounted on a small MDF board to form a port assy. The port assy was screwed to the backboard. Since it is detachable, it is easy to replace the port.
The port tube for the midrange box, Fostex P49, was cut to be 60mm long (the inner diameter is 49mm). It is mounted on the backboard with double-faced adhesive tape.

 Port tubes for the woofers   Port assy mounted on the woofer box 
The deadened port tubes    Port assy mounted on the woofer box

The last step was installing the drive-units and the filling. The detachable side boards were removed temporarily.

The drive-units were all removed from SS-307, and the cables remained soldered to the drivers' terminals. I mounted the drivers on the baffle with wood screws, after I made the cables pass through the holes for the cable. At that time, I was very careful so that I won't damage the drivers with the tip of the screw driver.
The diameters of the holes for the cables were matched to the external diameters of the cables, so that it is not necessary to seal the holes with putty.

I lined the filling, not filled, since both the woofer and midrange boxes are bass reflex type.

By the way, I mounted the drivers and ports to the midrange boxes temporarily before I painted the boxes, and played music for adjustment of the filling. Then, I removed the drivers and ports, and painted the boxes.

Completed baffle of the woofer box




The completed baffle of the woofer box

Completed woofer box (front)   Completed woofer box (back)     Woofer box with the grille attached    Baffle with Fostex grille attached
Completed woofer box (front)
(This photo was taken in 2009)
  Completed woofer box (back)
(This photo was taken in 2013)
  Woofer box with the Grille attached   Woofer box baffle with Fostex Grille


 Completed midrange box (front)    Completed midrange box (back)    Midrange box with the grille attached
Completed midrange box (front)   Completed midrange box (back)   Midrange box with the grille attached



Return to Contents


Install and tuning

Midrange unit before paintingTuning of the midrange module preceded painting. The driver, port and filling were installed temporarily.
I set up the midrange module as a full-range speaker, and played various kinds of music. Its crystal-clear sound was so impressive that I can never forget it. The tonal quality was totally different from SS-307. I keenly felt that different enclosures have so different tone even with the same driver.

When I listened very carefully, I felt a bit unpleasant echo, and supposed that it was resulted from standing wave in the box. I came up with a countermeasure; I crumpled up a piece of Micron Wool into a ball, tied it with a string and hanged it from the top wall of the box.
This countermeasure was more effective than expected. It made the sound clearer furthermore. This midrange module was the the finest one-way speaker I had ever heard, if its poor bass response could be neglected.

Standing wave stopper

I didn't care much about bass response, since this midrange module would be used mainly as the midrange speaker. And the driver 6N-FE88ES was designed for a backload horn system, so I accepted a certain level of non-flatness in the bass response. As a result, the bass response was not bad, except for poor deep bass response. The upper bass response was so good.
I couldn't measure the characteristic, as I didn't have required measurement instruments and environment. The only data I could measure was impedance vs frequency response. The curve had two peaks, which is typical for a vented-box.
Impedance vs frequency response of the midrange box
Although I was once satisfied with the midrange module from the bottom of my heart, I was disappointed later. It was painting that slightly lowered the sound quality. After painting the enclosure, I set up the midrange module the same as the first trial listening. I played various kinds of music again. The sound was no longer as clear as it had been.
I hadn't known the fact that painting influences tonal quality. The influence might be significant particularly to sound-absorbing materials like MDF. I suppose the painting makes the surface solid like plastic, and that increases unwanted radiation from the enclosure.

I tried affixing a piece of felt to the surface of the baffle in order to absorb the unwanted radiation. But it didn't work. On the contrary, it worsened sound quality. I had to remove the felt for the better result. At that time, the surface layer of the varnish was removed too. The appearance got worse.

Later, I found an article in IPSJ (Information Processing Society of Japan) Magazine that focused on the relation between tone of musical instruments and their measured data. The article showed an interesting example; one craftsman make two identical violins; the only difference between them is varnish; these violins have different tones to human, though their characteristics measured by the best measurement instrument are the exactly the same.
It is necessary to study the relation between the enclosure finish and tonal quality furthermore.

I determined the amount of filling by intuition and lined it on the inner walls of the woofer boxes. Soon I set up the woofer modules as full-range speakers. The first impression was very good. I said to myself, "It's perfect!" I really felt so without any concrete grounds. Therefore, I decided to use them as they were without any actual tuning.
I have been using the woofer modules without tuning since then, though I once replaced the port with a shorter one to boost bass. After all, I replaced the port back with the original one. As of August 2013, I still use them in the initial setting. I don't feel like tuning them, because I am afraid that I would spoil these perfect woofer modules if I change anything about them.

The impedance vs frequency response of the woofer modules were nearly ideal curves with two peaks. The Helmholtz Frequency fb was approximately 26Hz. The response was extended down to the 25Hz band at full level.
ウーファー部インピーダンス特性
At last, it was time to install SS-309s. I placed the woofer modules where the pair of SS-307 were installed (the corner of the living room of my condo). The speaker bases of SS-307 were reused. They were merely old magazines put on the floor. They were pressed by the weight of the speaker system and made flat and solid. Since a pile of papers is a good sound absorber, it might be an ideal speaker base.

A vibration proof rubber mat was put on the woofer module, the midrange module was placed on it, and the tweeter module on the midrange module. The midrange module was designed asymmetry. It was placed so that the driver was positioned near the center of the system. The tweeter was placed just above the midrange driver. The tweeter module was composed of the tweeter T925 and a stand that came with T925. A vibration proof rubber was affixed to the bottom of the stand.
The speaker systems were toed in a little.
Positional ralation of the drivers


The positions of the drivers

The right-hand channel speaker:
The midrange driver was positioned near the left end of the enclosure.
The tweeter was positioned above the midrange driver.







Installed SS-309The configuration of the amplifiers remained unchanged. The setting of the network CD-206 was also unchanged. The crossover frequencies were 800Hz and 8000Hz.

A loudspeaker sounds differently according to its position in the room. Especially, bass response is greatly influenced by the distances from the boundaries of the room. But I couldn't choose the best position because my room was not spacious. I had no option but to place the SS-309s at the corners of the room; each speaker was placed 20cm from the wall behind it and 10cm from its respective sidewall.
The bass was boosted and sounded so rich. But it was not unpleasantly boomy bass. I liked this rich and pleasant bass.

It took me a lot of time and labor to adjust the position of the tweeter back and forth. The crossover was 8000Hz, and the wave length at the frequency is about 43mm. If the tweeter was moved by 21mm (the half of the wave length) back or forth, the phase difference between the tweeter and the midrange would be 180 degrees, and the tone would change significantly. While playing music, I moved the tweeter little by little until the sound became the most natural.

The photo on the right side shows SS-309 (R-channel) just after the installation. (June 2001)


Return to Contents


My Own Review

Looks

SS-309 without the grillesThe main purpose of this project was make good-looking loudspeakers. I flattered myself that the purpose was achieved. My wife, too, admitted that, and said, "It's OK to install them in the living room."
She hates naked loudspeakers (the drive-units are not covered with the grilles). On the contrary, I feel SS-309 looks good without the grilles. Besides, the grilles slightly impair sound quality.



Woofer FW305Midrange unit 6N-FE88ESTweeter T925

Sound quality

The woofer FW305 delivered its potential performance compared with SS-307. It exhibits good transient response at low frequencies. It was so brisk that I felt so good when I heard the sound of a bass in jazz. Though the volume of bass was high, it was not pushy, and sounds of low-pitched instruments were never exaggerated.
In addition, the woofer module can produce deep bass that is felt more than heard and dynamic bass like rambling of the earth. For example, the sound of a big taiko drum of Ondekoza (its fundamental frequency is 28Hz) made me feel so good (CD, Ondekoza, Ondekoza (I), Victor VICG-5373). I felt a wind pressure radiated from the taiko drum, and the realistic sound made me feel as if I had been listening to a real live performance. And, when the scene of launching the rocket in the movie APOLLO 13 was played (DVD, APOLLO 13, SUD-29954), I really felt rambling of the earth. I thought SS-309's ability of reproducing deep sound favorably compared with a movie theater's public address system.
Though FW305 does not employ any novel mechanism nor special materials, I love it. I like a simple, back-to-basics audio parts like FW305. I would like to keep using it as long as possible.

The midrange module offered pretty high quality in midrange and upper midrange, though painting had slightly impaired the clarity of the sound. For most full-range drivers, treble is rolled off at a relatively low frequency, or trebles are distorted, so it's better to add a tweeter in many cases. But trebles of 6N-FE88ES were so good that I even thought it could be used as a tweeter.
Although full-range drivers perform very well when they are used as a full-range driver, they tend to lower the system's total sound quality when used as a midrange driver, because their performance are not as good as the woofer and tweeter. But 6N-FE88ES was excellent as a midrange driver too. I believed that's because its performance from upper midrange to treble is so good that it could be used as a tweeter.

ミッドレンジ部のポートをふさいだところWhen the midrange module is used as a midrange speaker, the bass reflex port should be stuffed up. The rear of the diaphragm emits the same sound pressure as the front of the diaphragm in opposite phase. The sound leaks from the port and interferes with the sound from the front. This distorts midrange sounds in particular. Although the midrange module sounds so good when it is used as a full-range speaker, when it is used as a midrange speaker with the port open, its sound quality is not as good as the tweeter and woofer because of the distortion. So the port should be stuffed up in order to balance its tonal quality with the tweeter and woofer.
I had expected closing the port would not be necessary because SS-309 had employed the rear port system. But it was necessary after all against my expectation. I decided to stuff up the port with a rolled urethane mat when the midrange module is used as the midrange speaker, though it is a little troublesome.
The midrange module can be switched to the full-range speaker in Gaudi (and vice versa). In that case, it is used with the port open. That is, it works as the bass reflex system (See 'Listening and BGM Modes').

I have been using the tweeter T925 since 1982. I don't think it has deteriorated by aging yet, and still have confidence in it. It is a super tweeter rather than a tweeter, so I set the crossover to 8,000Hz. Although it reproduces only upper treble, T925 contributes a lot to the system's overall sound quality . It greatly affects not only treble but nuance of midrange and bass. SS-309's high quality of bass sounds is resulted from not only the high performance of the woofer but also that of T925.

This tweeter's most attractive feature is its capability to offer the texture of the sound source (instruments and voices). For example, it reproduces sounds of cymbals in jazz not as mere, rhythmical treble sounds but sounds produced by hitting a heavy metal disc. T925 sounds as if you could see the instrument. I would like to use it until its life will end.

Low directivity is known as the shortcoming of horn tweeters. But, on the contrary, I take it to be an advantage. A horn tweeter can deliver acoustic energy toward the listener's ears with a least loss. I believe this fact is deeply related to the above-mentioned sound quality that makes the listener feel the texture of instruments and voices.
A dome tweeter, which has high directivity, cannot offer enough acoustic energy directly to the listener, and a fair amount of sound is reflected by the boundaries and reaches the listener indirectly. Indirect sounds become beautiful echoes in a very large room like a concert hall, because the indirect sounds significantly delay to reach the listener. But in domestic rooms, the time difference between the direct and indirect sounds is so small that the sound may be heard distorted and/or the sound image may be blurred.
I firmly believe that horn tweeters excel dome tweeters, unless the room is very spacious or the walls and ceiling are made of sound absorbing materials.

So far, I reviewed each module. It is a tough question to evaluate SS-309 as a whole. SS-309 is composed of the self-contained modules, and it is not a complete speaker system until combined with the network and power amplifiers.
Compared with SS-307, I thought SS-309 was at much higher grade. SS-309 sounded more vividly. Especially, sounds of brass instruments were remarkably lifelike, maybe because the tweeter was horn-loaded. In addition, I felt SS-309 could reproduce very low level sounds so precisely. I realized I almost achieved my goal: dynamic, vivid, yet precise sound.

Usability

The speaker cables are soldered directly to the terminals of the drive-units except the tweeters, so it takes a lot of labor to change the cable. It is a weak point, in terms of usability. However, in the world of audio, it happens quite often that high usability worsens sound quality. This is the case. Through my long experience, I know (a) contact(s) between the power amp output and drive-unit blurs the sound. I believe the direct soldering is the best solution in spite of its low usability.
And, experience shows that if a cheap cable and an excellent cable are connected in series, the sound quality will turn to be the same as the cheap one only. In the future, a more excellent cable may be invented. If only the external cable is replaced with the new cable, it will not bring about a good result. The internal cable is also to be replaced with the new one. As for SS-309, the cables directly connect the power amps and the drive-units. There is no distinction: 'internal' and 'external'. When the cable is replaced, it means the entire cable (both the internal and external cables) is replaced. This makes it sure to maximize the performance of the new cable.

When I was designing SS-309, I already had a plan to move to a single-family house. So I designed the woofer box so that the Fostex genuine grille could be attached. The grille was made of metal and so rigid. It was so useful when we moved.
I bought Fostex genuine grilles for 6N-FE88ES too. It was attached to the midrange box with adhesive tape. The box itself wasn't designed to attach the genuine grille.

Summing up

I think I have achieved the goals for both the appearance and sound quality. I feel Gaudi got another step closer to the high-end audio system.
However, Gaudi hasn't exactly conformed to the original system design yet. The full-range units are used in the midrange modules as substitutes of midrange units. Even though a full-range unit sounds good in its original usage, in general, it is inferior to midrange units when used as a midrange driver. The same thing applies to SS-309. The performance of 6N-FE88ES is apparently insufficient compared with FW305 and T925. It is desirable to upgrade the midrange modules as soon as possible (as described below, I upgraded the midrange modules by changing the drive-units to horn-loaded midrange units at last in 2012).


Return to Contents


Improvements

Move to my new house (May 2003)

My new house is a well-insulated, air-tight, two-story wooden house. Though it is not particularly constructed soundproofed, its thermal insulator and double-paned glass in the windows excel in soundproofing too.
The living room, where Gaudi is supposed to be installed, is located at the north end of the second floor unlike ordinary houses. There is no attic above the living room, so the ceiling is high and slanting. Because sound has the property of going upward, the high ceiling reduces the distortion caused by reflection. In addition, the slanting ceiling prevents standing waves from occurring, since it is not parallel with the floor.
The living room is designed so that Gaudi is installed against the north wall, which has no window. I put each SS-309 250mm away from the north wall and distant from its respective side walls (see the figure below). The distance between the speakers is not long (1560mm). I wanted to put them more distant from each other, but I had to compromise because the room is the living room and a TV set is also installed there.

The listening position is 2.1m away from each speaker. The distance is relatively short, though, it is desirable because the percentage of reflection from the boundaries in the total of sounds the listener hears is kept low. And, there is a large space behind the listening position. This is also desirable for a good sound quality.
Listening room layout
I spent much time and effort to adjust the distance between the speakers and the wall behind them, because it is an important factor for the volume of bass sounds. In practice, however, the volume didn't vary much according to the change of the distance. The volume was always a little short.
The low frequencies might not be reflected by the wall but go through it, because the inner side of the wall is made of ordinary plaster boards. And, I've heard that bass sounds of a rear-vented speaker is not boosted even if the speaker is near the wall behind it.
I attempted to boost bass by shortening the port (130mm->113mm). This countermeasure was a little effective, but not enough. I took another measure at a later date.
ポートの交換スピーカーと壁のすき間
I thought using old magazines as a speaker base is not suitable for the brand-new house, and decided to use a piece of carpet and four bricks for each speaker. This resulted in the obviously worse sound quality than the old magazines. I knew bricks are so easy to resonate.
I planned to add spikes under the speaker enclosure to reduce the resonance. I looked for good spikes at an audio shop where I was a regular customer, and chose Dali's spike set because it was reasonable and the height was adjustable.
The spikes improved sound quality to some extent, compared with putting the speakers directly on the bricks, but I was still not satisfied. Spikes may not be so effective to prevent vibration from propagating as their claims. (2015/01/27 corrected) {I used the spikes in a wrong way. For further details, see 'Correct use of the spikes (Nov. 2013)'}
DALIのスパイクを導入

Improvement of the speaker base (Nov. 2004)

I considered using heavier objects instead of the bricks. I looked for a shop that sell stones in Rakuten Ichiba (http://www.rakuten.co.jp/) and found a good shop that cuts stones like granite and marble stones according to the customer's plan at reasonable prices. The shop was MG Sekigahara (http://www.rakuten.co.jp/mgs/). I ordered two granite boards of which dimension was 20x401x390mm. The total price was about 8,500 yen.
The granite board is rather heavy, so it acts as a mechanical earth. It improved sound quality. I felt the speaker became able to offer higher-resolution sound.

改良したSPベース



Improved speaker base


From the bottom, a piece of nylon carpet, 20t granite board and DALI spikes.
A piece of filling (Micron Wool) is placed between the enclosure and the granite board.






I considered changing the insulator between the woofer and midrange boxes from the vibration absorbing rubber mat to spikes too. But spikes are so large that the gap between the woofer and midrange boxes becomes rather wide. I decided to use a substitute.
I found an interesting information on Internet that a 10-yen coin is a good insulator. I tried it out soon. It was true. It improved sound quality a little. I tried out a 100-yen coin too. It was not as good as the 10-yen coin. I experienced such a thing many times that the more expensive an audio accessary is, the poorer sound quality is.
I used the 10-yen coin insulator for some months, but I re-considered it and reached a conclusion that the point contact is better than the surface contact. I decided to use small pointy stones. I picked up about 20 small stones that are all the same in size from my graveled backyard. Then, I selected two quads of the stones that make the midrange box not shaky. I put each quad under each midrange box's four bottom corners. This led to better sound quality, and I was satisfied at last.
ミッドレンジ部用スパイク(小石)ミッドレンジ部用スパイク(拡大写真)ミッドレンジ部スパイク使用状態

Upgrade of crossover network (Feb. 2005)

I upgraded the crossover network from CD-206 to CD-211 A-NET.
The lower crossover of CD-206 was 800Hz (18dB/oct) and the upper one was 8000Hz (12dB/oct). As for CD-211, they were 720Hz (18dB/oct) and 7200Hz (18dB/oct), respectively. The filter for the tweeter cut off more rapidly below fc. This reduced the current flowing the tweeter voice coil below fc, and improved clarity of treble. The potential performance of T925 was successfully drawn out.

I used CD-211 in that setting for a while. One day, I found a bump at the fc region between the woofer and midrange bands, and I decided to change the characteristic of the filters. I changed the -3dB cutoff frequency of the woofer filters from 720Hz to 530Hz, whilst the lower cutoff of the midrange filter remained unchanged (720Hz). The attenuation at fc became -6dB. The continuity from the upper bass to the lower midrange was much improved, so that SS-309 played sounds neutrally as if it had been a one-way speaker.
In theory, the attenuation at fc should be -3dB in order to flatten the system frequency response. But the bump would be developed unless the midrange unit is separated from the woofer by a large spacing.

Upgrade of the speaker cable (Sept. 2009)

As for the speaker cable, I tried out various types of many manufacturers since I built SS-307. But I didn't think any one of them was obviously superior to others. Though each manufacturer's catalogue or something like that said how excellent their products were with many plausible advertising statements, I felt they were sort of extravagant advertisement. I began believing a simpler cable was better than high-tech cables these days just like amplifiers.
One day, I looked at the web site of Oyaide Denki (an Akihabara-based cable vendor, http://oyaide.com/catalog/) and was attracted to a 0.4mm solid wire manufactured by 47 Laboratory (http://www.47labs.co.jp/). In fact, I knew the wire a long time before, and had e-mailed an inquiry about it to Oyaide. At that time, I judged it was not good because it was made of very cheap materials. This time, I was attracted because most of the user reviews about it gave it five-star praise, and decided to employ this simplest wire. FYI, the conductor is tough pitch copper without plating and the sheath is 1mm-thick polyethylene. No other material is used. So simple!
All the speaker cables were replaced with the new ones. However, I chose 0.65mm solid wire for the woofer cables, because the 0.4mm wire has a little too high resistance and that may reduce the damping of the woofer. The low damping makes bass sounds boomy.
I bought the wires online from Oyaide. The price of the 0.4mm solid wire was 273 yen per meter, the 0.65mm was 315 yen (for the simplest wires, the prices were not so cheap, were they?).

I made the speaker cable by twisting the two wires at 30cm (1 foot) pitch. The end of the wires was only stripped because they were going to be soldered to the terminals. The only exception was the end that was going to be connected to the midrange amp Flying Mole DAD-M100Pro. Because this amp's terminals were unable to be soldered, I decided to attach banana plugs to the end of the cable that was connected to this amp. A special banana plug for the 0.4mm wire was available from 47 Laboratory. I bought four of the plugs for the cable.
This plug is designed based on a seemingly crazy idea; contacting the wire core directly to the terminal of equipment. The plug is made of Delrin and hollow to insert the wire in it. The tiny hole is bored at the tip of the plug to draw out the wire core through it. The wire core is bent 180 degrees and looped around the part of the plug that is inserted into the terminal. Inserting the plug makes the wire core contact the metal part of the terminal directly.
I realized this plastic plug had several problems. First, bending the wire core 180 degrees produces unwanted inductance and some of the current is reflected at the bending point. Those lead to distortion of the signal. Second, since the wire core is not plated, it corrodes rapidly even at room temperature. And, the terminal of DAD-M100Pro is plated with nickel. Contacting different metals accelerate corrosion. Third, the dimension accuracy of this product is so low. One of the plugs was a little too thick and could hardly be inserted into the terminal. Another one was so thin that it was not secured in the terminal and came out easily.
In spite of these problems, I used the plastic plugs for a few months. When one of the plugs became faulty because of corrosion of the wire core, I stopped using the plugs. I am so disappointed since the plug was not cheap (483 yen at Oyaide). I used banana plugs for measurement instruments, instead. Though it was cheaper than the plastic plug by far, it excelled the plastic plug in sound quality. I experienced that again, "The more expensive an audio accessary is, the worse its sound quality is." (2015/02/14 added) {I stopped using the banana plugs. The cable cores are directly connected to the terminals now. Wako Technical Titan Audio Oil are applied to the cable cores to prevent them from corroding.}

I read an explanation on 47 Laboratory's Web site that cables were so vulnerable to vibration. At that time, I didn't think speaker cables were so microphonic, because they conducted large signals and microphonic noise was too minute to negatively affect sound quality. But, as a trial, I followed 47 Lab's recommendation; hanging the cables in midair to prevent vibration. The speaker cable I made weighed only a dozen grams or so per meter, so hanging it was easy. I fitted up hooks on the wall near the ceiling, hanged strings from the hooks, and hanged the speaker cables with the strings. For the parts of the cables that inevitably touched the floor or the speaker enclosures I put vibration proof rubber mats beneath them.
This method was more effective than expected. The grade of sound rose one step higher. I really felt Gaudi sound was approaching that of high-end audio further.
SPケーブルの空中配線SPケーブル防振対策

Gaudi was successfully upgraded. It was proved that my motto, "the best is simple," was applied to speaker cables too. In addition, it came to light that the following theory was not right; the thicker the cable core is, the better sound quality is.
I guess too thick cable core distorts the signal, because its surface area is so large that high frequency currents that flows the surface due to skin effect causes eddy current and produces distortion in treble band.

It is often said that a thin speaker cable does not conduct deep bass. The same opinion was written in the homepage of 47 Laboratory too. However, I don't feel Gaudi cannot reproduce deep bass after the speaker cables were changed. I believe the theory that reproduction of deep bass needs a thick cable is a superstition. Perhaps, the fact that to produce bass sounds by playing a stringed instrument the thick string is used leads to this superstition.
In practice, a thin cable makes deep bass boosted, because its relatively high resistance weakens damping of the woofer just like using a power amplifier with low damping factor.

Renewal of the woofers (Oct. 2009)

ウーファーの交換The surrounds of the woofers were broken due to aging. I had to repair them.
I checked up Koizumi Musen's Web site (http://www.koizumi-musen.com/), and found that fortunately FW305 was still in production and available on the Web site. I decided to buy new FW305s instead of repairing the surrounds, because I was anxious about the result of the repair. Besides, the new drivers would extend the life of SS-309. I think Fostex Company should be praised for it produces products like FW305, which are excellent in spite of the old designs, continually for long term.
The price was raised compared with the old FW305. I bought the old ones also at the Koizumi Musen, a loudspeaker shop at Akihabara district, and the price was about 18,000 yen per unit, if I remember correctly. The new one's price was about 22,000 yen. But I was convinced when I saw the new FW305 for the first time. The many parts, at least visible ones from outside, like the frame, surround and terminals were improved.
After replacing the drivers, I realized sound quality was apparently improved. What I did can be said to be upgrade rather than a mere repair.

I replaced the port back with the original one (68mm-diameter x 130mm-long) after the new FW305s were well broken in.

Installation of acoustic panels (Sept. 2012)

Though I was satisfied with the volume of bass sounds when SS-309 played very loud, I was a bit dissatisfied at normal listening volumes. The interior wall behind the speakers is an ordinary one and made of 12.5mm thick plasterboards with wall paper on them. It reflects mid to high frequencies while gets low frequencies through it. To solve the problem, I hit upon an idea of making acoustic panels that absorbs mid to high frequencies and reflects low frequencies, and installing them behind the speakers.

The base was a 18x1220x910mm MDF board. Sound absorbing boards made of rock wool (Yoshino Gypsum (http://yoshino-gypsum.com/en/index.html) Solaton Cube Cross 6T) were nailed on it.
The acoustic panels were installed as they leaned against the wall behind the speakers.
自作調音パネル調音パネルの設置設置済み調音パネル

These acoustic panels improved not only bass response but also stereoimaging. Before that, sound localization tended to lean to the left-hand side. But now the localization was more accurate.

Rev. A: Upgrade of the midrange module (Nov. 2012)

Finally, the time came to replace the midrange modules with horn-loaded drivers. It was the time the loudspeaker system I had considered as the ideal since 1974, when I was 17 years old, became reality.

The LS units I employed were Fostex H400 wood horn and Fostex D1405 driver. These were the LS units I had already chosen when I designed SS-309.

H400前面H400上面D1405背面D1405前面

The two pairs of H400 and D1405 cost me about 180,000 yen. It was very unusual decision for me, who don't like spending much money for audio, to buy such expensive audio components. The point is that I believed (now still believe) the loudspeaker dominates the system's overall sound quality and the midrange unit is the most influential component in the speaker system. My theory is, "The midrange unit determines sound quality, and the tweeter determines tone." I think it is a waste of money to spend much money for power amplifiers, because they are less influential in the system's overall sound quality. In the audio system, it is the midrange unit that is really worth a high price.

The recommended crossover frequency fc of D1405 was 850Hz or higher, but the lower cutoff frequency of the midrange filter of the network CD-211 A-NET was 720Hz. I customized CD-211 to change the cutoff to 880Hz, and that of the woofer filter from 530Hz to 720Hz.
CD-211 had two more features to be customized.
One was adding attenuators to the midrange outputs. The gain of the midrange amp, Flying Mole DAD-M100Pro, was so high that the signal had to be attenuated quite a bit with the input attenuator. Until then, the attenuation was about -20dB. The sensitivity of 6N-FE88ES was 90dB/W(@1m), whilst that of D1405 was 104dB/W. The attenuation had to be increased by 14dB, but the maximum attenuation of DAD-M100Pro's input attenuator was -30dB. Therefore, I added the attenuators to the midrange outputs. The attenuation was fixed to -20dB.
The other one was a functional change. Until then, the midrange outputs of CD-211 was switchable from midrange to full-range, and vice versa. Adding full-range signals possibly damages D1405, so I customized the function so that the outputs are switchable from midrange to midrange + treble, and vice versa. When the switch on the front panel is set to '3-way', 880-7200Hz band is output from the midrange outputs. When the switch is set to '2-way', frequencies above 880Hz are output.

It took a lot of time and labor to customize CD-211. Though it was Oct. 2011 when I bought D1405s and H400s, they couldn't be installed till Nov. 2012, when the customize was completed and the suffix A is added to the model number (CD-211A).
CD-211A

I had one more thing to do before installing the wood horns. There was a possibility that leakage flux from the driver's strong magnet adversely affects the pickup of the analog disc player PS-104. To prevent this, I attached a magnetic shield board made of Permalloy to the audio rack.
磁気シールド板



The magnetic shield board attached to the audio rack

The black thin board attached to the right side of the audio rack is the magnetic shield board made of Permalloy that covers the analog disc player.








he installation of the horns itself was easy. First, I screwed the drivers to the horns. Then, I put vibration proof rubber sheets on the woofer boxes and the midrange horns on them. On the midrange horns I placed the tweeter modules. Finally, I made cables with 47 Laboratory 0.4mm solid wires, and connected the midrange drivers to the power amplifiers, Flying Mole DAD-M100Pro.
SS-309A midrange & tweeter完成したSS-309A(Nov. 2012)

The tuning of the speakers was less troublesome than expected.
First of all, I adjusted the level of each drivers. While playing a test CD that contains sine waves of ISO recommendation 2661/3 octave frequencies (20Hz, 25Hz, 31.5Hz, ..., 16kHz, 20kHz), I measured the relative SPL with a microphone (audio-technica AT-822) and an IC recorder (TASCAM DR-1) and adjusted the levels to flatten the frequency response. Next, while playing various kinds of music, I fine-tuned the levels such that they sounded maximally natural to my ears. It took me about one week to achieve a satisfactory result.
When 6N-FE88ES was used as the midrange driver, the adjustment of the levels was a very troublesome task. The method using a microphone and a VU meter was unreliable. The only dependable measurement instrument was my ears. It took me a month or two to tune the speakers those days (see 'Tuning').
At the end, I adjusted the position of the horns in front-back direction. It was unexpectedly easy too. But there's a possibility that the horns are still not placed at the best position due to my lax adjustment.

As for the point of interest, which is sound quality of course, I got as good a result as I had expected.
My first impression was, "Beautiful!". Beauty has been added to Gaudi sound that was attractive thanks to its vibrancy and vividness. Especially, harmony of strings or chorus is breathtakingly beautiful so that I am entranced with the music without knowing. For example, the sound of cascading strings performed by Mantovani (CD, Mantovani Orchestra, Mantovani, Encore!!, Victor VICP-5618) is so beautiful that it makes me filled with a feeling of happiness. Of course, not only strings but every kind of musical instruments such as wood winds, brass winds and the percussion sound better than before. I feel their sounds became more precise and lifelike.

I had been anxious that the sound might become a little 'honky' colored, but I realized it was a pointless concern. And, though I had been worried that sound quality wouldn't be good until the drivers were aged enough, it was a needless concern, too. I felt the new midrange drivers sounded okay from the beginning just like the new FW305s installed in 2009. Fostex might ship its products after aging them enough in its factory.

One more thing I noticed is that the quality of bass sounds has been improved. Although the woofer modules and their related components haven't been changed at all, bass sounds became more dynamic and clearer. Unlike typical commercial loudspeakers of which bass character is intentionally boosted and boomy, SS-309A reproduces comfortably sharp and lean bass sounds. Though they are dynamic, they do not sound exaggerated but precise and realistic. The sound of a wood bass in jazz music are so realistic that I feel like I were listening to a live performance.
Another merit of SS-309A that deserves special mention is clarity of both speaking and singing voices. It is so easy to catch lyrics of songs. As for English songs, the pronunciation is heard so clearly that I can catch the meaning of the lyrics without effort. I often find myself immersed in a world an English lyric depicts insomuch that I forget the lyric is written in English without realizing. Now Gaudi is the best hi-fi system I've ever heard when it comes to ease of catching lyrics.

I think it was a wise decision to employ the horn-loaded drivers. The horn-loaded driver can reproduce minute sounds precisely because its ultra light diaphragm and voice coil are driven by the very strong magnetic circuit. And it also can play super-loud with ease because the horn drastically increases the efficiency with which electrical power is converted into acoustic power. Its high sensitivity reduces the burden of the power amplifier and speaker cable, and makes it possible to simplify them.
Nowadays horns are largely disappeared from domestic audio use, though, I believe they deserve to be revalued.

This upgrade resulted in a big success, but one problem was revealed; residual noise comes from the midrange. It includes susurrus that is nearly pink noise and 'gee, gee, gee, ...' noise that might be harmonic components of hum noise.
I was surprised because the specification of the midrange amp, Flying Mole DAD-M100pro, is excellent (SNR:120dB, residual noise:25uV, etc).
The sound of piano is so realistic, but when I listen to it very carefully, I feel its lingering sound is a bit dull. This might be an adverse effect of the residual noise. (2015/01/26 corrected) {I found out later that this was caused by the lack of break-in of D1405. Noises, which are components not relating to the signal, doesn't make the sound dull. The real cause is distortion, which is deformation of the signal.}

If the noise includes hum, it may damage the midrange drivers. I observed the waveform and frequency spectrum of the residual noise with the instrument (PicoScope 6) at once. I did it with the input of the power amp short to ground and the output connected to the driver D1405.
The peak frequency was 393kHz and the level is 171mVrms. Probably, it is the clock frequency of DAD-M100Pro. The noise level in the audioband was quite low in spite of my concern. Hum was not found. The noise level is so low that it won't possibly damage the drivers, but it is surely a problem not to be overlooked.


Waveform of residual noise








Waveform of the residual noise













Frequency spectrum of residual noise








Frequency spectrum of the residual noise














I consulted some books and learned that the problem is not unique to DAD-M100Pro but common to all class-D amplifiers. The conclusion is that class-D amplifiers shouldn't be connected directly to horn-loaded drivers.
Class-D amplifiers have a passive (inductor/capacitor) low-pass filter between the final stage and the output terminal in order to cut off frequencies above the audioband (RF). But the passive filter cannot eliminate RF to zero. A small amount of RF, especially the clock frequency of the amplifier, pass through the filter as shown in the figure above. The direct radiator won't convert RF into acoustic wave, because its moving parts such as the diaphragm and voice coil are too heavy to respond it. In contrast, the horn driver's moving parts are vibrated by RF, and RF is modulated to audible frequencies. So the noise is heard.
For example, the user's manual of Rasteme RDA-560, a 6-channel class-D amplifier for tri-amplified audio system, includes a caution that says the amp may generate white noise or damage the driver, if it is a horn-loaded super tweeter (FYI, I heard Rasteme went bankrupt).

The midrange amp must be a linear amp. After all, I already had a plan to design and construct a new midrange amplifier, MA-215 Arabesque. I made up my mind to give the first priority to MA-215 and suspend the construction of a headphone amplifier HA-213 and the design of a phono equalizer PE-114 Petit. Of course, MA-215 will be an ultra low noise linear amplifier.

Correct use of the spikes (Nov. 2013)

(2015/01/27 added)
The spike without the pedestalI noticed that pedestals that came with the Dali's spikes were still placed under the spikes. They are needed when the loudspeakers are installed directly on the floor. They are not needed for SS-309 because it is installed on the stone board. But they are placed under the spikes since November 2004 (see 'Improvement of the speaker base (Nov. 2004)').
Using these pedestals spoils the effect of the spike (reflection of vibration). The vibration of the speaker used to propagate to the floor such that I felt the vibration with my fingers when the speaker played very loud and I touched the floor in front of the speaker.
I noticed this mistake at last, and removed the pedestals. The photo on the right-hand side shows the removed pedestal behind the spike. The vibration of the floor significantly reduced.
I felt self-hate for this entry-level mistake. However hard you work on designing and building amplifiers and loudspeakers, a small mistake like this impairs sound quality. I won't repeat a mistake like this.

Replacement of the tweeters (Dec. 2013)

(2015/01/30 added)
I felt inadequate HF attenuation and suspected the tweeter T925 had reached the end of its life. I measured the HF response with my portable digital recorder (TASCAM DR-1), microphone (Audio Technica AT822) and a test CD. As I felt, the response above 12.5kHz was steeply cut off. I decided to replace the tweeters as soon as possible.

The T925 was already discontinued. I had to choose another tweeter. The candidates were three of Fostex T series horn tweeters: T900A, T925A, T90A. I wanted only horn tweeters and there were only few manufacturers that were producing horn type. In addition, T925 kept me satisfied more than 30 years. So I didn't consider horn tweeters other than Fostex T series.
I excluded T500A from beginning, because it was too expensive and I doubted the quality of the pure magnesium diaphragm. I once attended a launch event for Fostex G2000 and it gave me a bad impression. I supposed the pure magnesium diaphragm couldn't offer vividness nor a feeling of vitality, which are the goal of Gaudi sound.

T925A was the most sensible choice, since it is the upper version of T925, which had been my favorite tweeter for more than 30 years. I expected it would offer refined sound quality with the same tone as T925.
I was also attracted by T900A. It was not only more expensive than T925A, but also it was designed more recently. I thought the higher sound quality could be expected. But its high price, nearly 100,000 yen for the pair, and its slightly worse spec than T925A made me hesitate to choose it.
The most attractive one was the most inexpensive T90A. It was apparently smaller than T925A and looked cheap, though, its spec was as good as the other candidates. Tonochi's way to select audio parts is "if function and performance are almost same, choose the smaller one." It is not unusual that the cheaper one excels the more expensive one.
I wavered over choosing which one, T90A or T925A. I could hardly decide. So I searched for other's reviews on Internet, though it is unusual for me. I found an American audiophile who owned all the T series tweeters and whose audio system was multi-amplifier system. I read his comment and other comments in a forum he administrated. It seemed that T90A was the best.

I almost chose T90A, but I changed my mind at the last minute. The reason was the appearance. The good appearance is the most important in the concepts of SS-309. I thought T90A would look too small on the wood horn H400. However highly evaluated T90A was, I thought the difference between T90A and T925A in sound quality was so slight that it would be OK to choose T925A for its good appearance.

FYI, the spec of each candidate was shown the table bellow. The spec of T925A is the same as T925 except the magnet type (the alnico magnet is used for T925A), and the horn of T925A is thicker than that of T925 though it was not shown in the spec. The list prices are as of Jan. 2015.

Model  T90A  T925A  T900A 
Impedance 8 ohm 8 ohm 8 ohm
Reproduction frequency response  5kHz-35kHz  5kHz-40kHz   5kHz-38kHz 
S.P.L.  106dB/W(1m) 108dB/W(1m) 106dB/W(1m) 
Music power 50W 50W  60W 
Magnet weight   100g (alnico)  240g (alnico)   240g (alnico)  
Net weight 800g 2kg 3.5kg 
Recommended crossover frequency 7kHz< (12dB/oct) 6kHz< (12dB/oct)  7kHz< 
Cutoff frequency 3.6kHz 3.5kHz  3.6kHz 
List price 19,000 JPY  34,000 JPY  47,000 JPY 

Fostex T925AI've bought all the LS units (except used one) from Koizumi Musen at Akihabara since I took up the hobby of building audio equipment in 1971. This time too, I bought the pair of T925A at Koizumi Musen as always. The purchase price was 27,800 per unit.

The installation was not troublesome. In the first place, the tweeters were merely placed on the wood horns. All I had to do is place the new tweeters at the same places. I used the wooded stands came with T925A, and I sold the old stands with T925.
Only item I prepared for the installation was Hanenite rubber sheets. I put them under the stands.
I didn't replace the speaker cables. The end of the cables used to be plated with silver solder and corroded because the major component of the silver solder was tin. I cut the cables by 10cm and applied Wako Technical's Titan Audio Oil to the stripped cable ends to prevent corrosion.

Only adjustment I did was adjusting the position of the tweeters. I didn't customize the network CD-211A A-NET, and even didn't adjust the levels of each tweeter. The lower cutoff frequency is 7.2kHz (18dB/oct) as it used be.
SS-309A with T925A
As for sound quality, my first impression was that it sounded beautifully but not powerfully. However, as break-in goes on, the sound of T925A is getting like T925. I wouldn't like to jump to the conclusion yet. (2015/02/26 added) {I found out the cause of the less powerful sound. I had connected the tweeters in antiphase.The fact that the power amplifier for the tweeters (MA-208) was an inverting amplifier slipped out of my mind, since I installed the tweeters for the first time in ages. The tweeters must be connected in the different polarity. After re-connection, the T925A came to sound powerfully and make me feel the textures of the musical instruments just like T925. Again, I learned that only one fault in the audio system impairs the sound quality of the system as a whole.}

As for looks, my wife complained that the unpainted stands of the tweeters were not matched with the color of the wood horn. I am reluctant to paint them because the paint might make them easy to resonate. But as my wife said, the stands are a little too conspicuous. I'm going to paint them in the same color as the wood horn in the near future.

I sold the pair of my old T925s at a nearby secondhand shop for 15,000 JPY. I was surprised that the 31-year-old half-broken tweeters were worth 15,000 yen. Maybe, T925 is more popular than T925A among audiophiles.


Return to Contents



Return to the top of this page