Thứ Tư, 15 tháng 6, 2011

The Sunshine

The Sunshine

By Holger Kraft
09 Aug 1997
Printed from, 15 Jun 2011 13:13
URL: http:///content/diy/1036/


First of all I would like to give you a short overview about the history of the project. The main reason why I began to develop my own speaker is, that I am interested in tube amplifiers, like many DIY-audio enthusiasts. Five years ago I built my first vacuum tube amplifier, an Orthophonic IV (this is a kit from a small German company called Audio Workshop). Since this time I was highly infected by the Vacuum Tube disease. Now you can imagine that this speaker is suitable for nearly all tube amplifiers and all other amplifiers that have a rather small output power.

Introduction and a short personal opinion

Because there are only very few commercial speakers available which are able to cooperate well with low output power around 7 Watts and a low damping factor and that do not look like a lorry after a heavy traffic accident, I decided to try my own way. Of course there are great projects in that field available on the Internet and of course you all know The ARIEL and The ME_2 by Lynn Olson and The Poly Natalia by Dick Olsher. But here in Germany there is no possibility to hear this speakers and I would recommend a speaker with a high impedance that will not drop under, let us be realistic, 7 Ohm, because all tube amplifiers can better deliver Voltage than current. The great designs mentioned above are both 4 Ohm designs, using two 8 Ohm woofers in parallel. For all of you who are puzzled about sensitivity and efficiency and all those stuff I will make the statement that an 8 Ohm speaker with an sensitivity around 90 dB has the same efficiency as a 4 Ohm speaker with an sensitivity around 93 dB.

First thoughts about choosing the right drivers

To reach my goals I began playing around with the new (at the moment they are not very new any more) Audax HD-A loudspeakers. My first choice was the HM 170 Z0, the 6,5" HD-A Driver. Here in Germany there is a quite successful 2-way design with this driver available on the DIY-audio-market. It is the Esprit, which was developed by the German DIY-audio-magazine Klang&Ton. In this design the Aduax driver is combined with the HRA 531, a ribbon tweeter from Expolinear which is very similar to the Jordanow ribbon tweeter HRA 531. I auditioned this speaker in a local speaker shop but I was not to impressed. There are also a few features in that design I don't like at all. First is they use a very large serial inductor for the low pass of the crossover (they choose 1,8 mH) which causes that the sound of the speaker is a little bit boomy like the sound of all the other speakers on the market. Speakers with a boomy sound are the worst choice you can make for SE Triode Amps. Because of there low damping factor the bass response will be gained and the resulting sound will be even more boomy than with a normal solid state amp. The second property of the design I dislike is the use of a very critical 12 dB filter for the high pass. Because of this, the impedance drops to nearly 4 Ohm in the region of 4 kHz and I think this is far to low. I wanted to use a 6 dB/6 dB crossover because over the years I discovered that this sounds best to me. That means I have to use a tweeter which is suitable for a 6 dB high pass. I decided to use a Dynaudio tweeter because they recommend their drivers for the use with 6dB filters and I like the sound of the Dynaudio tweeters (I own a pair of Dynaudio Contour 1.3's and I am very satisfied with their sound, but the heights , the mids and the bass...., I could imagine a little progress in every frequency range). There are two tweeters that sound really good to me:
  • first is the Esotec D 260 
  • and second the famous Esotar D 330-T
Every tweeter of this two has its special problem. The Esotec still has little problems with the sibilance. I hear this problem when I listen to my Contour 1.3 which has a tweeter that is on a midway level between the Esotec and the Esotar according to Dynaudio. The Esotar has only a "minor" problem: It is very expensive, especially here in Germany (around 670,- DM / piece). But the Esotar has a few advantages over the Esotec:
  • First of all the sound quality. I heard this tweeter 3 years ago at the High-End in Frankfurt. The Dynaudio people made a presentation of the new Confidence 3 speaker and I am still very impressed by this presentation. 
  • The second advantage is, that the efficiency of the Esotar is 4dB higher than that of the Esotec and I need this gain in efficiency for the design ( I used a special circuit in the high pass crossover to correct the phase between woofer and tweeter and this circuit wastes 2 dB of efficiency). 
After simulating around with the HM 170 Z0 and the Esotar D 330-T I recognized, that this two drivers do not cooperate in the best thinkable way. By this time I discovered the parameters of the HM 210 Z0 looked even better then those of the 170 Z0. The bigger chassis gives me an increase in efficiency around 2 dB and the frequency response showed a very good-natured behavior. After simulating for a few rounds, it seems that the 8" Aerogel driver only needs a very small inductor around 1 mH to give a reasonable frequency response. In connection with the Esotar tweeter the final result of the frequency response looked like this Fig. 1. Frequency response.

There has to be some bass too !!!

By now you only got a guess about the frequency response in the higher octaves. I have to mention here, that I am no bass freak at all. But every speaker should have some fundamental sub tones. Therefore and because the parameters of the Audax chassis point in this direction, I constructed a bass reflex cabinet. I know, this is not very creative and new, but I think a well done bass reflex cabinet performs better than a bad closed one. Because of the low damping factor of the famous single ended triode amps I decided to stay a little bit on the lean side of life with my bass reflex design. The result looks like the picture in Fig. 2. Bass response. As you can see on the graphic, the bass response reaches down to approximately 43 Hz. The netto box volume is around 47.5 liters and the tuning frequency is 35 Hz. This tuning frequency corresponds to a port with a diameter of 6.6 cm and a length of 13.0 cm.

The crossover construction

I mentioned above, I am a fan of 6 dB/6 dB crossovers. Because of this I decided to build one Fig. 3. The crossover. The major advantage of this filter type is that everything, like phase, frequency response and complex impedance is very smooth and the whole filter behaves very good-natured. One big disadvantage is, that the two drivers work in parallel for a long frequency interval . Therefore you have to take special care about the phase and because of this, I added some phase control circuit. I used a circuit very similar to the phase correction circuit used by Dynaudio. But take care: I connected the tweeter in the reversed polarity than Dynaudio! (This means that the tweeter in my approach is not reversed at all!!) The second big difference is the corner frequency of the circuit. Dynaudio uses in all their designs a corner frequency around 2 to 3 kHz. In opposition to this I use a corner frequency around 500 Hz. The only problem with this is, that it leads to bigger and more expensive component values. The advantage is, that the phase behavior of the tweeter and the woofer are nearly the same. This means, the impulse answer is better than the approach with the reversed polarity tweeter used by Dynaudio.

How the phase control circuit works

I will try to explain how the phase control circuit works (together with the whole design of the Sunshine). As you can imagine, when you are a experienced speaker builder, the cabinet and especially the placement of the drivers on the cabinet is very important for a perfect addition of sound between the two drivers. If you look at the picture below, you can see that the kind of placement used in the design of the Sunshine provides, that the way from the tweeter is approximately 2 cm longer than the way from the woofer to the ears of the listener. This works for approximately 2 meter listening distance Fig. 4. Distance from drivers to ear.
As you can imagine now, the crossover of the Sunshine is optimized for a listening distance of 2 meters and a ear level of 1 meter. This should be average values for most conditions. Now to the phase control circuit. As you might know from school or university there is a special way in AC-technology to display current and voltage. This is called a pointer diagramm. Two alternating voltages are displayed as two pointers rotating counterclockwise. The phase between the two pointers corresponds to the angle between the two pointers. You can translate this directly to the problem of two drivers in a speaker. Of course you would like the two pointers in a speaker to point in the same direction for a perfect addition of sound. Nature is cruel! A low pass always gives you a shift in the clockwise direction. A high pass gives you a shift in the counterclockwise direction. If you now add some phase shift from the drivers itself, a woofer always tends to add some more clockwise shift and a tweeter with ferrofluid tends to give you zero or counterclockwise shift. This means, the situation gets even worse. If you now consider, that in most designs the tweeter is located on top of the woofer, you have to add some more phase shift in the crossover region, corresponding to my little drawing. This means the situation gets even worse than before. In the design of the Sunshine I tried to get rid of this problems as good as possible. I did this with two main design features:
  • First of all I optimized the Sunshine for a listening distance of 2 meters. This covers the problem of aligning the acoustical centres of the woofer and the tweeter. It works in the way explained in the drawing above. 
  • Because I do not like tweeters connected in reversed polarity to restore the phase alignment of the drivers, I use a phase control circuit. The tweeter connected in reversed polarity does not really restore the phase alignment of the drivers anyway. Only the average sum of the tweeter and the woofer, the amplitude answer will look OK. In reallity the tweeter will begin a half period earlier than the woofer to radiate a special frequency and, of course, the woofer will end to radiate a special frequency a half period later. The phase control circuit slows down the tweeter. This means, that there is no phase shift at very low frequencies and 180 degree phase shift at very high frequencies.
As you can see on the schematic Fig. 5. Response of the phase control circuit I designed the circuit to have a corner frequency around 500 Hz. That means a phase shift of 90 degree ( 1,57 ) at 500 Hz. Like I mentioned before I slowed down the tweeter. Therefore the real phase shift is in the negative direction ( The possibility with the to big and to expensive condensators and inductors described below will alter the shape of the curve, but not the corner frequency ).

Dimensioning the phase control circuit

If you look at the crossover schematic you will find the phase control circuit behind the filter condensator (L2, L2', C3, C3'). To provide perfect working conditions for the circuit, you have to take care for a linearized tweeter impedance. Also the 6dB filter will love the linearized tweeter impedance. The linearized tweeter has an average impedance around the DC resistance of the tweeter voice coil. This DC resistance will be the starting value for calculating the phase shift circuit. First step now is, to choose a corner frequency: Lets take 500 Hz like in the design of the Sunshine. The second step is to calculate an inductor and a condensator which have the same resistance like the DC resistance of the tweeter at the corner frequency. The calculating formulas are:
C = 1 / (2 * PI * corner-frequency * DCR-tweeter)
L = DCR-tweeter / (2 * PI * corner-frequency)
With this formulas and a DCR of the tweeter from around 5,3 Ohm you end up at a value of 60 uF for the condensator and a value of 1,68 mH for the inductor. For the Sunshine I chose 60 uF and 1,5 mH for the part values. You can design the circuit with the inductor and the condensator having only the half value of the DCR of the tweeter, but then you will end up with 120 uF condensators and 3,4 mH inductors. This seems to be a little bit to expensive, even for me, and a little bit to big too.

A few words about parts quality

I used Solen Fastcaps in the whole crossover except for the 82 uF cap in the impedance linearizing circuit of the tweeter. I used them, because I like their sound. For further improvement I added a 0,1 uF Sn foil cap (IT) in parallel to the 8,2 uF filter cap in the high pass and a 0,2 uF Sn foil cap in parallel to the 60 uF (made of four 15 uF caps in parallel) phase shifter caps. After all I did not want to destroy the abilities of the Esotar tweeter. For all inductors I chose a rather large diameter, because I think in connection with a 8 W amplifier you should not waste a little bit of power in the crossover. The second reason is, that in my opinion inductors with a small diameter and big resistance do not their best to enhance the dynamic abilities of a speaker system. The third reason is that I did not want to lose a little bit of the very small damping factor of my poor SE amp. Lets say I want to make life easy for it, because one design goal was to make the speaker suitable for all 'poor' amps (SE tube, SE solid state, class a solid state, ...). OK, I need some attenuation for the tweeter and that also 'wastes' power, but.... For all the resistors I used ten 0,5W metal film resistors in parallel, like mentioned on the crossover schematic.
After all I did not design a impedance linearizing circuit for the whole speaker system. There are different reasons for it:
  • I recommend bi-wiring. I have always problems to decide where the circuit should be placed. Maybe directly at the amplifier or at the entry of the low pass? 
  • Like I said above, I did not want to waste power and damping factor. Using an impedance linearizing circuit means, that you bleed power on a resistor in a certain frequency range. 
  • Impedance only rises to, lets say, 20 to 25 Ohms and the crossover is a phase corrected 6 dB/6 dB design and does not behave in a very nasty way. 
Last thing to mention here would be, that I prefer, like many other popular developers here on the Internet (Dick Olsher, Lynn Olson, ...), an extra housing of the crossover because of the same reasons like the other guys mentioned before: Less vibrations on the crossover parts, better possibility for tweaking around and so on.

Construction of the cabinet

Pictures of the speaker:
  • Fig. 6. Cabinet cross section 1
  • Fig. 7. Cabinet cross section 2
  • Fig. 8. Speaker in room
  • Fig. 9. Speaker pair in room
Like you have seen on the pictures above the cabinet is constructed of birch multiplie. I chose it because of the great sound I expected from this kind of wood. It is very stiff and I did not want to use many internal bracings inside the cabinet because they do not kill the wall vibrations at all. As you know a bracing only kills one mode of wall vibrations. Another reason for this choice is the look of the speaker. I didn't like to spend many money on colour or veneer. It is better to spend some money on the structure of the wood than on the surface. Here in Germany the price for veneered MDF is the same as for birch multiplie. Yet remember out of what kind of wood pianos are made. There are surely not made out of MDF! As you can see in the picture below the cabinet is divided into two chambers. The upper chamber, nearly triangular shaped, is the effective volume for the woofer. The lower chamber, also triangular shaped, is for sand filling Fig. 10. Cross section explained. This sand filling damps the whole sidewalls and the free area of the front panel. The filling will provide that the speaker is very stable positioned in the room. This is true for two facts: The first is, the whole thing weights around 75 Kg and you will be very glad when you have found the right listening position for it in your listening room. The second is, there is only a very small impulse transfere from the driver to the cabinet. The most important reason why to use sand filling for a part of the cabinet is, that the sand filling acts as a trap for all the resonances in the cabinet. Its similar to for example damping a tonearm with silicone fluid. As you can imagine from the pictures, the cabinet gets softer from the front panel to the rear end. It is very important to make the front panel as stiff as you can and mount the drivers as fixed as possible. Do not use any softener between the tweeter and the cabinet. Look also at the bracing in the upper chamber. Later on the cabinet plans, you will see that this part is U-shaped. It is not only a bracing for the front panel and the sidewalls. The second important task of that piece of wood is, to prevent a direct sight from the woofer to the bass reflex port. Although the port is placed on the rear, you should try to minimize the ammount of midrange that disappears out of the bass reflex port, especially when you use as less damping material as I used in this design. I only covered the top part of the upper chamber with self adhesive roofing felt and covered all walls of the upper chamber with 15 mm felt lining, also over the roofing felt. (You can see the coverage of roofing felt in the top part of the effective volume in the picture above.) There are many additional design features in the enclosure, I will not discuss in details here like the off-axis tweeter, that reduces the standing waves between the tweeter and the edges of the front panel and the positioning of the tweeter under the woofer which provides a runtime correction and proper phase aligment between the tweeter and the woofer ( partly covered in the chapter: Crossover construction ). If you are interested in this design you can contact me and we can discuss some further details. But now lets stop talking and have a look at the nasty plans. Again you can click on the thumbnails to get a bigger image. (Please excuse the poor handdrawing quality of the schematics)
Cabinet drawings:
  • Fig. 11. Front and side
  • Fig. 12. Back wall
  • Fig. 13. Brace shelf
  • Fig. 14. Back inner wall
Of course, if you are interested in this speaker project and you have some additional questions that I did not cover with my pages, feel free to contact me.

A few words about listening impressions

As you can imagine, I am totally satisfied with the final result. But that is no wonder, because I built the speaker around my needs and my imagination how things should be. After all, let us try to go a little bit more in detail. The efficiency of the speaker is high enough to cooperate well with a single ended 300B tube, anywhere in the 90 dB range. The impedance is on a real high level, between 7,5 and 25 Ohms. These are the right circumstances to enjoy the typicall full bodied 300B sound with its realistic expression on voices and instruments. On the other hand the speaker is able to deliver a soundstage and a resolution like the best minimonitors. The bass response of the Sunshine is very deep and natural and misses the typical 60 to 100 Hz overshoot of the most commercial designs. Midrange is very open and detailed but very critical in terms of choosing the right electronics. Thin and sharp sounding transistor equipment will not perform well with the Sunshine, because you will hear exactly how poor they sound. The heights are outstanding good in my opinion. As I told, I love the sound of the Esotar tweeter. After a 'warm up' time of 15 to 30 minutes you will not hear that there is a tweeter at all. As a conclusion you can say the Sunshine is on a midway between the best big hornsystems and the best tiny minimonitors.
Of course, this is not a budget speaker project. The price for the complete parts, including wood, screws and damping material is around 3000 to 3500 DM. In my opinion the speaker is worth any penny of it and I try to sell my Contour 1.3 now. If you compare the material used in this project to the material used in commercial products you will recognize, that a commercial speaker made in the same way as the Sunshine would be in the price region between 10000 and 20000 DM. Remember, D. Burmester takes 3500 DM extra charge for the outboard crossover. 'so far so good' (direct translation from german).
All rights on this project do belong to the owner of the project: Holger Kraft, Talbergstr.15, 35644 Hohenahr, Germany. It is free for non commercial use only!!
If you have questions or problems please refer them to me.
Holger Kraft


Fig. 1. Frequency response
Frequency response, red and blue corresponds to the electrical filter slopes, green and yellow corresponds to the frequency answer of the driver with filter and purple is the sum of the two filtered drivers; the 1 on the y-axis corresponds to 90 dB.
Frequency response

Fig. 2. Bass response
Bass response

Fig. 3. The crossover
The crossover

Fig. 4. Distance from drivers to ear
Distance from drivers to ear

Fig. 5. Response of the phase control circuit
Response of the phase control circuit

Fig. 6. Cabinet cross section 1
Cabinet cross section 1

Fig. 7. Cabinet cross section 2
Cabinet cross section 2

Fig. 8. Speaker in room
Speaker in room

Fig. 9. Speaker pair in room
Speaker pair in room

Fig. 10. Cross section explained
Cross section explained

Fig. 11. Front and side
Front and side

Fig. 12. Back wall
Back wall

Fig. 13. Brace shelf
Brace shelf

Fig. 14. Back inner wall
Back inner wall

Copyright © 1995-2001 Roy Viggo Pedersen. All rights reserved.