This is an English version of the article from L’Audiophile no.25. september 1982.
Translated using source scans from:
La Petite Onken article scan in french at Onken.info
La Petite Onken article scan in french from http://6bm8.lab.free.fr
Original text © By Jean Hiraga
L’Audiophile
In issue no. 2 of L’Audiophile, we described a loudspeaker enclosure that enjoyed, and still enjoys in Japan, enormous success: the Onken enclosure, derived from the Jensen principle, but recalculated and optimised. Despite its imposing volume, it achieved great success in France since, according to our surveys, nearly one thousand audiophiles are said to have built it. This time, readers will become acquainted with another enclosure, still of the Onken type, but of reduced volume and equipped with a reference Altec 414-8B loudspeaker. Without sacrificing subjective quality, the result obtained is remarkably close to the first Onken version, within a much more acceptable physical size.
Eijiro Koizumi, of the Onken firm, once made a comparison between spaghetti served in several restaurants and hi-fi components. First, there is the raw material, its price, the ease or difficulty of obtaining it, and the choices to be made.
Then there are the different ways of making flour and, above all, of working the dough, of “drawing” it to make spaghetti. On one side, one can see the waste of good raw material through processing on industrial machines. On another, there will be the craftsman who is unable to control the raw material perfectly, or who is not skilful enough with his hands to ensure homogeneous production. At the opposite extreme, there will be the “purist”, who will do his best without thinking about yield, cost price, or the time spent kneading the dough by hand.
But for the customer, prices that vary too much will cause hesitation, bring some disillusionment and disappointment, until the day he finds, in a small alley, a little restaurant in front of which people queue every day, and where it is possible to taste the “best spaghetti in the world”. Finally, a restaurant where one can find pasta whose consistency, taste, and many other inexpressible qualities will make every newcomer a loyal customer.
Between two systems of the same price, the same differences may exist, the important point being the degree of satisfaction of the customer. At audiophile gatherings, one finds among them the happy ones and the unlucky ones, the latter often being more numerous. However, the degree of satisfaction or dissatisfaction with a system also depends greatly on oneself, something one does not always realise, not to mention the influence of one person on another. As a result, these remarks might remain incomprehensible in the eyes (or ears) of a third party. The important thing is to be satisfied with the spaghetti from the little restaurant or with component X.
Yes, but to think that the basic wheat or the resistors are almost the same in all cases clearly means that nothing has been understood. It is like someone who would buy the first acoustic guitar that comes along simply because it is made of wood, like the others. Eijiro Koizumi once had his Denon DL-103 cartridge exchanged five times in succession in order to recover the same timbres, the same nuances, the same ambience. Cost: 1,600 francs. Ridiculous, one person would say. At that price, another cartridge could have been bought, someone else would say. But conversely, one could say that the joy of rediscovering what one believed lost goes beyond the material question of price, whether it be 5 francs or 100,000 francs.
Whether starting from known or unknown factors of success, or from both, the fact remains that a “best seller” knows how to attract and please a very large number of people. One person will feel “very small” and deeply moved in front of the beauty of a painting, the warmth of a singer’s voice, or the naturalness of an interpretation. Another, the one who doubts, will tear his hair out trying to determine, with the help of a computer, whether the thousands of proportions of the Venus de Milo are really “the most perfect”.
For an acoustic enclosure, one can speak of art, something that one feels without being able to explain clearly: the art of knowing how to combine thousands of technical parameters in order to make something as “true” as possible on a subjective level. Something that can reproduce not only two sinusoidal signals perfectly in phase, but also the inimitable timbre of a Buffet-Crampon instrument or the hoarse voice of the Japanese singer Mori Shinichi.
It is quite incredible that, in 1982, certain loudspeakers, perfectly developed from a technical point of view and refined using the most modern computers, produce only poor subjective results, reproducing vocal timbres so distorted that they would be infinitely more faithful if reproduced over a telephone. How can one explain that a loudspeaker developed in a modest way, 25 years ago, by G. A. Briggs, can sound unquestionably “true”? Conversely, one can imagine what he might have been capable of if he had been able to benefit from our latest technical innovations and our powerful computers.
For Eijiro Koizumi, a good acoustic enclosure represents the highest level of subjective optimisation. It would not be a question to fall back once again on the so-called “subjective” component for which 15 % distortion at 60 Hz, linearity within +12 dB, or wow of 0.35 % could nevertheless make it “the best in the world”. A component can be remarkable both in measurement and in listening, and Koizumi knew how to demonstrate this.
The Altec 414-8B
The success of an older sibling, the 416-8B, could have prevented the younger one, the 414-8B, from being recognised. Yet, upon reflection, the 414-8B proves superior in several respects.
The 414-8B has a diameter of 30 cm, whereas the 416-8B measures 38 cm. However, their effective radiating surfaces and construction are very similar, as shown in Figure 1.
Figure 2a shows the free-air impedance curve. For a new 414-8B, one notes an instability in the measurement of the resonance frequency.
After several days of operation on a music programme, preceded by excitation using a signal generator (29 Hz, 1 W, for 3 to 4 hours), the measurement becomes more precise.
It is observed, during measurement, that simply placing the loudspeaker on the floor, cone facing upwards, is sufficient for the floor effect to be reflected in the impedance curve.
This measurement must be precise. It is not sufficient to rely solely on the characteristics given by the manufacturer. Doing so may result in an error of 20 to 30 % in the enclosure calculation.
Although Altec publishes an impedance curve that is very homogeneous between 100 Hz and 1 kHz (Fig. 2b), Figure 2 reveals two small resonances, one at 180 Hz and the other around 420 Hz, due mainly to the peripheral suspension.
In an anechoic chamber, these two resonances will have an effect on the level-versus-frequency curve, in the form of small irregularities.
In a normal listening room, they will pass unnoticed.
| Characteristics | 416-8B | 414-8B |
| Diameter | 38 cm | 30 cm |
| Impedance | 8 Ω | 8 Ω |
| Admissible power | 75 W | 50 W |
| Sensitivity | 102 dB/W/m | 98 dB/W/m |
| Frequency range | 20 – 1600 Hz | 30 – 4000 Hz |
| Resonance frequency | 25 Hz | 30 Hz |
| Voice coil diameter | 7.62 cm | 7.62 cm |
| Q₀ | 0.24 | 0.25 |
| Fixing diameter | 35.9 cm | 28.3 cm |
| Weight | 7.9 kg | 6.8 kg |
Figure 1 — Characteristics of Altec loudspeakers 416-8B and 414-8B
Source: Article page 70, bottom-right table
Two small irregularities appear around 180 Hz and 400 Hz. These are clearly caused by the peripheral suspension.
Source: Article page 71, top-left
Under normal listening conditions, these defects are not perceived. The dispersion of energy of the cone between 28 Hz and 32 Hz remains small. On the two 414-8B loudspeakers used, one resonance appeared at 28.5 Hz, the other at 31 Hz.
The mechanical damping coefficient Q₀ is obtained from the voltage measured at the resonance frequency f₀. By locating two points on the impedance curve corresponding to frequencies f₁ and f₂, and knowing both the DC resistance Rₑ and the impedance at resonance Z₀, Q₀ is given by:
Using the values taken from Figure 2a:
The value of Δf (f₂ − f₁), equal here to 4.1 Hz, shows that this loudspeaker is overdamped.
Next, the value of the equivalent moving mass M₀, calculated according to the method shown in Figure 3 and using formula (2), will be:
Whereas for a 38 cm loudspeaker the values of m were between 30 and 50 g, for this 30 cm loudspeaker values of m between 10 and 40 g will be used, which will give four values for M₀:
For m = 10 g:
For m = 20 g:
For m = 30 g:
For m = 40 g:
The response is linear between approximately 300 Hz and 800 Hz. Efficiency is slightly higher in this region.
Source: Article page 71, right-hand composite figure
The masses of 20 and 30 g giving the same value, M₀ will therefore be, for this 414-8B, equal to 35.6 g.
The magnetic circuit being close to that of the 416-8B, and M₀ being nearly half the value of the M₀ of the 416-8B, Q₀ is low and the damping is high.
However, M₀ does not correspond to the weight of the moving assembly alone. It is necessary to take into account the acoustic load and the equivalent compliance, which depend on the active surface area of the diaphragm.
Source: Article page 72, left-hand graph
Source: Article page 72, lower-left
It is observed, on the one hand, that the spiders of the 414 and 416 series are the same, and that the peripheral suspensions are of the same type, made from the same materials.
The values of Q₀ and M₀ being lower than those of the 416 series, one must expect a level-versus-frequency response curve that falls more rapidly at low frequencies.
One sees here the contradictions that exist between the parameters of transient response, low-frequency resonance, extension of the level/frequency response below 100 Hz, moving mass, and efficiency.
To determine the value of the useful radius of the diaphragm, one can refer to Figure 4, which makes it possible to determine the equivalent active radius “a”, according to formula (3):
which gives, for the active radius of the 414-8B:
Between the 416-8B and the 414-8B, the following differences are noted.
For the 416-8B, f₀ is 24.8 Hz, Q₀ is 0.28, the equivalent mass is 61 g, and the active radius is 16.3 cm.
For the 414-8B, f₀ is 28.5 Hz, Q₀ is 0.254, the equivalent mass is 35.6 g, and the active radius is 13.3 cm.
Low-Frequency Limit and Enclosure Volume
Low-frequency response is normally limited by the low-frequency resonance f₀. However, it is also necessary, in each case, to take into account the value of Q₀, which, together with f₀, will determine the lowest frequency that can be reproduced. This limiting frequency is called fₗ.
The value of fₗ is determined by f₀ × K, K being a factor related to Q₀ and to the enclosure volume, as well as to another quality parameter concerning the rate at which distortion rises below 100 Hz (often very pronounced, except for the S15B, 416-8A-B-C and 414-8B-C Altec, and a few rare exceptions).
According to this factor K, determined by a formula which Eijiro Koizumi apologises for not being able to disclose here, it is possible to determine the limiting frequency fₗ as well as the optimum volume, without neglecting the factors of transient response and overhang.
It is also observed that an excessively low Q₀ systematically causes fₗ to rise and rapidly reduces sensitivity below this frequency. Apart from coupling by a horn, a bass-reflex type enclosure cannot modify the value of fₗ.
For the 414-8B, Q₀ being 0.254, the value of K will be close to 1.57. The calculated value of fₗ will therefore be:
that is, 45 Hz, at −3 dB relative to the average level.
Below this frequency, one can already anticipate a very rapid drop in acoustic level. It would be incorrect to say that this drop will pass unnoticed by the ear. By using heavier diaphragms, it would be possible to lower the value of fₗ, but this would be at the expense of sound quality.
As for the frequency fᵣ, it is located, for a bass-reflex enclosure, approximately at the position of the anti-resonance on the characteristic impedance curve, composed of two resonances of fairly close or similar amplitude.
The enclosure is of a type inspired by the Jensen “bass-Ultraflex” version, although the latter is not calculated in the same way.
Compared with the Jensen enclosure, the Onken-type enclosure is characterised by a very linear level-versus-frequency response, a very low distortion rate up to the low-frequency cutoff, and an exceptional transient response.
Compared with the common bass-reflex enclosure, or even the Jensen-type enclosure, the subjective improvement is obvious. In addition, it is an enclosure that adapts very well to horn-loaded loudspeakers, a combination which, in 98 % of cases, provides with these drivers bass reproduction that is much “faster”, much less “slow to start”, and less “soft”.
The enclosure described here is therefore of the Onken type, baptised for this version “Petite Onken”, “I.P. Ultra-bass”, I.P. meaning “Impédance Porte” or “Impedance Vent”.
These Onken enclosures are specifically and exclusively adapted to loudspeakers with very low Q₀. It would therefore be out of the question to adapt just any 38 cm loudspeaker. The chosen combination makes it possible to reach a very high level of quality, provided that the indications given by Koizumi are respected.
Optimisation of the I.P. Ultra-bass enclosure
The enclosure described is designed to allow access to a very high level of low-frequency reproduction quality, down to 45 Hz at −3 dB, with an efficiency exceeding 95 dB/W/m.
This represents a major performance compared with the 337-litre Onken enclosure equipped with the 416-8A and having a cutoff frequency fᵣ of 35 Hz, because here one will see that, with only a 1 dB loss in efficiency, despite the use of a 30 cm loudspeaker, an fᵣ of 45 Hz is obtained with an enclosure volume between 160 and 200 litres.
In a larger volume, one would obtain a “descending” response curve and a troublesome low-frequency resonance, as well as a further general loss of efficiency in the bass, not to mention subjective results which would necessarily be inferior.
On this point, Koizumi again refers to the parameter K, the key, the principal point from which the enclosure is conceived in view of an optimum qualitative result.
The volume V₀ is obtained using the formula:
which gives: 1.55 × 10² dm³ that is, 155 litres
It is therefore necessary to have a volume of 155 litres, which must be distributed over three dimensions. This volume, small in comparison with the well-known Onken enclosure, is the minimum that makes it possible to reach this level of quality.
It would therefore be ridiculous to look for another solution aiming at a greater extension of the low-frequency response curve in a smaller volume, since this is impossible. Even at the cost of a loss of efficiency, the gain of a few hertz at the bottom would translate into an enormous loss of quality over the rest of the low-frequency response.
The essential point is not to obtain an ultra-flat level-versus-frequency curve down to 16 Hz, but rather to obtain the maximum of quality over the widest possible frequency band.
It is certain that with a sealed enclosure of a volume three times smaller and a 30 cm woofer with a very heavy diaphragm (120 to 140 g), it is possible without difficulty to reach frequencies as low as 40 Hz. But it is obvious that the quality of the bass will be very clearly inferior, and the loss of efficiency greater than 10 dB.
Such a solution would furthermore require very powerful amplifiers (more than 200 W), as well as loudspeakers whose voice coils would be capable of withstanding very high thermal dissipation. A solution that is ultimately more expensive and with limited objectives.
In fact, one should be able to adapt the enclosure and the loudspeaker to the listening room and have a loudspeaker whose parameters could each be pre-adjusted. It must also be considered that a loudspeaker that is “too good” does not always perform well in domestic use because it can be too demanding to use.
If one is dealing with a good woofer, overdamped, with a very low Q₀, there is a strong chance that it will produce a light, dynamic sound, full of vitality and energy, and perhaps little coloration if a good model is chosen.
Whereas in the past amplifiers were of insufficient power to control loudspeakers, the current trend consists in “assisting” a loudspeaker of mediocre quality (but sometimes extending very low in the bass due to a relatively heavy diaphragm) with an amplifier of enormous power and very low internal resistance, high fidelity being, in this unfortunately very common case, reduced to a question of bandwidth extension and damping factor.
Too much evidence shows that these “revolutions” have simply caused progress toward the highest levels of high-fidelity reproduction to stagnate. For while technological progress is always possible, it would not be a question of changing acoustic laws and of “forcing” 20 Hz at 130 dB sound pressure level into a 50-litre enclosure by any means.
The commercial failure of almost all bass enclosures and “subwoofers” is precisely due to a significant loss of quality relative to the rest of the system (even when that rest is already of average or mediocre quality).
This is why it must be understood that a volume of 155 litres, half that of the Onken enclosure, represents a very good compromise. The net volume V₀ is 155 litres, to which 37 litres (vents, loudspeaker, etc.) must be added, giving a total volume Vₜ of 192 litres.
The surface area of the vents was chosen to be fairly close to that of the active surface area of the diaphragm. As for the vent depth, since the calculation is not the same as for a bass-reflex enclosure and must be followed by optimisation (±5 % on either side of the calculated value), a vent depth of 32 cm is obtained.
In practice, the depth will be adjusted to 36.5 cm. Although Koizumi does not give his calculation formula, he attributes this difference to slight measurement errors, such as the calculation of the true active diameter (in the frequency band concerned) of the loudspeaker, the calculated value being only a good approximation.
Adjustments
The enclosure is presented as shown in Figure 5. The wood used has a thickness of 25 mm (Nantex-type). One will note dimensional ratios that differ from those of the Onken enclosure. Apart from a possible adjustment of the vent depth, the dimensions cannot be altered, nor (and above all) can the positioning of the loudspeaker on the front panel be modified.
This is, as with the Onken enclosure, a point that must be respected and must not be altered. One should bear in mind, for example, that for the Onken enclosure some 120 different positions were tested before arriving, six months later, at the optimal positioning. This long effort was carried out in collaboration with the research department of Japanese broadcasting and a professor of acoustics from the University of Tokyo, who had realised, together with Eijiro Koizumi, the well-known Onken enclosure.
One therefore understands that certain formulas cannot be disclosed to the reader. The essential point, however, is to be able to build a high-quality enclosure.
In the “I.P. Ultra-bass” enclosure, as in the Onken enclosure, one notes an impedance characteristic, after optimisation, that is slightly different from that obtained with a well-tuned bass-reflex enclosure. Indeed, one observes that the two resonances do not have exactly the same amplitude, which is therefore not a cause for concern.
The resonance f₂ is of greater amplitude than f₁, the effect of the acoustic load on f₁ being much less pronounced than in a bass-reflex enclosure.
Figure 6 shows the impedance characteristic obtained after optimisation. Figure 7 shows the frequency response curve obtained in a semi-reverberant room, where one should not take into account the slight irregularity at 200 Hz (room standing waves).
Source: Article page 74, upper-right graph
Source: Article page 74, lower-right graph
Source: Article page 76, top-right
Source: Article page 76, bottom-right
Other tests of vent depth were carried out, including one giving the impedance characteristic obtained for a depth of 37.5 cm (instead of 36.5 cm). The impedance curve varies slightly, as shown in Figure 8, and the level-versus-frequency response curve, shown in Figure 9, exhibits a slight gain in level below 60 Hz.
However, a subjectively inferior result (sound less light, more muffled) in the room in question (25 m²) led to the choice of the 36.5 cm vent depth as the best compromise.
Concerning vent depth, Figure 10 presents three examples: vent too deep, vent too shallow, or correctly adjusted, and their repercussions on the response curve. If the enclosure volume is modified on either side of an optimal value, a curve similar to that of Figure 11 is obtained.
The internal walls are covered with soft felt of average thickness 15 mm. Internal damping has the effect of rounding off the two resonances of the impedance curve. A brace is used between the front panel and the rear panel, as in the classic Onken enclosure.
In the case of using a horn for the midrange, it is recommended to place the horn as shown in Figure 12. The vertical dotted line corresponds to the position of the sound source image, located approximately at one third of the depth of the woofer diaphragm. It would be pointless to attempt to align the voice coils, since this method recommended by Onken provides very good alignment with the measurements.
By listening, the final positioning adjustment can be made using music: slow piano notes and fairly deep bass are suitable test material. Poor phase alignment will hollow out part of the 400–700 Hz band or produce an abnormal timbre.
Source: Article page 77
Listening
Overall, this enclosure is at the level of the Onken enclosure equipped with a 38 cm woofer. The faster roll-off in the bass is not very noticeable, and this is even an advantage in small rooms (positioning and adjustment are less critical). In this regard, one might even say that it is superior to the standard Onken enclosure.
The organ (Saint-Saëns) descends a little less deeply but remains perceptible and clean. On the cello, the results are very good, with little difference compared to the Onken enclosure.
On the TBM-63 record “Orpheus”, the double bass is reproduced perfectly, just as in the Dittersdorf concerto. At the moment, the 30 cm driver seems advantageous compared to the 38 cm. Large diaphragms tend to produce more overhang. The enclosure offers very good balance, without excessive heaviness between 60 and 100 Hz (a fairly common case).
Contrary to certain claims, the loss of a few hertz in the bass is not a disadvantage in the “petite Onken”. On a musical level, the 414-8B often behaves better than the 416-8A as soon as frequencies above 150 or 200 Hz are involved.
This enclosure was built in 1976 by Eijiro Koizumi. It also achieved great success in Japan despite the higher price of the 414-8B loudspeaker compared to the 416-8B. On the other hand, Altec models have evolved since then and certain older references may no longer exist.
It was not simply a matter of describing, without modification, the Onken enclosure designed in Japan in 1976.
Adaptation to another loudspeaker
This study is ongoing and concerns the adaptation of a Focal loudspeaker designed by Jacques Mahul, the 25 cm high-efficiency model 10C01. The aim would be to obtain a smaller enclosure volume and a low-frequency cutoff of approximately 55 to 60 Hz, while maintaining efficiency on the order of 94 to 95 dB.
[…] Construction of the ONKEN bass enclosure – Jean Hiraga, L’Audiophile 1977 Le Petite Onken – Jean Hiraga, L’Audiophile 1982 […]