Skip to main content

Full text of "BBC Research and Development Report Number 1966-28"

See other formats


^■<^£: 




RESEARCH DEPARTMENT 



The design of a low-frequency unit 
for monitoring loudspeakers 



TECHNOLOGICAL REPORT No. L-065 

UDC 621.395.623.742 1966|28 



THE BRITISH BROADCASTING CORPORATION 
ENGINEERING DIVISION 



This Re 


3 r t 


i 


s the 


pro 


per 


t y of 


the 


Bri ti sh 


Bro ad c 


as t i ng Co 


rporation 


an d 


may not 


b e 


r e 


P r d u 


c e d 


1 n 


any f 


r m 


wi thou t 


the 


wri 1 1 en 


J e rm 


I s s 


ion of 


the 


Corp or at Ion. 















Technological Report No. L-065 
THE DESIGN OF A LOW- FREQUENCY UNIT FOR MONITORING LOUDSPEAKERS 

Section Title Page 

SUMMARY 1 

1. INTRODUCTION 1 

2. SCOPE OF DESIGN 2 

3. EXPERIMENTAL DETAILS 2 

4. TESTS IN LS5 /I AND LS3/1 CABINETS 6 

5. RESULTS OF LISTENING TESTS 7 

6. COST 7 

7. CONCLUSIONS 8 

8. REFERENCES 8 

APPENDIX 8 



lune 1966 



Technological Report No. L-065 
UDC 621.395.623.742 1966/28 



THE DESIGN OF A LOW- FREQUENCY UNIT FOR MONITORING LOUDSPEAKERS 



SUMMARY 

The present state in the design of low-frequency loudspeaker units is 
reviewed and the areas where improvement is desired are indicated. Experi- 
mental details are given leading to the design of a 12 in. (305 mm) unit incor- 
porating a vacuum-formed cone of toughened polystyrene with a p.v.c. (poly- 
vinyl chloride) surround, and it is shown by objective and by listening tests 
that this design is superior to existing units. An analysis of the price indi- 
cates that the new unit should not cost any more than those at present in use. 



1. INTRODUCTION 



employing smaller diameter units. 



Wide range loudspeakers, such as are employed 
for quality monitoring, generally consist of a low- 
and high-frequency unit mounted in a cabinet to- 
gether with a crossover network. In the past 
colouration* has been so prominent in the repro- 
duction from low-frequency units that the choice 
of unit has been made on the basis of comparative 
freedom from this effect rather than on that of 
power handling capacity. As an example, a 15 in. 
(380 ram) unit is employed in the type LS3/1 loud- 
speaker when a unit of smaller diameter would have 
been chosen if one of the necessary quality could 
have been found. In addition, owing to the re- 
stricted working frequency range of the high-fre- 
quency units available, it has been necessary to 
use low-frequency units beyond the frequency range 
in which the cone and surround behave as a simple 
piston, i.e. up to about 500 c/s, and into the region 
in which the amplitude/frequency response is 
irregular and dependent on the modes of cone 
resonance and their degree of damping. Further- 
more, in existing loudspeaker units the frequency 
range over which the response is smooth appears, 
for reasons not fully understood, to be almost inde- 
pendent of cone diameter and from this aspect 
there is therefore no advantage to be obtained from 



* By colouration is meant a characteristic timbre im- 
parted to the reproduced sound by the loudspeaker; it is 
believed to arise from excitation of mechanical reso- 
nances. 



Cones have generally been made of a paper 
felt material, but in practice the characteristics of 
this material, especially the damping coefficient, 
are not accurately reproducible in large scale 
manufacture, and therefore the frequency charac- 
teristics are variable in the region of resonance 
modes. In an effort to improve matters some manu- 
facturers have turned to materials having a higher 
stiffness to weight ratio than is obtainable with 
felted paper, the idea being to make the cone so 
stiff and light that the inevitable resonances lie 
outside the frequency range of interest. For this 
purpose expanded polystyrene has been employed, 
generally with a reinforcing skin of some other 
material such as aluminium. The results are rather 
disappointing as resonances are found to occur 
within the middle-frequency band and by its very 
construction the cone is of such a high mechanical 
impedance that it is very difficult to secure ade- 
quate damping. 

In the BBC, the loudspeakers types LS5/1A, 
LS5/2A and LS3/1A all use a special Goodmans 
15 in. (380 mm) diameter low-frequency unit and 
have a crossover frequency of about 1600 c/s, and 
some difficulty has been found in obtaining units 
which will meet the BBC test specification in the 
500 to 1600 c/s region where various resonances 
occur; furthermore, the axial frequency characteris- 
tic in this region is not as smooth as could be 
desired. It was therefore decided to see whether 



it would be possible to make, for future designs, 
loudspeaker units which would have more uniform 
and more reproducible characteristics than those 
of the type at present in use. 

One of the difficulties restricting the develop- 
ment of paper cones has been the fact that the cost 
of a new mould has been in the region of £200, 
making experimental procedure very expensive. It 
was therefore decided to investigate the use of 
thermoplastic materials which can easily be made 
into cones by vacuum forming. For this process 
changes in mould shape and even new moulds can 
be made quite cheaply and easily; furthermore, as 
the raw cone material is made in the form of flat 
sheets, it should be very uniform and repeatable. 



2. SCOPE OF DESIGN 

It was explained earlier that the existing low- 
frequency units were chosen on the basis that they 
were relatively free from colouration although in 
fact they were unnecessarily large. It was therefore 
decided that the new units should be of 12 in. (305 
mm) diameter as this size should afford adequate 
power handling capacity to meet all requirements. 
In order to restrict the investigation as much as 
possible, it was decided to use commercially avail- 
able chassis and magnet systems, leaving open the 
choice of voice coil diameter and length, spider 
constants and the design of the cone and surround; 
for the latter two items, the influence of shape, 
thickness and material were to be examined. 



3. EXPERIMENTAL DETAILS 

During the period of roughly forty years in which 
moving-coil loudspeakers have been under develop- 
ment, very little has been published on the various 
factors which influence the frequency characteris- 
tics. One factor which is known, ^ however, is that 
cones with straight sides are much more likely to 
generate subharmonics than those which have 
curved sides and it was therefore decided to start 
with a cone shape having slightly curved sides, as 
shown in Fig. 1; the voice coil diameter was 2 in. 
(51 mm). 

The primary criterion which was applied to the 
choice of material was that it should possess a 
high degree of mechanical damping, for it was 
argued that since resonance modes were almost 
certain to occur in the frequency range of interest 
it was essential that they should be well damped if 
a uniform frequency characteristic was to be ob- 
tained. 




1 1-875 iadid. 
(302 mm) 



Shape of first mould 



The first material to be tried was expanded 
polythene which is available in sheet form in 
various thicknesses from 1/16 in. (1*6 mm) up- 
wards. This material is very light and is characte- 
rised by an extremely high damping coefficient. 
The first experimental models showed axial fre- 
quency characteristics which fell off above 500 c/s 
owing to insufficient stiffness of the material; this 
result was not altogether unexpected and steps 
were taken to stiffen the cone. A coat of polyure- 
thane varnish was applied to each side of the 
material and as a result the frequency characteristic 
was extended to about 1 kc/s. It will be noted from 
Fig. 1 that there is a sharp bend in the cone shape 
near the voice coil, and it was felt that flexure was 
taking place at this point. A further mould was 
therefore made, of the shape shown in Fig. 2, in 
which the sharp bend was replaced by a gradual 



curve, and this resulted in a wider frequency range 
but the frequency characteristic was rather irregu- 
lar. Coating the cone again with polyurethane 
would have improved matters but as more promising 
results had in the meantime been obtained with 
other materials, further experiments with this mate- 
rial were abandoned. 

Concurrently with the experiments described 
above, tests were carried out on cones made of 
0*02 in. (0*6 mm) thick unplasticised polyvinyl- 
chloride (p.v.c), which is a horny type of material 
and also with a polystyrene material of the same 
thickness which had been toughened by the addition 
of a synthetic rubber and possessed a higher degree 
of damping than did the p.v.c. Cones were made 
with the mould shown in Fig. 1, and the frequency 
characteristics were measured with the units moun- 
ted in an enclosed cabinet similar in volume to that 
of the type LS5/1 loudspeaker. These characteris- 
tics are shown in Figs. 3 and 4 respectively. It is 
evident that the high-frequency range covered was 
in both cases adequate for the purpose in hand and 
that the additional damping in the polystyrene was 
advantageous; further experiments were therefore 
confined to this material. 

All the experiments so far described were made 
on cones having a surround made of the same mate- 
rial as that of the cone and the irregularities which 
are seen in Fig. 4 above 500 c/s are due to the 
presence of resonance modes. The cone can be 
regarded as a transmission line and resonance 
modes can occur with the wave motion either in a 
radial or circumferential direction if it is not 
properly terminated in a resistive surround; as the 
required impedance for these two directions is 
different and the termination must occupy a distance 
small compared with a wavelength, it will be seen 
that the problem of designing a good termination is 
difficult. 

The first surround tried was of plasticised 
p.v.c. 0*02 in. (0'6 mm) thick of the shape shown 
in Fig. 5, this profile being chosen to allow for 
fairly large excursions of the cone at low frequen- 
cies. The surround was substituted for the integral 
surround on the polystyrene cone previously used 
to obtain the curve in Fig. 4 and the resulting axial 
frequency characteristic is shown in Fig. 6. It will 
been seen that the curve is considerably smoother 
than that of Fig. 4 but that the high-frequency 
response is reduced, probably due to the surround 
damping out resonance modes; on the other hand, 
as would be expected, the bass range is extended 
to lower frequencies. The fact that the axial 
characteristic rises with frequency is largely due 
to the directivity increasing with frequency and the 
concentration of more of the sound energy on the 



325 in. 
(82Smm) 




10-25in.diQ 
(260mm) 



Fig. 2 - Shape of second mould 

axis. Experiments with a cone material of twice 
the thickness, i.e. 0*04 in. (1*2 mm), showed that it 
was possible to recover the high frequency res- 
ponse, but the response was more irregular and the 
sensitivity lower owing to the greater mass. Cones 
were then made with 0*02 in. (0*5 mm) material to 
the second shape mould, shown in Fig. 2; as with 
the polythene material, the change in shape resulted 
in an increase in the high frequency response, as 
shown in Fig. 7. The dip in the curve at 250 c/s 
was thought to be partly due to a circumferential 
mode and this was checked by stroboscopic exami- 
nation. Further evidence was obtained by making 
a cone with a small turnover at the edge; this' had 
the effect of stiffening the cone edge, thereby 
increasing the Q and producing an increase in the 
depth of the dip. 













f\ 












y^ 


1. 


1 


\\ 












P-li 


k 




\ 








1 


^ 


'In 




5dB 


/ 










\ 




t 


J 
















1 












A 


/ 














IVi//l^ 



50 100 200 



500 1000 2000 

frequency, c/s. 



5000 



Fig. .? - Axial frequency characteristic of unplasticised p.v.c. cone from mould No. 1 















f\ 












^ 




/-' 


V 






\ 




^ 


^^ 


y 


i/\ 


\r\|\ 




5dB 


y 


/ 








I 




t 


/ 










\ 






/ 










1 






/ 










\ 




> 

















50 100 200 500 1000 2000 5000 

frequency, c/s. 

f'lg. 4 - Axial frequency characteristic of Bextrene cone from mould No. 1 



0-375 in.rad, 
(9-52 mm) 




'p.vc surround. 



0-0626 in. rad. 
(1-59 mm) 



Fig. 5 - Shape of first p.v.c. surround 





























A 






i 






^ 


f """ 


"^ \ 






5dB 




^ 


--^ 






K 




t 


^ 










A 
















\ 
















\/ 


A 














V 


\ 



50 



100 



200 



500 1000 

frequency, c/s 



2000 



5000 



big. 6 - Axial frequency characteristic uf Bextrene cone jrom mould No. 1 
fitted with p.v.c. surround of shape shown in Fig. 5 



















































i 








y-^ 


A 


f\ 




5dB 






vJ'^ 




u 


A 




t 


^ 




•"-^ 










^ 


y 


























1 


KA^\ 



50 



100 



200 



500 1000 

frequency, c/s 



2000 



5000 



Fig. 7 - Axial frequency characteristic of Bextrene cone from mould Mo. 2 
fitted with p.v.c. surround of shape shown in Fig. 5 



01875 in, 
\. (4-76mnn) 




flat region 
-polystyrene cone 



- p.v.c surround 



Fig. 8 - Shape of second p.v.c. 
surround showing flat region 



The effects of small changes in the shape of 
the cone and in the diameter of the voice coil were 
investigated and it was found that neither of these 
two factors was critical. 

A large number of experiments were then 
carried out, using surrounds of differing materials, 
thickness and profile in an attempt to damp out the 
mode at 250 c/s. It was finally discovered that 
with a suitable surround material better damping 
could be obtained if, as shown in Fig. 8, a small 
flat region was left before the turnover of the sur- 
round commenced. This flat region has the effect 
of introducing a shunt arm, as indicated in Fig. 9, 
consisting of a resistance and compliance, in 
parallel with the mass, compliance and resistance 
of the surround proper. The axial characteristic 
with this surround, shown in Fig. 10, is appreciably 
smoother than that obtained from commercial 12 in. 
(305 mm) units, especially in the region above 500 
c/s; the sensitivity is about the same as that of 
the Goodmans 15 in. (380 ram) unit referred to 
earlier. The power handling capacity and transient 
response were then tested. Mounted in a closed 
cabinet, the unit was able to take the full output of 
a 25 watt amplifier down to 70 c/s without obvious 
amplitude distortion when the waveform was obser- 
ved on an oscilloscope. Chopped-tone transient 
response tests showed the unit to be free from 
serious resonances below 3 kc/s. 

Four units were then made to check the repro- 
ducibility of this form of construction; the axial 



to cone edge 

— L 



rasistanca . 

flat region 

compliance ; 



curved region 
J compliance 



I 



Fig. 9 - Mechanical circuit diagram of surround 



frequency characteristics did not differ from one 
another by more than tVi dB from 75 c/s to 1250 c/s 
and ±1 dB from 30 c/s to 2 kc/s. It was therefore 
decided to design a complete loudspeaker employ- 
ing a unit of this type for the low frequencies and 
to carry out listening tests. Manufacturing details 
are given in the Appendix. 



4. TESTS IN LS5/1 AND LS3/1 CABINETS 
(a) LS5/1 

The Goodmans 15 in. (380 mm) unit in an LS5/1A 
loudspeaker was replaced directly by the new 12 in. 
(305 mm) unit. A slight excess of output in the 
middle frequencies was corrected by means of a 
resistor which was originally designed to be adjust- 



















































+ 






> 


^ — ~-^ 


r\ 


^^ •v 




5dB 






y^ 






- v'X 




t 




^ 


/ 






\ 




^■^ 




/ 


/ 










\ 




/ 

















50 



100 



200 



500 1000 

frequency, c/s 



2000 



5000 



Pig. 10 - Axial frequency characteristic of Bextrene cone fitted with 
p.v.c. surround of the type shown in Fig. S 




10-375 in alQ 
(263 mm) 



Fig. 11 - Shape of final mould for cone and surround 



5. RESULTS OF LISTENING TESTS 

The two loudspeakers described were given 
listening tests in Kingswood Warren A.F. Section 
Listening Room using recordings of speech from 
dead surroundings and recorded orchestral items; 
they were judged to be significantly superior to 
their LS5/1A and LS3/1A counterparts and were 
therefore offered to 0. and M. Department for an 
extended field trial. Reports have been very favour- 
able and in particular comments have been made 
regarding the freedom from colouration of the bass 
response compared with the corresponding loud- 
speakers employing the 15 in. (380 mm) Goodmans 
unit. 



6. COST 

The cost of the materials for the cone and sur- 
round is only a few shillings, which is a small 
fraction of that of the complete unit. 



9 in, 

diQ. 
C229mm) 



T 



L- -A /,r. 



0-0625in; 
(1-59mm) 



(a) 




<b) 



able for this purpose. A small dip in the axial 
response at 1750 c/s was traced to the effect of 
the 7 in. (178 mm) wide slot in front of the unit. 

(b) LS3/1 

When the Goodmans 15 in. (380 mm) unit in an 
LS3/1 loudspeaker was replaced by the new 12 in. 
(305 mm) unit, the response in the region 400 c/s 
to 800 c/s was found to be somewhat excessive as 
with the LS5/1 cabinet. To overcome this, it was 
found necessary to change the values of several 
components in the crossover network. 



pvc. surround 




00925in 
(238mm) 



Evo-si.ck ^X'^^e 



C^^^ r^ 



<c) 



01876 in 
m) 




Araldita rasin AV100 
- hordenar HVtOO 



voica coil /spidar sub-assy 



tig. 12 - Assembly of cone, surround and voice coil 



7. CONCLUSIONS 



8. REFERENCES 



Experiments have been described which have 
led to the production of a 12 in. (305 mm) low- 
frequency unit of performance believed to be 
superior to that of any known commercial product. 
The cost of production of the cone and surround is 
only a small fraction of that of the magnet system 
and the price of the complete unit should be no 
greater than that of corresponding commercial 
products. 



1. TIEDJE, J.Q. 1936. Speaker Design, 
Engng, N.Y., 1936, 16, 1, p. 11. 



Radio 



2. The development of high-quality monitoring 
loudspeakers ; a review of progress, Research 
Department Report No. L-041, Serial No. 1958/31. 



APPENDIX 



MATERIALS 

The cone is made from Bextrene sheet type 
234/2437, 0-02 in. (0-5 mm) thick, obtainable from 
Messrs. BX Plastics Ltd., Higham Station Avenue, 
Chingford, London, E.4. The surround is made from 
Nappatex 0-02 in. (0*5 mm) thick, obtainable from 
Commercial Plastics Ltd., Berkeley Square House, 
Berkeley Square, London, W.l. 



COMPONENTS 

Chassis and Magnet System 

The chassis and magnet system is made by 
Goodmans Industries Ltd., Lancelot Road, Wembley, 
Middlesex. 

Voice Coil/Spider Sub-Assembly 

This also is made by Goodmans Industries 
Ltd. 

Cone 

The cone is shaped by vacuum forming, employ- 
ing a drape process to the mould whose shape is 
given in Fig. 11. Prior to forming, the material is 
heated for 20 seconds by a radiant heater which, 
at the working level, gives a temperature of 180°C 
at the front and 160°C at the rear of the sample. 
After cooling, the cone is removed and trimmed as 
shown in Fig. 12(a). 



Surround 

The surround also is shaped by vacuum forming 
using a mould of the profile given in Fig. 11. The 
heating time, for the same radiant heater as des- 
cribed above, is 18 seconds. After cooling, the 
surround is trimmed as shown in Fig. 12(b). 



ASSEMBLY 

Voice Coil/Spider Sub-Assembly to Cone: 

These components are mounted on a jig to 
ensure concentricity and are fixed together with 
Araldite resin type AV 100 and hardener type HV 
100. The position of the voice coil former on the 
cone is indicated in Fig. 12(c). 

Cone to Surround: 

The position of the cone relative to the sur- 
round is also indicated in Fig. 12(c); a thin layer 
of Evostick type 528 is used as the adhesive. 

Cone Assembly to Chassis: 

The cone sub-assembly is mounted in the 
chassis with the voice coil concentric in the magnet 
air gap. The surround and spider are fixed to the 
chassis by adhesive or clamps. If all previous 
assemblies have been correctly carried out, the 
spider should be undeflected. 



CHD 



Printed by BBC Research Department, Klngswood Warren, Tadworth, Surrey