Talk:KEF

This article seems to have been almost entirely lifted from here - surely a copyright violation? Since the person who added all this content is an anon, it will be difficult to contact them to ask. Unless anyone knows otherwise, I'll flag this page up as needing a rewrite in a few days time. WikianJim 15:06, 2 August 2006 (UTC)

Agreed. The content seems more like a promo for the company's products than a factual article. There are some very dodgy statements in this piece, particularly with reference to the original 104, coupled cavity and conjugate loading. No mention of Laurie Fincham. And the company's bankruptcy is glossed over. (Why exactly did they go bust?)


 * I see that as of the 14th March 09 the whole page is now replaced with a copyright warning. I work for KEF and so can most likely get a version of the text in question released under a GFDL if required. I completely appreciate that the text in question is from our web site and is not be entirely appropriate for wikipedia, I have also refrained from starting a new version myself because of NPOV problems. The history that we publish on the web site is factually accurate although it does go into more detail on some points and skims over some others (such as the change of ownership). If there is a neutral party willing to have a shot at rewriting this article I would be more than happy to help them out with facts and figures from KEF wherever possible. Jackocleebrown (talk) 17:24, 15 March 2009 (UTC)


 * Hi. I don't see any sign that a GFDL release has been acquired, so I have replaced the text with a clean stub that can be expanded on by interested editors. Meanwhile, if you do choose to seek release of the website's text, you can read the procedure at Donating copyrighted materials. Thank you for being mindful of COI concerns. :) --Moonriddengirl (talk) 11:59, 23 March 2009 (UTC)

KEF or KEF Audio
I think KEF Audio --87.65.138.252 23:19, 11 December 2006 (UTC) Comments to eegrek39@gmail.com

I was employed by KEF electronics as the first dedicated production engineer in 1972. During the following years I started developing the Production Engineering Department. Eventually becoming Production Controller responsible for all engineering services including Purchasing and finally ending my career with KEF as Production Manager. I left the organisation in 1990 after 18 years of rewarding and interesting employment.

The following description of many aspects of KEF at this time is an engineer’s viewpoint of the culture and effects of the drive to improve the product and give our customers a better experience. This narrative quite possibly belongs to the document by Dr Watson outlining the various drive units that KEF produced during this period since he was only able to show the visible differences of developments whereas the main developments occurred in the hidden components such as the voice coils and the magnet assemblies. As well as the changes in culture brought about by the introduction of the Fourier analysers and computer matching techniques.

Much of what I have read in the history books written about KEF at this time highlights the many advances made by the organisation and quite rightly concentrates on the work done by the design department under Laurie Fincham. However, there seems to be little, if any mention of the work done to meet the challenges faced by the production department to ensure the customer received the benefit of these advances. Once the design was completed or at a stage when many of the parts were designed it was necessary for the production facility to either start the purchasing of the parts or designing the equipment and tools to assemble the new units and systems. We had to come to grips perhaps with new methods of assembly and then train the operators to produce the assemblies accurately and efficiently. KEF was also a commercial organisation required to provide a product at a price that allowed the customer good value for money, and also to ensure we made a profit to carry on in business. The real challenge was to bring Laurie Fincham’s brilliant designs to the market place.

Some of this narrative may be of use to those interested in the development of speaker systems at this time and the problems encountered along the way. I am not an acoustic engineer nor competent in electronics but my skills were employed to troubleshoot both production problems as well as design shortcomings that some customers experienced while using the product, many of which were a reflection of technology in many source industries such as Copper wire or Adhesive suppliers. My skills also included many years of special purpose machine design which I found to be important when improving existing production techniques as well as developing new methods and products.

My first project was to develop methods to manufacture the BD139 (bass drone). It would seem to be a simple technique requiring only mechanical skills without the problems of either electrical or fine tolerance assembly. However the Diaphragm, made from polystyrene was always going to be a difficult material to bond to the surround, because it is easily attacked by solvent adhesives which would have provided a good strong bond while water based acrylic adhesives did not provide the required strength. The designed alternative was an epoxy which proved to be very irritating to sensitive skins and also requiring very long drying and curing times. We built many wooden forming jigs to hold the components together while they were curing but the technique was not very efficient and applying the adhesive required special jigs an adhesive dispensing equipment. After a year or two of persevering with this technique we finally reverted to the original method of assembly of the B139 bass unit where a groove was made in the side of the polystyrene diaphragm and the surround was jammed into this groove with string and plenty of acrylic adhesive making the joint a shear stress bond rather than a peel stress joint (When bonded to the front of the diaphragm). This technique was reverted to when we introduced the newer painted chassis die casting rather than the original unpainted design first used on the B139 bass unit.

Adhesive technology was important since most components were bonded to each other during the assembly process. It became necessary to be aware, if not an expert on adhesives and their characteristics. The production of assemblies, at this time, was very vulnerable to the technology development by suppliers. It was not unusual at this time to find a complete previous day’s work subject to quarantine owing to adhesives not performing correctly. For example magnet assemblies would fail after the assembly process, during the magnetisation of the ceramic magnet. The pole assembly would move to one side of the magnet gap showing a shear failure. This was bad enough in production but it showed the vulnerability of the assembly to stresses possibly occurring during transportation of the System to our customers. The adhesive system used was an anaerobic adhesive where once the joint was made and air was excluded from the surfaces, then the adhesive would cure over a period of a few hours at least to the handle able stage. Shock testing proved that not only was the strength of the adhesive important but the ability of the adhesive joint to flex very slightly to absorb the kinetic energy of the unit being dropped from say waist height. Many adhesive systems were tried and after some considerable time we managed to find a flexible epoxy system to meet the requirements. The main drawback was the time required for the adhesive to cure. During this period the magnet assembly required to be held accurately in place to maintain the gap between the pole and the top plate in which the voice coil was to operate with a 0.2mm clearance. This gap was maintained during assembly by an expensive jig. So we were restricted economically to about 50 jigs. This restricted our output while this problem was solved. Eventually a machine was designed to move the assembly into an induction heating coil which raised the temperature of the assembly to 90 degrees centigrade within 30 seconds. This vastly improved the curing time and solved the problem for a number of years. The last improvement was the development of the tough acrylic adhesive with an accelerator which allowed assemblies to come off the production line in about 20 seconds. A special purpose machine was designed by an outside company in 1986 and was a successful technique for the rest of the time I was at KEF. This machine used collets to maintain a perfectly concentric gap, independent of any tolerances of the components.

Centre termination is a method of connecting the coil to the terminal panel on the drive unit via a flexible copper braid. Early examples of nearly all speaker units took the braid from the coil and anchored the braid though the front of the diaphragm and back again underneath the diaphragm through holes made in the diaphragm. This method is unsightly and also not acoustically good for reproduction as it makes the diaphragm vibrate inconsistently and interrupting the flow of vibrations outwards to the surround. Laurie Fincham mentioned this to me and suggested we change the method to terminating the coil just above the rear suspension and anchoring the braid against the coil former with adhesive. This method was adopted about 1975 where appearance changes can be noticed both in the diaphragm and the improved gloss of the diaphragm coating.

Diaphragms for bass units were largely made from Bextrene which is a sheet of plastic vacuum formed in the shape of a cone. Acoustically this is as fairly lively material and required to be damped down by coating the back of the diaphragm with two layers of plastiflex (an acrylic adhesive). Improvements in this area included the measuring if the basic bextrene material before vacuum forming and predicting the weight of the finished component. The coats of palstiflex applied during the preparation of this component were carefully controlled so that the final weights of the moving parts could be maintained within certain tolerances. This improved consistency especially important for the reference series of speakers (More on this later). The voice coils made in 1972 were constructed by producing a flat coil former made from a single sheet of Nomex. The former was then coated with a layer of very thin epoxy resin. This former was then wrapped around a PTFE mandrel and held in place by two end cheeks on the coil winding machine. The coil wire was at that time, made by BICC and coated at their factory with a heat reactivated adhesive. The coil was then wound on the former and before removing from the machine was subject to an electric current for a short time which heated the coil and reactivated the adhesive. This also caused the epoxy resin on the former to cure making a bond between the coil and the former. There were a number of weaknesses in this method which resulted in the coil not consistently reactivating the epoxy resin and the coil falling apart when it was subject to high levels of electrical power by the customer. The coil adhesive was literally reactivated and fell apart by the power of the amplifiers.

The period from 1960 to about 1972 amplifiers were typically only up to 20 Watts per channel. Then the electronics revolution started to make 50 to 100 watts per channel very affordable. Clearly our voice coils were not going to meet these higher specifications. The first improvement in 1972 was to wet wind the coil onto metal mandrels with the existing nomex coil former. A bath of epoxy was attached to the coil winding machine and the wire ran through this bath and onto the coil former. The mandrel was removed from the machine with the coil still not bonded in any way either wire to wire or wire to coil former. The tension during the winding process was increased to ensure the assembly was held together and as long as the coil wire was not touched. The mandrels were then loaded into an oven and raised to 200 degrees centigrade to cure the epoxy resin. The main improvements were that the wire insulation specification could be raised to a class H and the epoxy resin would also meet the same high temperature specification without degrading.

Up to 1972 the bass unit surrounds which attached the cone to the front of the speaker chassis were all made of neoprene. This material was tough and flexible but not so acoustically satisfactory as plasticized PVC. With the introduction of the SP1039 bass unit designed for the first Reference Series system it was decided to try forming a PVC surround in house and attaching this to the Diaphragm during the vacuum forming process. Plasticized PVC has some difficult characteristics when being used in this manner. The material has a poor ‘memory’ and tries to revert back to the sheet it was formed from so tending to flatten back out. The plasticizer also attacks any solvent adhesives used to bond the material to other substrates. A special purpose machine was designed to form the surround and attach itself to the diaphragm at the same time. This technique was developed ensuring the correct temperatures were reached prior to forming and by using an acrylic adhesive the characteristic problems were overcome.

The SP1039 coil was wound on a new nomex former which was supplied as a tube and made with two spiral wound layers of the material. The first drive units were produced using this tube and all seemed well until the first systems were introduced to the market. Unfortunately, it was soon discovered that the nomex tube, under certain conditions could not withstand the high temperatures the coil could be subject to. The former tended to de-laminate and blister in such a way as to rub on the magnet pole or the magnet top plate. Experiments were carried out and it seemed that if the coil was forced out of the magnet system by perhaps an excessively warped vinyl record or high output levels, then the coil would reach very high temperatures. The alternative was to use a hybrid former by winding the coil onto a short aluminium former and extending this by use of a short length of nomex tube to act as a heat insulator between the coil and the bextrene diaphragm. This was adopted in the summer of 1973 and overcome these problems.

The Reference Series Speakers of which the first system was the model 104 was a highly successful concept. The method of design and manufacture previous to this was probably endemic to many other manufacturers. The development of a new product would be initiated by the Design engineering department where the cabinet of the speaker system was designed and the drive units to be used were decided upon. Typical units were diverted from the production line and checked for acoustic response using B & K equipment to ensure that the units used for the development system conformed to reasonable requirements. At this time in 1972 we did not have a development anechoic chamber. The crossover board was designed to match the units and the cabinet characteristics. Once the system was approved for production then purchase of bought out items were commenced and production began. For the ‘C’ series of product there were no regular checks made to ensure that production units and systems conformed to the original design concept. Acoustic measurement was extremely laborious and could not be undertaken regularly. However, all drive units and systems were checked for acoustic problems caused by mechanical interference using B & K equipment to produce a frequency sweep from 20 to 20 kHz. With the introduction of the Fourier analysers it was possible to economically measure many drive units in a short time. The method also saved the response data on a hard disc which enabled the data to be stored and manipulated. The main aim was to be able to choose very quickly the units with the responses within certain parameters and further to match units together which had similar response patterns. It turned out that the high frequency units were the most difficult to match and at first only between 30% and 40% of production met the requirements. The T27 chosen for the model 104 system was the most popular unit for both our own use and that of our OEM market so we were making the unit in relatively high numbers. Although the yield was low for the reference series, the remainder of the units still had a market use. The situation changed when the Model 105 was introduced. This system used the T52 high frequency unit which was once used on the Celeste speaker system and had reached the end of its product life cycle. Consequently production of this unit was geared only to a few Spares requirements. There was no large pool of production to choose from. Consequently the low yield had an effect on our ability to meet our customer’s requirements. We were reduced to taking apart what at one time was a perfectly good unit and rebuilding it with a new diaphragm assembly in the hope that this would pass the analysis test.

Ironically, the very equipment that was rejecting the drive units was particularly the very tool that was going to help solve the problem. We were able to design experiments and get the results of those changes very quickly. We checked every small discrepancy from magnet strength, coil heights in the gap, flexibility changes in the diaphragm, and any other likely culprits. We could not find any correlation to these small differences and the drive unit responses. In desperation we changed many of the adhesive systems used to assemble the unit. It was not until we changed the adhesive joint between the coil former to the diaphragm that we suddenly found all units in this experiment displayed the consistency we required. The cause was obviously a variable decoupling of the coil to diaphragm due to the heat reactivation of the adhesive during assembly. Two main changes were the adoption of a cyanoacrylate adhesive together with the design of a special purpose machine to automatically make the joint and apply pressure to the joint while the adhesive cured. This technique was then applied to the T27 HF unit with the same improvement for the Model 104 system. Certain changes had to be made to the dividing network to compensate for the different response of the T52. We all learned that consistency was going to be the key to a successful product range in the future. Many of our production systems were scrutinised for inconsistencies and as control of both the components and production methods were adopted.

Capacitors were a purchased component where we had little control of the quality. Commercial pressures meant that the most economical specification was to buy capacitors with a plus or minus 20% tolerance. This was not acceptable for the reference series where the dividing networks were paired with the drive units. The solution was to check all capacitors before use and separate into 1% bands. Then the inductors were wound to suit the values of the capacitors available. This provided the means to present a more stable load for amplifiers to drive throughout the frequency range.

This is my interpretation of what happened to KEF Electronics over the years I was employed by them and with an outsiders point of view after I left the organisation.

The Speaker industry is slightly seasonal aligned to the seasons of the northern hemisphere despite the value of product being exported to the southern hemisphere. In other words when autumn and winter were upon us the sale activity seemed to increase and the production facility began to work to maximum capacity. However, during the summer months there was a noticeable drop in activity. This change was to some extent automatically compensated by the taking of summer holidays by the workforce. Another mechanism to help this situation was to recruit a number of outworkers whose activity could be regulated without causing too much disruption to our employees when times became tough. Most of the outworkers were wives of existing employees and were paid strictly by the numbers of assemblies made by that outworker. This proved to be popular with our employees and helped to provide a welcome addition to the family’s budget. KEF also seemed to be vulnerable to the market conditions in both British and World economic cycles - the boom and bust culture of the 70’s to 90’s. When this happened in ’79- ‘82 we found that after a struggle, we had to make some redundancies. This left the organisation with low moral and insecurity. Eventually the existing Board of directors changed by the addition of Peter Gaskarth as Managing Director and Bill Angel as Accountant (who previously were employed by Celestion of Ipswich). This coincided with an injection of capital to tide us over and a change in the commercial culture to make sure the company could meet its financial commitments. Changes in the way our components were sourced and the way our stock control worked were instrumental in improving our commercial competitiveness

As the early ‘80s progressed the country began to experience a boom which allowed the company to once again expand and increase its turnover.

Some of the expansion began by exploiting the American market and eventually opening a small production facility in Richmond, Virginia. There was also some acquisitions in the form of Meridian Audio, a CD Player/Amplifier manufacturer near to Cambridge, England. While the organisation was trying to consolidate these changes another recession hit both Britain and other countries. Once again KEF found itself overstretched and this time could not recover without external help. I had left when this resulted in the sale of the assets to Gold Peak (after KEF Electronics went into administration in early 1992). The Board of directors in the main did not survive this change in the management structure but the backing of this organisation seems to have stabilised KEF to enable it to look forward to a bright future.86.14.70.156 (talk) 14:16, 3 October 2014 (UTC)