The Little Legend

The Little Legend written by Trevor Butler was published in Hi Fi News & Record Review in January 1989. It is reproduced here with kind permission of Steve Harris of Hi Fi News & Record Review. Please note that for copyright reasons the original photographs cannot be included here.

Conceived originally as a ‘Grade II monitor’ for the BBC’s own use, the LS3/5A has been in production for a dozen years and remains as popular as ever. Trevor Butler unravels the true story. There can hardly be another single box which as provoked as much emotive comment, or given rise to so many myths and misunderstandings, as the BBC LS3/5A. The very name has caused confusion, but in fact it simply follows the BBC’s coded equipment format. In this, cabinets have the prefix ‘CT’, filters ‘FL’, and loudspeakers ‘LS’, hence the first two letters. The figure ‘3’ in the code indicates that the design is primarily for outside broadcast (OB) use. A figure ‘5’ would mean studio broadcasting, as LS5/9. The number ‘5’ after the stroke is the model number, the LS3/5 supplanting previous OB speakers like the LS3/1. So we arrive at LS3/5 which was the title of the initial model. Later the ‘A’ was added to indicate the first and only design alteration to the original specification. Any further specification change would result in a ‘B’; but that hasn’t happened, and isn’t likely to, for reasons which will become apparent.

The concept of this speaker was to suit those BBC environments where monitoring on headphones was not satisfactory and yet there wasn’t sufficient room for a ‘Grade I’ monitor. A Grade I monitor can be used for critical tonal balancing of programme material, setting of microphone positioning, etc. Current Grade I monitors are the LS5/8 and LS5/9. A grade II monitor may be used for checking the quality of programme, but balance and mic positioning are normally Grade I-checked unless there is no alternative. It was recognized that what would be required was a small Grade II unit with some sacrifice of bass response and loudness reproduction, this being justified for the sake of achieving compactness. The likely users were, for instance, the production control areas of television OB control vans, where there producer needs to listen at a lower level than that used for the actual mixing.

There was no suitably available commercial unit, and so the Research Department of the BBC was asked to design one, at its headquarters in Kingswood Warren. Less than a week elapsed between this request and the first prototype being offered for field trials and evaluation. this came about because the LS3/5 was based around an experimental loudspeaker which had been developed for some preliminary work on acoustic scaling tests at Kingswood.

By using one-eighth scale models, and one-eighth wavelengths (IE, 8-times frequencies), recordings can be made by which the merits of particular acoustic techniques can be assessed without the expense of a life-size environment. Naturally, though, this implies that the entire model reproduction chain, of tape player, loudspeaker, microphone and recorder, must be capable of operating at very high quality with a frequency range of 400Hz to 100kHz, in order to be able to model accurately a typical bandwidth of, say, 40Hz to 15kHz.
Pictured is the model speaker the BBC used from 1972 until 1980, which represented a major step in the modeling process. Although it had its limitations, primarily in terms of its tonal quality and maximum power handling, components of this small loudspeaker were found to be of a sufficiently high quality to help fulfil the demand for the unusually compact monitor needed in outside broadcast use. This was therefore packaged and called the LS3/5 and showed the model experimenters to have achieved good results.

A small number of these units where made in house by the BBC and used in television mobile-control- rooms where they gave satisfactory service. The BBC got to the stage of inviting applications to make the LS3/5 under license by outside manufacturers. Indeed, Rogers issued a press release on 19th February 1974 which proudly announced that they would be exhibiting the new design at SONEX ’74, an exhibition to be held that April. They offered a photograph and a provisional specification: 25W power handling with +/-3dB 80-20,000Hz and +/-4dB 60-20,000kHz. Crossover frequency was quoted at 3kHz and the units were a 110mm bass driver with a Plastiflex doped Bextrene cone and a 27mm Mylar dome tweeter. The price, £52 each plus VAT!

Alas, though, there were to be early problems. Although the BBC had confidence in the KEF B110 bass unit at the time, when a subsequent batch of in-house LS3/5s was needed it was found that the low and high frequency units had undergone significant changes and a re-design of the speaker would be required before production could resume. Accordingly, the speaker was passed on to the BBC’s Designs Department, then located in Great Portland Street, with a request to modify it so that it would once again be suitable.

The trouble was in three areas; the B110 had changed, exciting the cabinet in a different way, producing a coloration both from the LF unit and the cabinet; and the HF unit had developed a pronounced ‘lipsy’ quality.

In an attempt to ensure that cabinet resonance would not cause problems, the side panels had each been damped with a bituminized pad, and the top and bottom panels likewise, except that two layers were required here. In addition, a PVC edging was applied to the chassis of the LF unit and to de-couple it from the front panel and seal the join. In order to damp the air modes of resonance inside, all internal surfaces were lined with polyurethane foam. The cabinet was sealed to prevent air leaks, which might produce extraneous sounds from the high pressures produced – indeed even the screw holes were made airtight. The cabinet problems were associated with the softwood parana pine fillets which connected the back panel to the front; these had an insufficient impedance and were therefore replaced with beech. The back panel was re-specified from the precious sandwich of spruce to multi-ply birch, thus obviating the voids often associated with spruce. This then became the CT4/11A.

The opportunity was also taken to sort out the treble lisp of the tweeter. It was felt that this unit, the KEF T27 SP1032, was vulnerable because it was exposed and could easily be damaged during rigging at a venue. It was decided to incorporate a protection grille and a suitable one was found, with some modification, from a Celestion HF2000. The tweeter is surrounded by a thick felt-strip mounted on the baffle in order to prevent acoustic discontinuity presented by the edge of the cabinet setting up an interference pattern, since the T27 radiating surface is small and the radiator nearly Omni-directional. The addition of the protection grille was wholly beneficial as it raised the output at higher frequencies and help cure the lisp. The effect of the Tygan cover on the front was taken into account in the design, and the loudspeaker should always be used with the grille on, to avoid discontinuities in the upper presence region.

Having carried out these changes, the crossover had to be adjusted to compensate and the new type, FL6/23 replaced the original FL6/16. Sporting these alterations the first variant of the LS3/5 was born – the LS3/5A. The ‘A; suffix was necessitated because although the LS3/5A sounded similar to the LS3/5, the differences were significant enough for it not to be possible to use a mixed pair for stereo listening. Since only small quantities of the LS3/5 had been built, this did not present much of a problem and only 20 units had to be considered obsolete.

For those not fully acquainted with it, the design is still now very much as it was then, with just a few slight enhancements which came about over the years, as will be explained.

There were a few noted difficulties in 1977, and almost every Summer saw the reject rate at the manufacturers rise slightly. The early 1980s saw a change when the surround-dip was seen to have moved slightly. This was resolved when the crossover resonator controlling it was tuned to a lower frequency and the damping resistor changed. Any alteration like this is issued by the BBC to the current manufacturers and is incorporated into their license. There have been also some adjustments to the value of the tweeter coupling capacitor; these are part of the relative balance adjustment. As the material of the tweeter alters, the ‘Q’ changes and there is a mild ‘tippling’ of the response. This coupling capacitor affect the shape of the network and decides whether it is under or over damped. To adjust the treble balance, an output from a tapped transformer on the FL6/23 is adjusted, and the coupling capacitor changed to maintain the crossover frequency.

Re-assessment

The major re-assessment of 1987 was widely reported in the specialist press; but this time the alterations had been requested primarily by the manufacturers. As one of the licensees, Richard Ross of Rogers, explained, over the years, whilst it had been possible to make an LS3/5A which sounded and measured within spec, and which was acceptable, the number of units, in particular the woofers which were within tolerance varied. Because of the particular design of the LS3/5A, the B100s, although being to KEF’s standards, happened to have the point where their tolerance was most variable occurring at the most critical point in the speakers design. In hot summers the reject rate of the bass units could be anything from 10 to 85 per cent, and, for the licensees, this was not satisfactory from an economic standpoint.

At the same time as the licensees were finding it difficult to make consistent units, the BBC was encountering their own problems. At the end of 1987 it was noticed that the units were tending to drift towards the limit of acceptability. Specifically, they were found to be about 2dB up at a octave around 1kHz, a particularly critical area for balance. There had always been a slight lift on the LS3/5A here, but now it started to become objectionable. It should be stressed that the units were still operating within limit for a Grade II monitor, but would have been unacceptable in a Grade I.

The crux of the matter was again the ‘surround-dip’, the precise frequency of which varies by a few hundred Hertz over time. This had become interactive with the box resonance to the extent that action was needed. A change had to be brought about so that speakers could be manufactured which the BBC would find acceptable, to provide them with a consistent loudspeaker they designed, and wanted.

The BBC referred the matter to KEF, who set their Special Products Division to work. The problem was traced to variations in the consistency of the Neoprene used in the making of the SP1003 B110 and so a new B110 was specifically designed for the LS3/5A. This uses a PVC surround, more consistent and temperature stable than the Neoprene but less compliant. A new spider voice-coil assembly was therefore required to give the new drive the same sensitivity as the older one, in order to achieve the bass performance. This is the B110 SP1228. Crossover design is always a source of great debate and this has certainly been the case with the LS3/5A. Changes have been made, as reported, but the main specification and overall sound balance of the speaker have remained the same.

Crossover

A book could be written on this subject as far as the LS3/5A is concerned. The complexity is indicated in the circuit diagram. L1 and R1 are employed to equalize the rising axial response and frequency characteristic of the bass unit, while the group C5, L2, and R2 compensate for a hump in this characteristic. For the tweeter, L3 acts as a shunt inductor and as an autotransformer to allow for differing levels of sensitivity. C2 is adjusted to keep the crossover frequency constant and R3 is there to prevent ringing, with R4 and C6 to adjust the frequency response at the upper end.

The SP1228 has a considerably smoother response than the older-type B110, and hence needs gentler crossover equalization. It was decided therefore that the crossover could not only be re-designed to compensate for the change to the drive unit, but could also incorporate new technology. In essence the equalization is very much the same with a similar looking circuit. The earlier design used the already mentioned taped auto-transformer as part of the high-pass filter and to adjust the level of the tweeter in 1dB steps, but it was noticed that as the tweeter sensitivity was becoming increasingly consistent (to within 0.5dB), a simpler constant impedance resistive attenuator could be used to give the +/-1dB of adjustment in 0.5dB steps. One consequence of the simpler equalization is that the overall impedance has been reduced from the previous 15ohm nominal to now 11ohm nominal.

KEF analyzed the effect of component tolerances on the overall response and found that by tightening these in the most critical areas an improvement in consistency could be achieved. In previous units it had been possible to obtain a 2dB variation. For example, if the network was +1dB in the critical area and the accompanying drive unit was also +1dB the two compounded. KEF decided that by opting for computer-aided matching, it was possible to nominally select components so that if the network was +1dB a drive unit with a -1DB figure would be chosen to give 0dB overall change. the new crossover is the FL6/38, SP2128/

Therefore KEF was able to offer matched components to the manufacturing licensees even though the company does not manufacture the speaker in its entirety themselves. It so happens that all three current licensees use the KEF matched parts. At the end of a day it has to be said that KEF did a superb job. They were issued with a BBC LS3/%a reference unit (no. 6) of 1975 vintage and produced a set of 1988 parts which sounded so similar that they were within Grade I limits, and could be made much more consistently. Hence LS3/5A remains.

Spendor used to make their own complete crossover units and say that they could start again if they situation called for it. They have now adopted the iron-dust inductors, not to be confused with ferrite, and have dispensed with their previous silicon-iron E & I transformer winding. The E & I laminates tend to be more costly, and the opportunity to buy the matched KEF sets was taken.

Rogers had for years used various ferrite and iron-dust cores, because their research showed that they were more economical and gave superior saturation qualities compared with the radio metal or grain-orientated strip steel, given that in the transformers there are quite small gaps with quite high fluxes, whereas in the ferrite’s there are big air gaps and a lot of magnetic material. Although there are probably a few more turns and the DC resistance rises slightly, the saturation properties are considerably enhanced. Rogers decided that with modern crossover techniques, it was simpler to make a 2 or 3% tolerance iron-dust core inductors than to make a gapped radio metal choke. This is partly because the design is complicated and any opportunity to make things less so should not be avoided.

It would perhaps have been simpler in the long term for the BBC to have completely re-designed the speaker, and to have produced a LS3/5B, but they did not want nor could they afford this option, since some three and a half thousand of the existing design were in use up and down the country. And design change had to produce similar results and the same response because future models would end up alongside existing ones. Even today, when a batch is taken to a location, it does not matter whether some are from one generation and some from another – they should all be capable of matching as stereo pairs. That in essence is the whole purpose of the design, consistency through production and during use, anywhere. To sum up, then, the recent changes have offered no enhancements in terms of specification, simply a means to provide better consistency in production to maintain the speaker’s standard. This is now back on ‘median’ and so may sound a little different from those latter ‘older models’ of just pre-Christmas 1987, but will sound the same as earlier production runs.

The impedance has altered slightly post modifications – this can be seen by the modulus shown, reflecting the Rogers’ two standard models. Whereas the original was a nominal 15ohms the newer model is nearer to a nominal 11ohms. Looking at the nomograph, we can see the original as 15ohms, using the old equalizer and with the old bass unit which had about a 7ohm resistance. The new bass unit has a lower DC resistance but, with the new equalizer, the curves are very similar in shape, although the average level is now more like 11ohms. This still means that the LS3/5A is voltage rather than current driven and needs an amplifier that will swing volts and produces reasonable power levels, but not into low impedances.

So an LS3/5A circa 1977/78 will not in any particular way be better or worse than one from 1983 or now. The consensus of opinion from those who understand balance, is that the system is the same and offers the same qualities as it always has in terms of its tonal characteristics. In detail, the latest models are now marginally smoother because irregularities of the high treble have been sorted out, and the coloration has been reduced by a small degree with the new surround, but not so as to affect the tonal balance, because the perception of localized coloration is all part of the balance of the loudspeaker system and its equalization. So if the tonal coloration was reduced to an extent in, say, the 1kHz region where it was noticeable, it could highlight colorations elsewhere.

The new design offers only, in real terms, a better consistency, executed by KEF to maintain the standard. No enhancement in terms of the specification was made although commentators say the loudspeaker now sounds a little better because it is back on the median. Essentially it produces the same sound. This has been backed by tests carried out at the BBC and with manufacturers, with regard to the units stereo capability.

The total number of LS3/5As made since the first license was granted is estimated to be around 60,000 pairs so it is not surprising that a huge interest has been generated around the world. In the BBC the speaker has found many uses from the one-man OB to the control and balance of experimental quadraphonic transmissions from the Promenade Concerts several years ago and in BBC Local Radio Stations, the design provides the main cubicle monitoring. Even so, the design has its fallbacks like the passive and complicated equalizer in the crossover to restore the loss of low frequency caused by the cabinet size and limited cone area.

Licensees

The very history of the speaker has interesting connections with licensees who have been permitted to manufacture them. After early production runs at the BBC’s own Equipment Department in Chiswick, the Corporation permitted a number of companies to apply to produce the speakers under license, as it is obliged to do under its Charter.

The license fee is a means of recouping the original development cost, not purely a way of cutting unit costs by mass-production, and the money is recovered by the Director of Engineering and not BBC Enterprises, the commercial arm of ‘Auntie’ which has taken over other money making areas like publications and records. Recently as many as four companies had licenses although in essence only two actually had a production line at that time. This situation came about when Harbeth, formed by HD Harwood who was working on the LS3/5A at BBC Research Department, was granted a license while Rogers. Spendor and Goodmans were also permitted to manufacture. Goodmans in fact had ceased production and did not re-apply for their license when the renewal time came, and Harbeth had yet to start making any.

Another licensee manufacturer, Spendor, founded by the late Spencer Hughes who had worked with Dudley Harwood at the BBC, did not apply for a license at the first opportunity. The company was then already working to full capacity, but when a ‘vacancy’ next arose it did apply, encouraged by the fact that Mr. Hughes knew the design and all that would be entailed by association with it.

Rogers were involved almost from the start and now own the Chartwell brand, a company who also previously licensed to make the LS3/5As. The parent company, Swisstone, acquired the defunct Chartwell, a previous competitor, in 1978 having merged with Rogers in 1975. Manufacturing to quantity demands, in 1979 Rogers announced that it had sold its ten thousandth pair and proclaimed LS3/5A the worlds most successful small loudspeaker. With the Rogers total to date standing at 33, 534 pairs.

The latest company to be permitted to manufacture is Harbeth, now under the control of Alan Shaw. They had samples from their first batch passed by the BBC and are now eligible for full production; all are of the computer-optimized type. This initial sampling process, a condition of all licensees, is to assure the BBC that their speakers are being made correctly. The licensee selects two units from the initial batch to send to the BBC. One of these is kept by the Corporation as that licensee’s working reference, while the other is tweaked so that it matches the BBC’s own reference, a unit from the very first in-house batch. This second speaker is then returned to the licensee and becomes the reference for all their subsequent units to be compared with.

Earlier companies to have been licensed by the BBC have not fared as well. It is agreed that the LS3/5A is not an easy product to make properly and must be marketed correctly. Some names have fallen by the wayside because they found the market was not as large as they forecast, and some found it too difficult to make to the BBC’s tight tolerances. These problems could be because people expect more than they see for a £250 loudspeaker, and that without the correct marketing, sales are just not realized.

Names like Audiomaster, who launched their models in April 1976, had a rather short lived license span while RAM never got any units out on the street: they went bankrupt before the production line got underway. Falcon, who were making crossovers for Goodmans, made a batch for ailing RAM company and then found themselves having to distribute the boards directly because RAM were not in any position to pay for them. A flurry of activity was seen from JPW but no quantities to speak of ever materialized. Constant checking of the licensees production is carried out, and there are some tales of ‘nasties’ being found. There’s the case of a unit coming in for evaluation with a thicker than usual grille material causing a two-octave 1 1/2dB dip. Needless to say it was swiftly returned.

Of the three current manufacturers, Spendor says that although a percentage of their production line ends up in the UK, 80% of their sales are to the export trade, the areas being mainly Europe, the USA, and the Far East, including Japan and Taiwan. Rogers also export to Japan, Europe, North America, Hong Kong and Singapore, with more sales to Japan than to the United Kingdom even, where, while a quantity go to the BBC, the hi-fi retail outlets get through a fair volume too. Despite the advances of other designs, the LS3/5A remains a reference standard loudspeaker in its class.

Acknowledgements

I am indebted to BBC Engineering for ensuring the accuracy of this article. Within the BBC the work of two people, T Somerville and D E L Shorter, must be considered as having laid the foundation for current standards. The particular work on the LS3/5A was carried out by Messrs H D Harwood of Research Department and both M E Whatton and R W Mills of Designs Department. It was with deep regret that we learned of the recent death of Mr. Whatton.