BBC LS3/5a

LS3/5a (39K)These loudspeakers should require no introduction, having acquired "classic" status. Or should that be "cult"?

I'm painfully aware that while the rest of this section has grown and grown over the years, I've written almost nothing about the LS3/5A. There are several reasons for this, but I should say that it's not because of a lack of personal enthusiasm about the model itself - I own several pairs, from Rogers, Charwell and Goodmans, and use at least two pairs every day. I like the LS3/5A. A lot...

But of course, like all loudspeakers, the LS3/5A is not perfect, and it does have shortcomings over and above the obvious bass extension and SPL limits. But experience has taught me that talking about these is risky - many LS3/5A fans are extremely loyal and don't tolerate criticism, not matter how reasonably presented. This is sometimes very frustrating, as I am an engineer first and foremost, and instinctively I want to deal with objective and verifiable facts. A rather old-fashioned notion!

So I could go into details about my extensive experience with this model, which includes lots of measurements as well as subjective listening tests, and I could describe the trends I've observed from different brands and different vintages, and discuss how unlikely it is that surviving examples are still in-spec. I could back up these claims with data. But, almost no-one would be interested - the LS3/5A is on its pedestal, and it's there to stay.

And that's fine. Technicalities aside, many people - myself included - love the LS3/5A. For some, that love is blind and unconditional. For me, I love the LS3/5A in full knowledge of its flaws. I think my position is definitely more healthy than the faith-based, quasi-religious stance adopted by some!

Of course, the other main reason is that so much has been said elsewhere, and often by people who are considerably more expert that I am - for example, Paul Whatton, who's father Maurice worked on the development of these in BBC Designs Department. Paul runs http://www.ls35a.com/, which is where you'll find everything you need to know about the LS3/5A, and much more besides, including pictures, reviews, history and links to manufacturers. Details of the crossovers are also there.

So it is with some trepedation that I expand this page. I will tread carefully!

History

As most know, the LS3/5A was derived from the acoustic scaling research that BBC R&D were doing. This involved making scale models of proposed studios to evaluate the acoustic properties of the space before commiting to building the real thing. The models were built to an 1/8th scale, and this means that the wavelengths of the audio signals in the space also had to be reduced by a factor of 8 - in other words the frequencies were multiplied by 8. The main frequency range of interest is 50Hz to 12.5kHz, meaning a loudspeaker (and measurement microphone) that can cover the 400Hz to 100kHz range was needed. This is easy to say but incredibly hard to achieve. And even if you can find transducers - loudspeakers and microphones - that can cover this frequency range, what about the electronics and recording devices? Remember, this was 50 years ago!

An overview of all this is included in BBC RD 1972-34. Of course, it's the loudspeaker that we're interested in, so jump to page 10 of that PDF.

When the loudspeaker developed for that project was auditioned using normal-speed programme, it sounded surprisingly good, despite not reaching much below 100Hz. And this planted the seed of an idea: would it be possible to build a really small monitor for use in confined spaces? We have to bear in mind that at this point in time, the BBC's idea of a "small" monitor meant something like an LS3/6 or Spendor BC1, which are basically 2 cubic feet in volume - about 56 litres. The LS3/5A occupes just a tenth of that volume!

Shortly afterwards, and with rather fortuitous timing, a request was made for a small loudspeaker for use in mobile television control rooms ("OB trucks" in modern parlance!) and the so the LS3/5 was born.

The LS3/5

Around 20 LS3/5s were built in-house. The enclosure was made from 9mm birch plywood, but according to Trevor Butler ("The Little Legend"), the rear panel was made from lower quality spruce ply. In the classic "BBC thinwall" tradition, bitumen damping pads were added to the panels, and foam was used for air damping. The front and rear panels were removeable, secured with screws. The tweeter was the same KEF T27 that was used in the acoustic scaling work mentioned above, but rather than the custom-made bass driver, the KEF B110 was selected. The complete design had a nominal impedance of 9Ω.

This initial batch was deployed in a number of OB trucks to good effect. But when more were requested, it was found that KEF had changed the design of both drive units, and a re-design was required, causing the LS3/5 to become the LS3/5A.

Today, almost no original LS3/5s exist, which proved to be something of a challenge when Graham Audio revived the design a few years back. I am aware of some of the fascinating research they did during that project, but can't say more here for obvious reasons.

The LS3/5A

While the LS3/5 was entirely the work of BBC R&D, the work to re-engineer it in light of the changed KEF drive units was a collaboration between BBC R&D and BBC Designs Department.

Several interesting things emerged from this work. It is not clear to me how much of this came about because of the drive unit changes, and how much was because we now had two different teams working together - personally, I think that's a fascinating question. After all, BBC Designs Department had Ralph Mills, who was referred to as "Golden Ears", and for extremely good reason! Also, as Paul Whatton explains in his personal history, his father's background was electronic design, and audio was a new area to him, and I suspect that this brought a fresh perspective to the team. Either way, some of the changes resulting from the LS3/5A development work included:

  • A PVC edging was required on the frame of the B110 where it met the rear of the baffle
  • The type of timber used for the battons was critical - using Parana Pine rather than beech caused audible colouration
  • Thick felt strips were added to the tweeter were to discourage diffraction at the cabinet edges
  • A protective grille was added to the tweeter - this came from the Celestion HF2000 - which had the effect of lifting the top end, which was seen as beneficial at the time
  • The cabinet wall thickness was increased from 9mm to 12mm
  • The rear panel is glued in place, and is definitely birch ply now

Drive unit changes

Of the two, it's perhaps the tweeter that changed the most. The original KEF T27 was type A6340 which, being made from injection-moulded plastic, has a very different appearance to its replacement, the SP1032.

At a glance, the SP1032 looks much more like the standard "flat plate bolted to a magnet" construction that so many tweeters share. But look more closely and you'll see that the diaphragm is applied to the front of the faceplate rather than being sandwiched between it and the magnet assembly. Indeed, the front plate is the front pole piece - normally that's a separate component bonded to the magnet. The diaphragm is made from Melinex polyester film, vacuum-formed into a 19mm dome.

The original B110 woofer was KEF type A6362. This was replaced by the SP1003, which became the A version (B110A). Compared to the tweeter, the visual changes are modest. The earliest versions of the SP1003 (and all of the A6362) had an unpainted chassis; subsequent examples were painted black. All versions of the B110 have a Bextrene diaphragm which was doped; both the A6362 and the SP1003 were doped on the front. Both also have a neoprene surround. For more information, see A History of KEF Drive Units.

B110 problems

Unfortunately, as Brian Pook explains, the rejection rate of the SP1003 was as high as 30% at the start of LS3/5A production, and this grew as production progressed. To make matters worse, it wasn't until the whole loudspeaker had been assembled and tested that the problem showed itself!

KEF were keen to help, and exchanged the out of spec woofers for Rogers, but it clearly wasn't a sustainable situation and so an investigation by Laurie Fincham (KEF's Technical Director) and Malcolm Sked (Rogers) found that the BBC's design work was carried out using woofers that were at one end of the normal production spread! As the BBC couldn't be persuaded to revisit the design, KEF agreed to pre-test B110s so they could supply woofers to Rogers that would meet the specification. You can imagine that other licensees were rather keen to know how Rogers were able to successfully manufacuture so many of them!

This is just one reason why drive unit replacement is a thorny area for the 15Ω LS3/5A. If the "donor" unit is from another LS3/5A of similar vintage you might be OK, but if it's a B110 from some other loudspeaker, then almost certainly not. Remember, the B110 was widely used by many other people at the time, including KEF, of course, and the fact it says "SP1003" on the magnet is not good enough!

As time went by, the B110 proved even more troublesome. Part of the problem was caused by the Bextrene manufacturers, who regularly made minor changes to the formulation. Their primary customer base - the automotive industry - would have been completely untroubled by this, but it caused many problems in the loudspeaker world. I well remember Derek Hughes telling me about a problem Spendor had with a batch - Spencer Hughes called KEF to see if they were having the same experience, which indeed they were. "Hold it up to the light" said Spen. The confusion at KEF HQ could practically be felt down the 'phone line in the resulting silence, which was broken by the sudden explanation "it's green!". Of course, the loudspeaker industry was only buying tiny amounts of the stuff - as little as 1 ton per 10,000 sold, according to some estimates I've seen - so they really were at the mercy of the manufacturers.

Another problem was caused by variations in the neoprene surround, which again got worse as the years passed.

The B110 has a peak in the 1.5kHz region. This is tackled by the crossover, but this part of the crossover is fixed. This means that if a particular drive unit has a different amplitude at this peak, then there will be a rise or fall in the overall freqency response at that point. In practice, it's usually a rise, as the peak got larger as time went by - this can be seen in most published 15Ω LS3/5A measurements to some extent. The exact level of this peak will depend on many variables, such as the Bextrene batch and the doping. Minor modifications (in the form of component value changes) were occasionally issued by the BBC - as detailed by Falcon Acoustics, but eventually it got to the point where something had to change.

The 11Ω LS3/5A

This work was done by KEF, who from this point forward supplied the drive units complete with crossover assemblies to the licensees - all pre-tested and ready to bolt into the cabinets.

The woofer changed to SP1228, which as far as I know was only ever fitted to the LS3/5A - it was the C version now (B110C). It looks rather different to earlier versions because the doping is applied to the rear of the cone before being assembled into the chassis. The surround is PVC. As this surround is less compliant than the neoprene original, the spider is different to ensure the same bass performance. The nominal impedance is reduced from 8Ω to 6Ω.

I'm not aware of any significant changes to the T27 over the life of the LS3/5A, and I don't think it was specially selected (though of course a replacement tweeter might need the level adjustment altered in the crossover, which needs a measurement microphone). Having said that, Falcon record that a minor crossover modication (C2 increased by about 10%) was issued in 1984 to correct for a rising response trend in the T27 - something to look for when replacing tweeters.

The new crossover was cheaper to make - instead of using expensive radiometal inductors, standard ferrite-cored inductors were employed and the auto-transformer was changed for a resistor ladder. As a result of the changes to the crossover and the slightly reduced impedance of the woofer, the nominal impedance of the complete loudspeaker went from 15Ω to 11Ω.

There was a version of the new crossover that supported bi-wiring, as the market demanded at the time - though of course the BBC would have ignored that. Indeed, most examples supplied to the BBC were fitted with a 3 pin mail XLR socket, but very occasionally you'll stumble across domestic connections like 4mm binding posts. Or if it's from Goodmans, it might have nasty screw terminals!

In theory at least, any version of the LS3/5A could have been made in this form, as it was just a question of PCB design. But that particular fad didn't emerge until the mid-late '80s, and the BBC didn't approve it until some time later (on the proviso that the speaker met the spec with the terminals bridged together).

It's interesting that all of these changes are really quite significant, but it's still an LS3/5A, not an LS3/5B. The idea from the BBC's point of view was that the new version should be close enough to the original that 11Ω versions can be mixed with 15Ω versions in a stereo pair. Of course the BBC didn't use the concept of "matched stereo pairs", as all examples of the same loudspeaker should match the original specification sufficiently closely to work together for stereo. If the changes meant that the new speaker was sufficiently different to warrant a new "B" designation, then that would have caused all sorts of confusion and inconvenience across the BBC - support engineers would now have to ensure that each stereo pair was the same version - despite looking identical from the front - and a failed speaker might mean that a pair would have to be swapped rather than just a single one, and this means that more spares would need to be kept in stock. When you think about it, the BBC's position made a lot of sense.

But having said all that, whenever I've tried mixing 11Ω and 15Ω examples in a stereo pair, the results were OK but not good, especially in terms of stereo imaging. It's hard to say if this was because the older 15Ω example was no longer in spec, or because the two versions just aren't quite close enough for stereo. Either is possible, but my research suggests that the former had most effect.

Historic LS3/5A Licensees

Many manufacturers built the LS3/5A. A full timeline of the history of the LS3/5A production can be found at LS3/5A.com so there's no point duplicating it all. I'll restrict myself to just a quick summary based on Paul Whatton's data - if anyone can add to this, please contact me. This table is sorted by order of introduction of production:

LS3/5A Production Overview
Manufacturer From To Comments Pairs sold
Rogers 1975 1993 The first to be awarded a licence, and the most prolific builder, making something like 60,000 pairs between 1975 and 1993. They made all versions of the model, including gearing up for the original LS3/5 before the model was changed to the A version. 60,000
Audiomaster 1976 1981 Formed by Robin Marshall, who worked at the BBC for a few years. ?
Chartwell 1976 1978 ?
RAM 1979 1983 RAM obtained their licence after Chartwell stopped production ?
Spendor 1982 1999 Spendor manufactured both the 15Ω and 11Ω versions. By tightly selecting drive units they were able to modify the crossover to do away with the complicated auto-transformer in the earlier model. ?
Goodmans 1984 ? Domestic versions came with cheap screw terminals. Crossovers were supplied by Falcon Acoustics. Only made the 15Ω version. ?
Harbeth 1988 ? Dudley Harwood originally obtained a licence in 1977, but didn't build any before selling the company to Alan Shaw in 1986. Only built the 11Ω version. ?
KEF 1993 ? Despite supplying the drive units (and the 11Ω crossovers) to all other licensees, KEF didn't start building the LS3/5A until 1993. 4000

Surviving LS3/5A examples

Here's where things get controversal!

These command high prices on the second-hand market. Inevitably, certain vintages are prized more highly than others, and the same applies to the brands too. Of course, there are many collectors out there, so scarcity value is a strong factor in auction closing prices. Taking an objective position - as you'd expect me to, I hope - let's look at some of the interesting trends that can be seen on the second-hand market:

15Ω or 11Ω?

There's no doubt the 11Ω versions are technically "better" than the 15Ω examples, and many of the reasons for that have already been discussed above. But the general trend is that 15Ω versions fetch higher prices.

As we know, it was hard enough to make 15Ω units that met the spec when new, especially towards the end of this model's production run, but what has happened to them in the decades since they left the factory? All loudspeakers change over time. These changes are gradual and often minor, and will usually go unnoticed. But in the case of the LS3/5A specifically, we're talking about a speaker that is extremely critical in many areas, and small changes in drive unit parameters can be more significant that would overwise be expected.

I can't get into the details here for a variety of reasons, but I have measured a fair few LS3/5As over the years, and relatively few meet the LS3/5A specification any more. In general, 11Ω examples are pretty good, but 15Ω models are less so. I'd understand anyone who didn't want to believe me here, but Derek Hughes has found the same, and has said so publicly. Of course, Derek has measured far more than me as part of his work with Stirling Broadcast and Graham Audio, and was building original versions of these when he was running Spendor, so I think we can trust him.

The fact that many prefer the 15Ω versions is interesting - as luck would have it, the deviations from the original standard are clearly appealing! There's nothing wrong with that - this is a question of personal preference after all - but just bear in mind that what you're listening to might not be exactly what the BBC originally intended.

Which brand?

Again, prices are driven by factors other than absolute quality.

When it comes to the later 11Ω version, the drive units and crossovers were matched by KEF, and as mentioned above, they have survived the passing of time rather better than the earlier versions. So from a technical point of view, they should be much of a muchness, ignoring minor variations in cabinet finishes, etc. So scarcity and condition usually determines the price, and as Rogers made the most, they tend to be the least prized, especially by the collectors. But for us non-collectors, that's no reason not to buy a pair made by Rogers if the price and condition is right.

For the older 15Ω versions, you could argue that Rogers were the best because of their secret deal with KEF to supply pre-measured B110s. Other licensees likely did not have this arrangement with KEF, so would have needed to reject a lot of the B110s and mess about finding suitable performers from the batches they'd been sent. Put yourself in their shoes - the temptation to let through marginal examples must have been immense (though I must say that I have no evidence of this happening!).

You also have to wonder how the quality control was managed in some of these companies. Back then, measuring loudspeakers was an expensive affair that required a dedicated room with specialist equipment. In my testing, I've noticed that some brands exhibit certain trends that is more likely explained by difference in testing rather than drive unit variations (and aging). And these findings also corrolate pretty well with subjective views commonly expressed on the audio forums. As in the 15Ω-vs-11Ω debates, it seems that many people prefer their LS3/5As to be out of spec!

Digging through my archives, I have just come across a scan of a Subjective Sounds column, written by Paul Messenger for the November 1976 edition of Hi-Fi News & Record Review. The scan was found on the Harbeth User Group, but I doubt I'd be able to find the thread after all this time - I won't include it here for fear of upsetting Alan Shaw, but suffice to say, it does mention some findings with a Chartwell sample that, despite being subjective impressions, do fit in nicely with my findings - suggesting that my measurements weren't of samples that were outliers.

Caveat emptor!

Most people know that LS3/5As are worth a lot on the second-hand market - and those who don't are usually able to use a search engine! As a result, the temptation to repair a broken example in order to increase the sale price is tremendous.

When buying a used LS3/5A, how will you know it still contains the original drive units? I've said a lot about this already - the SP1003 B110 must be towards one end of its tolerance curve, and those that were went to Rogers by default. An SP1003 from any other speaker will not meet the LS3/5A specification.

So provenance is paramount when buying LS3/5As. Ideally you want to find a pair that were bought for gentle domestic use and have never had their grilles removed - let alone the baffles. If they come with the original packing and receipts, so much the better! But such examples are few and far between, and people bidding know that.

You might think that buying ex-BBC would be a good plan. There is an assumption that BBC equipment has been meticulously maintained by highly qualified staff, but that simply isn't the case - at least, not in recent times. There are fewer engineers than ever before, and they have far more important things to worry about than obsolete teak boxes! And quite a few of them actually have little to no knowledge about speakers in general, let alone the LS3/5A. Of course, this is not intended as critism - it's just the reality of working in a difficult and demanding environment where there isn't time to lavish love and affection on fully depreciated assets!

Speakers brought from the BBC - usually via a 3rd party asset disposal company who will buy vast quantities of redundant kit from the BBC for a fixed price and auction off what they can - will be scratched and dented, covered in stickers, and likely faulty. If they work at all, there's a strong chance they have been repaired badly at some point in their life.

A couple of examples - I've found several pairs of LS3/5As fitted with Vifa tweeters. These are decent enough in their own right, but nothing like a KEF T27. At least the engineer added stickers to alert users that they were non-standard. We took them out of service a few years back, and that means I have a nice pile of slightly used obsolete Vifa tweeters to use in some DIY projects!

Second example: when attempting to repair a 15Ω example with a dead B110, a colleague took a look at our small stocks of broken LS3/5A in the hope of finding a working woofer. At this point I wandered in - I'd been tipped off that some loudspeaker maintenance was happening, and that I might be interested (my reputation preceeds me!) - just as he was about to fit an SP1228 from an 11Ω donor! When I asked him to explain his thinking, it was a case of "this one is more shiny than the original". I kid you not! I tried to explain that this a completely different woofer with a different impedance and it was shiny because it was rear-doped, unlike the original SP1003 that was doped on the front. However I could see that all this wasn't going anywhere, so I took a look through our graveyard and found one with a very close serial number that still had the original working woofer (you can usually tell if the components are original by closely inspecting the soldering). Once installed and working, I offered to measure it to make sure it was close enough (as these things get!) to the original spec, but this was considered to be over the top - and in business terms, that was probably correct as we're not using the LS3/5A for critical or even semi-critical work any more.

So imagine you were the lucky (?) winning bidder on that speaker. Having won the auction - and then realised that as this wasn't eBay you have to add the buyer's premium and VAT to the closing bid amount - you get the speaker home (collection only), pop off the grille, and find the wrong woofer (and chewed up screw heads and air leaks from the gasket that wasn't checked before reassembly and goodness knows what else!). And being a commercial auction rather than eBay, you have no rights whatsoever...

If there was ever a product where "buyer beware" really mattered, it's the LS3/5A!

Some 25 years ago I was shown the back room of a second-hand hi-fi dealer, and was shocked to see about a dozen pairs of speakers, all in pieces, and someone desparately swapping parts around in an attempt to make working speakers that could be sold. There was no measuring equipment, and no attempt to check anything beyond how it looked from the front. While the person doing this work was (just about) able to solder, it was clear that he lacked any real knowledge about speakers. For the vast majority of hi-fi speakers this is perhaps a reasonable approach, but this is not something you can do to an LS3/5A.

Still, if I ever decide to sell any of mine, you can be assured that they will be decent!

Current LS3/5A licensees

If reading the above has scared you off the idea of investing in a pair of LS3/5As, there's good news: you can have a brand new pair, complete with a warranty.

At the time of writing, there are 4 companies making the LS3/5A. In alphabetical order:

I haven't been able to closely inspect all of these, and can't be drawn into conversations about which version is the best to buy - that's what the discussion forums - like the LS3/5A Groups.io - are for. I say a bit more about the licensees and my reasons for being impartial elsewhere in this section.

LS3/5A Reviews

There are countless LS3/5A reviews out there, but for many years I've carried a review by Martin Colloms (from the 1986/87 Hi-Fi Choice Loudspeaker Yearbook) that I haven't seen elsewhere. It is interesting because it is one of the less favourable reviews that has been published over the years - and of course, comes from near the end of the 15Ω model run. Click to read in a new window.

Owner's manual

Here are scans of the manual that came with my domestic 11Ω examples from Rogers - click to enlarge...

Please note: The copyright of this manual probably belongs to Rogers International Ltd, and while I believe that the inclusion here constitutes fair usage given their historical value, please be aware that they may be removed at any time.