Harbeth BBC LS5/12A loudspeaker
I was reminded of chocolate ice cream when I unpacked the loudspeakers I'm reviewing this month. Both are finished in chocolate—I mean, black ash veneer. Both are two-way ported designs from designers who learned their speaker engineering at the British Broadcasting Corporation. The Epos ES 14 was designed by Robin Marshall after he'd left the government-controlled broadcasting organization; Robin had played a small role in the development of the BBC's classic LS3/5A design. The Harbeth BBC LS5/12A was developed by Graham Whitehead at the BBC's Design & Development Department.
Both speakers are also typically British in that they are not time-coherent. In the December 1994 Stereophile, I reviewed two small two-way speakers from American designers—the Spica TC-60 and the Dunlavy Audio Labs SC-I—for which time coherency had been a major design goal. The fact that both speakers had superbly flat, uncolored midranges also revealed their designers' not forgetting the importance of the traditional frequency-response measurement. In the UK, however, while flat frequency response has always been a prime goal, for a speaker to be time coherent has never been considered a necessary indicator of goodness.
Harbeth LS5/12A: $2200/pair
This tiny monitor was developed by Graham Whitehead of the BBC's Engineering Department, a team famed for its work on monitors in the '70s, and which designed the classic LS3/5a, among other loudspeakers. Sadly, to save money the BBC has axed the staff in this department, meaning that the LS5/12A might be the last loudspeaker design to come from the BBC. As with all BBC designs, however, actual manufacture is licensed to outside companies; the LS5/12A is being produced by Harbeth Acoustics, the company founded by ex-BBC engineer Dudley Harwood and now run by LS3/5A enthusiast Alan Shaw (see "Manufacturers' Comments," January 1994, p.235).
The design brief for the LS5/12A was to make a monitor loudspeaker around the same size as the LS3/5A, with a tonal balance similar to existing larger monitors, but to have a much greater dynamic range and a more extended bass than its predecessor. The result is an elegant little loudspeaker, with the front baffle almost entirely occupied by two high-quality Dynaudio drive-units compared with the '3/5A's KEF units. The 1" soft-dome tweeter is an expensive Esotec unit, the D-260, which has a vented pole-piece, while the woofer, a Dynaudio 15W-7508, features a plastic cone with a radiating diameter of 4", a large voice-coil, and the big dustcap typical of this company's driver designs. The woofer is reflex-loaded by a large port on the cabinet rear, this about 5.5" deep and 2" in diameter. The port opening to the outside world is flared to minimize wind noise at high levels.
There are two pairs of metal binding posts on the cabinet rear to allow for bi-wiring. The LS5/12A can also be special-ordered with Neutrik Speakon connectors. Internal wiring is good-quality multistrand, and all the joints are soldered. The crossover appears to be fourth-order, and is constructed on a glass-fiber printed circuit board attached to the terminal posts. It is said to be relatively simple, due to the Dynaudio drive-units having naturally smooth responses, but still features a large number of parts—six inductors, eight capacitors, and seven resistors. These appear to be of high quality; for example, plastic film capacitors are used (with one exception in the woofer circuit) rather than electrolytics. However, although air-cored inductors are specified in the promotional literature, all the coils in the review samples featured ferrite cores. (This is an approved production change.)
I was interested to note in the LS5/12A's publicity material that "the final balance of the system was optimized after extended listening trials in BBC studios, comparing the output of the loudspeakers with that of live program. Thus the balance has been set to re-create reality in a room, not a theoretical anechoic measurement ideal." This sounds like a praiseworthy goal, but it does throw the responsibility for the speaker sounding neutral on the methodology used for the listening tests. The very different directivities of loudspeakers and "live program" can add a significant disturbing variable into a listening test which, if not accounted for, can lead to a preference for a decidedly non-flat balance, in my experience.
One QA aspect of the speaker should be noted. After six months, the black ash veneer on one cabinet split vertically and started peeling away from the carcase. This is probably just a sample fault, but Harbeth should look into it.
All during the review period, I kept flip-flopping over the merits of the Harbeth LS5/12A. When it was good, it was very, very good. When it was bad...well, it was never really bad, it just failed to light my fire.
First the good news: Harbeth's LS5/12A has a simply stunning midrange—significantly more neutral than the LS3/5A, which always had a bit of a nasal honk at the top of the woofer's passband. Voices, while appearing to be a little more forward than I'm used to, sounded very natural. Midrange colorations there were none! And on rather complicated recordings, such as the superb new Dave Matthews Band CD Under the Table and Dreaming (RCA 66449-2), sonic soup was never in evidence, every little detail in the mix easily apparent.
The '5.12As' imaging was also precisely focused and very stable with both frequency and lateral position, if not sounding quite as palpable as the Spica TC-60 and Dunlavy SC-I, both champs in this respect. There's that lack of time alignment, I guess.
The LS5/12A also sounds considerably more dynamic than the LS3/5A. An astonishing amount of sound comes from these two tiny boxes. While its sounds clogs up early compared, for example, with the much larger Epos ES 14, in this respect this is a loudspeaker to confound critics of "minimonitor sound."
Naturally recorded classical piano came over with a considerable amount of lower-frequency authority—perhaps too much, as recordings that I knew to be flat in the bass tended to sound too rich. I'll give you an example: For the next Stereophile Robert Silverman release—a tumultuous performance of the Liszt Piano Sonata in b—we used two pairs of microphones and recorded with 20-bit resolution on the four channels of a Nagra-D open-reel digital recorder. The first pair were spaced omnidirectional B&K 4006s, the mikes used for our recent Concert release, while the second pair of mikes was actually a single stereo model, the Schoeps KFM-6U "Sphere." The B&Ks have terrific low-frequency extension but rather hazy imaging; the Schoeps Sphere has much more solid imaging but lacks low-frequency weight. Except when I played 16-bit DAT backups of the Nagra tapes over the Harbeths, the Schoeps sounded perfect, the B&Ks too fat. The naïve listener might well conclude that the LS5/12A has "great bass"—me, I started worrying that the speaker's low-frequency alignment might be a little too good to be true.
Naturally, I reached for Stanley Clarke's East River Drive album (Epic EK 47489), particularly track 5, "Africa I'm Home." Oh yes! These little speakers can play loud, reaching spls of 100dB at my listening position before wind noise from the port and distress signals from the woofers caused me to take pity on the little beasts. But the mondo bass riffs on this track were both absent any real low-bass content and tended to blur into a roar, the spaces between the notes filling in. Some would call this behavior "slow," others would refer to a shortfall in the pace'n'rhythm department. I just felt that the music passed me by. The same could be heard on acoustic double bass, which lost its "tunefulness," if you will.
This bass quality was annoying, but I could minimize its effect by careful experimentation with the speaker positions. What I could not live with was the '5/12A's treble balance. A consistent feature of its presentation was a prominent mid-treble, a boost in just the right region to make everything sound bright. At first, this added a quite impressive degree of verve, but it rapidly outstayed its welcome.
I asked Tom Norton over, primarily to share Thanksgiving dinner, but also to take a listen to the Harbeths. He sat for while. "They're a little, er...bright, aren't they."
Indeed they are, Major Tom! In fact, after my listening was done and I was writing up my notes, I took a gander at a review of the LS5/12A by one Eric Braithwaite in a British magazine, Hi-Fi News & Record Review (footnote 1). "[The LS5/12A] is markedly bright-sounding...A slight but perceptible emphasis in the treble produces the studio monitor hallmark of extreme crispness and clarity..." he wrote.
Indeed it does, Eric. I looked up Ken Kessler's "Industry Update" report on the launch of the LS5/12A at the 1993 Heathrow Penta Show (footnote 2), in which he reported the LS5/12A's designer, Graham Whitehead, as saying that whereas the LS3/5A makes everything sound nice in the midband, the new one was a "merciless bitch." Well, perhaps I wouldn't go that far, Graham, but your speaker is certainly on the way to the mercilessness 'hood.
As suits a monitor, which is intended to help mixing and balance engineers hear what's wrong with recordings and microphone feeds. But in the home, you need to hear an emphasis on what's right with your CDs and LPs if you want lasting musical pleasure from a loudspeaker. And in that area the LS5/12A falls down—perhaps only slightly for some, and not as much as I'd expected from its other reviews, but enough to rule it out of contention for me.
Being a long-term BBC LS3/5A owner, I really wanted to like the Harbeth BBC LS5/12A. As a professional nearfield studio monitor, where its balance will be revealing of poor tape splices and disturbing background noises, and its increased dynamic range compared with the LS3/5A will be an advantage, I'm sure it has a successful career ahead of it. But as a domestic loudspeaker, I'm less impressed. Yes, the new speaker will play louder than its BBC predecessor, and it has a superbly neutral midrange balance. But Homo audiophilus doesn't live by the midrange alone. Ultimately, the LS5/12A's bass and treble balance irritated me, particularly on rock music, to the point where I was happy to stop listening to my music.
When you consider that Harbeth's own superb little HL-P3, which I reviewed in December 1993 (Vol.16 No.12, p.189), is both more neutrally balanced in the treble and has a particularly well-tuned bass, and that it costs only half as much as the LS5/12A, you'll see why I don't think a recommendation for this final design to come from the BBC's Engineering Department is appropriate. Sorry, Auntie Beeb!
Footnote 1: Hi-Fi News & Record Review, September 1994, Vol.39 No.9, p.46.
Footnote 2: Stereophile, November 1993, Vol.16 No.11, p.39.