Harbeth blog


Alan Shaw, the designer and owner of Harbeth, the high end speaker manufacturer, has started a new blog on the Harbeth User Group: http://www.harbeth.co.uk/usergroup/forum/designer-s-blog/75232-ahoy-from-the-engine-room
He discusses audio and particularly speaker design, but also the science of hearing and the methodology for testing and listening to gear. Anyone who is interested in the scientific approach to audio engineering (and sensible purchasing decisions) will enjoy his writings. Subjectivists may be annoyed, and so will be the snake oil sellers.
willemj
Thank you willemj for the link to Alan Shaw's new blog "ahoy-from-the-engine-room". Always good to have an "objective" focus in discussions instead of "my golden ears tell me it is so"! Scientific truth always trumps magical beliefs!
He is a curious mix of engineering and listening. He designs everything himself and tunes with his ear alone. He dismisses “Rock Speakers” entirely. Definitely quirky as his musings tend to convey. His interest is acoustic and classical at low to moderate levels. This has had a big influence on design.

Alan’s big focus is on damped cone material. I know of only one other speaker maker with such a focus - ATC. All the rest are chasing light weight rigid cone materials that look great under frequency measurements but sound terrible due to added intrinsic coloration from cone vibration. Latest ATC designs include CLD - constrained layer damping - this places damping material in a sandwich between two light weight cones.

http://atcloudspeakers.co.uk/technology/cld/

The key understanding that Alan latched on to early in his career with Dudley Harwood is that all cone materials impart a coloration to the sound (as seen on waterfall plots but best assessed with critical listening). The key realization is that you use engineering design for a flat frequency response but you use intrinsically damped cone material to reduce subsequent tonal coloration after the transients. The sole purpose of the radial damped cone was to get away from polypropylene foggy or nasal sound.

If you examine Harbeth and ATC mid range and tweeters you will find they are damped with highly viscous semi-solid semi-fluid material. ATC mid range are sticky.

This approach differentiates Alan’s work from all other transducer makers except ATC. The difference between ATC and Harbeth is that Harbeth are engineered (voiced) to play at low volumes where they sound best. ATC are engineered for realistic loud volumes where they sound best. This is a design choice based on equal loudness contours and preferred listening level.

Anyhow, a thread about Harbeth must necessarily start with a recognition and realization about what makes Harbeth sound so amazing. It is all about intrinsic damping within the cone material - this is what makes Harbeth sound so natural and revealing - the usual clutter from cone vibration after the initial transient is absent. Quad electrostatics achieve similar results in mid range clarity but they are hopeless at providing a wide even sound field/sweet spot (too much beaming)  and are dynamically severely limited (much more than Harbeth.)

Two cents
@willemj

This TAS interview with Alan Shaw might interest you!

Interview with Alan Shaw on the Genesis of the Monitor 40.2

Please compare the Monitor 40.2 to earlier iterations.
The main improvement in the Monitor 40.2 is its completely redesigned crossover network. The obvious functions of the crossover are the division of the audio spectrum among the drive units and the level adjustment of the contribution of those drive units. The less obvious function is that the crossover overrides the ear’s sensitivity to sound coming from more than one source. In nature, the sounds we hear emanate from point sources that give them directionality. For me, seventy-five percent or so of the total design for any new Harbeth model is precisely this issue of making a convincing junction between the sonic contributions of the drive units in order to simulate a point source. Also, the work in developing the SuperHL5plus opened up techniques that I could weave into the Monitor 40.2. It’s often about small, incremental steps that can cross-fertilize from one model to another if they’re appropriate.

One difference I immediately heard between the new model and the original is the slightly deeper bass extension of the 40.2 but slightly less overall bass around, say, 100Hz, the so-called warmth region.
The Monitor 40 was originally conceived as a drop-in replacement for the Rogers LS 5/8, a speaker in wide use in the late nineties in the BBC and elsewhere in British broadcasting. It was quite a shock to procure a pair from a disbanded studio and measure and listen to them under domestic conditions, which they were never designed to be used in. The entire frequency response had been tuned to the highly absorptive acoustic environment of “old school” studios of the era. Nobody could foresee how the architectural minimalist trends would become so popular among consumers, and even among studio designers. Since the Monitor 40 project now covers over fifteen years, the bass has been progressively optimized, that is, made drier, for those less damped acoustics. So the 40.2 is better balanced for the majority of modern rooms.

I’ve always found your speakers to be notably coherent, yet so far as I am able to tell, you don’t employ special methods to achieve this, such as physically staggering the drivers so their voice coils line up, etc. Can you comment on this?
Coherency is primarily a crossover issue. Conceptually the difficulty lies in the way the ear/brain interprets the junction between the two or more sound sources. It doesn’t take much of a mismatch in level or phase for the subconscious to needle the brain with awareness that there is something wrong. I was lucky to be sensitive to this matter as a rookie designer some thirty years ago. By trial and error I developed an approach to getting the best out of the drive units, an approach I’ve basically replicated in all subsequent Harbeth designs. I’m sure other loudspeaker designers have their own lexicon of tricks.

Both the coloration and integrality of Harbeth speakers is really low, yet your drivers are not made from the same materials.
Coloration takes many forms. I recall asking my predecessor, Dudley Harwood, from whom I purchased Harbeth, as he handed the keys of the company to me, for a definition of coloration. Taciturn at best, he replied, and I quote, “You will know it when you hear it.” When I started designing, coloration seemed so extremely obvious, to my ears anyway, that even though I didn’t technically understand where it came from, I kept designing until I had eliminated it. Loudspeaker drive units are energized by the music, which causes their diaphragms to be shocked into motion. These shockwaves substantially radiate through the diaphragm and generate sound waves in the room. Unfortunately, a proportion, a very small proportion, bounces around inside the diaphragm itself and interferes with successive musical events. You can imagine that within the first thousandths of a second after the music starts that a background of residual energy will have built up in the diaphragm and will be topped up by successive notes. It’s this sonic mush that at best fogs the overall sound and at worst introduces audible coloration, where some notes are dominant. The pioneering research that we, in collaboration with government funding, conducted in the 1990s proved beyond doubt that all commercially available materials, including all the popular ones, used in loudspeaker manufacturing are really unsuitable for the task. Their molecular structures, particularly the inter-chain bonds, have characteristics that nip energy from the music they’re attempting to reproduce, particularly in the presence and lower treble region. The low coloration that you hear in the Harbeth loudspeakers, especially in that critical musical band, is a direct result of us conceiving, proving, and blending different materials for their acoustic properties.

Do you still employ recordings of your daughter’s voice to do a final voicing of the speaker?
My daughter is now in her early thirties and has quite a different voice to the nine-year-old that I recorded all those years ago. Good news, though—my granddaughter is nearly four years old and I am grooming her for a life in loudspeakers! Seriously, the ear/brain is highly optimized for detecting subtle nuances in human speech. If we guess that our ears have been under development for some millions of years, we know that the first musical instrument appeared around fifty thousand years ago. This is far too recent to have had any physiological impact on the development of the human ear. It follows then that to use our ear as an analytical instrument when grading loudspeakers, it’s the reproduction of voice that can tell us a lot about the mechanics of the loudspeaker. Note that the human vocal tract is a soft tissue structure with plenty of “damping” thanks to being nourished by warm blood and elastic tissue. All of the undesirable characteristics of loudspeakers that are commonly mentioned such as spitty, ringing, wiry, harsh, biting, gritty, bright, brittle, and so on are likely to be the consequence of hard materials in undamped resonance. No wonder then that convincing natural sound is so elusive in home hi-fi.

I have always been curious about this whole matter of voicing. How do you “voice” a speaker system without the use of, say, an equalizer, whether analog or digital?
“Voicing”—I don’t like this word and don’t use it. All it means in a fancy way is of setting the contribution of the drive units so that they are blended adequately to fool the listener’s ear into thinking that he is actually in front of the performers, live. Present one hundred loudspeaker designers with a cabinet fitted with drive units and a box of crossover components and you will end up with one hundred different voicings. Which one is correct? That’s a tough question because those one hundred designers will have two hundred different ears. They also evaluate sound differently, different instruments will appeal to them or not, they’ll be sensitive to different colorations and some may see themselves as wizards with the power and right to “interpret” the recordings. Some may use test and measurement equipment that will guide them towards a relatively neutral contribution of the loudspeaker, others may voice entirely by ear. Whatever the strategy, expect a wide variation in sonic performance. If, however, a degree of objectivity is introduced, those speakers could be graded. One attack would be to record a human voice under non-reverberant conditions and to switch between that human sitting next to the loudspeaker and his or her voice reproduced over the loudspeaker. My experience is that ninety percent of the candidate loudspeakers would be dismissed as having characteristics not at all present in the live voice. It’s a great pity that the word “voicing” is rarely associated with the concept of listening to a human voice over the loudspeaker!

Also, how do you control the dispersion of the response?
In reality there is not much that can be done to control the dispersion of loudspeaker drive units unless they are fitted with horns or similar diffusers. The BBC’s view was that it’s the on-axis response that’s paramount. This has merit, providing the listening environment can absorb the off-axis sound that is splashed onto the sidewalls. In the domestic environment often the sidewalls are untreated, even though there may be snug carpet on the floor. We’re back to the issue of careful selection of crossover frequency and contouring the fade-out and fade-in of drive units so that off axis, where they are becoming beamy due to their diameter relative to the frequencies they are reproducing, the transition is smooth.

You still use a relatively thin-walled enclosure with lots of bracing for support and stability, but no heroic measures, so far as I can tell, to dampen resonances as such with the use of synthetic materials or super-rigid construction.
Cabinets: We do indeed live in a world where visual impressions seem to count for so much. The physics of panels—those forming a loudspeaker cabinet—say very clearly that thickness and stiffness do not guarantee low sonic contribution. Indeed, rigid panels can move the latent resonances away from the bass region and up into the midrange where they are energized by the music and can sound extremely objectionable. After extensive research the BBC concluded that a relatively thin-walled but sturdily braced cabinet could be steered by the application of damping into a state of relative inertness, in a way that no thick panels could be. This gives the thin-walled cabinet designer a whole armory of tricks for better sound.

One of the arguments you make in Harbeth literature is that exotic parts and wire are not necessary for state-of-the-art performance, merely parts and wire of requisite specification that will be reliable under dynamic conditions and for a very long time. I’m sure you know that many, perhaps most, Harbeth users ignore this when it comes to selection of speaker cable (and interconnects).
During my teenage years, music was my escape, and my interest in radio and broadcasting led to an involvement with the local BBC radio station. It was then that I was introduced to the BBC monitor. What impressed me was the pragmatism of the BBC designs, a total focus on simplicity, cost-effectiveness, and real, solid, honest engineering. I’m well aware that consumers can select whatever they like, but I worry when the hardware becomes more important than the music and seems to trap consumers in a cycle of dissatisfaction. Having said that, I’m acutely aware that it’s the ongoing sale of accessories to consumers, who rarely change their core system components, that helps keeps the audio dealer in business. I’d just make a plea for common sense. If the consumer has the interest and the cash to invest in exotic audio accessories, do so. But do so without feeling compelled to apply what may be pseudo-science.

The 40.2 is among the first Harbeths to use drivers made from Radial2, a new formulation of your proprietary RADIAL material. What’s been changed?
A small few-percent adjustment of the ratio of the key elements in the diaphragm polymeric compound.

What do you say to those audiophiles who ask, “Well, if you had no constraints as regards price or size, what would an all-out, no-holds-barred Alan Shaw speaker look and sound like?”
Looking back I can see how fortuitous it was that the BBC control rooms are approximately the same size as a typical British living room. Had the BBC control room been three or four times the size of the home listening room, the magic simply wouldn’t have been translatable. But your question is about my so-called “Magnum Opus.” Assuming I had the time to develop a speaker for myself, with no need to be concerned about commercialization, what would it look like? Actually, now that I have dictated that sentence I can’t actually answer my own question. What I do know is that the speaker would unquestionably sound like a Harbeth of today, but I can’t decide what to sketch on that blank sheet of paper when it comes to its physical configuration. Perhaps things will become much clearer after a couple of pints—they usually do!

Thanks Yogiboy! That interview really exposes Alan’s critical realization and understanding about transducer material. Alan really is quite exceptional. I have never owned a Harbeth but I have listened extensively and hear exactly why everyone raves about them. Great unique speaker designs that refuse to follow the trendy exotic rigid lightweight cones espoused by the rest of the industry. Harbeth sales keep growing because Harbeth do much more right than most everyone else’s latest and greatest diamonds and metals.
@shadorne

Harbeth and ATC are not the only manufacturers using "intrinsiaclly" damped cones. Far from it. It's called plastic.

Stirling Broadcast, Spendor, Graham, and Falcon are just a few that used damped cones in addition to lossy, thin-walled cabinets. As for other manufacturers using damped drivers, that’s a very long list.
Thanks Helomech. I suspected there were a few more. Yes plastic/polypropylene is intrinsically damped and was popular in the 80’swith Mission. Harbeth mid range cones are even more damped - so it has been taken further than the old Rogers Ls3/5 design. All those you happen to mention are UK BBC style. I have heard of Spendor but not the others. I still think I am correct that the vast majority of designers have gone in the direction of lightweight rigid material over the last 30 years - ceramic or metal etc. as it is cheaper and it uses smaller drive units (more efficient)

I guess Harbeth is quite a bit less unique than I stated. Maybe it is a UK thing as these styles of heavily damped mid range design aren’t often found in North America.
There are indeed quite a few more. Harbeth, however, are unique in that with a government research grant they developed a special polymer (which they call Radial) that was optimed for this application. Others mostly used ordinary polypropylene. There is quite a bit of documentation about this on the Harbeth website.
@shadorne 

I suppose it really depends on what one considers "intrinsically damped." One could argue that only plastics fall into that category, but a case can be made for some alloys and composites as well.

The issue I have with Harbeth is they dominate the BBC market on somewhat false pretenses, as though they're the only ones who know how to make a stiff plastic driver or lossy cabinet. Don't get me wrong. I think they're good speakers and I'd recommend an audition to anyone shopping for a BBC inspired monitor. However,  the others I listed are equally competitive and worthy of a listen. Unfortunately for them, they lack the marketing skills of AS.
@willemj

No, the others don’t just use generic polypropylene.
Sorry, that is what I thought. Anyway, the Harbeth material is a proprietary compound to have just the right mix of properties. See here for some of their explanation: http://www.harbeth.co.uk/usergroup/forum/the-science-of-audio/213-harbeth-radial-v-other-cone-materi...
Here is a close up of Radial mid cone

http://www.harbeth.co.uk/usergroup/forum/the-science-of-audio/1119-fast-and-slow-speakers-should-fas...

Here is is a close up of ATC mid dome

http://www.stereomojo.com/ATC%20SCM50ASLT%20SPEAKER%20REVIEW/images/ATC-Soft-Mid.jpg

Harbeth is injection molded (entirely plastic). ATC apply the “dope material” to a woven fabric. Both have a similar look and feel. Both are hydrocarbon based plastics. Both are softer than polypropylene and cannot be used for large cones (not stiff enough). Specifically designed material just for the mid range. What ATC use is even softer (tacky to touch) but the fabric and dome shape provides the necessary rigidity.

I am not aware of any others using this type of material but after so many years of use I can only assume the cat is long out of the bag and others do use it.
My Vienna Acoustics uses xp3 , Ithink that is the term they use for their plastic derived driver cones. They use some form of that through out their line.