Low wattage tube amplifier combined with high sensitivity speakers

[Sebastien]

Hi Alan,

I'm actually contributing on a french forum specialized in tubes (history, manufacturing, identification, sound, etc.). Many contributers on this forum, like in audio in general, think that the nirvana of sound is obtain using a very low wattage tube amplifier (SET) with very high sensitivity (100 dB and more) and easy load speakers.

With a speaker designer's perspective, can you tell me your opinion on that practice? As far as I know, the BBC never took the road of high sensitivity speakers. Rational explanations will help me to better understand Harbeth's demarche.

Thank you,

Sébastien

{Moderator's comment: this sonic dead-end of high sensitivity and low mass has been covered here before. Links anyone?}

[GregD]

SET amplifers and high sensitivity horn speakers are - regardless of how nice some people think they sound in their homes - ALWAYS high in coloration. There are technical reasons for this. Because of this they would be useless as monitors, which need to be highly accurate and neutral. This is why the BBC never designed any and Harbeth do not make any.

I think I remember reading that the word "coloration" puts a shiver down Alan's spine!

[Sebastien]

...{Moderator's comment: this sonic dead-end of high sensitivity and low mass has been covered here before. Links anyone?}

Hi Mod,

Yes I know that it was already touched in the past but some questions remain. Here's from Alan Shaw, sorry if I don't have the link, it was from an email exchange with friends. For sure, it can be found somewhere on the HUG. From Alan:

"...I suspect that whilst the sensitivity of Harbeth speakers remained static in the mid 80dB's over the past 30 years, I guess that the average sensitivity of speakers has nudged up. However, to achieve a smooth, well balanced sound there probably *has to be* a significant proportion of moving mass in the (heavy) rubber surround, accounting for about 40% of the total. Reduce the rubber mass, and yes the efficiency increases, but the sound quality degrades. Since watts are cheap, I think sensitivity is the least important design goal - it's fine in the 80dB's because even modest amplifiers will generate a really loud sound.

Of all the design objectives that the BBC-speaker designers of yesteryear or me now, sensitivity is the least important. Everything else is so much more relevant to great sound..."

I'm still wondering why "the sound quality degrades" with more sensitive speakers. Plus, is it only the "moving mass" of rubber that counts in a sensitive speaker?

All in all, I'm looking for more argument on the side of "low sensitivity" speakers. It looks like low sensitivity start below 90 dB.

Sébastien

[A.S.]

...From Alan:

"...I suspect that whilst the sensitivity of Harbeth speakers remained static in the mid 80dB's over the past 30 years, I guess that the average sensitivity of speakers has nudged up. However, to achieve a smooth, well balanced sound there probably *has to be* a significant proportion of moving mass in the (heavy) rubber surround, accounting for about 40% of the total. Reduce the rubber mass, and yes the efficiency increases, but the sound quality degrades. Since watts are cheap, I think sensitivity is the least important design goal - it's fine in the 80dB's because even modest amplifiers will generate a really loud sound..."

I'm still wondering why "the sound quality degrades" with more sensitive speakers. Plus, is it only the "moving mass" of rubber that counts in a sensitive speaker?

All in all, I'm looking for more argument on the side of "low sensitivity" speakers. It looks like low sensitivity start below 90 dB. Sébastien

There really isn't anything to add to my reported statement apart from this: be aware that a drive unit is nothing but a lump of individual parts (cone, surround, dust cap, glues, wire etc.) that are accelerated backwards and forwards by the coil/magnet thousands of times every second. These individual parts do not 'want' to co-exist as a whole entity: they were quite 'happy' to sit on a bench as a pile of parts. They are not natural partners. But we have forced them to come together into what we call a drive unit. If we are very lucky (at the design stage and in production) these parts will behave as one contiguous whole up to a certain mid or upper frequency when mechanical chaos takes over: each component starts to rebel and even though seemingly strongly glued to adjacent parts, components start to sing or buzz on their own. And that's called mechanical resonance.

Common sense tells you that a) it is only the glue that keeps them relatively but reluctantly linked together and b) the heavy parts have inertia which inhibits their ability to move (which we call damping), and the lighter parts will accelerate much more readily. But .... and this is the core problem .... when the music commands the cone to cease moving or reverse direction, the parts with less inherent damping are not able to come to a dead stop. They will bounce backwards and forwards just as a glass marble does on a hard surface.. And we can see that demonstrated perfectly in the other thread running here about driver resonances. What we see at the top end of the driver is that although these attractive units appear to be solid, indivisible and one whole, above a critical frequency they cease to be - the result is that the frequency response 'takes-off' as some parts sing extremely efficiently. Do we at Harbeth want our drive units to whistle along with the music? I don't think so.

That's the long and short of it. If you want the drive units to generate tones which are not actually in the music, then find ways of stripping out damping from your drive unit. Yes, the lower weight will increase the efficiency perhaps by several decibels. And the resulting atonality may be attractive on music, but on speech the reduced damping will be very obvious. Can't add anything to this.

[Sebastien]

Hi Alan,

Thanks for your answer. It is more detailed than what I have already read on the HUG. So, BBC designers chose to go with a construction that offers a balance between a drive unit that can starts and stops quickly. The best of two world. In the end, this offers less coloration.

About the glue, is it such a critical aspect that in the design process you tested different kinds and the speakers sound or react differently depending on the one you used?

Sébastien

[Gan CK]

I couldn't agree more with all that has been mentioned above with regards to high efficiency spks. That's where the irony is. Many have said that low powered SET (Single Ended Triode) amps are inherently high on musicality, tonal accuracy & tonal purity but their mandatory marriage with high efficiency spks defeats the very purpose of the design intention of such amps.

I for one can never comprehend why so many people can ever like such a combination. Any decent British integrated driving any Harbeth will yield much better results in terms of musicality & tonal purity. Period.

[A.S.]

... So, BBC designers chose to go with a construction that offers a balance between a drive unit that can starts and stops quickly. The best of two world. In the end, this offers less coloration.

About the glue, is it such a critical aspect that in the design process you tested different kinds and the speakers sound or react differently depending on the one you used? Sébastien

All mechanical design is about compromise. You can build a fast car (an Aston) or one that is extremely economical (a Micra), or has a super-smooth ride (a Bentley). But you cannot design a car that is simultaneously fast and economical and smooth. At best, as a designer you can aim to satisfy two of the three objectives. A better all-round plan may be to aim to balance the three objectives into a working engineering compromise providing that you can find customers who seek that particular compromise. That's what inspired the BBC monitor speaker: regardless of whether it was used to record rock music, or drama, or jazz or classical music, it performed well within it's entirely adequate power capabilities. But if the user needs head-banging loudness and extreme durability under stress, then as we know, the BBC monitor was never engineered for that workload, and the user would be best advised to look elsewhere.

About glues .... this is a really interesting and important subject.

When we at Harbeth started making drive units some fifteen years ago, we used 'soft' glues like Uhu, Bostik or similar. In fact, the BBC's pioneering drive unit design work (in the 1960s) would surely have used such adhesives. Considering that the drive unit is very much 'alive' and is not a rigid, static structure, you can imagine that different elements of the assembly are under vibrational stress. The problems are in the interface areas where one part abuts another: for example, the edge of the cone bonding onto the rubber surround. You can imagine that this joint is under real stress as the drive unit moves in and out at high power levels. If you look at old BBC monitors made 20+ years ago, you will see the tell-tale problem of 'glue creep' where the surround has slowly pulled away from the cone leaving a horrible brown ring around the cone edge. Eventually the surround will completely separate from the cone and that will be the end of the driver's life.

It is almost certainly to be expected that changing glues will influence the measurable response and/or the sound to a small degree. You may be aware that during the early/mid 2000s we made many thousands of drive units for KEF UK from kits of components that they free-issued to us on the same production line that we made our own Harbeth drivers. Included in this technology transfer was the most superb documentation covering in minute detail (for their own internal training) how to make these drivers, and what glues to use and where. I'm not sure that we made any money out of this venture, but what we learned was priceless: it opened our eyes up to a whole world of specialist adhesives. As a small company, Harbeth wasn't able to engage with the major international adhesives suppliers, but using the relationship with KEF we were very soon in direct contact with the majors, including the Loctite parent company, Henkel. And, they have the laboratory facilities to examine every glue joint in a drive unit and recommend the best adhesive. And in their view the optimum glue joint means the longest lasting, most secure bond regardless of the cost of the adhesive. If a standard product is not quite right, they'll compound one specifically for your needs. The downside is that one speciality glue we buy from them costs GBP500 for one litre!

So, whilst we still use some contact adhesives, we have moved towards superglues over the years. That way we can be very confident that providing you don't overload your woofer, it should outlast you.

BTW: there are two downsides of using superglue to assemble drive units. These hard glues are completely unforgiving: if we make a tiny assembly error on a RADIAL woofer which we could easily rework, we can't. We can't separate the permanent glue bonds no matter how hard we pull. We have to throw the whole drive unit away. Also, this means that we can't (as some speaker companies do) supply recone-kits because you too won't be able to disassemble the driver.

[GregD]

That's a fascinating story about glues, Alan. I never cease to be amazed by how much effort Harbeth go to in order to make the best loudspeaker possible. You really go the extra mile.

As I look across at my P3ESR now - thinking of all the varied scientific and technological expertise needed to produce the many parts - I feel both proud and humbled to own them.

Glue at £500 per Litre? I'd like to know which Champagne you serve at the Harbeth Christmas party!

{Moderator's comment: Just like champagne we keep it in a fridge to prolong its shelf life.}

[EricW]

Glue at £500 per Litre? I'd like to know which Champagne you serve at the Harbeth Christmas party!

I bet a litre of superglue goes a lot further than a litre of champagne (especially at Christmas!).

But seriously, knowing the price of the glue Harbeth uses sheds some additional light for me on the whole "wire" issue (which I'm not trying to revive, just bring a perspective to).

Can anyone seriously doubt that a company that will spend £500 per litre on the world's most technically perfect glue would not hesitate to spend a bit more on internal wiring if there were even the slightest performance advantage involved?

[jplaurel]

£500 is not out of line for specialist cyanoacrylate adhesives. Some of the glues we use to construct competition-grade composite model gliders are in that price range.

Glue at £500 per Litre? I'd like to know which Champagne you serve at the Harbeth Christmas party!

{Moderator's comment: Just like champagne we keep it in a fridge to prolong its shelf life.}