Trends in speaker design. What to listen out for?

[A.S.]

It's true. Once you've heard it, nothing else sounds as clean. Every other speaker (I've heard) seems either a bit fuzzy or a bit over-aggressive by comparison, even the very good ones.

There's one ... speaker company that seems to be well-regarded (I haven't heard their speakers, but they're well-reviewed) that seems to follow a design principle of crossing over to the tweeter at a very low frequency (seems usually to be between 1.6 and 1.8 kHz, depending on the model). Presumably the intent is also to have the woofer extend as little as possible into even the upper midrange, presumably on the theory that the tweeter will have faster and better response and reveal more detail?

That could well be in the designer's mind. But, as with most technical issues involving loudspeakers it is 'one step forwards and 1.2 steps backwards': really not advised.

A tweeter is like a car with an F1 racing suspension: great when the surface is mirror smooth, but hopeless on a bumpy road. The reason is that unlike the bass/midrange cone which has a generous (rubber) surround which permits the cone to move in and out several mm (the equivalent of shock absorbers on a bumpy road) the tweeter has no real surround. It is virtually a rigid diaphragm that perhaps at best has a tiny fraction of a mm of permitted forwards/backwards movement - perhaps only the thickness of a sheet of paper. That's absolutely fine when the crossover is nice and high (say, > 3kHz) because there is virtually no energy in music at those frequencies (see another thread). But we know that the movement of any diaphragm is directly linked to the frequency that it is reproducing, and a small change in frequency can give a big change in necessary amplitude. So, in the real world, every tweeter has a naturally optimum crossover frequency which is non-intuitive: you have to find it by hours/weeks of experimentation and listening.

What about distortion? That, like amplitude, is directly correlated with the amount of forwards/backwards movement. So again, a small lowering of crossover frequency results in a potentially large increase in distortion. How does a tweeter that is being asked to operate below its naturally optimum crossover frequency actually sound? - i.e. a unit designed for 3.5kHz being operated at a ridiculously low 1.75kHz (one octave below?). Easy to demonstrate that ....

Loading the player ...

Clip A8: sound of tweeter at different crossover frequencies

In the above clip ...
0:00 - 07:00 normal speech, full bandwidth
07:00 - 18:00 - speech with bass/midrange driver disconnected - we hear the tweeter via 3.5kHz crossover (i.e. remove biwire links on Harbeth speaker)
18.00 > - speech with bass/midrange driver disconnected; same tweeter but crossed-over at 1.8kHz (never used by Harbeth)

As you can hear, when a tweeter is pushed too low it barks. There is no solution to this other than raising the crossover frequency to the tweeter's 'natural' crossover point. If, as you astutely point out, the designer is trying to mask the coloration at the top end of the bass/mid driver by handing that frequency band over to the tweeter rather early, he's traded one material coloration problem for perhaps an even more serious mechanical coloration problem: nothing is gained. This new coloration may be somewhat covered by music (esp. pop music which is multi-processed and devoid of reality) but on speech - example above - it is very obvious anywhere in the listening room. The human voice box doesn't sound like a strained mechanical resonator.

The only escape from this is to solve the bass/midrange driver's material coloration issue by developing a better material and then working that unit right up to the point (around 3kHz) where the tweeter can comfortably take over without straining. Oh, and it goes without saying that the fragile tweeter's life expectancy is directly linked to how hard it is working; the lower crossover frequency will inevitably shorten its life. These are the penalties of using bog-standard, mass produced, outsourced woofers purchased for $3 in so-called high fidelity loudspeakers.

Hope this makes sense.

[EricW]

Hope this makes sense.

Absolutely. That too-low crossed-over tweeter sounds dreadful. You say "bark"; I say "honk" might be another possible descriptor. Either way, it doesn't sound good: the distortion is extremely easy to hear.

[STHLS5]

I've kept on playing all the clips over and over again. Something, I generally avoid for the fear of being turned over by another person’s view no matter how benign it maybe. And where did it lead me too?

Well…, I started last night with the so called reference SACD which I said sounded fabulous in another thread and by the time I reached the 3rd track, I asked myself why am I feeling fatigue with the vocal. I went back to listen to Clip P and Clip S and then Clip Z and A1. Now I suspect that the vocal in the reference SACD resembles the “clarity” of Clip Z [boosted presence] that is usually associated as high fidelity recording and it is not pleasant even though was captivating in the beginning.

Instead of looking for trends in speakers design I am now looking for trends in recording equalization. Perhaps, this is inconsequential but this thread is useful in its own way for my needs. Thanks.

ST

{Moderator's Comment: look no further than the marketing department's input. Ok? There you will find your answers. All of them. It' about making money. Nothing to do with fidelity.}

[BAS-H]

This might have just become the most useful thread on the HUG. Expertly informative. One day I too will enjoy the sonic signature of Radial.

[A.S.]

No matter how clean and life-like the middle and upper frequencies are reproduced in the listening room, it is the bass range which underpins our immersion in music. If the bass is not of sufficient quantity and quality, and not adequately blended with the mid/top, then the illusion of 'being there' cannot be achieved.

Unfortunately, reproducing 'bass' in a normal domestic listening room is a real challenge. Not only is the room dimensionally far smaller than the wavelengths of the low (but vital) bass notes* but the negligible absorption at those low (and middle) frequencies encourages the low notes to bounce around the room, unattenuated, and long after the original note has ceased. The fact that we can enjoy a somewhat life-like experience listening at home is a tribute not to high fidelity equipment designers, but to the poor discrimination of the human ear especially in the lower registers.

Few - almost no - audio enthusiasts will treat their rooms to dampen-down the prominent reflections, and from that we must deduce that almost every listener can either ignore of somehow hear-through the chronic 'distortion' that the normal room imposes on the sound generated. It is just as well, because if that were not the case, high fidelity listening at home would be impossible except in a millionaires fully treated HT room.

Speaker designers are obliged to design high fidelity speakers as if they were to be used in a controlled, damped acoustic simply because there isn't a standard domestic listening room to design for. And even if such a target room is specified by bureaucrats in Brussels down to the last nail and screw, in the real world there would be no exact examples of it - it would be just another manifestation of the issue of designing a speaker in the anechoic chamber but to be used at home.

So how should the speaker designer approach this challenge? The answer is, remove the room from the equation. That solves lots of issues. And the technology has been available for generations. It's called .... headphones! OK, so it's not a vote winner, but that really is the truth .... if we really want to experience what the microphones captured we have to completely null-out our room's destructive contribution and really good headphones (example, Stax) are the only way to achieve that. But we know from the commercial ups and downs of the headphone business that there is substantial consumer resistance to serious listening over headphones. That's the reality; given that audiophiles are wedded to the idea of (two) loudspeakers in the room, how should those speakers generate low frequencies that are as well as possible integrated with the mid/top sound?

Speaker solution to bass output No. 1: the sealed box

Surprisingly the sealed box was not invented until the 50s. Before that, loudspeaker drive units were either mounted on a large plate (a baffle board) or mounted in a baffle with sides/top but open at the back. The issue with such open baffle arrangements is that due to the forwards/backward motion of the cone producing low notes, the sound bleeds from the back to the front and the resulting full or partial cancellation gives a weak bass. That pulls against our basic requirement that a solid bass is necessary for realism. Then someone suggested fitting a back panel to the box to seal it and - shock, horror! - dramatically reducing the size of the box .... and the modern loudspeaker was born. This coincided with a proper electro-mechanical study of the drive unit's parameters (magnet strength, cone weight, surface area etc.) and led to an understanding of how to tune the drive unit to the box (or vice versa).

The sealed box designer can swing the design from a peaky one-note bass response to a weak bass output by adjusting the numerous variables of the drive unit or the box/box stuffing or somewhere in the middle: it depends upon his design brief. Typical design simulator tool which saves 'cut and try' by predicting how the low end will behave is here.

Speaker solution to bass output No. 2: the vented (ported) box

Who wouldn't want something for nothing? The vented box can theoretically extend the bass down to a lower frequency and simultaneously relieve the bass unit from working so hard. Sounds like a win-win. Well, almost.

Blow across a bottle (demonstrated in this video here) and the plug of air in the neck of the bottle bounces on the larger lump of air deeper inside the bottle and we hear a note. That's exactly how ported speakers work: the air volume in the port bounces on the much larger air inside the box, and if these are proportioned somehow, the tuned note and the magnification of the note can be adjusted by the designer. The designer now has added some additional variables to those he had with the sealed box, so his armoury of technical tricks has been enhanced. However, the frequency range where the vent is normally tuned (say, 40Hz) can coincide with domestic room standing-wave issues, and when it does the designer is faced with a variable over which he has no control at all. So it is recommended that speakers should be chosen with the target room in mind. Or when there is an unwelcome coincidence between the port tune frequency and the room, either more the speakers around until that note is tamed or plug the port to impede air flow so that the port contributes little or no output .... which converts the speaker system to a sealed box with a dryer bass.

The same software simulators are available for vented speakers as sealed ones. I have been using software simulators to predict how woofers behave in sealed/vented boxes since about 1990. I can assert that if the speaker designer is very careful in gathering data and feeding it to the simulator, bass performance can be predicted to better than +/-1dB error below about 200Hz. This means that there is no need whatsoever to actually cut-and-try speaker enclosures: the bass end of the design can be perfected on the beach at San Tropez with a laptop and amply supply of Pino Colada (dream on!). But you do have to be really very alert - there is a natural human temptation to believe in the simulator's fancy print-outs. It's essential to build regular reality-checks into the software model, so that variable are checked against reality. Or you can be led far up the garden path ....


* To 'hold' a bass note at the intended loudness, the room has to be dimensionally big enough for the wave to sweep through its entire 360 degree cycle. We know that sound travels at about 340m/s in air. Using this on line calculator here, if you enter 40Hz on the left side (considered to be the lowest important audio note) you can see how big the room must be to 'hold' that note. How many of use listen in rooms that big?

[Pencey]

I find this thread one of the most informative and interesting I've read on the Harbeth forum. More please!

[A.S.]

Speaker solution to bass output No. 3: the pipe labyrinth

The idea here is that the woofer is mounted at one end of a tunnel or pipe. The far end is either open to the room or sealed. The basic concept goes something like 'since we know that organ pipes are associated with generating deep bass, if we replicate the organ pipe concept and mount the woofer in the pipe we can produce deep bass'. On the face of it this sounds plausible - but upon close study, the analogy fails. There are a number of serious and insurmountable problems to do with the physics of pipes, not specifically woofers mounted in pipes.

First, real pipe organs have one pipe for every note, but the speaker-pipe is attempting to use one 'average' pipe length to cover a wider range of notes. If the designer could predict with certainty what note in the music listeners found most alluring he could optimise the design (i.e. the pipe length) to reproduce that note really well. But as we know, one of the least attractive characteristics of loudspeakers in real rooms is the horror of one-note-bass, so even that isn't an option. Secondly - and this is really very interesting although difficult to visualise: the problem of pipe length and folding the pipe. As you can see from these pictures here, organ pipes are long and straight, and hence need to be mounted (vertically) in very large buildings and the overall length is related to the single note the pipe is optimised to play. The real killer property is that the pipe must be straight. Straight means just that: millimetre perfect from top to bottom - this straightness parameter is absolutely crucial. If an organ pipe is damaged - for example, someone accidentally puts a dent in the pipe, the tonality of the reproduced note will change a little or a lot depending upon how deep the bruise is, where it is along the length and what shape it is. This is just one reason that church organ pipes are well away from little fingers. You do not see organ pipes that bend around corners or fold on themselves (like a tuba) - they must be straight to have the necessary purity of tone. That's the long way of saying 'fundamental and harmonic evenness'.

Few domestic rooms are big enough to mount an organ pipe of the length necessary to generate deep bass (that is, the single note bass that a single pipe would generate) so the labyrinth-speaker designer cunningly resorts to folding the pipe, perhaps several times. Such a pipe then occupies less height, and some more depth. Since the problems is lack of height in the room, and we have plenty of depth, that sounds like a good approach. However, this is the start of the insurmountable problems. Let's imagine that the top end of this pipe has the drive unit mounted in it at ear height, and the bottom end is down near the floor. The relatively small cabinet is designed so that, just like a car exhaust silencer box, slats of wood are arranged inside the box that guide the rear sound from the woofer along a definite tunnel - a wooden pipe in effect. Since the cabinet is probably taller than wider, the slats will be carefully arranged suspended from the top and bottom such that the air passageway inside is as tightly folded on itself as possible, minimising the cabinet size and volume. The rear sound wave from the woofer will typically be led up and down the cabinet as it works along the pipe. So far, so good. Let's remove the woofer so we can just have a look at the acoustics and obtain a starter's pistol, used on the sports field and a couple of firing cartridges. They'll make a nice loud impulse.

Let's place into the void a few inches from the gun (woofer has been removed) a microphone: just a cheap one, connected to our audio measuring equipment and pull the gun's trigger. What does the microphone hear? The first sound to hit the microphone is that of the starting pistol which is adjacent to it and very loud. We're not interested in that sound at all: that's what we expected to record. What then? Well, something really non-intuitive happens .... a little time passes and we see on our screen a small blip which looks like a miniature version of the original impulse ... then a little later another one, slightly smaller ... and another one ... and another and another. And we scratch our heads and wonder where that series of mini-impulses have come from. We conclude that we must have made a mistake, so with our second cartridge, we reposition the mic, check the recording gear and fire the second shot. Same result: the initial impulse followed by a series of mini blips which look like ... a series of precisely timed echoes. How very unexpected. What's going on?

What we have discovered is the same issue as bedevils the organist who mandates perfectly straight, unbruised pipes with a perfectly smooth inner surfaces. By taking a straight pipe and making numerous folds in it (with the slats) we've assumed that sound wave in the pipe sees the same volume of air (which it does) but we have not appreciated that at every change of direction or cross-sectional area of the pipe, every corner, every nail or screw, a proportion of the air passing the 'disturbance' in the pipe (example: a 90 degree bend) won't play ball: it about turns and heads back to the source. And that's what our microphone hears. So knowing that, we can anticipate that for every fold in the pipe the mic at the source end will receive a pressure pulse, which will diminish in intensity (probably) the deeper its source along the folded pipe. How can we completely eliminate these unwanted reflections? We can't: acousticians have been searched for that answer for over fifty years. What we can do is to try and fool the sound wave in the pipe that it's not really going around lots of transitions (bends) - oh no - it's just sliding along a perfectly straight, smooth metal pipe. And how would we do that? We'd have to radius every bend to that there were no knife-edges along the pipe (sharp edges = strong reflections). That's the easy bit. Then, appreciating that even radiused bends are reflective, we'd have to try and fill the pipe with some sort of damping material to reduce the strength of the reflections. Serious problem! Whilst the damping material would indeed reduce the reflections, it can't be arranged like a one-way valve: air passage friction in the pipe would increase in both directions which would lead to a dramatic reduction in efficiency.

We are now in a real engineering dilemma. Our need to reduce the echoes from the pipe means we have to damp the pipe. But when we damp the pipe its desired efficiency diminishes. So it produces less bass .... less bass than a reflex enclosure. There is no solution to the classic efficiency v. damping conundrum: the designer has to make a choice between echoes in the pipe (and their coloration) or good bass output in a limited 'average' sort of way. Or maybe his marketing department will make the decision for him. But this basic engineering problem is insoluble.

DIY article which shows evidence of reflections manifest in the frequency response of the speaker (bottom graph, mouth output, series of ripples)

More here showing variations on internal structure - page 100.

Empirical adjustment of pipe to "reduce hangover" here

Famous 1970s UK pipe design: note critical positioning of damping material in pipe

The 'magical' Dr. Bailey's Long Haired Wool was a 70s solution to pipe damping. Sadly, no longer available .... read here.

We have not looked at another serious issue, that in domestic application, the pipe is far too short to properly reproduce deep bass (as you can see for the size of church organ pipes needed for real bass) and that the output from the pipe is very wide-band. Bad enough in a tall-boy speaker system, but in a really small pipe system a disaster. The consequence - and this is probably the most serious shortcoming of the pipe technology - the output from the mouth of the pipe and the woofer are wholly or partly out of phase in the midrange. This means that the sound from the mouth and the drive unit are fighting each other, and the result is a leanness of sound that gives the pipe speaker a typically 'gutless' presentation and absence of proper instrumental warmth. Cellos and piano in particular sound unnaturally thin. This suck-out is self-evident from a frequency response measurement of the speaker with the microphone receiving sound from the end of the pipe and the woofer simultaneously as it would in the normal listening position. Solution: completely stuff the pipe with absorbing material! Seal the end off! (Then you have a closed box ....).

The masters of the pipe speaker were surely the British IMF company (long gone - why?). The frequency response curves from an old brochure show an extremely well integrated bass/mid/top on axis performance doubly so because the drive units were spaced-out on a large baffle. However, close examination of the curves in the PDF (below) indicates what looks like the tell-tale dip in the low bass where the pipe and the drive unit are partially out of phase - the very region which defines the subjective 'weight'.

Pipe speaker software simulators: Sealed and vented box speakers benefited from a rigorous engineering analysis in the 1970s (The Thiele-Small parameters) and that led to the first generation of software simulators able to predict how a given driver would behave in a sealed or vented box. Now, there are numerous fifth generation sealed/vented box simulators available with remarkable accuracy. Unfortunately, the pipe speaker has not had the same scientific attention and we are still at the first generation of software modelling. Whether they can accurately predict the behaviour of a pipe speaker I cannot say. Here is an example of a pipe speaker software simulator from Japan. You can see from Fig.1 and 2 that the simulator predicts that the result of mounting a drive unit in a pipe is not the smooth frequency response we associate with sealed or vented boxes, but (as explained above) a series of sharp peaks and troughs due to echoes from within the pipe.

Original Wireless World labyrinth speaker DIY design (c. 1973) page one attached.

>

[KT88]

Very enlightening, Alan, thank you.

A question: does Harbeth supply foam bass plugs for the speaker ports, and if not, what would you recommend? I've wanted to try this with my Monitor 30's.

Thanks!

Bob LaBarca
USA

[Spindrift]

As fas as I know, Harbeth does not supply port-bungs but one can easily make them out of a foam paint-roller.

This leaves a small hole in the centre where the centre-pin of the paint roller sits but that aperture is in fact beneficial, as it leaves a minimal opening for the port to 'breathe' through.

Lining the port's inside with a 1 cm thick piece of foam (forming a foam cylinder inside the port) is a good, benign starting point. If that does not dampen things sufficiently, the foam paint roller is the next step upwards in damping. From the roller cylinder, one could slice off discs of different depth and according damping values.

{Moderator's comment: brilliant idea!}

[A.S.]

Other speaker solutions: electrostatic speakers

The last two opportunities I had to listen to a electrostatic speakers were on my Far East tour a couple of years ago and before that, at an exhibition somewhere overseas years ago. It's important to appreciate that the electrostatic speaker first appeared about one hundred years ago at about the same time as the first moving coils speakers, so the technology is far from new. Indeed, N.W. McLauchlan's seminal technical book Loudspeakers published in 1934 mentions the type.

I must admit to some bias. The very first high-fidelity stereo speaker system I ever heard (at about 11 years old) was a pair of full-range electrostatics playing Arrival of the Queen of Sheba (Marriner, Decca) and this was a fantastic experience for its openness - I still recall it 40+ years on. However, my subsequent encounters with electrostatics have not been quite so encouraging. In both instances I mentioned, switching over the SHL5s (they happened to be handy) seemed to give a much better overall balance, a more true to life presentation and crucially, more clarity. This was highly counter-intuitive to me as a cone speaker designer. One could reasonably expect that a weightless diaphragm should outperform a cone loudspeaker hands-down in every area, except perhaps the deep bass, because the promoted advantage of the 'stat is the low mass and hence low intertia - its diaphragm should be able to accelerate and stop in a tiny fraction of a second, theoretically vastly outperfoming the heavy, comparatively sluggish cone speaker's diaphragms.

As you will know from reading my posts here, with the exception of our RADIAL cones which I strongly advocate, I do not overtly push or plug Harbeth speakers. What matters is your opinion after an audition, not mine, so even in this post, you must set aside my views and go listen for yourself. Your taste, your music, your room, your expectations may give you a very different perception about what is 'right', and I absolutely encourage and respect that. So, that said, let's look at the electrostatic speaker from the ground up.

Believe it or not, you could make an electrostatic speaker on your kitchen table. When I was about 16 I made a pair of electrostatic headphones (scan attached) - a frightening experience resulted on the test-run when my long hair was dragged into the highly charged diaphragms - and in the few minutes they worked before shorting out irreparably, I enjoyed The Carpenters in fantastic stereo. I must have been mad: I could easily have electrocuted myself such was my curiosity to chase the high fidelity dream!

So - how to make DIY electrostatic headphones or speakers. Very simple in concept. You'll need a roll of wide cling film (supermarket), some graphite powder (art store), cotton wool balls (wife/girlfriend, but how to steal 100 unnoticed?), a drop or two of washing up liquid and clean water. Step 1: thoroughly clean the kitchen table. Stretch about 1m or so of cling film onto the table, minimise air bubbles as best you can and be sure that the central area is as smooth as possible even if the edges are a bit crinkled. Step 2: mix a desert spoon of the fine graphite powder with an egg-cup of water (it won't disolve) and drop in one or two small drops of washing-up liquid - agitate. The graphite is now held in a soap/water mix. Step 3: Soak a couple of cotton balls in the graphite mix, take the phone off the hook and a day off work and proceed to rub the mixture into the cling film in small circles of perhaps 2 inches starting in one corner and working down to the far corner. It will take hours. Maybe days. Or maybe you'll be lucky. What you are trying to do is to rub graphite (a conductor of electricity) into free hooks in the molecular backbones of the cling film. They must not just lie on the surface, they must be embedded into the molecules or they'll jump out as soon as the music starts (due to stroing attractive forces) and you'll have to repeat the process properly.

Think you're finished? Wrong! When you think the film has taken an even light-greyness (or measure the resistance if you have a very, very high reading resistancer meter) peel the film off the table and .... turn it over and repeat the entire process If your arm isn't hurting at this stage it definitely will be when you've finished side two! At the end of this you have now created a diaphragm that can sustain (or hold) an electric charge evenly across its entire surface, front and back. Now the fun really starts .....

To be continued ...

P.S. This morning I went and bought a pair of recently serviced full range electrostatics to be in a better position to appreciate them. I'd forgotten during typing of the above post (here at the Old Barn) that I'd switched to the P3ESRs so now I've just switched back to the 'stats. I didn't realise just how engaging the P3ESRs are: this is the first time I've listened to hifi this year (it's 27 March).

>

[Spindrift]

Hi Alan,

it is nice to read about your DIY ESL experience - at a young age as well. Now if that isn't proof of your true commitment to designing and building the most honest acoustic transducers, I don't know what is!

I too built my own HiFi gear many, many years ago, from DIY tube-preamps (with MC phono stage - dead quiet!) to DIY tonearms & powersupplies but never loudspeakers. I was content with my QUAD ESL's and LS 3/5a's and why shouldn't I? Few speakers from the past were better at reproducing the human voice or an acoustic guitar than the venerable QUAD's and not many mini-monitors were able to throw as convincing a soundstage as the little Baby Beebs could.

Since then, other ESL's have come and gone (Martin-Logan's etc.) but the QUAD ESL63 was perhaps the one that will remain my all-time ESL favourite.

Still, it's P3ESR's that have been making sweet music in my listening room for over two years and last year, my oldest pair of original ESL's went away too, so no electrostatic remains here. I must admit to missing having a pair of original Peter Walker speakers in my home but one can't have it all -not with my income at least.

I look forward to hearing/reading about your experiences with your newly aquired pair of ESL's. May I ask which model/make they are?

{Moderator's comment: thanks for that! To remain as objective as possible, we'd prefer not to name the specific model except to say it is not brand new.}

[A.S.]

Making an electrostatic speaker (or headphones) continued ...

I described in post #70 how we make the diaphragm, using nothing more than cling film, graphite and some washing up liquid. It's the diaphragm itself which we will force into motion to trace the musical signal, and the diaphragm will move and push sound towards us.

Have you experienced the effect of combing your hair and finding that the comb statically attracts your hair ... or taking off a sweater or shirt and hearing or seeing a small static spark? Or walking across a certain type of floor with certain shoes on, and getting a small static shock from the door handle? All of those effects are exactly the process we need to get the diaphragm moving: a controlled attraction and repulsion between the thin, flexible diaphragm and two rigid, perforated plates, held extremely close to the diaphragm and either side of it. When I made my schoolboy electrostatic headphones, the making of the plates was by far the most difficult part and I never got it right through choice of the wrong materials (too bendy) an poor technique on my part.

In essence, we can think of the plates as two metallic sheets with as many holes drilled in them as possible. We need the holes to allow the sound wave generated by the cling film diaphragm to radiate away and towards us. But every hole weakens the plate and makes is more prone to bending. Keeping the plates absolutely parallel to the diaphragm over the entire surface is all-critical. If the gap between plate and diaphragm becomes greater or lesser - even for a mm or so - then big problems. To persuade the diaphragm to be attracted to (or repelled from) the plates we have to mix the incoming music signal with an very high, and potentially lethal permanent voltage: around 4000v. It's this voltage potential which causes the attraction and repulsion of the charged diaphragm (remember: that's why we rubbed graphite into the cling film - so that it would hold a charge) and that attraction and repulsion magnified over the entire surface causes a useful total amount of motion - and we hear sound. So far so good.

The problems are really those of the structure of the diaphragm and the plates. Electrical charge between the plates and diaphragm is weak, and unless we increase the charging voltage from 4kv to perhaps 10kv or more (which would be extremely dangerous) we are going to have to strike some engineering compromises as we do with all transducers. As I see it these are some of the issues ...

• If we bring the plates closer to the diaphragm we can increase the loudness
• If we bring the plates too close, we will limit the loudness because the diaphragm must not touch the plate
• We know that bass notes move the diaphragm more than high notes ....
• and this means that the primary loudness limitation will be in the bass ....

So now we appreciate that electrostatic speakers and bass loudness are not easily achieved, which is why the hybrid electrostatic (conventional bass unit, electrostatic mid/top) are a good work-around. If the electrostatic panel is designed for 'full range', it inevitably means that pop/rock music with a strong bass content cannot be played loud without the frightening reality of the diaphragm actually touching one of the the plates and there being an electrical short circuit of the high voltage supply. Very nasty, and depending upon how the diaphragm protection circuit is implemented (if there is one) the diaphragm could be ruptured at the point of contact with a small hole, or even a tear. This really does limit how loud you 'dare' play a pure full-range electrostatic speaker.

More later

[A.S.]

The importance of damping in mechanical systems and the difficulties of damping electrostatic systems ...

The electrostatic transducer has been around for about 100 years. It is not a new technology: the concept of forcing an almost weightless (and very low cost) diaphragm into motion appealed equally to our grandparents. So if the concept is so alluring, why has it just not caught the imagination of the public? What hidden problems exist that make the electrostatic speaker (or headphones) a niche market solution?

I mentioned above that the bass output is necessarily limited by the awkward and insurmountable complications of the very high polarising voltages and the small gaps between the moving diaphragm and the perforated conduction plates which push/pull the diaphragm and create sound. There are other issues, some not so obvious. When I started this thread - and before I actually purchased a pair of electrostatic speakers to remind myself of their character - I had considered the conventional loudspeaker with its woofer and tweeter as being at a theoretical disadvantage to the wide-range 'stat. I'm not so sure now. In fact, I now clearly appreciate why a well executed conventional speaker is really unbeatable - more on that in a minute.

Back in post #70 I mentioned my schoolboy efforts to make DIY electrostatic headphones. Trawling my archives, I've found a follow-up Wireless World article Electrostatic headphones (N. Pollock) published in 1979 (which I was unaware of at the time) with improvements in loudness output due to simpler/better construction and clearances techniques, and using solid state driving electronics. But most importantly, the article touches on what is perhaps the single most important aspect of the design of any mechanical (transducer) system: pickup, tone arm, microphone, loudspeaker or headphone: the ugly word .... damping. In short, all mechanical systems have a tendency to oscillate (we've seen that with the defective shock absorbers in post ...) at their natural frequency or frequencies which are related to how stiffly held the parts are together, and the weight of the parts. A heavy car body presses on the very stiff shock absorbing dampers and the result is a smooth ride. If the shock absorbers are defective and leaky, their stiffness diminishes, and we feel every bump: their damping effect has reduced.

What's the relevance to loudspeakers? Loudspeakers are resonant systems. The moving parts have weight, and they have stiffness. Just look at the bass unit and you can see that a semi-flexible rubber surround runs around the rim of the cone and permits the cone to move inwards and outwards. There we have the two components that form a mechanical resonator: the weight of the voice coil + cone (and surround) bouncing against the flexible, compliant rubber surround. There is not much we can do about the weight of cone etc. but if we're clever, we can manipulate the compliance of the surround to engineer the natural oscillation frequency of that moving combination to push it down into the low frequencies, where the effect is to augment the bass output of the speaker. We can also adjust the damping of the surround by chemical changes to the materials from which it is made, we can adjust the vulcanisation and curing process, we can introduce oily, viscous chemicals to permit the surrounds molecules to slide more easily over each other and be more elastic .... there are many, many variable the moving-coil speaker designer can tune. And, of course, no perfect solution: the skill is in deciding what compromises to make for the design objective to be achieved.

We've discussed the effect of the surround at very low frequencies (say, 50Hz) but the surround is a critical part of the overall sonic quality of the speaker all the way across the bass/mid driver's output. It is, after all, glued rigidly to the cone, and like it or not, will be damping the sound up to where the tweeter takes-over. The chemical composition of the surround is hyper-critical, and this is an area where UK made - UK engineered - surrounds win because we can sit with the chemists and discuss small changes to the chemical composition which they can prototype for the customer to evaluate by measuring and above all, careful listening. And yes, even a few % change in the materials make a marked change in sound quality in the middle and upper working frequencies*. So, optimum damping, by chemical means of the molecules of the surround can perfect a well designed cone, or if excessive, kill the life in the cone.

Even though the electrostatic transducer has the theoretical advantage of a super light-weight moving diaphragm, it should be obvious that the damping component is non-existent. The diaphragm has no rubber surround or equivalent. The cling film sheet is rigidly clamped between the electrodes. Indeed, were the electrostatic diaphragm to have (conceptually) a rubber surround, the weight of the rubber relative to the almost zero weight of the film would dominate by a factor of 1000 or so - the diaphragm would be so inert that no amount of safe electrical charge would set it into motion. And even if it did, as noted previously, the diaphragm and plates must be perfectly parallel to each other. If the diaphragm film had a flexible rubber surround, gravity would pull the diaphragm out of true and it would be impossible to centre the diaphragm precisely between the plates. So what? Well, the inability to add mass to the electrostatic diaphragm means that the designer cannot introduce mechanical damping (i.e. a shock absorber) into the electrostatic speaker as the moving coil speaker designer can. Why would he need to? Surely, the principal of the electrostatic is that the light diaphragm is completely driven by, and controlled by, the electric charge - end of story. Sadly, not completely true.

Just because the electrostatic diaphragm is made from almost weightless cling film does not means that it has zero mass. It definitely does have measurable mass. And it also has very high stiffness: you try stretching a piece of cling film and until you use considerable force, it has little 'give' which is the same thing as saying high stiffness. Now, that stiffness is a property that we definitely do want and do need: we are relying upon the low mass and the high stiffness to guarantee that the diaphragm is perfectly positioned and parallel to the plates. But the issue is this: unlike the speaker designer who can manipulate the stiffness of his bass/midrage surround, there is no surround in the electrostatic diaphragm - just the naked diaphragm itself - and that means that if there are frequencies at which the electrostatic diaphragm will self-oscillate due to it's mass and stiffness, they are likely to be much higher up the audio scale than the conventional moving coil speaker with it's heavy, damped cone/surround.

What does that imply? It implies that the conventional con-speaker designers ability to inherently self damp the oscillation of the moving air-generator (the cone) is not available to the electrostatic diaphragm designer. If he finds resonances (and inevitably he will) not only with they be higher up the audio range (and more audible) he cannot manipulate the diaphragm itself to introduce damping - he must rely on 'external' solutions. That is indicated in the attached PDF which shows the measurable sonic peaks (hence, highly colored sound) of the naked diaphragm (Fig 7a: note 10dB is about three times the sound pressure output) versus the diaphragm with an additional external damping layer of foam either side of the panel. Still the prominent fundamental 100Hz mass/stiffness peak, but at least the chronic peaks further up the band are damped.

So - whilst the electrostatic transducer is theoretically a great way of generating sound, the difficulties in taming its natural tendency to ring at certain frequencies due to little inherent self-damping is significant. Optimum damping is one of the most useful design variables. Some electrostatic speaker systems use a cloth sock to cover the entire diaphragm/electrode plate assembly, front and back to provide protection and dust filtering. Some enthusiasts recommend removing the sock for 'more transparency'. As we have seen, the sock is likely to offer a degree of damping to the diaphragm (just a cone speaker's grille covers and damps the vent). It would be highly ill advised to remove this sock, as the perception of increased clarity is most likely to be that of increased, undamped resonances: peakiness. The foam dampers in my STAX electrostatic headphones have perished, changing the sound.

A DIY test: go find a roll of cling film in your kitchen. Stretch it tight over a bowl. Gently heat it with a hair dryer to tension the film to simulate the manufacture of an electrostatic diaphragm. Let it cool. Now drum your fingers on the 'diaphragm' and you hear the characteristic cling film sonic 'twang'. If the film had more damping that character would be suppressed but as it has negligible damping, you hear coloration.

More later
>

[A.S.]

Conventional 'moving coil' sealed box or vented loudspeaker systems really are a remarkably versatile, simple and cost effective way of generating high quality sound. That versatility has made them the pre-eminent loudspeaker system by far.

I mentioned a few posts back that I have an intellectual respect for the electrostatic speaker/headphone based on my formative exposure to the Quad 57 some ten or so years after its introduction, driven by Quad valve amps. A wonderful stereo sound in the book lined study of my English teacher. Who couldn't fail to be attracted to the idea of a weightless diaphragm, pushed and pulled under electrostatic charge over its entire surface as it flexes and created sound. The very notion is extremely seductive: the thought that the diaphragm could be accelerated to trace the highest harmonics and then brought to a dead stop thanks to the virtually zero inertia. We've seen that an unwelcome characteristic of the wightless diaphragm under tension is that it has a tendency to 'twang' and as a result of the low damping. We also know that over the years the maufacturers of electrostatic panels have used-up the entire stocks of a particular cling film and when a new film has been sourced, the sound of the speaker has changed. That shouldn't be the case if the diaphragm is under total control of the static forces, but it seems to be. Exactly the same issues have been experienced by cone tweeter suppliers when cloth diaphragm material supplies change.

Elecrostatic panels are open at the front and the back. The sound radiating from the rear is of exactly the same loudness and character as that from the front, and quite unlike a conventional speaker, the electrostat pair could be turned around, back to front, and would, conceivably both measure and sound the same. The diaphragm doesn't have a listener facing and rear wall facing side: it doesn't know which way round it is.

Think about this for a moment: the total amount of sound sprayed into the room is of a very different nature to that of the conventional cone speaker. We know that the conventional box speaker has the tweeter on the front face, and if we listen behind the speaker, the HFs are greatly lowered in loudness. Not so the panel speaker: there is as much HF spraying from the back of the speaker onto the rear wall as facing the listener. This must mean that the total energy 'splatter' from the electrostat into the entire room is radically different from that of the cone speaker, and that can bring its own issues. It implies that if we exchange a conventional speaker for a panel speaker in our listening room that we could reasonably expect the 'stat to sound different because the untreated rear wall will reflect the back-wave towards the listener even if the on-axis response measures exactly the same as the box speaker. In short, the perceived sound experience of the electrostatic speaker at home is going to be different to the box speaker not necessarily because of the technology of the cling film diaphragm v. the cone speaker but because of the way they differently illuminate the room Does it mean that if you listened to a good panel speaker and a good cone speaker in a completely dead acoustic (such as an anechoic chamber) that they would (conceptually) sound the same? Yes it probably does.

So, we know that the conventional speaker designer divides the sound spectrum between drive units: usually, a bass/midrange driver and a tweeter, crossing over the signal from one to the other in a circuit of coils and capacitors - the crossover network. When I stated writing about electrostatic panels in this thread I imagined that the frequency division and separation of the bass/mid and high frequencies into physically separate and differently sized (moving coil) drivers units was not ideal compared with the theroretical all-sound-from-one-unit diaphragm panel. But after careful listening, I can hear that the opposite is true: the moving coil speaker designer has far more tricks available to him to select the best bass, mid and HF drive units and many clever ways of sending the signal to these drive units, flattening their outputs, adjusting their sonic contributions, phase and blending. The panel designer has far fewer options: he's nailed his colours to the 'panel mast, although, from what I hear, perhaps his best sountion (were his marketing to allow it - shock! horror!) would be a super-hybrid solution of moving coil bass unit, electrostatic panel for the midrange and handing over to a conventional tweeter at perhaps a rather lower frequency than good tweeters like to operate at. I can't see the marketing boys going for that one!

The reason - based solely on my own critical listening as a virtual electrostatic virgin - is that the panels certainly 'do something nice' in the midrange but they have (or at least, this respected design examples do) a fatiguing glassy sheen which adds an edge to voice and makes brass instruments really rather unpleasant. This is a real surprise: I had anticipated that the frequency band in which the panel would excel would be the HF region, but perhaps not so. Although there were conventional speakers (P3ESRs) as references to hand, there was no need to switch over to them: once the listener locks-onto some aspect of sonic performance that he's not completely happy with, he cannot ignore it. In my case, as every brass instument had this character my pallate was quickly sensitised and my observation just wouldn't go away.

So what was I hearing? What was it about the higher tones which were so irritating? Three things I suspect. A) I refer you to my last post: perhaps the damping is inadequate in certain higher frequency bands so that there is some latent ringing in the panel and/or B) the absence of a conventional crossover/equaliser has deprived the panel speaker designer of frequency response shaping tricks to trottle-down the energy in certain HF bands where there is, in fact, an excess output and/or C) the much greater HF spray front and back into the room draws attention to itself. One work around could sidestep these issues: the use of a first class moving coil (conventional) tweeter via a suitably complex crossover to cover the higher frequencies, facing the listener with perhaps just a little fill-in tweeter behind the speaker with an adjustable level control.

More later ...

[A.S.]

Ribbon speakers

Another light weight diaphragm way of generating sound with important differences.

We know that the electrostatic diaphragm is an extremely light sheet of cling film, charged up, and held in a static field between two electrodes. It is induced to move when the charge on one plate pulls it towards that plat and simultaneously the charge on the opposing plate weakens. When that process is repeated exactly according to the music signal, the diaphragm produces a sound wave that is (approximately) the same as the incoming voltage from the amplifier.

And the ribbon uses a similar push/pull arrangement, but this time using magnetic force not static force. The ribbon diaphragm is made from a tissue-like strip of extremely thin aluminium foil, a thinner version of kitchen cooking foil. That foil - or ribbon - is held close to an extremely powerful magnetic field and surprisingly for a non-ferrous material, the foil will actually move if the magnetic field is strong enough. The same principal applies: the foil diaphragm will generate a sound wave according to the music input.

There are some important differences between the ribbon and electrostatic transducers that are likely to effect their sound quality and durability. First, the cling film diaphragm is relatively robust providing that little fingers do not pierce it (could be dangerous due to high voltages) and it is naturally protected by the perforated plates. Providing that the electrostatic diaphragm is not over-driven with excessively loud signal, when the diaphragm may slap onto one plate and not bounce back, and providing that the diaphragm doesn't arc and burn a hole in it, and providing that the diaphragm tension doesn't sag, it should have a long service life.

The ribbon diaphragm is quite a different animal, and much more vulnerable. First, the ribbon is hanging limply under gravity, and unlike the 'stat diaphragm is not stretched or tensioned in any way. It is literally a piece of cigarette-packet thickness foil suspended in the air and clamped at both ends but unsupported at the sides. If you blow on it, you will permanently bend it out of shape - it's that thin and malleable - and moved away from the normal rest position in the middle of the magnetic field, it's sonic performance will change dramatically. If you drop it or the speaker falls over, it will also be damaged. If curious little fingers get near it, it will be completely destroyed. Secondly, the ribbon diaphragm is rather heavier than the 'electrostatic diaphragm (size for size) and that limits it's ability to accelerate (more inertia) and that limits its high frequency output. Third: the fact that it is clamped only at the top/bottom and unsupported at the sides means that there is a natural limit to how big the diaphragm can be made before the influence of gravity and the reality of damage combine: and than means, in practice, that ribbon technology is only suited to upper midrange and tweeter applications, when we know that there are full range electrostatic speakers.

The essentially interesting point to me is that the electrostatic diaphragm is under extreme tension, and the ribbon diaphragm is under no tension whatever. This is going to have a sonic influence. We mentioned damping previously, and you can visualise that these two solutions will invoke entirely different damping effects. Rapping your fingers on the electrostatic diaphragm (which should spring back under tension) would produce a definite note, but if you dared to touch the ribbon diaphragm (which would damage it) you would hear nothing. For this reason alone, setting aside all other technology, size, cost, cosmetic or marketing reasons, we can expect these technologies, even with an identical frequency response to have a somewhat different latent sonic character due to the differences in damping. Which is likely to impose less of a character on the music?

Here surely is an expert on the restoration and renovation of electrostatic panels. You will read about the all-critical tensioning requirements and of potential multiple-resonances due to tensioning problems. You will also read of his suggestion of a damping film on the diaphragm and see evidence of the different resonance frequencies due to panel geometry. Don't assume that any one speaker technology is necessarily superior to any other - they all have strengths and weaknesses and the end result is always a compromise!

Conversely, here you will see the naked corrugated ribbon diaphragm relaxed and under zero tension.

More later

[witwald]

The masters of the pipe speaker were surely the British IMF company (long gone - why?). The frequency response curves from an old brochure show an extremely well integrated bass/mid/top on axis performance doubly so because the drive units were spaced-out on a large baffle. However, close examination of the curves in the PDF (below) indicates what looks like the tell-tale dip in the low bass where the pipe and the drive unit are partially out of phase - the very region which defines the subjective 'weight'.

Attached below are brochures for the IMF TLS50 Mk II and the IMF Reference Standard Professional Monitor (RSPM) Mark IV loudspeakers. The TLS50 utilised a 205 mm Bextrene bass unit, while the RSPM (and also the TLS80) used the larger 300 mm X 210 mm flat polystyrene bass unit. The TLS50's cabinet had a gross volume of about 88 litres, while that of the TLS80 was much larger at 185 litres, and that of the RSPM was 217 litres in size.

Like the TLS80, the TLS50's sine wave frequency response curve also shows a dip in the 100 Hz to 200 Hz low bass frequency band. The depth of this dip is about 4 dB or so (relative to the midrange level, which is quite flat). The size and shape of this dip is very similar to that in the TLS80, even though the TLS50 uses a different bass unit to the TLS80.

However, unlike the TLS50 and TLS80, the RSPM's sine wave frequency response shows no clear sign of any dip in the 100 Hz to 200 Hz region.

The question I have is why would the TLS50 and TLS80 have this response dip while the RSPM doesn't seem to? If the dip is caused by a phase-related cancellation between the output from the mouth of the transmission line and the output from the bass unit, why isn't it also evident in the RSPM's response?

[witwald]

.... if we set the audio analyser to take the audio equivalent of a long exposure picture of the audio over an entire performance (say, the complete 23 minutes of Symphonic Dances) we can determine how much of each frequency is present across the whole piece.

Here is computer-generated pink noise...

OK, now let's open the shutter in the audio analyser and make a long-exposure capture of pink noise playing. ...

Next I played into the audio analyser the entire 23 minutes of Symphonic Dances. Ignoring the difference in vertical offset between Dances and pink noise (I didn't calibrate the analyser as we're working in dBs so no need to) I've plotted Dances on the same graph as the pink noise (ref: dances-pink-sc.jpg). ...

For my own curiosity, I applied a filter having the overall shape (the three orange lines) of the Symphonic Dances profile onto the pink noise (as per clip P above) and made it have the same sonic spectrum.

I am interested to learn what software was used to generate the pink noise, to capture it, filter it and perform a spectral analysis on it, and also to capture and analyse the piece of music. Can you provide some details?