My years of experience, fiddling give me the perspective that getting good subjective sound is equivalent to accumulating points towards some necessary total number, like showing enough ID material to get a passport, or opening a bank account - unless you reach 100 points, say, you don't pass the requirement ... nor do you get satisfying sound.
IOW, a brilliant speaker, worth say 70 points will never do it, if amps of 20 points worth or less are used to drive it. But, attach an amp of 30 points .. and bingo!!
Very rough analogy, but you get the idea ...
IOW, a brilliant speaker, worth say 70 points will never do it, if amps of 20 points worth or less are used to drive it. But, attach an amp of 30 points .. and bingo!!
Very rough analogy, but you get the idea ...
Okay, try again!!
Let's say subjectively satisfying sound can achieved with a system of quality 100 - just points, it's not a percentage.
The very best source in 'perfect' condition could be 75 points, best amplification 75, best speakers 75. Connect them together 'perfectly', you get 225 points, way, way better than you need - should have excellent sound,! But wait, there's more!! Subtract points for interference effects, connection problems, all sorts of niggling problems - very easy to lose 150 points in a blink of an eye, no longer have 225, you're down to 75 in real terms, what the ears hear. Somehow, you have to recover 25 points to get the good sound - this can be the hard bit!
This also means that 'mediocre' gear, worth only 35 points in each category, adds up to 105 - but end system is very, very carefully tweaked so that only 5 points are lost - with cleverness and understanding it has just made the grade ... it's worth listening to.
This type of balancing is how I see the process of achieving worthy sound ...
Let's say subjectively satisfying sound can achieved with a system of quality 100 - just points, it's not a percentage.
The very best source in 'perfect' condition could be 75 points, best amplification 75, best speakers 75. Connect them together 'perfectly', you get 225 points, way, way better than you need - should have excellent sound,
! But wait, there's more!! Subtract points for interference effects, connection problems, all sorts of niggling problems - very easy to lose 150 points in a blink of an eye, no longer have 225, you're down to 75 in real terms, what the ears hear. Somehow, you have to recover 25 points to get the good sound - this can be the hard bit!This also means that 'mediocre' gear, worth only 35 points in each category, adds up to 105 - but end system is very, very carefully tweaked so that only 5 points are lost - with cleverness and understanding it has just made the grade ... it's worth listening to.
This type of balancing is how I see the process of achieving worthy sound ...
I don't think adding subjective numbers together tells us much; the amplifier/speaker interface determines much of the resulting sound quality, and the standard measurements do not really address interface issues that well.
Some amplifiers are more compatible with some speakers than others; this is due to settling time of the amplifier when driving reactive loads, transient output Z variations as transistors pass through the Class A/AB switching region, current pulses from +/- supply rails getting into input stages, AM radio/RFI pickup from the speaker wires, etc. etc. There are many subtle faults in amplifier design that are more readily exposed by some speakers than others.
Loudspeaker faults are grosser, and more easily measured, although there is much controversy about which measurements correlate most closely with subjective impressions.
Some amplifiers are more compatible with some speakers than others; this is due to settling time of the amplifier when driving reactive loads, transient output Z variations as transistors pass through the Class A/AB switching region, current pulses from +/- supply rails getting into input stages, AM radio/RFI pickup from the speaker wires, etc. etc. There are many subtle faults in amplifier design that are more readily exposed by some speakers than others.
Loudspeaker faults are grosser, and more easily measured, although there is much controversy about which measurements correlate most closely with subjective impressions.
Last edited:
yeah, what if my 100 is your 32? Or visa versa? I follow this website to see what other people do to try and get their fix not to satisfy my thirst for perfect music reproduction in my systems. Sometimes I get ideas but most of them are really not worth pursuing. Even my preference for headphones is probably different then 99.9 percent of the people here. I know what's right and sounds realistic but do you?
I have a very firm idea of what 100 is, and this relates to the 'invisible speaker' thing that Pano understands - the system, as a whole, works well enough that it becomes impossible to localise the speaker as a source of the sound, no matter what you do to 'tune' into the drivers as being the devices creating the sound field. This is a very definite transition in performance, it's like an on/off switch when it happens - psychoacoustics is doing its job of maintaining the illusion no matter what.
A system can do many things brilliantly well, but still fail, in fact nearly always, on this test. Key is that certain types of distortion are kept low enough in level, and that the system can do SPLs cleanly, the ability of the sound to build in intensity without 'revealing' that just the speakers are doing it is extremely important.
Hence the idea of a number, say 100, just pull one out of the hat, could use 500, say. The way it works, is that at 99 you can still 'see' the speakers working - and we're forgetting "sweet spots" here - but anything above 100 the speakers disappear and the illusion of a 'real' musical event is always maintained ...
A system can do many things brilliantly well, but still fail, in fact nearly always, on this test. Key is that certain types of distortion are kept low enough in level, and that the system can do SPLs cleanly, the ability of the sound to build in intensity without 'revealing' that just the speakers are doing it is extremely important.
Hence the idea of a number, say 100, just pull one out of the hat, could use 500, say. The way it works, is that at 99 you can still 'see' the speakers working - and we're forgetting "sweet spots" here - but anything above 100 the speakers disappear and the illusion of a 'real' musical event is always maintained ...
Yes, the whole business of 'debugging' subtle faults is downright painful, very irksome, irritating, tedious. But so far I've found no shortcuts, just chucking an ultra-good speaker into the mix almost never will work - slogging at it, grinding away the little defects, one by one, is the only method that I've discovered to work, so far ...
I don't get brilliant ideas that often; the patent for Shadow Vector was filed a long time ago. There can be a lot of overlapping small ideas; that's how the Ariel was designed, nibbling away at the challenge for a year or so.
The big idea for the LTO, or "Beyond", or whatever it gets named, was a better horn with improved settling times, compared to commercial offerings. An all-horn system was never considered; too big, not gonna happen. I'm looking now at the vector addition of paired bass drivers and how they interact with the floor image; with any luck, phase coherence in the working range should be possible.
Dr. Geddes did a beautiful job of system optimization of a high-performance 2-way system, which is intended to complemented with a subwoofer array. Congratulations are in order for accomplishing a very difficult goal ... generous headroom and response symmetry over a 60 to 90 degree arc.
My goals are a bit different. The Ariel and the new speaker are optimized for rapid settling time (like an electrostat) at the listening position, freedom from narrow "spikes" in the polar pattern (unlike an electrostat), and a good interface with low to moderate power amplifiers.
I prefer to overdesign the critical 200 Hz to 5 kHz region, giving the drivers plenty of overlap room. (The drivers have useful and well-behaved response outside the system crossover.) Not everyone does this; many designers take drivers right to the limit of what they can do, and sharp-cut at that limit. Speakers that are designed that way sound "nervous" and a little unsettled to me; I prefer a "relaxed" sound from a loudspeaker.
The amplifiers carry forward a similar philosophy. Many of the high-end amplifiers on the market clip all at once; when the output devices hit the wall, the drivers also hit their limits too, which greatly lengthens recovery time (this applies to both solid-state and vacuum-tube amplifiers). The Amity and Karna amplifiers have overdesigned drivers with 3 to 5 dB of additional headroom, the capability to drive the outputs into the A2 region when necessary, and isolated power supplies. This prevents the outputs from modulating the lower-level sections of the amplifier.
The new speaker has a similar overdesign in the MF section, with headroom in the 700 Hz to 5 kHz region that will never be used in domestic applications. A similar degree of headroom in the 200 Hz to 700 Hz region unfortunately requires a large bass array or even larger bass horn, so there's an inevitable compromise. But two 15" drivers with co-operation from the floor image should give reasonable radiating area to complement the 16.5" MF horn.
People sometimes ask why I left the dipole approach; it was the requirements for equalization (and power) that did it. Audiophile speakers are already woefully inefficient; 0.3% to 0.8% are very typical figures. All the rest of the power simply heats the voice coil. Dipole cancellation makes this worse, at a rate of 6 dB/octave below the baffle peak.
By nudging conversion efficiency into the 3% to 5% range, modest Class A1 amplifier power is used more efficiently, maintaining Class A quality over the entire dynamic range of typical program material. By careful selection of discrete transistors and/or vacuum tubes for low-level parts of the system, the entire chain from digital or analog sources can be Class A all the way through.
Last edited:
Lynn, have you considered putting a crossover point at 200Hz (or thereabout) and using a dedicated midbass driver in the 200/700Hz range?
This would let you use a more efficient 15" driver, or even a 12".
I find the 200Hz-300Hz range ideal for a crossover point, for many reasons (human voice fundamentals limit, Schroeder frequency in typical rooms, floor bounce, baffle step, etc.).
This is also a very difficult crossover point, for exactly the same reasons, and one where a linear-phase crossover comes in handy.
The exact frequency of the crossover in that range should ideally be let to the end user or installer, because it is room and position dependent...
This would let you use a more efficient 15" driver, or even a 12".
I find the 200Hz-300Hz range ideal for a crossover point, for many reasons (human voice fundamentals limit, Schroeder frequency in typical rooms, floor bounce, baffle step, etc.).
This is also a very difficult crossover point, for exactly the same reasons, and one where a linear-phase crossover comes in handy.
The exact frequency of the crossover in that range should ideally be let to the end user or installer, because it is room and position dependent...
Lynn, It is a good simple brief of where its at, with the two, lets call them ideologies, if we aspire them to be that.
Just to say that I keep a good stock of cheap valves
200 HZ -5 khz........
Example........ Lowther (95-100db)
I prefer a "relaxed" sound from a loudspeaker............
Example........... Horning Loudspeakers
There cannot really be a points or percentage rating. You can get into the A is better than B is better than C is Better than A or however this false paradigm can go.
If you take the original sound whether at a concert, in the studio, or anywhere else as the standard to achieve, you may say that ii is nearer or less near than another system. You could do this for 1000 systems say for opera recorded from previously attended venues eg. Madison Gardens or RAB in London etc. If you then say repeat this for some jazz and say drama your order of preference will probably be different.
This assumes top quality recordings without sound reinforcement at the venue. i.e all real sound.
Now if you play back at low to high levels and compare you may judge differently.
It gets more complicated if you try to be too analytical.
I am going to be tough and say CLass A only period. Oe exception may be extremely fast devices with very low switching times in nano seconds. If you cannot hear the difference then the rest of the reproduction chain has some problems or you dont have perfect hearing.
It is not just about dynamic contrast and power which is important. We know it is a matter of compromise through the whole chain of reproduction. Even the room temperature and humidity can have soime effect.
John Atkinson had it right when he sted in stereophile about hearing a guitar being play directly through its amplifier and speaker (with its limitations) it still sounded better than any recording of the same played back through a then especially esteemed system.
Everyone will have a different and interesting view on this. This is not intended to be daunting and as DIYers we are really trying to get closer to fidelity or our own personal preference.
If you take the original sound whether at a concert, in the studio, or anywhere else as the standard to achieve, you may say that ii is nearer or less near than another system. You could do this for 1000 systems say for opera recorded from previously attended venues eg. Madison Gardens or RAB in London etc. If you then say repeat this for some jazz and say drama your order of preference will probably be different.
This assumes top quality recordings without sound reinforcement at the venue. i.e all real sound.
Now if you play back at low to high levels and compare you may judge differently.
It gets more complicated if you try to be too analytical.
I am going to be tough and say CLass A only period. Oe exception may be extremely fast devices with very low switching times in nano seconds. If you cannot hear the difference then the rest of the reproduction chain has some problems or you dont have perfect hearing.
It is not just about dynamic contrast and power which is important. We know it is a matter of compromise through the whole chain of reproduction. Even the room temperature and humidity can have soime effect.
John Atkinson had it right when he sted in stereophile about hearing a guitar being play directly through its amplifier and speaker (with its limitations) it still sounded better than any recording of the same played back through a then especially esteemed system.
Everyone will have a different and interesting view on this. This is not intended to be daunting and as DIYers we are really trying to get closer to fidelity or our own personal preference.
That's why we build our own systems. For example i found only a couple of triode class A amplifiers I like to drive my multi-amped system (most are so colored they are a joke) but only in a certain range above 100 cycles. Pentodes serve their purpose too as well as class AB sand amps. There really is a benefit to having a lot of power below 100 cycles and class tube or SS A down there (250 WPC or more) isn't in my budget or game plan. Ceartainly there is a difference in AB amps driving bass horns and I'll take tight powerful control over tube roundness anyday. I do agree with Lynn stating where he likes to use wider bandwith drivers in their relaxed range. It's a no brainer once you set something up like that right and live with it for a while.
Your design rational is conflicted system of subjective impressionism with a dash of technical merit tossed in for flavor.
Overdesign? An imaginary line of headroom drawn in to separate vague transition from linear to non linear from a sharper one.
Making decisions about amplifier choice based on comparisons with speaker designed to flop about on its own v damping that is coupled to amplifier output impedance?
Yes, there are many speaker faults readily exposed by some amplifiers and not by others.
Yes, the algebra of subjectivity is tricky. With it rules of manipulation, and definitions of objects and operators, all proofs become random collection of symbols. Each is at best an impressionist's rendering, open only to interpretation of each observer's mind.
Dr. Geddes design approach is very simple. It is based on direct radiator's transitional behavior from omnidirectional with spherical wavefront to that of planar wavefront, and on horn loaded radiator's transition from omnidirectional behavior to that of bounded projection as a sector of a spherical wavefront. Metrics for shapes of wavefront for two systems are found, and used for selections of direct radiator, horn radiator, and crossover point that is chosen. Equalization applied by crossover filters determines acoustic performance of each drive element in targeted listening field, and are generally chosen for a smooth frequency response. Design constraints based on complexity/cost of filters v performance in shaping response of drive elements are made. With design constraints defined, building and optimizing system is relatively straight forward process.
Neodymium/conductive magnets v ceramic/non-conductive magnets:
Differences are readily observed by simply dropping sample slugs of each through copper tube. Magneto restrictive damping inherent to conductive magnet is a primary consideration for employment in transducer motor. Think about dropping magnets through solenoid coil. Behavior with coil terminals shorted v open are completely different. Think of complex impedance of such coil with complex drive signal; or even with high level high frequency signal. Think about damping system across coil terminals, and notion of feedback.
Perhaps think about Barkhausen Noise Analysis. Behavior of magnetic field of magnet immersed in external magnet field; displacement of field about magnetic domains; and ability to flip or rotate magnetic domains. Hysteresis involved with repeated application and removal of external magnetic field are quite telling, and repetition with increasing external field strength reveals even more.
At some point build something and compare thoughts about how predicted and measured responses compare; see how well thought process correlates to physical process. This correlation is best improved by modifying thought process.
No speaker to amplifier interface exists other than wire. Once connected together a single system is formed. It is best described as convolution of impulse responses of each system. Absolutes are calculated through application of reciprocity principle to each degree of freedom for all elements within measurement system and system under test. Primary example: Standardization of reference microphones. Same techniques may be used for each individual systems; but results from reciprocity based reference may be used as comparative standard allowing matching methods that greatly reduce efforts required of practitioner in the field using measurement qualified microphone associated amplifier.
An artist is grounded by real behavior of physical media. Art is refinement in practiced technique with input from the imagination. Loads of room for art in speaker building, but making art requires doing the work too.
Hmmm - thanks for the info !
Which type of driver were they, Pano ?
The 288H has the extended response that goes up to 12kHz, but it's not smooth on the way up to 12k . I'll still try to give this a try if I can . Will fit the small choke this evening which should at least confirm if it's the 10k+ area that's causing the shriek .
If you worked for Boeing then you had a head start. At least they know how to use Titanium and Beryllium
Nicely put. It's what's frustrating about the overly artsy types- many who have eventually become decent designers started out by tweaking, but they only get really good when they become more serious.
There are of course the opposites, engine-ears who jump all too readily to conclusions about what's important, and wind up designing systems that meet their criteria and sound bad. They need to learn the art, the artists need to learn the engineering, and large portions of either cult fail to progress beyond the half of the equation they're comfortable with.
I like that analogyOkay, try again!!
Let's say subjectively satisfying sound can achieved with a system of quality 100 - just points, it's not a percentage.
The very best source in 'perfect' condition could be 75 points, best amplification 75, best speakers 75. Connect them together 'perfectly', you get 225 points, way, way better than you need - should have excellent sound,
This also means that 'mediocre' gear, worth only 35 points in each category, adds up to 105 - but end system is very, very carefully tweaked so that only 5 points are lost - with cleverness and understanding it has just made the grade ... it's worth listening to.
This type of balancing is how I see the process of achieving worthy sound ...
It works for me