I agree, diy cannot compete with budget speakers in terms of value for money.
'tis more fun though...
Trying to stay within budget one could pair up the LSR305 with Tang Band W69-1042 ($56 from PE) and end up with quite decent, slim floorstanders.
One would need to build a 2way active xover (Elliot Sound Products?) and I assume that a suitable stereo amp to drive the woofers is present.
'tis more fun though...
Trying to stay within budget one could pair up the LSR305 with Tang Band W69-1042 ($56 from PE) and end up with quite decent, slim floorstanders.
One would need to build a 2way active xover (Elliot Sound Products?) and I assume that a suitable stereo amp to drive the woofers is present.
Creative Sound - Product Details
The JBL's may be the more versatile of the two (and perhaps better resale if I decide to go that way later). How would I hook up active speakers like these to FM radio (I don't have streaming and such and still listen to FM as my primary source of radio). The same thing goes for hooking up a cd player? Would I be better off with passive speakers if these are my primary sources (CD and FM)?
Every manufacturer today claims there speaker is reference grade and designed for accurate reproduction. Far more people believe their speakers are accurate, or that such a thing exists. This is even more common today, in a niche industry filled with many designers that rehash common designs and use off the shelf parts. It goes without saying that there are no accurate speakers, since all speakers produce linear and non linear distortion. The room and ancillary equipment also play major parts in the goal of accurate reproduction, but this discussion is focused on the speaker itself. Very few speakers come close to being accurate, including electrostatics, ribbons, and moving coil types. There are good and poor examples of each.
So, what behaviors criteria play a part in building an accurate speaker?
Put into terms for the sake of simplicity, the absence of linear and non linear distortion. It isn't any one parameter of behavior, but each of them, and how they come together to form the total speaker:
- Frequency response (ref. 1, 2, 12) is claimed to be the first thing we notice about the sound of a speaker. 1db is the smallest change we can reliable detect in steady state stimulus, and 3db is the smallest we can recognize in musical playback. The flatter the response, the more true it is in that respect to the signal. The problem with FR measurements on blogs and audio mags is they smooth the measurements out too much.
- Resonance is present in every technology where some mechanical part is put into motion, and how that resonance begins and ends can vary to some extent. When it's not visible on a a frequency response graph as a peak, or a bump in the impedance, it can sometimes be seen on cumulative spectral analysis. Resonance is the accentuation of a frequency or continuation of a tone after the input signal has attenuated.
- Frequency response is how equal in amplitude all frequencies will be reproduced. Even the best speakers have several decibels of ripple over their bandwidth, although according to some researchers changes in response amplitude under 3dB are difficult to detect. The absolute threshold of change is between .1dB and 1dB (ref. 1, 2).
- Power response refers to frequency response off axis. There are two parts to this topic. Because lower frequencies are largely omnidirectional, the acoustic power is divided over the dispersion area, resulting is a 6dB loss where the wavelength exceeds the baffle width. Where high frequency drivers are more directional, their output is higher and there is less loss to off axis. There was great deliberation between researchers about how high frequency power response should be in practice. Floyd tool, Martin Colloms, and several vocal designers were at stark odds about baffle widths and their opinions about wide and narrow dispersion.
- Harmonic distortion is an interesting area of discussion. When marketing literature talks about distortion, they tend to focus on it as though it and frequency response where the only constituents to audio reproduction. Most speakers produce harmonic distortion in the 1% or higher range, enough to mask the harmonics of the electronics (ref. 3-11, 13). Some of the better developed speakers have distortion in the -46dB (0.5%) range, but few can accomplish that from their lowest frequency to their top end. A few expensive drivers have come about with harmonics in the -53dB (0.2%) range.
- Impulse and step response performance indicate the maximum rate of change possible by a moving mechanism. A sharp pulse is presented to the driver and quickly removed, leaving the moving mass to the devices of its own suspension and air loading. It also reveals pre and post ringing, along with timing alignment and phase behaviors. Impulse tests on woofers can reveal overshoot and ringing tendencies, although it's important to also note that woofers don't need faster rise times than the maximum frequency demands. For example, 200Hz, only requires that a woofer rise at a rate within 1.25ms for each quadrant of a full cycle. The actual maximum rate of change is resides at .707 of the total magnitude. Putting this into perspective, one can see that "slow bass" has little to do with a driver's rise time. Hint. hint.
- Phase shifts are often sated to be inaudible in scientific doctrine (ref. 14-15), and it's the effects on timing between frequencies that becomes audible through harmonic misalignment.
- Losses can take place within the speaker driver's magnetic motor, the diaphragm material, the suspension or edge terminations, and the crossovers. Loss affects efficiency and small signal information, although the two are not intrinsically connected, ie: low sensitivity does not imply loss of detail or distortion of signal information. To state otherwise would be a gross but common oversimplification.
- The load put on the amplifier is important, because low impedance loads require higher wattage due to increased current flow. This results in high power consumption, increased distortion, and heat generation. Low impedance loads are also often accompanied by reactance, with capacitance being the big offender. High capacitance taxes the stability phase margin of the amp and can be detrimental to the slew rate of the amplifier. Low impedance loads are usually considered improper designs today. Speakers with serious design flaws like the Wilson Mini Watt (.5 ohms @~200Hz) and CR resonant bass enhancement circuits such as the Infinity Kappa 9 (.7 ohms through the bass) limit the variety of amplifiers available for the desired playback levels. Many amplifiers for of driving .5 ohms are unnecessarily expensive and distorted in their own right, although the audibility of this is yet to be determined. There is no reason to build speakers will almost fault impedance anymore, since frequency augmentation can be carried out in the line-level domain with no ill effect on the power amplifier.
So, what speakers are closest to the goal of accuracy?
Tough question. I've kept up on speaker tech but I haven't seen a lot of genuine advancement in any area. Everything I see is based on the knowledge dating back over some 40 years. What's special about these modern times is that it's finally affordable and practical to build what they were conceptualizing back then.
Duntech was said to have the most accurate speaker in the world, the Sovereign, but no measured evidence was produced. Bowers & Wilkins claimed it was their Nautilus. The list goes on and there is something each of the pricey companies has in common - they don't provide measurements. That leaves a quick-witted fellow questioning exactly how they arrived at their conclusions. Hmm.
I can give a couple pointers. The most empirically accurate speakers I've seen and heard included the Quad ESLs and Yamaha models with the beryllium dome drivers. The Quad ESL-63 electrostatics and Yamaha's NS-1000Ms have a reputation that precedes them. For many years the NS-1000M was the reference speaker of the Audio Engineering Society, a few designers, and several reviewers. Both perform exemplary in each department, both the Quad USA-63 monitors and NS-1000Ms had and continue to see recording studio use. Both the Quad ESLs and Yamaha beryllium drivers have some of the best impulse responses one could hope for. Harmonic distortion for both Quad ESLs Yamaha speakers are still benchmarks. In the case of the ESL-63 and-2000, harmonics for both were to the tune of -70dB in the midrange. Nothing else I've seen anywhere can do that. DSP and active powering could take either to a whole new level, far surpassing anything.
(1)"A change by 1 dB is about the smallest change a human being can detect." Sound
(2) Dr. Floyd Toole: "There is some evidence that we can detect slopes of about 0.1 dB/octave, which translates into a 1dB tilt from 20Hz to 20kHz - not much. Such a spectral of error, is likely to be quite benign and subject to adaption."
(3) Auditory Perception of Nonlinear Distortion, Audio Engineering Society, Geddes, Earl R.; Lee, Lidia W.
AES E-Library Auditory Perception of Nonlinear Distortion
(4) Audibility Perception of Nonlinear Distortion, Audio Engineering Society, reprint 5891.
(5) Harry F Olson, JDD level of .7% using 40 Hz to 14 kHz bandwidth test system
(6) D.E.L Shorter, "Just perceivable distortion values of 0.8% to 1.3%".
(7) P.A Fryer, 2% - 4% distortion
(8) James Moir, Just detectible distortion (JDD) "level can be no lower than 1%, ie -45dB"
(9)Von Braunm ü hl & Weber, 1% - 2% at frequencies > ~ 500 Hz
(10) M. E. Bryan & H.D.Parbrook also embarked on tests involving the audibility of harmonics.
(11) Just Audible Thresholds for Harmonic Distortion, By Bryan, M.E.; Parbrook, H.D.
Just audible thresholds for harmonic Distortion: ingentaconnect
(12) Audibility of linear distortion in loudspeakers, Sylvain Choise
http://www.almainternational.org/ya...audibility_of_linear_distortion.106172825.pdf
(13) Measurement of Harmonic Distortion Audibility Using a Simplified Psychoacoustic Model. By Temme, Steve; Brunet, Pascal; Qarabaqi, Parastoo
https://secure.aes.org/forum/pubs/conventions/?elib=16446
The Acoustics and Psychoacoustics of Loudspeakers and Rooms.
(14) Dr. Leach, The Differential Time-Delay Distortion and Differential Phase-Shift Distortion as Measures of Phase Linearity, Audibility of Phase Shift, Journal of the Audio Engineering Society, vol. 37, no. 9, pp. 709-715.
(15) M. Hawksford, Audibility of Phase Distortion, Audio Engineering Society
So, what behaviors criteria play a part in building an accurate speaker?
Put into terms for the sake of simplicity, the absence of linear and non linear distortion. It isn't any one parameter of behavior, but each of them, and how they come together to form the total speaker:
- Frequency response (ref. 1, 2, 12) is claimed to be the first thing we notice about the sound of a speaker. 1db is the smallest change we can reliable detect in steady state stimulus, and 3db is the smallest we can recognize in musical playback. The flatter the response, the more true it is in that respect to the signal. The problem with FR measurements on blogs and audio mags is they smooth the measurements out too much.
- Resonance is present in every technology where some mechanical part is put into motion, and how that resonance begins and ends can vary to some extent. When it's not visible on a a frequency response graph as a peak, or a bump in the impedance, it can sometimes be seen on cumulative spectral analysis. Resonance is the accentuation of a frequency or continuation of a tone after the input signal has attenuated.
- Frequency response is how equal in amplitude all frequencies will be reproduced. Even the best speakers have several decibels of ripple over their bandwidth, although according to some researchers changes in response amplitude under 3dB are difficult to detect. The absolute threshold of change is between .1dB and 1dB (ref. 1, 2).
- Power response refers to frequency response off axis. There are two parts to this topic. Because lower frequencies are largely omnidirectional, the acoustic power is divided over the dispersion area, resulting is a 6dB loss where the wavelength exceeds the baffle width. Where high frequency drivers are more directional, their output is higher and there is less loss to off axis. There was great deliberation between researchers about how high frequency power response should be in practice. Floyd tool, Martin Colloms, and several vocal designers were at stark odds about baffle widths and their opinions about wide and narrow dispersion.
- Harmonic distortion is an interesting area of discussion. When marketing literature talks about distortion, they tend to focus on it as though it and frequency response where the only constituents to audio reproduction. Most speakers produce harmonic distortion in the 1% or higher range, enough to mask the harmonics of the electronics (ref. 3-11, 13). Some of the better developed speakers have distortion in the -46dB (0.5%) range, but few can accomplish that from their lowest frequency to their top end. A few expensive drivers have come about with harmonics in the -53dB (0.2%) range.
- Impulse and step response performance indicate the maximum rate of change possible by a moving mechanism. A sharp pulse is presented to the driver and quickly removed, leaving the moving mass to the devices of its own suspension and air loading. It also reveals pre and post ringing, along with timing alignment and phase behaviors. Impulse tests on woofers can reveal overshoot and ringing tendencies, although it's important to also note that woofers don't need faster rise times than the maximum frequency demands. For example, 200Hz, only requires that a woofer rise at a rate within 1.25ms for each quadrant of a full cycle. The actual maximum rate of change is resides at .707 of the total magnitude. Putting this into perspective, one can see that "slow bass" has little to do with a driver's rise time. Hint. hint.
- Phase shifts are often sated to be inaudible in scientific doctrine (ref. 14-15), and it's the effects on timing between frequencies that becomes audible through harmonic misalignment.
- Losses can take place within the speaker driver's magnetic motor, the diaphragm material, the suspension or edge terminations, and the crossovers. Loss affects efficiency and small signal information, although the two are not intrinsically connected, ie: low sensitivity does not imply loss of detail or distortion of signal information. To state otherwise would be a gross but common oversimplification.
- The load put on the amplifier is important, because low impedance loads require higher wattage due to increased current flow. This results in high power consumption, increased distortion, and heat generation. Low impedance loads are also often accompanied by reactance, with capacitance being the big offender. High capacitance taxes the stability phase margin of the amp and can be detrimental to the slew rate of the amplifier. Low impedance loads are usually considered improper designs today. Speakers with serious design flaws like the Wilson Mini Watt (.5 ohms @~200Hz) and CR resonant bass enhancement circuits such as the Infinity Kappa 9 (.7 ohms through the bass) limit the variety of amplifiers available for the desired playback levels. Many amplifiers for of driving .5 ohms are unnecessarily expensive and distorted in their own right, although the audibility of this is yet to be determined. There is no reason to build speakers will almost fault impedance anymore, since frequency augmentation can be carried out in the line-level domain with no ill effect on the power amplifier.
So, what speakers are closest to the goal of accuracy?
Tough question. I've kept up on speaker tech but I haven't seen a lot of genuine advancement in any area. Everything I see is based on the knowledge dating back over some 40 years. What's special about these modern times is that it's finally affordable and practical to build what they were conceptualizing back then.
Duntech was said to have the most accurate speaker in the world, the Sovereign, but no measured evidence was produced. Bowers & Wilkins claimed it was their Nautilus. The list goes on and there is something each of the pricey companies has in common - they don't provide measurements. That leaves a quick-witted fellow questioning exactly how they arrived at their conclusions. Hmm.
I can give a couple pointers. The most empirically accurate speakers I've seen and heard included the Quad ESLs and Yamaha models with the beryllium dome drivers. The Quad ESL-63 electrostatics and Yamaha's NS-1000Ms have a reputation that precedes them. For many years the NS-1000M was the reference speaker of the Audio Engineering Society, a few designers, and several reviewers. Both perform exemplary in each department, both the Quad USA-63 monitors and NS-1000Ms had and continue to see recording studio use. Both the Quad ESLs and Yamaha beryllium drivers have some of the best impulse responses one could hope for. Harmonic distortion for both Quad ESLs Yamaha speakers are still benchmarks. In the case of the ESL-63 and-2000, harmonics for both were to the tune of -70dB in the midrange. Nothing else I've seen anywhere can do that. DSP and active powering could take either to a whole new level, far surpassing anything.
(1)"A change by 1 dB is about the smallest change a human being can detect." Sound
(2) Dr. Floyd Toole: "There is some evidence that we can detect slopes of about 0.1 dB/octave, which translates into a 1dB tilt from 20Hz to 20kHz - not much. Such a spectral of error, is likely to be quite benign and subject to adaption."
(3) Auditory Perception of Nonlinear Distortion, Audio Engineering Society, Geddes, Earl R.; Lee, Lidia W.
AES E-Library Auditory Perception of Nonlinear Distortion
(4) Audibility Perception of Nonlinear Distortion, Audio Engineering Society, reprint 5891.
(5) Harry F Olson, JDD level of .7% using 40 Hz to 14 kHz bandwidth test system
(6) D.E.L Shorter, "Just perceivable distortion values of 0.8% to 1.3%".
(7) P.A Fryer, 2% - 4% distortion
(8) James Moir, Just detectible distortion (JDD) "level can be no lower than 1%, ie -45dB"
(9)Von Braunm ü hl & Weber, 1% - 2% at frequencies > ~ 500 Hz
(10) M. E. Bryan & H.D.Parbrook also embarked on tests involving the audibility of harmonics.
(11) Just Audible Thresholds for Harmonic Distortion, By Bryan, M.E.; Parbrook, H.D.
Just audible thresholds for harmonic Distortion: ingentaconnect
(12) Audibility of linear distortion in loudspeakers, Sylvain Choise
http://www.almainternational.org/ya...audibility_of_linear_distortion.106172825.pdf
(13) Measurement of Harmonic Distortion Audibility Using a Simplified Psychoacoustic Model. By Temme, Steve; Brunet, Pascal; Qarabaqi, Parastoo
https://secure.aes.org/forum/pubs/conventions/?elib=16446
The Acoustics and Psychoacoustics of Loudspeakers and Rooms.
(14) Dr. Leach, The Differential Time-Delay Distortion and Differential Phase-Shift Distortion as Measures of Phase Linearity, Audibility of Phase Shift, Journal of the Audio Engineering Society, vol. 37, no. 9, pp. 709-715.
(15) M. Hawksford, Audibility of Phase Distortion, Audio Engineering Society
Quad ESL-63
ESL-63 harmonic distortion, this distortion graph goes to -60db, .1%. Harmonics in the bass are about -45db, 0.56%. Distortion above that is better than -60dB, .1%. Predominantly 3rd ordered harmonic since the driving mechanism is push pull, which cancels second order:
ESl-63, harmonic comparison between 86db and 96db playback levels:
ESL-63 impulse response:
ESL-63 CSD, 30db and 60db:
Quad ESL-57
Between the ESL-63 and ESL-57, the model 57 had the highest distortion. The 10khz area becomes larger at higher listening levels. Overall, its still impressive:
More harmonic distortion results for the ESL-57
The ESL-57 impulse and CSD plots, 3ms timestep. Settling takes about 6.5ms to mostly cease and the signal bounces back across the membrane. This is a common behavior among panels, since their surface area is so large. CSD see settling to -26dB in 3ms (3ms graph timestep) from 100Hz upwards. This is exceptional.
Square wave responses for the ESL-57. This isn't exclusive to the ESL-57, Acoustats also display lots of ringing, too.
Polar response pattern of the ESL-57. Quads had a narrow figure 8 radiation pattern that resulted in narrow forward dispersion that could (would) cause treble attenuation when listening far off axis:
Yamaha beryllium series speakers:
The most distortion in the NS-1000M was -45db, 0.56% at 40Hz. In practice, third party measurements found it to be closer to -50db (.31%), and the midrange was far lower. The NS-2000 was even better in the mid and treble. They all used different drivers that looked similar (and are commonly mistaken on blogs as all being the same) but were not the same. I came to learn this and it's true. I had a set of measurements of the NS-2000 and I can't fathom where they went. Most of my measurements are printed and I am not sure how to get 'em here. The NS-1000X and NSX-10000 graphs (below) will suffice until I find them. Notice they are similar...but not the same:
Yamaha JA-0548 impulse and CSD plots. Tweeter polarity is inverted to replicate use in the speaker. Very fast and clean rise, very fast return to resting. CSD reveals -30dB in 1 ms above 4khz. This is exceptional.
Yamaha JA-803 impulse and CSD plots, 2.5ms time step. Note: In an actual speaker, the leading edge of impulse will be created by the tweeter. Overshoot past the rest position makes half a cycle under 30% total impulse magnitude, then stops by 1.5ms. -30dB in about 3ms (2.5ms graph time step) from 325Hz upwards. In terms of moving coil midranges and ESLs, this is revolutionary. It's better than some planar midranges, evidence in hand:
No set of measurements would be useful with looking at what a woofer and passive crossover do otherwise excellent mid and treble drivers. Impulse trigger was a bit delayed by the capture and started at 1.75ms, so start at that point. The resulting transient response is actually excellent for a non time aligned and non phase coherent speaker system. This is better than I've seen in $20k USD speakers, take the Bowers and Wilkins 801 for example. CSD is hashier than the speakers driven directly from an amp, but setting is still under 3ms above 3kHz. This is the "sound" of the speaker. It's either the crossover or the woofer doing it. Given the blistering fast impulse response of the tweeters and midranges, an active crossover could integrate and eliminate the remaining timing and phase mis-alignments that are inherent in the passive crossovers. The result would sound like an ESL from 500/600Hz upwards.
Polar response between the different beryllium series models was quite different, but none show any signs of crossover lobing. Yamaha's had an essentially perfect wide response that could (would) cause problems in commonly untreated rooms. Below is the NS-1000X. I haven't heard of many speakers out there that can do this:
ESL-63 harmonic distortion, this distortion graph goes to -60db, .1%. Harmonics in the bass are about -45db, 0.56%. Distortion above that is better than -60dB, .1%. Predominantly 3rd ordered harmonic since the driving mechanism is push pull, which cancels second order:
ESl-63, harmonic comparison between 86db and 96db playback levels:
ESL-63 impulse response:
ESL-63 CSD, 30db and 60db:
Quad ESL-57
Between the ESL-63 and ESL-57, the model 57 had the highest distortion. The 10khz area becomes larger at higher listening levels. Overall, its still impressive:
More harmonic distortion results for the ESL-57
The ESL-57 impulse and CSD plots, 3ms timestep. Settling takes about 6.5ms to mostly cease and the signal bounces back across the membrane. This is a common behavior among panels, since their surface area is so large. CSD see settling to -26dB in 3ms (3ms graph timestep) from 100Hz upwards. This is exceptional.
Square wave responses for the ESL-57. This isn't exclusive to the ESL-57, Acoustats also display lots of ringing, too.
Polar response pattern of the ESL-57. Quads had a narrow figure 8 radiation pattern that resulted in narrow forward dispersion that could (would) cause treble attenuation when listening far off axis:
Yamaha beryllium series speakers:
The most distortion in the NS-1000M was -45db, 0.56% at 40Hz. In practice, third party measurements found it to be closer to -50db (.31%), and the midrange was far lower. The NS-2000 was even better in the mid and treble. They all used different drivers that looked similar (and are commonly mistaken on blogs as all being the same) but were not the same. I came to learn this and it's true. I had a set of measurements of the NS-2000 and I can't fathom where they went. Most of my measurements are printed and I am not sure how to get 'em here. The NS-1000X and NSX-10000 graphs (below) will suffice until I find them. Notice they are similar...but not the same:
Yamaha JA-0548 impulse and CSD plots. Tweeter polarity is inverted to replicate use in the speaker. Very fast and clean rise, very fast return to resting. CSD reveals -30dB in 1 ms above 4khz. This is exceptional.
Yamaha JA-803 impulse and CSD plots, 2.5ms time step. Note: In an actual speaker, the leading edge of impulse will be created by the tweeter. Overshoot past the rest position makes half a cycle under 30% total impulse magnitude, then stops by 1.5ms. -30dB in about 3ms (2.5ms graph time step) from 325Hz upwards. In terms of moving coil midranges and ESLs, this is revolutionary. It's better than some planar midranges, evidence in hand:
No set of measurements would be useful with looking at what a woofer and passive crossover do otherwise excellent mid and treble drivers. Impulse trigger was a bit delayed by the capture and started at 1.75ms, so start at that point. The resulting transient response is actually excellent for a non time aligned and non phase coherent speaker system. This is better than I've seen in $20k USD speakers, take the Bowers and Wilkins 801 for example. CSD is hashier than the speakers driven directly from an amp, but setting is still under 3ms above 3kHz. This is the "sound" of the speaker. It's either the crossover or the woofer doing it. Given the blistering fast impulse response of the tweeters and midranges, an active crossover could integrate and eliminate the remaining timing and phase mis-alignments that are inherent in the passive crossovers. The result would sound like an ESL from 500/600Hz upwards.
Polar response between the different beryllium series models was quite different, but none show any signs of crossover lobing. Yamaha's had an essentially perfect wide response that could (would) cause problems in commonly untreated rooms. Below is the NS-1000X. I haven't heard of many speakers out there that can do this:
Last edited:
Do you have a pre amp or if not does your integrated amp have pre outs?
What comes out of either is the signal you want to feed active speakers.
Depends on the system. In my office I use an old Luxman R-1040 receiver that does not have any pre-out. In the basement (the other likely listening space) I use a pre-amp.
A problem with Behringer products is that their resale value is close to zero due to their reputation for ripping off other peoples designs and getting away with it.
Their first victim was Mackie followed by DBX and Genelec.
There may be others but Mackie took them to court which decided that visuals and the internal electronic design are not protectable under the law.
Genelec opted to change the design (visual not internal) after Behringer copied it for their own monitors.
DBX stopped producing their mic pre.
For that reason alone I'd go for the JBLs. At least to the best of my knowledge JBL has not gone for dirty tricks.
I'm familiar with Geithains. However, there is a bit of discrepancy with the .25% distortion form 30hz - 20khz. The graphs provided by the manufacturer do not support it:
The RL 801k is stated as ≤ -45db from 100hz - 10kz at 1m, 10W input. Distortion at 900hz and 1.3khz is approximately .3%, ie -50db. The RL 901k is stated as ≤ -44db from 100hz -10kz at 1m, 5W input, also approximately .3%.
It appears to drop lower at certain frequencies and is really good, but it's smoothed so much that we can't see what else is hiding there. It really should have been 1/12 or 1.24th octave smoothing (or none at all), not 1/3 or 1/6th. What we can say is that these graphs aren't the same as .25% (-53db) 30hz-20k, as their distortion graphs are only 100hz-10khz.
They look good in these respects, as expected from studio monitors. However, it's been smoothed alot and I suspect something is being omitted to accomplish 2 good looking measurements. Notice that they do not provide an actual measurement of other behaviors described in Post #45; impulse and energy storage (FR energy vs time).
It goes without saying that there is more to a speaker than just response and harmonic distortion. Even so, after 40 years it appears as though many designers still ignore that speakers not only reproduce steady state sine waves. They also have to start and cease with the input signal. Therein resides where the lasting great speakers are engineered. Some important behaviors include internal losses, which come about with motor design, materials' propagation velocities, mass, and edge termination. There is also energy storage which results from the moving mass, moving assembly's geometry, rigidity, the rear cavity, effectiveness of damping, last but not certainly not least - acoustic propagation velocity. The higher the velocity, the easier it is to accomplish lower loss since less damping is needed. The result of loss is the distortion of signal information in small signal impulse response tests, usually also through energy storage and masking in the form of resonance.
Geithain RL 922K, third party impulse response. The RL 992K has a distortion profile similar to the 801 and 96db testing level as the 901:
Quad ESL-63
ESL-63 harmonic distortion, this distortion graph goes to -60db, .1%. Harmonics in the bass are about -45db, 0.56%. Distortion above that is better than -60dB, .1%. Predominantly 3rd ordered harmonic since the driving mechanism is push pull, which cancels second order:
ESl-63, harmonic comparison between 86db and 96db playback levels:
ESL-63 impulse response:
ESL-63 CSD, 30db and 60db:
Quad ESL-57
Between the ESL-63 and ESL-57, the model 57 had the highest distortion. The 10khz area becomes larger at higher listening levels. Overall, its still impressive:
More harmonic distortion results for the ESL-57
The ESL-57 impulse and CSD plots, 3ms timestep. Settling takes about 6.5ms to mostly cease and the signal bounces back across the membrane. This is a common behavior among panels, since their surface area is so large. CSD see settling to -26dB in 3ms (3ms graph timestep) from 100Hz upwards. This is exceptional.
Square wave responses for the ESL-57. This isn't exclusive to the ESL-57, Acoustats also display lots of ringing, too.
Polar response pattern of the ESL-57. Quads had a narrow figure 8 radiation pattern that resulted in narrow forward dispersion that could (would) cause treble attenuation when listening far off axis:
Yamaha beryllium series speakers:
The most distortion in the NS-1000M was -45db, 0.56% at 40Hz. In practice, third party measurements found it to be closer to -50db (.31%), and the midrange was far lower. The NS-2000 was even better in the mid and treble. They all used different drivers that looked similar (and are commonly mistaken on blogs as all being the same) but were not the same. I came to learn this and it's true. I had a set of measurements of the NS-2000 and I can't fathom where they went. Most of my measurements are printed and I am not sure how to get 'em here. The NS-1000X and NSX-10000 graphs (below) will suffice until I find them. Notice they are similar...but not the same:
Yamaha JA-0548 impulse and CSD plots. Tweeter polarity is inverted to replicate use in the speaker. Very fast and clean rise, very fast return to resting. CSD reveals -30dB in 1 ms above 4khz. This is exceptional.
Yamaha JA-803 impulse and CSD plots, 2.5ms time step. Note: In an actual speaker, the leading edge of impulse will be created by the tweeter. Overshoot past the rest position makes half a cycle under 30% total impulse magnitude, then stops by 1.5ms. -30dB in about 3ms (2.5ms graph time step) from 325Hz upwards. In terms of moving coil midranges and ESLs, this is revolutionary. It's better than some planar midranges, evidence in hand:
No set of measurements would be useful with looking at what a woofer and passive crossover do otherwise excellent mid and treble drivers. Impulse trigger was a bit delayed by the capture and started at 1.75ms, so start at that point. The resulting transient response is actually excellent for a non time aligned and non phase coherent speaker system. This is better than I've seen in $20k USD speakers, take the Bowers and Wilkins 801 for example. CSD is hashier than the speakers driven directly from an amp, but setting is still under 3ms above 3kHz. This is the "sound" of the speaker. It's either the crossover or the woofer doing it. Given the blistering fast impulse response of the tweeters and midranges, an active crossover could integrate and eliminate the remaining timing and phase mis-alignments that are inherent in the passive crossovers. The result would sound like an ESL from 500/600Hz upwards.
Polar response between the different beryllium series models was quite different, but none show any signs of crossover lobing. Yamaha's had an essentially perfect wide response that could (would) cause problems in commonly untreated rooms. Below is the NS-1000X. I haven't heard of many speakers out there that can do this:
The ESL-57 and ESL-63 Quads and the Yamaha NS-1000M's are still benchmarks for the rest of the industry to reach; they outperform (in an objective sense) nearly all of the modern Stereophile "Class A" speakers selling for $25,000 to $150,000/pair.
Other speakers may be more efficient, have more dynamic range, or have polar patterns that conform to modern ideas of directivity, but the Quads and Yamaha's have been used for research and in recording studios for a long time. Very few DIY speakers reach the performance levels seen in the graphs above ... and if they do, they aren't cheap to build. The visual equivalent is 4K digital cinema, 70mm widescreen, and large-format still photography; the pinnacle of the art.
DSP, multi-amping, and carefully selected Chinese drivers let you get closer at a much lower price than when these models were still in production. (Most speakers on the Sixties and Seventies were pretty terrible, particularly the mids and tweeters, which is why these two models stood out.)
Going back to the original poster's question, is this standard of accuracy available for less than $500 if you build your own (and know what you're doing)? Probably not. For $1500 to $2000, maybe, just maybe.
Last edited:
The guy wanted suggestions for a neutral sounding speaker to listen to that was not expensive. The beringer2031p actually has measurements from a neutral party which I pointed towards,
I had not realized the price of the thing had actually gone up and that it doesn't look too available, anyway. Doesn't look like there's many used for sale
Spare us the sermon. Nobody can have an intellectual monopoly on the shape of an integrated waveguide speaker baffle with diffraction smoothing, or any sort of conventional crossover design.
To update, since starting this thread I lucked into a pair of JBL LSR32 monitors which I have come to love so much that I recently bought a pair of LSR308's as well. Each has its own virtues, of course, but I'm very, very pleased with both sets of speakers. The sound quality is absolutely amazing for the price.
Speaking of which, I'm confused as to why the LSR308's haven't been a bigger hit within the 'audiophile' community. I actually posted this very question on another audio site and have had very little response. I wonder if it might have something to do with the 308's (and 305's) being powered monitors, leaving the audiophile's cherished (and likely expensive) amplifier with nothing to do. Just curious and wondering if anyone here might have some thoughts on this.
Speaking of which, I'm confused as to why the LSR308's haven't been a bigger hit within the 'audiophile' community. I actually posted this very question on another audio site and have had very little response. I wonder if it might have something to do with the 308's (and 305's) being powered monitors, leaving the audiophile's cherished (and likely expensive) amplifier with nothing to do. Just curious and wondering if anyone here might have some thoughts on this.
i think you hit the nail on the head there.
the 308's being both powered and class D (and unlike the 32's the wave guide is non-rotatable,which for studio use in the classic on the meter bridge position is an asset)along with low WAF (very industrial looking)are probably the biggest factors for them being relatively unknown/unpopular.
i like the Control 1's as a nearfield (at the console) monitor, but that's just me.
the 308's being both powered and class D (and unlike the 32's the wave guide is non-rotatable,which for studio use in the classic on the meter bridge position is an asset)along with low WAF (very industrial looking)are probably the biggest factors for them being relatively unknown/unpopular.
i like the Control 1's as a nearfield (at the console) monitor, but that's just me.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.