Ok, on a totally different note, I was wondering- the new speakers need a name. So far they have only been the 'Beyond The Ariel' system. But they need a nice name, like Ariel or Amity or Karna. They could be called the Karna speakers (as opposed to the Karna amp), and then from there flow a whole system (Karna pre, etc.), but I leave that up to Lynn. They sound so special, I just don't want to call them 'those speakers' anymore. 
Deon
Deon
Depending on where the subs are positioned, this is not true. For example if you had 4 subs, each in one of the corners of a rectangular room, two of the fundamental room modes are not excited at all - the length and the width modes. (The height modes still are excited unless you had another 4 subs in the corners of the ceiling...)
This is because the wave leaving one woofer travelling to the other end of the room arrives there 180 degrees out of phase with the wave emanating from the 2nd woofer so they cancel instead of reinforcing.
This is particularly helpful for the lowest mode of the longest dimension of a room which typically puts a big peak in the bottom end of the bass response, but only when you're sitting near either end of the room - at the middle of the room there is no reinforcement at that frequency.
This leads to the situation where if you EQ the low end bass (with a PEQ notch) to be flat when listening near the back of the room, there will be a lack of low end bass at the middle of the room, conversely if you EQ for flat near the middle of the room the bass will be very flabby and boomy at the bottom end near the ends of the room.
With subs in phase at both ends of the room you actually get a reduction in bass at the bottom end when listening at the end of the room so more power is required (because the modes aren't boosting the level) but when EQ'ed for a flat response it is far more consistent at both ends and middle of the room...
Mode cancelling only works at fundamental and odd order modes mind you, it does nothing for even order modes such as 2nd order (one wavelength) but they're usually less of a problem than the fundamental modes.
The sharp dips that can be so problematic aren't due to room modes but boundary cancellations. In the 60-100Hz region the reflection from the ceiling is a major cuprit, from 100-200Hz its usually the front wall (behind the speaker) and side walls that are the guilty party.
Dispersing subs around the room makes it far less likely you will get deep nulls because the geometry makes it almost impossible for the various different arrivals and their reflections to cancel out, due to the distribution in phase and amplitude.
Do you have any measurements to back that up ? Sounds like just an intuitive (but probably wrong) speculation to me.
One of the problems with bass produced from only one or two discrete locations at the far end of a room (such as 2 main speakers) is that reflections can and usually do cause frequency regions where there is a large amount of excess phase (as much as 80ms is common) - meaning that the response is non-minimum phase, with sharp phase transitions near notches in the response.
This screws up the phase response no matter how much minimum phase EQ you apply to try to sort out the amplitude response. One of the things that properly placed multiple subs can do is eliminate these large peaks of excess group delay, bringing the bass response at the listening position back to being near minimum phase, which can then have EQ successfully applied to it to correct both the amplitude and phase response.
Another point to consider is that if you compare a system with woofers only at the far end of the room to a multi-sub one that also includes woofers at the listener end of the room at low frequencies, the net effect of that is to add a slight time lead to the low bass frequencies. (some of the bass producing speakers are closer to the listener than the main speakers)
This is a good thing, because the bass response of most speakers has considerable group delay at low frequencies, anywhere from 10-30ms or so...so this can help to reduce the effective group delay at low frequencies, which should in theory lead to a better phase response...(provided these rear woofers aren't crossed over too high where the group delay is lower than the path length time delay of the rooms length...)
Last edited:
Simon,
What you say is about the peaks and dips is true. However, please be reminded that we are talking about flattening the response, so it is very unlikely that you will place subs like that.
About the phase not being coherent with the original instrument, this is most likely to be the case in most minimum phase systems unless effort is made in linearize the phase. Multiple subs just further trades this off for a flatter in room low frequency response. Some subs will be closer to the listener, this is true, but generally the low frequency end already has a phase lead, plus the fact you now have multiple phase of the same pass band, which really messes the perception further.
I feel that group delay data is very confusing, I have not yet been able to relate such data directly with specific sonic coloration when I look at the full audio band. When I only look at the lower frequencies, the room reflections make the measured data meaningless.
What you say is about the peaks and dips is true. However, please be reminded that we are talking about flattening the response, so it is very unlikely that you will place subs like that.
About the phase not being coherent with the original instrument, this is most likely to be the case in most minimum phase systems unless effort is made in linearize the phase. Multiple subs just further trades this off for a flatter in room low frequency response. Some subs will be closer to the listener, this is true, but generally the low frequency end already has a phase lead, plus the fact you now have multiple phase of the same pass band, which really messes the perception further.
I feel that group delay data is very confusing, I have not yet been able to relate such data directly with specific sonic coloration when I look at the full audio band. When I only look at the lower frequencies, the room reflections make the measured data meaningless.
Last edited:
Ok, on a totally different note, I was wondering- the new speakers need a name. So far they have only been the 'Beyond The Ariel' system. But they need a nice name, like Ariel or Amity or Karna. They could be called the Karna speakers (as opposed to the Karna amp), and then from there flow a whole system (Karna pre, etc.), but I leave that up to Lynn. They sound so special, I just don't want to call them 'those speakers' anymore.
Deon
Well, I've considered using my initials - LTO - as a placeholder until I can think of something better. Karna is my wife, partner, and all-around best friend of 25 years, and Amity is my daughter. After that I run out of names.
The auditioning I've done so far is with a collaborator in another part of the country, and the system was fairly unfamiliar and not what I'd use - a 1-watt SET amplifier with a DAC based on the old 16-bit Philips chipset (non-oversampling and all that).
After the Rocky Mountain show here in Denver, I plan on moving forward with my own pair of loudspeakers, connected to the original Karna amplifier, Monarchy DAC with PCM 1704 Burr-Brown chipset, and eventually get a phonograph going as well (Technics SL1205 Mk II and a moving-coil cartridge). I'm entirely happy with the amplifier as-is, as well as the Monarchy DAC; no interest in a different amplifier or DAC, after comparing both with many other amplifiers and DACs. Definitely not a fan of sigma-delta converters, with the exception of the ESS Sabre 9018.
Moving on to Soongsc's question about group delay, I don't find it particularly audible below 500 Hz, except perhaps as a mild timbre shift. Above 500 Hz to 1 kHz, though, it starts to be audible with some kinds of broad-spectrum music. Mid-to-high group-delay variations are not easy to hear with rock music, but are quite noticeable with pink-noise, choral, or large-scale symphonic music - subjectively, it sound like peaking, or too much energy concentrated in a narrow part of the spectrum.
Group delay at very low frequencies is intermingled with room effects, since the ear can no longer separate out the direct-arrival sound. Room nodes - both peaks and nulls - subjectively sound like overhang on rapid drum passages, and interfere with the sense of rhythm of the music. Power-supply problems can also sound fairly similar in this frequency range - blur and overhang.
Last edited:
I was not convinced that group delay was significant for low frequencies, until I sat in a very interesting demo, which I've written about here:
Red Spade Audio: Bathurst 2011 - the audio event of the year
Scroll down to DEQX demo and you will see. In a nutshell, a zero group delay response was dialed in and instantly switched from one to the other with a remote. It was tested blind, level matched, and also with EQ to ensure the response was closely matched. I was the first guinea pig. Everyone who sat through the test heard the same things, and could easily pick which version was corrected. The bass was much tighter, and sounded as if a slightly bloated loudness control had been defeated. The sound stage depth and width was dramatically increased. It was quite an improvement, the uncorrected version sounded broken in comparison. This has led to my interest in Bodzio's Ultimate Equaliser which I see as the way forward, and a more affordable alternative to DEQX.
At some point I'll be repeating this test in my own room and system, as I think the combination of a point source horn with this kind of processing, would be very impressive indeed.
Sorry about the off topic!
Red Spade Audio: Bathurst 2011 - the audio event of the year
Scroll down to DEQX demo and you will see. In a nutshell, a zero group delay response was dialed in and instantly switched from one to the other with a remote. It was tested blind, level matched, and also with EQ to ensure the response was closely matched. I was the first guinea pig. Everyone who sat through the test heard the same things, and could easily pick which version was corrected. The bass was much tighter, and sounded as if a slightly bloated loudness control had been defeated. The sound stage depth and width was dramatically increased. It was quite an improvement, the uncorrected version sounded broken in comparison. This has led to my interest in Bodzio's Ultimate Equaliser which I see as the way forward, and a more affordable alternative to DEQX.
At some point I'll be repeating this test in my own room and system, as I think the combination of a point source horn with this kind of processing, would be very impressive indeed.
Sorry about the off topic!
I sometimes get a bit of heat for not adopting linear-phase crossovers. Unfortunately, 1st-order passive crossovers have quite poor IM distortion performance. The problem is mostly with the highpass portion; since direct-radiator excursion increases at a rate of 12 dB/octave as frequency is decreased, tweeter excursion actually goes up in the band-reject region below crossover. Things are a little better on the lowpass side of things, but LF drivers can have a surprising amount of high-Q breakup crud above 1 kHz. 1st-order crossovers have nice pretty square waves, but IM distortion and power handling are not good - which doesn't show up in time-domain measurements.
Many real-world drivers have substantially more time-domain energy storage than an ideal LR4 crossover. The worst case, unfortunately common in high-end speakers, are poorly designed crossovers (of any slope) combined with rigid-cone drivers with strong breakups. Ripples in the driver response move the acoustical (real-world) crossover away from the ideal, which degrades group-delay response, and also contributes to rapid phase shifts between LF and HF drivers. Poor control of erratic driver rolloffs result in long time-domain decays that almost resemble noise; this type of decay is undesirable, and draws attention to the crossover region.
I take a compromise position of selecting drivers with the smoothest possible band-reject regions, compensating the crossover for ripples in the transition region (mostly to control the phase angle between drivers), and then lowering the Q of the 2nd to 4th order crossovers so group-delay peaking is not audible on sensitive program material. In other words, what I care about is the phase angle between drivers and the overall system settling time.
Many real-world drivers have substantially more time-domain energy storage than an ideal LR4 crossover. The worst case, unfortunately common in high-end speakers, are poorly designed crossovers (of any slope) combined with rigid-cone drivers with strong breakups. Ripples in the driver response move the acoustical (real-world) crossover away from the ideal, which degrades group-delay response, and also contributes to rapid phase shifts between LF and HF drivers. Poor control of erratic driver rolloffs result in long time-domain decays that almost resemble noise; this type of decay is undesirable, and draws attention to the crossover region.
I take a compromise position of selecting drivers with the smoothest possible band-reject regions, compensating the crossover for ripples in the transition region (mostly to control the phase angle between drivers), and then lowering the Q of the 2nd to 4th order crossovers so group-delay peaking is not audible on sensitive program material. In other words, what I care about is the phase angle between drivers and the overall system settling time.
Last edited:
Well, you're up as late as I am!
What do you think of well designed series crossovers? I have a local Boeing engineer friend who takes inexpensive speakers and other than the extra -50db of coherent signal that EnABL drivers have, equals the performance of my speakers. He has won the speaker builder contest that the Pacific Northwest Audio Society holds every two years twice. This is a pretty impressive contest with great care taken in matching every measurable parameter to insure equal opportunity, but the judging is done double blind.
Bud
What do you think of well designed series crossovers? I have a local Boeing engineer friend who takes inexpensive speakers and other than the extra -50db of coherent signal that EnABL drivers have, equals the performance of my speakers. He has won the speaker builder contest that the Pacific Northwest Audio Society holds every two years twice. This is a pretty impressive contest with great care taken in matching every measurable parameter to insure equal opportunity, but the judging is done double blind.
Bud
Series crossovers don't seem to work for me. My high and lowpass sections are almost never symmetrical, and above all, I don't want IM distortion from the LF driver creeping into the HF driver. I take the isolation a little bit further with bi-wiring the crossover, only joining at the speaker terminals of the amplifier.
Many crossover designers seem to have trouble with driver integration, which has never bothered me. Then again, my crossovers have always had the drivers within 10 degrees of each other, and use response shaping to control phase angles in the hand-off region between drivers. I can't take any credit for this approach; Laurie Fincham of KEF taught me this in 1975, and I've stayed with it since then.
Strictly from a nuts and bolts technical perspective, what are the benefits of a series crossover? I really don't want the drivers to interact electrically with each other, and that's what I see when I look at a series crossover. Spreading the distortion signature of one driver across both bothers me at a philosophical level; why would I want woofer distortion in the tweeter, or vice versa?
True, they would sound more alike if they shared distortion spectra, but there are cleaner ways to assure driver integration. If a series crossover sounded better subjectively, I would be very curious exactly why that's happening.
Many crossover designers seem to have trouble with driver integration, which has never bothered me. Then again, my crossovers have always had the drivers within 10 degrees of each other, and use response shaping to control phase angles in the hand-off region between drivers. I can't take any credit for this approach; Laurie Fincham of KEF taught me this in 1975, and I've stayed with it since then.
Strictly from a nuts and bolts technical perspective, what are the benefits of a series crossover? I really don't want the drivers to interact electrically with each other, and that's what I see when I look at a series crossover. Spreading the distortion signature of one driver across both bothers me at a philosophical level; why would I want woofer distortion in the tweeter, or vice versa?
True, they would sound more alike if they shared distortion spectra, but there are cleaner ways to assure driver integration. If a series crossover sounded better subjectively, I would be very curious exactly why that's happening.
Last edited:
It is true that there are lots of time delayed release of stored energy in drivers. One thing that I have found when using the Ultimate Equalizer from Bodzio Software is that the process of designing crossover using that actually reduces the effect of this delayed release of stored energy resulting in a much cleaner CSD performance. I have to credit Joachim Gerhard to bringing this to my attention.
Series XOs really sound nice and seem to integrate the drivers together. I cannot really say why it is, but I was inspired to try it after reading reports on The Ultimate Monitor (if my memory had not failed me). I was impressed with the results.
Series XOs really sound nice and seem to integrate the drivers together. I cannot really say why it is, but I was inspired to try it after reading reports on The Ultimate Monitor (if my memory had not failed me). I was impressed with the results.
The driver stored energy thing had been discussed long before we had the UE. I presented these examples many years ago. If you can eq the driver to a specific target then the CSD of the acoustic output will be the same as that of the ideal target, at least at the design point. This was one of the things I used to get Bohdan to pull the DSP out of SoundEasy leading to the UE.
Good thing to point out, and one of the key reason huge high output visually dominating subs tend to disappear sonically. This is also one of the reasons isobaric subs work so nicely- even though the even order distortion that they cancel is not tonally offensive, it DOES draw attention to the location of the sub and thus any reduction is highly desirable.
How about the name Leira. As it happened, you went back - almost to the very beginning of the hi-fi era in order to advance the art from your previous effort (I am thinking here of the Altec large format two-way). So would it be "back to the future" for the future arieL?
John, while letting the acoustic target match the ideal can improve CSD performance, the ideal target is not ideal. I think this would be more evident if you used a full range driver to do testing. I would like to see someone do this with a full range driver without any DSP.
An externally hosted image should be here but it was not working when we last tested it.
Well, there certainly is a "Back to the Future" aspect to my projects. It's not anything I planned, it just came out that way. The Ariel drew on the TL loudspeakers I did for Audionics back in 1975 as well as the crossover techniques imparted by Laurie Fincham, the Amity and Karna drew on research about the Western Electric 86A and 92 amplifiers, and the LTO on research about Bell Labs, ERPI, and the original Altec team during the 1930's and 1940's.
What was surprising about these various audio-archeology projects was how much knowledge had been lost over the decades. The monoculture of the 1947 Williamson wiped out prior art in balanced-topology direct-heated amplifiers, and more importantly, the Bell Labs' "Harmonic Equalizer". Some of the more subtle aspects of the early Altec work was lost with the introduction of the cost-saving sectoral horn, followed by JBL's transition from underhung to overhung voice coils, along with the transition to ceramic magnets. There was also a massive loss of fidelity (and visual quality) as the 1950's 70mm 2.35:1 Technicolor widescreen with six-magnetic-track audio was replaced by 1970's 35mm matted 1.85:1 Eastmancolor semi-widescreen with mono optical soundtracks. The THX team at Lucasfilm discarded nearly everything known about quadraphonic sound in the early Seventies, leading to the absurd proliferation of channels and asymmetric soundfields created by an array of loudspeakers with dissimilar radiation patterns.
We can certainly measure and simulate far better than ever before, but many of the technological changes were done for purely cost-savings reasons, and were not technological improvements at all. I also discovered a number of interesting little byways - that Ampex nearly introduced a discrete 3-channel prerecorded tape format in the mid-1960's, that the Altec team went back and forth on the merits of circumferential vs radial phase-plugs, that the lineage of Radian went back to Emilar, and before that, to patents filed by two Altec engineers (Hilliard and Renkus) in 1966, and that Paul Klipsch was using autoformers for horn attenuation as early as 1959.
Speaking of "Back to the Future", here's a neat little picture of Les Paul, the pioneer of multitracking, with a Lansing Iconic in the background. Although taken in the early 1950's, multitracking was to become the standard of the industry in decades to come, and the Iconic harks back the late 1930's.
What was surprising about these various audio-archeology projects was how much knowledge had been lost over the decades. The monoculture of the 1947 Williamson wiped out prior art in balanced-topology direct-heated amplifiers, and more importantly, the Bell Labs' "Harmonic Equalizer". Some of the more subtle aspects of the early Altec work was lost with the introduction of the cost-saving sectoral horn, followed by JBL's transition from underhung to overhung voice coils, along with the transition to ceramic magnets. There was also a massive loss of fidelity (and visual quality) as the 1950's 70mm 2.35:1 Technicolor widescreen with six-magnetic-track audio was replaced by 1970's 35mm matted 1.85:1 Eastmancolor semi-widescreen with mono optical soundtracks. The THX team at Lucasfilm discarded nearly everything known about quadraphonic sound in the early Seventies, leading to the absurd proliferation of channels and asymmetric soundfields created by an array of loudspeakers with dissimilar radiation patterns.
We can certainly measure and simulate far better than ever before, but many of the technological changes were done for purely cost-savings reasons, and were not technological improvements at all. I also discovered a number of interesting little byways - that Ampex nearly introduced a discrete 3-channel prerecorded tape format in the mid-1960's, that the Altec team went back and forth on the merits of circumferential vs radial phase-plugs, that the lineage of Radian went back to Emilar, and before that, to patents filed by two Altec engineers (Hilliard and Renkus) in 1966, and that Paul Klipsch was using autoformers for horn attenuation as early as 1959.
Speaking of "Back to the Future", here's a neat little picture of Les Paul, the pioneer of multitracking, with a Lansing Iconic in the background. Although taken in the early 1950's, multitracking was to become the standard of the industry in decades to come, and the Iconic harks back the late 1930's.
Attachments
Last edited:
3" full range driver, since the driver is made to custom order, none are commercially available. I think one could give the Jordan driver a try.
Frequency and phase response after UE
An externally hosted image should be here but it was not working when we last tested it.
CSD
An externally hosted image should be here but it was not working when we last tested it.
Measurement point is same as design point, 1M on axis.
Last edited: