My sentiment too in my own domain. Even a cross over design can take over 6 months <and then> more.
Don't you automatically have that when you do frequency response measurements?
Uh, no. A minimum-phase system has a fixed identity between the impulse response and the frequency response, so correcting frequency errors via equalization also corrects time errors as well. This is true of RIAA equalization and FM pre-emphasis and de-emphasis; the pre-EQ and de-EQ process result in flat time response. Outside of the HF breakup region, single drivers mounted on large flat panels are also pretty close to minimum phase.
However ... once reflections, diffraction, and time-alignment problems enter, then the system is no longer minimum-phase. When this happens, correcting these errors through equalization makes the time errors worse, not better.
A rule-of-thumb in audio is to correct problems in the domain where they occur. Frequency response deviations that are the result of resonance, or an overall tilt in response, lend themselves to correction via equalization in a passive crossover or an active, analog crossover.
Reflections, diffraction, and time-alignment errors are another matter. Frequency-based equalization does not remove time echoes or driver misalignment.
Digital correction via FIR algorithm is not a minimum-phase process, so digital correction can correct time and frequency-domain errors independently.
Although digital FIR correction has the charm of solving time and frequency-domain problems, this is only true for one point in space. One of the annoying problems with reflection, diffraction, and driver time-alignment errors is a significant spatial component to the error. In other words, if one point in space is fully corrected, another point a foot or two away could be quite bad.
This is why it's a very good idea to slowly sweep the microphone while looking at the impulse response, or measure at 5-degree intervals. It's been my experience that diffraction from sharp cabinet edges does not uniformly spray off the cabinet edge, but is concentrated at certain angles, and can be readily detected if you are looking for it. Since cabinet-edge diffraction creates false secondary images, as well as comb-filtering in the frequency domain, it's worthwhile to take steps to minimize the unwanted emission.
Returning to FR measurements, it has to be remembered that FR measurements discard time and phase information. If you see a comb series (a periodic series of peaks and dips), it might be caused by a reflection, but it might not. You don't actually know until you look at the impulse response, where reflections are pretty obvious. If you know the time delay of the reflection, you can then look around at the driver, horn, or enclosure, figure out what's causing it, and take steps to remove it.
Last edited:
How can this new speaker have a fast decay and a tidy impulse? The distances between the drivers are huge. Even if you get some kind of phase overlap between the horn and the woofer, that RAAL supertweeter you have is going to completely mess up your impulse response. The audibility of fast decay and a tidy impulse is debatable (above the room transition frequency, we hear the frequency response bump not the time domain ringing), but you've certainly made it a priority. I just don't see how the current design can achieve it.
In my understanding of the phenomenon and of how the human perception of sound works, phase matching and time alignment only matter at those frequencies at which we are sensitive to phase, i.e. up to ~3 or 4 kHz.
The impulse of a supertweeter crossed over at >7kHz may well arrive a few ms before the main impulse of the Woofer + Horn (within reason), without impairing the subjective perception of a well-behaved behaviour in the time domain.
In my experience, we are also completely insensitive to the ensuing comb filtering in the combined frequency response, again provided that this happens at very high frequencies only.
Marco
Last edited:
Marco,
Thanks! This has been mentioned before but I haven't seen any evidence of it. Got some sources? Remember that we are sensitive to the frequency response. Comb filtering due to reflections may be inaudible, but this is not comb filtering. This is the direct sound arriving at vastly different times resulting in a ragged response.
Besides, it will completely ruin the impulse response tidiness that Lynn is after. There are better approaches to getting a tidy impulse response.
Thanks! This has been mentioned before but I haven't seen any evidence of it. Got some sources? Remember that we are sensitive to the frequency response. Comb filtering due to reflections may be inaudible, but this is not comb filtering. This is the direct sound arriving at vastly different times resulting in a ragged response.
Besides, it will completely ruin the impulse response tidiness that Lynn is after. There are better approaches to getting a tidy impulse response.
ra7,
I don't think that an absolutely fast decay in a waterfall is what you are looking for though that would be an added bonus if it can be achieved. What I think, or at least I am looking for is the lack of any small band that is hanging on, this will show up in the frequency response as a peak. You don't want these high Q resonances to hang in the waterfall, that is something to avoid.
I also agree that all the talk about phase alignment and time alignment in the system that Lynn is talking about just don't seem to align with what I understand. I agree that with the disparate center to center distances between devices that you could only make this work if at all at one exact location, this has always been the argument and problem with so called time aligned and phase aligned systems. The only way that I know of to make that work and even then you need some electronic correction is with a coaxial speaker arrangement. A Tannoy or something like that comes as close as we can get but you still have the depth issue with the cone and compression driver, that can be taken care of by careful design in an active crossover or perhaps with very difficult tuning of a passive crossover. Having flat phase and time alignment in a distributed speaker system is almost impossible really.
This is my only argument against multiple round waveguides, the center to center distances become to great, those who say you can't hear that I just don't understand unless your listening distance is very large. Anything practical in a room for listening distance you can hear, that is if there is more than one round horn and the lower one is rather large as many seem to use.
I don't think that an absolutely fast decay in a waterfall is what you are looking for though that would be an added bonus if it can be achieved. What I think, or at least I am looking for is the lack of any small band that is hanging on, this will show up in the frequency response as a peak. You don't want these high Q resonances to hang in the waterfall, that is something to avoid.
I also agree that all the talk about phase alignment and time alignment in the system that Lynn is talking about just don't seem to align with what I understand. I agree that with the disparate center to center distances between devices that you could only make this work if at all at one exact location, this has always been the argument and problem with so called time aligned and phase aligned systems. The only way that I know of to make that work and even then you need some electronic correction is with a coaxial speaker arrangement. A Tannoy or something like that comes as close as we can get but you still have the depth issue with the cone and compression driver, that can be taken care of by careful design in an active crossover or perhaps with very difficult tuning of a passive crossover. Having flat phase and time alignment in a distributed speaker system is almost impossible really.
This is my only argument against multiple round waveguides, the center to center distances become to great, those who say you can't hear that I just don't understand unless your listening distance is very large. Anything practical in a room for listening distance you can hear, that is if there is more than one round horn and the lower one is rather large as many seem to use.
I agree with your observations, kindhornman. I had built a similar system to Lynn's and I could easily hear the disparate sources. You need a very large listening room for such a large system to work. In my space, which is quite large actually and dedicated, it simply isn't working. It is true that the SPL horns can produce is addictive. But the problems are there as well. I am working on a coax, corner loaded expanding array (array gets longer with reducing frequency). The coax unit (KEF Q100) has a very smooth frequency response, and augmented by low-frequency support sounds better than anything I've built before. Now, if you can remove the objectionable floor and ceiling reflections, you have actually done something to improve the sound.
Yes, you don't want resonances to hang on, but they are revealed in frequency response as well. As I understand it, this is one of Lynn's goals for each driver, but he also wants a fast decay and tidy impulse for the whole system. What is the point of having fast decay for each driver and having a wretched system impulse response?
Danley's Synergy and the Dunlavy speaker achieved the nearly impossible goal of flat phase. But is it audible? As long as you get good phase overlap through the crossover region, you are good.
Using the Rephase tool by member Pos here on diyaudio, you can flatten the phase. Give it a try and see if you can reliably detect flat phase. But do it blind. I thought I was reliably hearing an improvement with flat phase, but the blind test revealed I was fooling myself.
Yes, you don't want resonances to hang on, but they are revealed in frequency response as well. As I understand it, this is one of Lynn's goals for each driver, but he also wants a fast decay and tidy impulse for the whole system. What is the point of having fast decay for each driver and having a wretched system impulse response?
Danley's Synergy and the Dunlavy speaker achieved the nearly impossible goal of flat phase. But is it audible? As long as you get good phase overlap through the crossover region, you are good.
Using the Rephase tool by member Pos here on diyaudio, you can flatten the phase. Give it a try and see if you can reliably detect flat phase. But do it blind. I thought I was reliably hearing an improvement with flat phase, but the blind test revealed I was fooling myself.
Last edited:
I had a pair of the big Dunlavy 5's for a while and found the only thing they did "good" was image - otherwise boring, sterile, over priced and surely over hyped by the magazine pimps. They barely hinted at realistic reproduction. I'm surely glad I ran away from cones and domes
If you are serious about great sound fully horn loaded is an excellent way to achieve it. The Danley horns I heard were home made. The brief listen i had was quite enjoyable.
Before we all go off the deep end, let's look at Dr. Geddes' Summa, a loudspeaker that has very low diffraction and superb measurements from nearly any viewpoint.
It has a 15" B&C woofer and an OS waveguide with a mouth just about the same size as the woofer (an important design feature). That puts the center-to-center distance between the woofer and the HF section about 16" apart. The crossover is at 950 Hz, giving a distance of 1.12 wavelengths at the crossover frequency.
The new loudspeaker has a 15" woofer and the outer diameter of the AH425 is 16.25". I find that 3" of spacing between the top of the cabinet and the outer edge of the AH425 is desirable, and the woofer is mounted close to the top of the bass cabinet. This results in a center-to-center distance of 20". The crossover is at 700 Hz, giving distance of 1.03 wavelengths at the crossover frequency.
The new loudspeaker, as has been pointed out earlier, is hardly novel. This format, and even the crossover frequency, go back to the Lansing Iconic of the mid-Thirties. There are limitations in cone materials, and the size of the MF/HF horn, that are the same now as they were then, so center-to-center distances and choice of crossover frequency are going to be similar, whether it was built by Jim Lansing, Dr. Geddes, or the collaborators in Seattle, Canada, or Australia.
If you feel that non-coaxial large-format 2-way systems are unworkable, or require very large listening rooms, be my guest. But plenty of other people feel otherwise.
It has a 15" B&C woofer and an OS waveguide with a mouth just about the same size as the woofer (an important design feature). That puts the center-to-center distance between the woofer and the HF section about 16" apart. The crossover is at 950 Hz, giving a distance of 1.12 wavelengths at the crossover frequency.
The new loudspeaker has a 15" woofer and the outer diameter of the AH425 is 16.25". I find that 3" of spacing between the top of the cabinet and the outer edge of the AH425 is desirable, and the woofer is mounted close to the top of the bass cabinet. This results in a center-to-center distance of 20". The crossover is at 700 Hz, giving distance of 1.03 wavelengths at the crossover frequency.
The new loudspeaker, as has been pointed out earlier, is hardly novel. This format, and even the crossover frequency, go back to the Lansing Iconic of the mid-Thirties. There are limitations in cone materials, and the size of the MF/HF horn, that are the same now as they were then, so center-to-center distances and choice of crossover frequency are going to be similar, whether it was built by Jim Lansing, Dr. Geddes, or the collaborators in Seattle, Canada, or Australia.
If you feel that non-coaxial large-format 2-way systems are unworkable, or require very large listening rooms, be my guest. But plenty of other people feel otherwise.
Last edited:
POOH,
I have often wondered about the actual sound of the Danley's. It just seemed that the idea was directed at constant coverage above all else. Your description confirms what I suspected. Perhaps for large spaces and PA applications they are good but at the loss of sound quality I think there are other products in Pro-audio PA that would have a better sound quality.
I have often wondered about the actual sound of the Danley's. It just seemed that the idea was directed at constant coverage above all else. Your description confirms what I suspected. Perhaps for large spaces and PA applications they are good but at the loss of sound quality I think there are other products in Pro-audio PA that would have a better sound quality.
I had a pair of the big Dunlavy 5's for a while and found the only thing they did "good" was image - otherwise boring, sterile, over priced and surely over hyped by the magazine pimps. They barely hinted at realistic reproduction. I'm surely glad I ran away from cones and domes
Horns are good when piercing Wax, much different after a good clean...
Lynn,
Agree on your choice of spacing and crossover frequency, though I am not a big fan of midrange out of a compression driver, that is your choice to make. The real problem comes when you add another horn to augment that horn for the top octaves, then the spacing becomes the issue. A two way system would be simpler and not have to deal with that problem. As long as you are basically listening on or close to the axis you can meet your goals with this type of system and I think that is your plan. Probably one of the reasons that I still have those old Altec Barcelona speakers sitting after all these years.
Ps. Sorry I read the Dunlevy as Danley. But I still wonder about the sound quality when the driving force of the Danley is coverage angle. I'll ask my ex brother in law about how they sound, he's still is in the Pro audio industry and has heard them all.
Agree on your choice of spacing and crossover frequency, though I am not a big fan of midrange out of a compression driver, that is your choice to make. The real problem comes when you add another horn to augment that horn for the top octaves, then the spacing becomes the issue. A two way system would be simpler and not have to deal with that problem. As long as you are basically listening on or close to the axis you can meet your goals with this type of system and I think that is your plan. Probably one of the reasons that I still have those old Altec Barcelona speakers sitting after all these years.
Ps. Sorry I read the Dunlevy as Danley. But I still wonder about the sound quality when the driving force of the Danley is coverage angle. I'll ask my ex brother in law about how they sound, he's still is in the Pro audio industry and has heard them all.
Last edited:
Well, Dr. Geddes doesn't make any claims for a fast decay or tidy impulse, nor does he employ an awkwardly positioned supertweeter. He prioritizes frequency response and off-axis uniformity, which are goals worthy of pursuing. The goal being pursued here is a fast decay. Even with the large CTC distance, you could get a reasonable impulse response if the phase between the two drivers is properly overlapped. But throw that supertweeter in there and it will without doubt result in a poor impulse response.
If you really wanted a fast decay and a perfect impulse, the two known approaches are Danley's synergy and the Dunlavy. If you want to play at realistic (loud) SPL this new speaker can do it. But it won't have a tidy impulse response. In fact, the Ariel could have a better impulse response than this speaker.
Can one of the builders of this system post the system impulse response?
Lynn,
I want to relate my experience to you and maybe you'll appreciate my viewpoint better. I had a system very similar to yours and I grappled with similar issues. The large horns sound very good. Very lifelike. But they have problems too. The dispersion gets narrow with increasing HF. Besides, you do need a solution to the higher frequencies. I like to prioritize flat frequency response on axis and a uniform off-axis response. So, a separate supertweeter is not for me. Therefore, large horns are out.
Then I tried constant directivity horns. I have everything from SEOS-12 to the SEOS-24 with a 1.4" throat that can go quite low, but somehow I could never get the balance right. To negate the increased power in the HF, you have to make the on-axis response droop. And this trade-off just doesn't sound right to me. Besides, the constant directivity horns require quite a bit of EQ to get a flat response and there's only so much you can do (even with DSP).
Then I had an experience that changed my outlook. I had built cabinets for the KEF Q100 driver. I set it up in front of the big horn system and it blew away the horn system. In realism, in soundstage depth, width, balance. It was so good that at reasonable volumes you couldn't tell whether the large horn system was playing or the little coaxes were playing. In hindsight, it was quite simple why they sounded so good. Flat, smooth frequency response, smooth off-axis response, and a small source. Yes, it doesn't go as loud as the big horn system, but who cares? It is very rewarding both spatially and on all kinds of music.
So, in the end, if you have a large room to fill or you like to listen loud, horns can work. But otherwise you are always grappling with CTC problems or with frequency response flatness and smoothness.
Last edited: