Directivity is a good thing providing that it is controlled and uniform. This is what wave-guide based, or coaxial orientated designs do well. In a coaxial design the cone of the mid range driver acts as a wave guide and provides a decent amount of directivity control.
The Tannoy dual-concentric designs are examples of this and Tannoy cross over at a frequency that results in a good directivity match between the falling response of the mid/bass unit and the region of the tweeters response that is falling at the same rate as the mid/bass units response. What happens here though is that the tweeter stops falling and then projects the sound out in a very specific cone like beam. This reduces early room reflections and makes designs very tolerant of the environment in which they are placed. The wave guide loading also helps to maintain the off axis response over the usual range of angles that one would listen at giving you a sound that doesn't change as you alter where you are sitting.
The other large advantage is that the directivity match means that the off axis response is smooth and even. The B&W present with the typical wide-narrow-wide profile associated with drivers that are crossed over too high. In other words the dispersion at low frequencies of the FST is extremely wide, the wavelength of the sound that its producing is huge compared to the cone diameter and it thus radiates out omnidirectionally. Then as you go up in frequency the wavelength reduces, the size of the cone becomes significant and the off axis response begins to fall and the driver starts beaming. Quite literally it turns into a lighthouse. Then the tweeter comes in, where it's now operating as the mid range once was, that is omnidirectionally, throwing sound out across all angles. This is what is meant by the typical wide-narrow-wide dispersion profile. You start off wide at the bottom of the mid range drivers range, then the pattern narrows as the driver starts to beam and then you crossover to the tweeter causing the dispersion to suddenly widen again.
While this goes on in the off axis, the on axis response remains perfectly flat and it's this reason, the difference that this creates between the direct sound and the reflected sound (created by the off axis reverberating around the room), why lots of people feel the sound improves significantly because of room treatment designed to absorb and reduce the effect of the reflected sound. When the off axis response mirrors the on axis response room treatment becomes less necessary and if you've got a system that uses controlled directivity, like the Tannoys, you often find you need no room treatment whatsoever.
It depends. As a first pass design, yes I could simply redesign the mid range crossover to control the issues that the FST presents with and most likely reuse most of the crossover components. Art is currently on the way to doing this quite successfully I believe, but I get the impression that the loudspeaker would benefit significantly from a redesign of both the tweeter and the mid range crossover.
The idea here would be to build a 'perfectionist' type crossover that aims to provide the flattest on axis response, controlling any resonances and ensuring excellent phase integration between the two drivers.
Primarily I would like to lower the crossover point down so as to achieve the above, whilst also vastly improving the off axis response. But that would require knowing how capable the tweeter is and would require a distortion measurement be made at a suitable drive level.
Exactly, the B&W is designed in such a way that makes the crossover easily accessible and easy to modify. If measurements are made in the correct way then there is absolutely no reason why I cannot design a crossover at a distance that one can use as the basis for a decent comparison.
For what its worth I have removed the phase plug from my FSTs and converted them into a coaxial 😀
I mean try a new XO for the midrange but I wouldn't want to push myself to start taking parts off the design of the drivers itself, otherwise I start doing like the other folk here who replaced all the midrange and tweeter with other drivers on his 801. Perhaps the new drivers he has are really better, who know, but if I would ever feel the need to do it would mean that I no longer like the sig800 and I would simply look out for a better speaker and sell mine.
Now regarding the XO, 5th you are a gentlemen in sharing your knowledge and your brain power.
LEt me ask this though: in taking the measurements for the drivers, how would you get a real measurement considering that I don't have a perfect room although it is treated.
I am assuming that when B&W takes measurements they have
1) precise and calibrated equipment.
2) controller room with no resonances nor reflections that will interfere with the measurements itself.
Does my objection make any sense?
Now regarding the XO, 5th you are a gentlemen in sharing your knowledge and your brain power.
LEt me ask this though: in taking the measurements for the drivers, how would you get a real measurement considering that I don't have a perfect room although it is treated.
I am assuming that when B&W takes measurements they have
1) precise and calibrated equipment.
2) controller room with no resonances nor reflections that will interfere with the measurements itself.
Does my objection make any sense?
😀
I figured why not give it a go eh? The FST seemed like the perfect choice for a decent coaxial as it has a very smooth profile without any irregularities, like half roll surrounds, to get in the way. The trick was to find a tweeter that would work suitably well. The ND16 from Dayton Audio worked as a proof of concepts, but wasn't ideally suited. I managed to procure some new old stock Aura NT1 titanium dome tweeters and after carefully milling them down, they fitted perfectly. These tweeters are little gems and have very low distortion within the range I wish to use them over and also have nice underhung motors. Plus the rigid dome gives pistonic behaviour which I am partial too.
Here's a shot of the loudspeaker itself.
I replicated B&Ws enclosure for the FST out of polyester resin filled with marble dust. This is very dense! The bass module uses Jantzens JA8008-HMQ in a ported, floor loaded, enclosure with the over sized port located in the base of the loudspeaker and firing down towards the floor. They provide solid extension down to 50Hz with 95dB re 8 ohm sensitivity up until the 300, where the FST takes over, offering similar or greater sensitivity, but at a 4 ohm load. Wave guide loading the NT1 with the FST cone gives it much higher sensitivity around the xover frequency, so it operates up to around 94dB sensitivity at its maximum (around the xover frequency) before gradually declining down to its usual 88dB by 20kHz. This set up is very dynamic with ridiculous levels of transparency and due to the coaxial, wave guide loading has awesome off axis response.
Here is how the on axis response looks, which is quite something considering the coaxial nature. The ND16 didn't look anywhere near as nice in the top octave, which is usually one of the accepted trade offs when using wave guides, you get suck outs, on axis, in the top octave. Not so with the NT1!
I doubt the drivers would be better. The FST is quite simply one of the best midrange drivers in the world, period and from what I've seen B&Ws tweeters are pretty good stuff too. The fact that chap brutalised his B&Ws in that way shows a clear misunderstanding of good loudspeaker design.
Now I know that the mid range drivers crossover is located in the plinth of the loudspeakers, and is therefore it is very easy to bypass, but isn't the tweeters crossover located in the same place? I had assumed that bypassing the tweeters crossover would be just as simple as bypassing the FSTs.
I am not asking you to physically alter the design in any way except for trying out different crossovers.
The measurement guide I linked to before in this thread details exactly how you go about taking measurements within the home environment so that they are suitable for xover design.
ARTA, the program I wrote the guide for, has decent help files to describe how you need to have your hardware configured. But for taking frequency responses all you really need is a microphone and line level out microphone pre amp.
Attachments
Quick reply I am on my way to Chicago:
1) beautiful job with ur speakers it really looks interesting I wish I could listen to it to give you my feedback but measurements are near perfect wow
2) xo on 800 is at the base of the plinth and all 3 XOs are there: low mid tweeter
1) beautiful job with ur speakers it really looks interesting I wish I could listen to it to give you my feedback but measurements are near perfect wow
2) xo on 800 is at the base of the plinth and all 3 XOs are there: low mid tweeter
I've been afflicted with a flu virus for the past few days, hence my absence.
Today though I got around to taking some measurements of the tweeter response with the (simple) B&W crossover. You'll see that it's not quite the ideal second order L-R response, but the main problem is the untreated resonance at 800Hz.
So I decided that some impedance correction might be worth implementing. (Note the before / after plots attached)
I'll hook the system back up a little later and will report any subjective changes. Overall, measurements wise it's not too different, although the response between 800 and 2000 Hz appears to be a bit smoother. (I'll provide the before and after's in a subsequent post)
Today though I got around to taking some measurements of the tweeter response with the (simple) B&W crossover. You'll see that it's not quite the ideal second order L-R response, but the main problem is the untreated resonance at 800Hz.
So I decided that some impedance correction might be worth implementing. (Note the before / after plots attached)
I'll hook the system back up a little later and will report any subjective changes. Overall, measurements wise it's not too different, although the response between 800 and 2000 Hz appears to be a bit smoother. (I'll provide the before and after's in a subsequent post)
Attachments
Last edited:
Hi Art...I was wondering what happened to you! Apparently this year lots of people have been infected with this flue virus...me included!! Glad to hear you are doing better!
Please let us know how the auditioning test went and what you are exactly auditioning i.e. modified XO for tweeter or only midrange.
Thanks for sharing.
Please let us know how the auditioning test went and what you are exactly auditioning i.e. modified XO for tweeter or only midrange.
Thanks for sharing.
Definitely worth implementing. Looking at the tweeters plot I am surprised that B&W didn't implement this themselves. Running a simple 1 cap xover on a tweeter provides minimal protection vs overload and this is especially true where the tweeters resonance is concerned. As the impedance increases at resonance the effect the cap has diminishes significantly, which is why the response deviates away from the L-R slope and exposes the tweeter to low frequency stuff that it should really not be seeing. Impedance compensation is usually employed specifically to prevent this from occurring.
Like in your case here it is usually out of band, so has minimal impact on the overall summed frequency response, but the tweeter will definitely appreciate it.
I wonder if the extra impedance correction will impact sonic performance and this is the reason why B&W didn't do it.
Although I am more inclined to think at this point they don't do it because they can cut costs a bit down and yet it won't show up on the audible frequency plot.
Although I am more inclined to think at this point they don't do it because they can cut costs a bit down and yet it won't show up on the audible frequency plot.
Not putting it in will only stress the tweeter more around resonance and increase distortion, especially intermodular.
This is relatively far out of band, but nevertheless the impedance compensation network cuts the output by 10db! Around resonance. This is not insignificant as the tweeter will most definitely not be happy playing signals that low and will already show increased distortion down there that'd you'd rather not be exposed to. Granted the diamond tweeter is likely going to have a rather beefy motor with a healthy amount of linear excursion, but nevertheless, at high listening levels it could quite easily be a problem.
One of the problems with shallow first order electrical filters is that they generally require a lot of additional components around them to ensure that they function correctly. Most of the time this is impedance compensation as resonances and rising responses prevent the single component high/low pass from doing what it's supposed to be doing, this is even if the frequency response is perfectly flat.
Dynaudio are famous for using 1st order filters and their poly cone drive units are certain suitable as far as frequency response goes, but their filters are anything but simple due to the compensation required to make the simple low pass/high passes work correctly.
It's not the end of the world that B&W haven't included impedance compensation to flatten the impedance at resonance on the tweeter. But including it will only improve performance.
It is a cost cutting exercise and even though the B&Ws cost £18000 here in the UK, due to the way the industry works it essentially means that B&W build the entire loudspeaker for around £2000. This isn't a lot of money and B&W are still going to build their loudspeakers with a budget in mind. It also helps to explain why some other loudspeakers end up costing £100,000. Of course I would much rather that B&W charge a bit more and include the necessary filters to improve the loudspeakers performance. B&W would not it seems and including an impedance compensation network out of basic, cheap, crossover components (because resonance/impedance compensation networks don't require anything more) wouldn't fly well with the marketing department.
yeah the ratio you mention makes sense. Any company has about 1:6-10 times the BOM price.
Mine is the Signature and doesn't have first order on the tweeter nor diamond tweeter. does this make any difference on mine as far the compensation network required for the Art's diamond is concerned?
I also get that by not having the tweeter undergoing peak resonance will lower its distortion abilities at high level on the audio band too..
Mine is the Signature and doesn't have first order on the tweeter nor diamond tweeter. does this make any difference on mine as far the compensation network required for the Art's diamond is concerned?
I also get that by not having the tweeter undergoing peak resonance will lower its distortion abilities at high level on the audio band too..
It makes a considerable difference actually although I'd have to try and model the way the tweeter xover performs. It's possible that B&W are actually using the same target acoustic slope (or similar) in the sig as they are for the standard dia 800, but they need the extra crossover components to get there with the metal dome and not the diamond. Still the extra xover components would help in hitting the target slope too.
It's not actually about the resonance itself, it's more about how the resonance interacts with the crossover filter. The impedance peak associated with the resonance causes the filter to stop working properly and as a result doesn't protect the tweeter as much as it should. This results in the tweeter getting more signal than it should down at resonance and this is what causes the increased distortion.
If one were to use an active crossover this would not happen because the loudspeakers impedance does not interact with the way the filter works.
This is one of the reasons why people traditionally say crossover at least one octave above resonance and preferably two. This is usually in conjunction with 2nd order or 4th order filters, so with their steeper slopes the tweeters resonance is pushed so far out of band that it doesn't matter how it interacts with the filter. If you go first order electrical, ie a single cap, then you need to compensate everything otherwise the filter will not work properly.
Can you spend a couple of words on how active crossover and multi-amping work?
One thing that I ignore is: provided I have 3 matching, gain-wise, set of monoblocks that for the sake of the consideration we will assume to be 26dB, assuming also that each monoblock drives one driver: 200W on the woofer, 30W on the mid and 30W on the tweeter and the active crossover will have 3 outputs which will have cut offs that match the existing crossover and same slope.
How do I manage to control the proper efficiency for each driver?
I am assuming the crossover will have attenuation where needed to have an overall balance so that the tweeter for example won't be 10db more efficient than the woofer, is that right?
My apologize in advance I don't have much clue as to how multi-amplifying a speaker and I got curious to know how it works and why the active crossover with multi amplification would be better than let's say bi-wiring the speaker with the existing crossover....which btw since b&w has 3 separate crossover on the inside of the 800 they could have also offered a triamping with 3 sets of connectors in the back...but then I am sure they would have gone too much out of the audio standard where people that buy expensive bi-wire cable don't want to think they have to jump a third set....I mean just my theory...which is pretty lame.
One thing that I ignore is: provided I have 3 matching, gain-wise, set of monoblocks that for the sake of the consideration we will assume to be 26dB, assuming also that each monoblock drives one driver: 200W on the woofer, 30W on the mid and 30W on the tweeter and the active crossover will have 3 outputs which will have cut offs that match the existing crossover and same slope.
How do I manage to control the proper efficiency for each driver?
I am assuming the crossover will have attenuation where needed to have an overall balance so that the tweeter for example won't be 10db more efficient than the woofer, is that right?
My apologize in advance I don't have much clue as to how multi-amplifying a speaker and I got curious to know how it works and why the active crossover with multi amplification would be better than let's say bi-wiring the speaker with the existing crossover....which btw since b&w has 3 separate crossover on the inside of the 800 they could have also offered a triamping with 3 sets of connectors in the back...but then I am sure they would have gone too much out of the audio standard where people that buy expensive bi-wire cable don't want to think they have to jump a third set....I mean just my theory...which is pretty lame.
In a normal passive loudspeaker the signal chain would look something like this.
Source>Preamp>Two channel power amp>Passive crossover>Drive units.
In an active loudspeaker it would look something like this
Source>Preamp>Active crossover>Multiple channels of power amplification>Drive units.
With the passive version you are filtering after the amplification and requiring components that work around very low impedances (the impedances of the drive units). With the active crossover you are filtering before the power amplification, requiring an amplifier channel per loudspeaker drive unit and working around high impedances, usually between 1 and 10k. This means the filter components are much smaller and can be made to much tighter tolerances and are also much less expensive. Of course with an active crossover you're introducing active elements, usually opamps, to the signal chain.
The design of high quality active analogue crossovers is not to be taken for granted, but there are resources out there that can offer guidance. It goes without saying that the simpler the active crossover the better, where keeping noise and distortion low is concerned, the more stages you have to use the harder it is to ensure state of the art performance.
If all your sources are digital though then the holy grail is to use a DSP based active crossover as this essentially removes all of the headache of analogue signal design. In this situation you do need as many channels of D/A conversion as you've got drive units and power amps, but this isn't really a problem if you're building the DSP yourself. In fact I would probably recommend using a DSP even if you've got analogue sources and simply invest in a top notch ADC.
With variable gain stages inside the active crossover. Usually you only want to reduce gain as the opposite can increase noise so you'd set unity gain on the channel that goes to the least sensitive drive unit and then turn the rest down to match.
Bi-wiring has been shown to be largely pointless, this is one of the reasons that some of the high-end speaker manufactures provide only one set of terminals to their loudspeakers. Its also because the designers want their loudspeakers to sound the way they intended them too. Giving the user the ability to bi-wire, or really bi-amp, could throw things out of balance and the designer doesn't want to take this chance. That and they could also be using series passive xovers instead of parallel.
B&W only offering bi-wire terminals is probably only because of convention.
There are many good reasons to go active, lots of which you can probably find with a little bit of google seaching 😀 For me though, the biggest reason for having an active DSP based system is flexibility. My system is fairly complex and needs the abilities of a DSP to do what I need it to do, plus it sounds better 😉
Of course you can't just go with any off the shelf DSP and end up with acceptable objective performance, which is why I built my own. DSP done right is an amazing thing, DSP done wrong isn't quite as good, but it's still a mile better than analogue active done wrong.
Looking at the crossover, they definitely haven't skimped on quality though. I doubt that I could buy those components for under $1000 per channel, so I'm inclined to think that the design engineer is a fan of the KISS principle.
I've optimized the correction a little further since the last plot but I can say with full confidence that I prefer the treble now with it installed, and surprisingly the mids appear to be more transparent too, which is an added bonus.
I would say that the treble is slightly smoother and sweeter than before, and the mids have a more bell-like clarity and focus. The soundstage is more 3D and better defined than I've heard previously too, so there are definitely no downsides at all.
I've attached the latest on axis freq response plot. Note that I've slightly padded the tweeter now too - to level match it to the FST.
Down the track I'll probably go to a full DSP crossover solution but for the time being I'm very happy with the slight modifications I've made, which now allow me to listen on-axis without any excess upper mid or treble prominence, and TBH I can't really imagine how I could improve upon what I'm now hearing. It's now as good or better than a friend's system which boasts Revel Ultima Salon2's, which I've always felt were among the best bang-for-buck high end commercial speakers going around.
Attachments
I recommend the use of a DSP active crossover highly. There are a handful to choose from, with MiniDSP (a few models) being the most popular. I personally use the Najda DSP which was developed by a member here. It is a well-spec'ed product, and it includes both analogue and digital inputs, as well as analogue volume control. Some info is available here:
Najda Overview
DSP Xover project (part 2)
To get a feel for working with a DSP crossover you can download XoverWizard from Ground Sound, who also make DSP XOs, from this page:
Ground Sound DCN23 (XoverWizard download)
Najda Overview
DSP Xover project (part 2)
To get a feel for working with a DSP crossover you can download XoverWizard from Ground Sound, who also make DSP XOs, from this page:
Ground Sound DCN23 (XoverWizard download)
5th...a comment on the biwiring. I can agree on your statement in the sense of that most of the people will mix and match cables and type of amplifiers for low and mid-high and this definitely ruins the coherence of the speaker the way it was intended.
In the other end though, if the bi-wire is done properly with identical monoblocks and same cables throughout, this will lower distortion and offer a far superior great dynamics. Granted with the active the amplifier only amplifies a portion of the signal which is a big plus. Nonetheless a proper biwire is far superior to a single amp provided that quality of both monoblocks is the same as the single amp. I have been to plenty of demonstration where soundstage and 3D increases along with clarity and control.
To the DSP for active crossover, yes DSP is very flexible but in my setup I ONLY believe and listen to the Analog with capitol A. This would point to a well design perhaps discrete crossover without using any opamp on the signal path. I might consider designing one in the future once my other developments are completed. For now I feel I would be happy to take a step at a time by going to passive bi-amping.
EDIT: regarding bi-amping you can see B&W has it implemented with identical monoblocks on their listening room and same setup is at the Abbey Road studio...just thought of mentioning that.
In the other end though, if the bi-wire is done properly with identical monoblocks and same cables throughout, this will lower distortion and offer a far superior great dynamics. Granted with the active the amplifier only amplifies a portion of the signal which is a big plus. Nonetheless a proper biwire is far superior to a single amp provided that quality of both monoblocks is the same as the single amp. I have been to plenty of demonstration where soundstage and 3D increases along with clarity and control.
To the DSP for active crossover, yes DSP is very flexible but in my setup I ONLY believe and listen to the Analog with capitol A. This would point to a well design perhaps discrete crossover without using any opamp on the signal path. I might consider designing one in the future once my other developments are completed. For now I feel I would be happy to take a step at a time by going to passive bi-amping.
EDIT: regarding bi-amping you can see B&W has it implemented with identical monoblocks on their listening room and same setup is at the Abbey Road studio...just thought of mentioning that.
ART that is interesting, would the crossover you have design work on my aluminum tweeter as well?
I will boost the measurements for 5th as soon as I am done finalizing an important project I have to deliver in the next month and half after which I will have time to play with my speakers.
Why did you pad the tweeter? Wasn't the efficiency properly matched on the original crossover?
Regarding the saloon 2, the are on the same price range of the 800, do you feel these speakers are better than the original B&W?
Interesting measurements, do you guys want to comment on these?
Revel Ultima Salon2 loudspeaker Measurements | Stereophile.com
I will boost the measurements for 5th as soon as I am done finalizing an important project I have to deliver in the next month and half after which I will have time to play with my speakers.
Why did you pad the tweeter? Wasn't the efficiency properly matched on the original crossover?
Regarding the saloon 2, the are on the same price range of the 800, do you feel these speakers are better than the original B&W?
Interesting measurements, do you guys want to comment on these?
Revel Ultima Salon2 loudspeaker Measurements | Stereophile.com
Hi Stefanoo, re tweeter - conjugate impedance matching, it's probably not worth the trouble with your 3rd order network because the filter would render it insignificant - unlike my single pole network, and I doubt you would hear any difference. Having heard my friend's S800's many times I would say that the treble is every bit as good as I'm now hearing from my D800's, which says a lot about that tweeter.
Wrt the tweeter pad, it's only 0.5dB, so is trivial in the big scheme although surprisingly quite audible. There's actually a marked pad on the tweeter's crossover pcb for the resistor, although you need to relocate 1 leg of the 4.7uF into an adjacent hole (which connects to the alternate trace) after you've installed it.
I've always held the Revel's in high regard and they are definitely on par with D800's overall but are IMO slightly more accurate in the mids and undoubtedly less sensitive to the listening environment.
OTOH, the Revel tweeter is very 'beamy' above 8kHz which means they can sound a little closed and lacking airiness in some rooms. The Revel's are not quite as capable in the bass either and there's a hump around 100 Hz that can become annoying at times, particularly if your listening room enhances it further.
Now that I've removed most of the 3.5kHz bump and and corrected the impedance etc, my 800D's are simply sublime through the mids, and tbh there's not another speaker that I would rather own at the moment. Btw, I decided to leave a slight bump at 3.5kHz because it seemed to add a hint of sweetness to female vocals, and I'm thinking of running a small twisted pair cable out of the plinth to a switch box so that I can select a resistor value that best suits what I'm listening to.
Thanks for the links. I'll go and check it out.
I was considering buying a DEQX, but that's obviously a high cost option.