I got it no problem, not questioning how this project sounds, there is huge following which I doubt wasn't there if it didn't. Never heard this or similar speakers, sims are just a study. I was checking out if the polar measurements of similar speaker arrangement would reveal anything and nope, there is not much to see. If the bottom woofers were muted or crossovers tweaked the directivity is roughly the same, like similar speaker arrangements in general. Joe doesn't need the data to know it behaves like this, even if this was part of speciality it is widely known what MTM or array response looks like. Obviously the project specialty is something else than directivity though, at least the directivity is not mentioned often even if it was. Narrow vertical directivity and contribution of vertical early reflections has been on my study list for a while now and hence wanted to check this one out.
If one checks out any speaker related design issues on frequency response plot it is most likely on the midrange like vertical early reflection related
interference. These are reduced with such driver configuration due to increased attenuation towards the specular reflection points for example. In this case roughly the whole important midrange. edit. There is some messages from 2019 where Joe and folks tested the sound went better by elevating the speaker. This would further reduce floor reflection for example, as the angle gets bigger toward first reflection point.
Please don't interpret this as if the polar data would somehow decrease the value of the project and Joes effort. People are blindly following the polar data measurements and it is not easy to relate the graphs into reality, what something sounds like. This is something I (and hopefully many others) are trying to explore, whats relevant, what is not, how to relate graphs to sound in the personal environment.
If one checks out any speaker related design issues on frequency response plot it is most likely on the midrange like vertical early reflection related
interference. These are reduced with such driver configuration due to increased attenuation towards the specular reflection points for example. In this case roughly the whole important midrange. edit. There is some messages from 2019 where Joe and folks tested the sound went better by elevating the speaker. This would further reduce floor reflection for example, as the angle gets bigger toward first reflection point.
Please don't interpret this as if the polar data would somehow decrease the value of the project and Joes effort. People are blindly following the polar data measurements and it is not easy to relate the graphs into reality, what something sounds like. This is something I (and hopefully many others) are trying to explore, whats relevant, what is not, how to relate graphs to sound in the personal environment.
Last edited:
No, what I said was quite clear - you have just taken it out of context.
For the sake of clarity: Most speakers (multi-way commercial and DIY) have poor vertical results (on any axis) because they aren't designed for a wide vertical "window".
"right away"????
How about "at all"? (..I can't think of many manufacturers that have this kind of equipment.)You don't need either Power Response or Directivity Index to make a good loudspeaker for a particular listening window. It would be nice to have, but it certainly isn't a requirement (..obviously).
I think people who are unable to hear this speaker would like to see some more measurements, that’s all.
CTA/CEA 2034 method has been ratified about 7 years ago.
It isn’t new, and it doesn’t need a Klippel or anechoic chamber:
https://shop.cta.tech/products/standard-method-of-measurement-for-in-home-loudspeakers
Audioholics has been doing it for awhile, and recently ASR and EAC.
Like all measurements, it’s a visual translation of an acoustic event: it’s not perfect; but it’s more complete than an on-axis line or in room measurement.
If it’s not able to be provided, I think should move on.
CTA/CEA 2034 method has been ratified about 7 years ago.
It isn’t new, and it doesn’t need a Klippel or anechoic chamber:
https://shop.cta.tech/products/standard-method-of-measurement-for-in-home-loudspeakers
Audioholics has been doing it for awhile, and recently ASR and EAC.
Like all measurements, it’s a visual translation of an acoustic event: it’s not perfect; but it’s more complete than an on-axis line or in room measurement.
If it’s not able to be provided, I think should move on.
I think some people would like to see more measurements - some a LOT more than others (..and those that need those measurements before making the build are very likely to "move on").
My guess is that many others just don't care.
I think the real problem is the tendency for being put-off on a build because of a certain expectation (as seeing a very narrow vertical window, when something less directive and more uniform is expected) that isn't ever going to be met with this design. This of course is despite the fact that there are many happy builders listening to these speakers (even if accommodating the narrow vertical listening range).
While you don't need a Klippel or an Anechoic Chamber, getting good consistent results (particularly lower in freq.) can be quite difficult (..and further complicated by desiring a multitude of measurements to create a full "balloon" result).
As evidenced by Erin's own thread, it's often more complicated than it seems:
https://www.diyaudio.com/community/...how-close-is-close-enough-to-anechoic.353347/
My guess is that many others just don't care.
I think the real problem is the tendency for being put-off on a build because of a certain expectation (as seeing a very narrow vertical window, when something less directive and more uniform is expected) that isn't ever going to be met with this design. This of course is despite the fact that there are many happy builders listening to these speakers (even if accommodating the narrow vertical listening range).
While you don't need a Klippel or an Anechoic Chamber, getting good consistent results (particularly lower in freq.) can be quite difficult (..and further complicated by desiring a multitude of measurements to create a full "balloon" result).
As evidenced by Erin's own thread, it's often more complicated than it seems:
https://www.diyaudio.com/community/...how-close-is-close-enough-to-anechoic.353347/
I think something has been forgotten here. I just posted the results of modeling and spent days on it before I built the crossover:
We are only hours away to listen and see if/what corrections need to be made to the crossover.
Now take a look at the two largest inductors on either side of the large Audyn True Copper Max capacitors. What these inductors do has a different kind of philosophy behind them. We can think of them as straight 1st order inductors, and that is just fine. But these are uncorrected in any way, so as to add the largest series inductance to the driver's internal but unstable inductance. Also that cap, using lowest possible value is a plus, the inductors, here the largest value is a plus.
The key word here is philosophy.
Every speaker designer needs to have one.
Some time ago on the John Curl Blowtorch forum, I noticed that they hardly ever talked about speakers and I pushed the subject. Initially, there was a lot of resistance, but eventually, the discussion took off. I took quite a bit of flack (they are conservative and I am viewed as not - but in fact, they got that wrong). During that time, John Curl himself posted this:
"I know how to design a loudspeaker, I just don't know how to design a great loudspeaker."
Wow! That is quite profound. And that shows an open mind. Using 'tried and tested methods,' it's not that hard to make a 'well-designed' speaker. Put some hours into the effort, follow the rules and you will get there.
Except there may be other rules that the mainstream hasn't quite figured out. But in time I hope they will and we will get more great sounding loudspeakers. There are so many speakers out there where I hear a sound that doesn't want to get out of the box. Constrained is the word. Smooth, possibly, but not engaging. The well-known John Atkinson says that speakers with correct step-response sound more accessible and I know what he means. The sound engages you!
I am of the view that if a speaker doesn't sound the way it should, then some kind of distortion is getting in the way.
BTW, John has been very supportive and he comes across as a decent guy, he even admits admiration for the sound of tube amplifiers and admits that there are things the best tube amps can do that his own design cannot. You have got to respect the man!
We live in an inherently conservative world. Many feel that the current rules are well set and no need to change. There are now well-established ways of designing loudspeakers. There are tools to accomplish that aim, now plentiful and available. But what has also happened is that there has become a "correct" way of designing loudspeakers. Maybe we can call it confirmation bias or cognitive dissonance, where we measure whatever the prevailing winds have decided is right or wrong.
I think something significant has been overlooked! And it has to do with those two inductors above.
In the Elsinore Mk6 "MFC", they are 4mH and 1mH respectively. What they do is different from what everybody else is doing and just looking at the crossover is not enough. It needs to be explained.
Let me finish with this point:
If this was a choice of linearising the current through the voice coil OR perfecting acoustic performance, I have made my choice already, both can be viewed as important, but one is more important than the other!
But then again, distortion trumps everything else, and let me make it clear, speakers can sound distortion-free, but they are not. I know, that is shocking!
Having said that, the Elsinore's vertical performance is not bad. For a start, it is symmetrical both laterally and vertically. It is better than the usual 1st order by a big margin.
The step response looks like this and comes out like that over a wider window than any other 1st order design like Duntech/Dunlavy.
Take any more fancier methods of measurements (which I still welcome), the above measurement tells me so much, but again a trained eye will know why. I know that this speaker will be friendly current wise (number one requirement) without being greatly deficient anywhere else on the acoustic side. Earlier on I mention Andrew Lipinsky, the designer of the L707 model - it too has that step response you see above. It will also have a wider window than any Dunlavy design (they worked together for a while) and Andrew came to this conclusion independent of mine.
That will have to do for now.
New concepts are hard to digest the first time around.
Cheers, Joe
PS: I suspect this is only Part 1 and that in Part 2, I will show you how a driver's impedance can modulate the current and distort the force factor that causes the sound we hear. This is a little know distortion mechanism, but one I know that Purifi is aware of (please read their blogs) and why choosing their driver fits into all of this. It will also explain why those two inductors are so critical as part of the Elsinore design philosophy.
.
We are only hours away to listen and see if/what corrections need to be made to the crossover.
Now take a look at the two largest inductors on either side of the large Audyn True Copper Max capacitors. What these inductors do has a different kind of philosophy behind them. We can think of them as straight 1st order inductors, and that is just fine. But these are uncorrected in any way, so as to add the largest series inductance to the driver's internal but unstable inductance. Also that cap, using lowest possible value is a plus, the inductors, here the largest value is a plus.
The key word here is philosophy.
Every speaker designer needs to have one.
Some time ago on the John Curl Blowtorch forum, I noticed that they hardly ever talked about speakers and I pushed the subject. Initially, there was a lot of resistance, but eventually, the discussion took off. I took quite a bit of flack (they are conservative and I am viewed as not - but in fact, they got that wrong). During that time, John Curl himself posted this:
"I know how to design a loudspeaker, I just don't know how to design a great loudspeaker."
Wow! That is quite profound. And that shows an open mind. Using 'tried and tested methods,' it's not that hard to make a 'well-designed' speaker. Put some hours into the effort, follow the rules and you will get there.
Except there may be other rules that the mainstream hasn't quite figured out. But in time I hope they will and we will get more great sounding loudspeakers. There are so many speakers out there where I hear a sound that doesn't want to get out of the box. Constrained is the word. Smooth, possibly, but not engaging. The well-known John Atkinson says that speakers with correct step-response sound more accessible and I know what he means. The sound engages you!
I am of the view that if a speaker doesn't sound the way it should, then some kind of distortion is getting in the way.
BTW, John has been very supportive and he comes across as a decent guy, he even admits admiration for the sound of tube amplifiers and admits that there are things the best tube amps can do that his own design cannot. You have got to respect the man!
We live in an inherently conservative world. Many feel that the current rules are well set and no need to change. There are now well-established ways of designing loudspeakers. There are tools to accomplish that aim, now plentiful and available. But what has also happened is that there has become a "correct" way of designing loudspeakers. Maybe we can call it confirmation bias or cognitive dissonance, where we measure whatever the prevailing winds have decided is right or wrong.
I think something significant has been overlooked! And it has to do with those two inductors above.
In the Elsinore Mk6 "MFC", they are 4mH and 1mH respectively. What they do is different from what everybody else is doing and just looking at the crossover is not enough. It needs to be explained.
Let me finish with this point:
If this was a choice of linearising the current through the voice coil OR perfecting acoustic performance, I have made my choice already, both can be viewed as important, but one is more important than the other!
But then again, distortion trumps everything else, and let me make it clear, speakers can sound distortion-free, but they are not. I know, that is shocking!
Having said that, the Elsinore's vertical performance is not bad. For a start, it is symmetrical both laterally and vertically. It is better than the usual 1st order by a big margin.
The step response looks like this and comes out like that over a wider window than any other 1st order design like Duntech/Dunlavy.
Take any more fancier methods of measurements (which I still welcome), the above measurement tells me so much, but again a trained eye will know why. I know that this speaker will be friendly current wise (number one requirement) without being greatly deficient anywhere else on the acoustic side. Earlier on I mention Andrew Lipinsky, the designer of the L707 model - it too has that step response you see above. It will also have a wider window than any Dunlavy design (they worked together for a while) and Andrew came to this conclusion independent of mine.
That will have to do for now.
New concepts are hard to digest the first time around.
Cheers, Joe
PS: I suspect this is only Part 1 and that in Part 2, I will show you how a driver's impedance can modulate the current and distort the force factor that causes the sound we hear. This is a little know distortion mechanism, but one I know that Purifi is aware of (please read their blogs) and why choosing their driver fits into all of this. It will also explain why those two inductors are so critical as part of the Elsinore design philosophy.
.
Just a few comments. The floor bounce is actually much better than usual. Below 500 Hertz all the drivers are working at four different heights from the floor and hence reflections are spread out to different frequencies.
Please take a close look at this, where Red is the bottom two drivers:
How can a single inductor cause such a low-pass function Red? The truth is that it cannot!
Again, some of the tricks used in the Elsinores are not obvious until they are pointed out. The initial roll-off looks normal and near 1st order as you would expect. These two bottom drivers need to perform very well in the midrange. But then the response goes to acoustic 3rd order until you see that dip near 4KHz. This goes back to the original Mk1 design that came out in 2006. The answer? It's a cancellation between the bottom two drivers and what you see above is at 2 metres and on the axis of the tweeter. The distance between the two drivers is what causes that null. The frequency is related to the sum of those differences.
But it is still electrically 1st order. I am actually quite proud of having achieved this trick. I nailed it.
As for MTM arrays, if they are 1st order Butterworth and sums at -3dB at the crossover, you get higher distortion, poorer power response, and of axis both laterally and vertically. Here we use -6dB and use another trick, make use of the tweeter as an acoustic inverter. Link below explains it.
I am not sure f you have across this, but it will explain further, even if it does need an edit or two to bring it up-to-date:
http://www.customanalogue.com/elsinore/elsinore_17.htm
http://www.customanalogue.com/elsinore/elsinore_19.htm
As for tilt and increasing the height, this is something I always suggest, almost no matter what loudspeaker.
I can certainly agree with that.
Hope you read my post just before this one. I would like to see a view comments first, but I have a Part 2 in mind.
Cheers, Joe
^^ This is something the CTA graphs don't tell, thanks writing it out.
The phenomenon, or case of boring/exciting sound can be made with DSP xo as well, although I'm not quite sure what eventually was the cause on my prototype speaker but was shocking to realize that two DSP presets can have very different sensation even though the magnitude graphs (in simulator) stayed pretty much the same between. Boring/exciting were the best words how to describe the difference and although the boring sounded just fine before and looked very good on simulator there really is no comparison how much more lively or exciting the later was after I discovered it. There might be even more exciting sounding systems, hence waiting eagerly more philosophy.
From this limited experience with little bit reasoning atop comes realization that boring/exciting and polar pattern are not excluding each other out because pattern is tied to the driver layout and structure in acoustic domain and livelines seems to be in the electric domain. Not tied together and separately adjustable, at least to some extent. Or perhaps they are not but at least any speaker can be ruined be boring with poor xo topology.
.
The phenomenon, or case of boring/exciting sound can be made with DSP xo as well, although I'm not quite sure what eventually was the cause on my prototype speaker but was shocking to realize that two DSP presets can have very different sensation even though the magnitude graphs (in simulator) stayed pretty much the same between. Boring/exciting were the best words how to describe the difference and although the boring sounded just fine before and looked very good on simulator there really is no comparison how much more lively or exciting the later was after I discovered it. There might be even more exciting sounding systems, hence waiting eagerly more philosophy.
From this limited experience with little bit reasoning atop comes realization that boring/exciting and polar pattern are not excluding each other out because pattern is tied to the driver layout and structure in acoustic domain and livelines seems to be in the electric domain. Not tied together and separately adjustable, at least to some extent. Or perhaps they are not but at least any speaker can be ruined be boring with poor xo topology.
.
Last edited:
^^ Thanks, I'll read the doc, gotta go outside first as it is nice weather today for longest time
Yeah, reducing floor bounce needs tight vertical pattern as it happens typically in very small angle. Array of drivers can do it in relatively compact form factor, a waveguide would need to be huge for same effect.
edit. lost long edit but short version is: Passive crossovers require expertise, simulators can get only so far. Must be a great feeling when it all comes together in a very nice way, certainly something to be proud of!
May I ask your view/philosophy regarding DSP systems? They can be assumed to be adjustable to anything which makes the crossovers trivial along with good simulator that can even export the settings for you. But, the implementation will be only as good as the targets of the designer so it is kind of no different to passive crossover other than in capability, designer makes good or bad use of it. I've never done passive crossover, perhaps should try a project with one, perhaps there is something to it, to affect drivers performance with passive filters?
Yeah, reducing floor bounce needs tight vertical pattern as it happens typically in very small angle. Array of drivers can do it in relatively compact form factor, a waveguide would need to be huge for same effect.
edit. lost long edit but short version is: Passive crossovers require expertise, simulators can get only so far. Must be a great feeling when it all comes together in a very nice way, certainly something to be proud of!
May I ask your view/philosophy regarding DSP systems? They can be assumed to be adjustable to anything which makes the crossovers trivial along with good simulator that can even export the settings for you. But, the implementation will be only as good as the targets of the designer so it is kind of no different to passive crossover other than in capability, designer makes good or bad use of it. I've never done passive crossover, perhaps should try a project with one, perhaps there is something to it, to affect drivers performance with passive filters?
Last edited:
I have a smile on my face, finished the work today, did the 'voicing' as some call it. The crossover modeling was right on and the last thing to sort out was the box tuning. The crossover is external and will run it like that for a while, but eventually will go inside on the bottom.
Very tired, very pleased.
Cheers, Joe
Very tired, very pleased.
Cheers, Joe
"ULD" Update:
The speakers are in my system and performing well. I know that the question will be, how do they compare to standard "MFC" and "NRX" versions of the Elsinores. Yes, there is a difference and I am sure you will like what it does. Indeed the Elsinore was always about low distortion and particularly keeping an eye on what the current does and the "ULD" builds on this and the improvement is indeed audible and perhaps the best description is 'more of the same.'
It will need good sources and amplifiers to hear the full potential of this speaker. Its abilities to unearth inner detail in an organic way, its seemingly unlimited transient capabilities and open-air purity, plus some things that remind me of the best electrostatic speakers, and more. As usual, balance is high on the list of requirements, that many high-end speakers seem to do some things very well, and they will be different from one expensive model from another and that is frustrating. This is not the case here.
Bottom line? If you got the money, it will be well spent. This time the extra dollars are really worth it. I don't think you will be disappointed if you are able to build them. Yes I know, cost.
To the guys who have the drivers already, I can't wait for you to get them up and running. Helmut, how long do you think it will be?
Now working on publishing the crossover which will be posted here shortly.
Cheers, Joe
PS: I hope those who cannot afford to build the "ULD" version, please do not feel any letdown. In no way does the affect what you already like and are listening to now. But... ahem..., if you can afford a bank loan...? Well, you know what I mean.
The speakers are in my system and performing well. I know that the question will be, how do they compare to standard "MFC" and "NRX" versions of the Elsinores. Yes, there is a difference and I am sure you will like what it does. Indeed the Elsinore was always about low distortion and particularly keeping an eye on what the current does and the "ULD" builds on this and the improvement is indeed audible and perhaps the best description is 'more of the same.'
It will need good sources and amplifiers to hear the full potential of this speaker. Its abilities to unearth inner detail in an organic way, its seemingly unlimited transient capabilities and open-air purity, plus some things that remind me of the best electrostatic speakers, and more. As usual, balance is high on the list of requirements, that many high-end speakers seem to do some things very well, and they will be different from one expensive model from another and that is frustrating. This is not the case here.
Bottom line? If you got the money, it will be well spent. This time the extra dollars are really worth it. I don't think you will be disappointed if you are able to build them. Yes I know, cost.
To the guys who have the drivers already, I can't wait for you to get them up and running. Helmut, how long do you think it will be?
Now working on publishing the crossover which will be posted here shortly.
Cheers, Joe
PS: I hope those who cannot afford to build the "ULD" version, please do not feel any letdown. In no way does the affect what you already like and are listening to now. But... ahem..., if you can afford a bank loan...? Well, you know what I mean.
Hello Joe,
good news, very good news. I am building a pair each Hamlet and Elsinores ULD in parallel, so it is time consuming. May be about a month? The Hamlets are closer to there finalisation. Front panels will be Panzersperrholz, These panels for the Elsinores have still to be milled. I will report.
good news, very good news. I am building a pair each Hamlet and Elsinores ULD in parallel, so it is time consuming. May be about a month? The Hamlets are closer to there finalisation. Front panels will be Panzersperrholz, These panels for the Elsinores have still to be milled. I will report.
Attachments
I’ve already mounted and have been using the Purifis in my mark V so I should be able to quickly make the changes to the xover.
Couple questions
—> I know you addressed this before and indicated that it shouldn’t make a large difference, but just to ask again now that you’ve finalized the design —> any reason surface vs flush mounting should make a significant difference in performance? I currently have the drivers surface mounted as per the Mark V
—> I’m obviously awaiting your final suggestion for port length but I did extend to 16cm when I modified my Mark V. However I added the additional port length externally for ease of conversion and for the ability to modify easily once your design was finalized. At the time I did this my internet searches indicated that internal vs external port doesn’t change anything in a vented box - rather only the length. Is this correct on your end or should I re-modify to internalize the port? Obviously internal is better cosmetically but in my application it’s not visible..
Thank you!!
Couple questions
—> I know you addressed this before and indicated that it shouldn’t make a large difference, but just to ask again now that you’ve finalized the design —> any reason surface vs flush mounting should make a significant difference in performance? I currently have the drivers surface mounted as per the Mark V
—> I’m obviously awaiting your final suggestion for port length but I did extend to 16cm when I modified my Mark V. However I added the additional port length externally for ease of conversion and for the ability to modify easily once your design was finalized. At the time I did this my internet searches indicated that internal vs external port doesn’t change anything in a vented box - rather only the length. Is this correct on your end or should I re-modify to internalize the port? Obviously internal is better cosmetically but in my application it’s not visible..
Thank you!!
Good news, the Mk-V boxes should work OK with Purifi drivers.
With the Mk-6 using SB drivers, they had to be rebated 7mm. But with Peerless drivers, they had the option of rebating or not. In my own case, I used no rebating.
I wonder if the same could be said for the Purifi drivers? Claus Tinggaard was with Peerless and a significant part of Purifi, so maybe we are seeing his influence and that it is OK to do, or maybe they have not given much thought to it?
I did a search on Google Images and looked for pictures of Purifi Speakers. They were all rebated, but there was one exception I saw that it wasn't. See below.
Will you be able to rebate your Mk-V boxes? The 4mm difference is OK, whichever you use. Without rebating, the top MTM array of three drivers focuses just bit closer, but in practice makes little difference. A lot of this has to do that we get 6dB summing at the crossover frequency and not 3dB.
Re port length, I suspect you had 80mm long ports if using 90mm diameter PVC pipe (inner diameter around 87mm). I ended up with 75mm, but 80mm will be just fine.
Look at the example above, the lower right hand side of the driver, it looks like this was not rebated.
Crossover will be posted within 24 hours.
What you are doing will be well worth it.
Cheers, Joe
Sounds good, the changes From Mk-5 I think there will be six component values changed plus add a Zobel network of 0.47uF and 8R2. You will see when I post it. The good thing is that 18mH and its Resistor in series, plus 300uF, these will not change and that will make it a lot easier.