I am sure I would be. I have a 2-Way design here based on the 7.5" version right here. But there are two additional things to be said: Time and cost. If somebody would come along and sponsor a Sartori version, then maybe I might consider it. And then I would have to find the time, it is not fast work.
But the drivers we have used so far are not inferior drivers, would a Sartori version sound better? Keep this in mind, let us say that it would take 1 nominal Watt (or 2.83V) to produce 92dB, that is 1000mW. Now put four in series parallel and to get to produce 92dB would mean 62.5mW per driver.
The reason is that only 1/4 Watt of a Watt, shared across four drivers, means that each driver only needs to produce 1/16th of a Watt nominally.
This is why if you get good quality drivers with Elsinores, that you will actually get next level performance out of them.
So all the Elsinore Mk6 iterations benefit from that and there are diminishing return even if you throw a LOT of money at them. Well, we tried that with the Purifi ULD drivers and they are even more costly than the Sartori drivers.
BUT WAIT:
Then something completely unforseen happened!
Going from the TMC to the Miflex capacitor took me completely by surprise. It made a much bigger difference that any choice between the main drivers and this is connected to the tweeter. Go figure, because that had me both scratching my head as well as put a wry smile on my face. The truth is that I loved it even if I had to admit the TMC was what I had recommended and we are not talking cheap caps here.
So whatever driver you end up with, just make sure that you budget for those Miflex caps, even if it is a little further down the line.
The NBAC that I now have in my system sounds better than my ULDs did. I had to do the same with the ULDs. Yes, in my system (which anybody can drop in for a listen and report here) the Miflex 1.8uF 250V caps made that much of a difference.
So there you are. 🙂
PS: My theory is that the difference in sound has to do with AM (amplitude modulated) distortion of the current to the tweeter. Also, would that be measurable? I am going to try at some point.
Last edited:
The Miflex took 1 day to sound good in my ULD's, 1 week to show, that they are better than the TMC. Difficult to describe what is better, but it definitely is. Cleaner? Maybe.
You are quite correct about the ULD drivers, their low Qt explains it, the response pulls it down and even noticeable above 100Hz. The shorter port is a recognition of that. The thing that I thought long and hard about was that lower Qt and below is the reason why I still went ahead with it, goes like this:
Here I want to mention something that the software does not reveal.
The fact that this is more a 'filled' box and not the usual 'lined' box means a different kind of behaviour is introduced. You can partly model for it by choosing a much low QL in the software, not the usual QL=7 (or sometimes higher) and more like QL=<3. Yes, less than 3. Now the difference between the Qt of the drivers is not as much as an issue, but maybe still some. Hence the slightly shorter port. Try QL=2 and check the difference. Even the MFC will start looking more like ULD. But it's OK.
But I think something even more is involved. Something happens to the quality of the bass going from fully lined to well above 50% of fill (the lining is only part fill). And something approaching 2/3rd fill if possible. The bass just seems to blossom. Also, I get the impression that the port is also better behaved, and I have not had trouble with port noise, even in smaller designs.
Keep up the good work.
Last edited:
David,
Congratulations on completing the build. The speakers have come out very nice, and are looking good in your room. Be proud, and enjoy the music!
Share more impressions with us when you have more hours on the drivers.
Congratulations on completing the build. The speakers have come out very nice, and are looking good in your room. Be proud, and enjoy the music!
Share more impressions with us when you have more hours on the drivers.
Joe,
I am so happy to have made your day! Thank you for your superb and continued support and evolution of your design, giving us the opportunity to take a leap of faith and come out with a completed project that exceeds our expectations. I spent most of yesterday just sitting listening and enjoying them during the break-in. They come across as faster and tighter sounding than the MFC, transients are really improved. So far listened with a KT66 Williamson amps for the first few hours and then my 211 Class A2 SET amp s(which really works well, there is a nice synergy with this amp and the Elsinores). I have a few more amps in my stable to try out including a Pass DIY FW F4 driven by an Aikido 12SN7 preamp.
Just one note on my system. For digital playback I use a Chord DAVE DAC (with the Hugo MScalar feeding it) as a digital preamp, so drive the amps directly, avoiding any additional gain stages. For analog, I use a passive resistive ladder attenuator in between my phono preamp and the amps (with a bypass switch so digital is straight through). This provides the most transparent, uncolored, sound In my system. There is no doubt that the ULDs extend that transparency.
Again, Thanks!
David
I am so happy to have made your day! Thank you for your superb and continued support and evolution of your design, giving us the opportunity to take a leap of faith and come out with a completed project that exceeds our expectations. I spent most of yesterday just sitting listening and enjoying them during the break-in. They come across as faster and tighter sounding than the MFC, transients are really improved. So far listened with a KT66 Williamson amps for the first few hours and then my 211 Class A2 SET amp s(which really works well, there is a nice synergy with this amp and the Elsinores). I have a few more amps in my stable to try out including a Pass DIY FW F4 driven by an Aikido 12SN7 preamp.
Just one note on my system. For digital playback I use a Chord DAVE DAC (with the Hugo MScalar feeding it) as a digital preamp, so drive the amps directly, avoiding any additional gain stages. For analog, I use a passive resistive ladder attenuator in between my phono preamp and the amps (with a bypass switch so digital is straight through). This provides the most transparent, uncolored, sound In my system. There is no doubt that the ULDs extend that transparency.
Again, Thanks!
David
I have to add something to your theory, the cone of a nbac is more pistonic than a ptt6.5w nfa paper cone ?
Not sure the motor quality is a key here. I don't think you have only one factor. The quality of the capacitor just put thing on the right place/way. Not sure you will have a better sound with the miflex in the ULD than the NBAC.
I was surprised to see the difference with good design between speakers is very small or nul. But the price is really different.
To measure such thing perhaps a simple square wave and see the quality of it harmonic contents ?
Other point yes the mw16p is a very good driver in the two impedance version to my ears. Note I prefer it in closed box, I think if you make something with it, the tunning of the reflex port should be very accurate.
Oh boy, you are opening the door to so many things that could be much discussed.
The Purifi approach to the motor is very different to anybody else. I has quite low inductance, but not as ultra-low as SB variants including the MW16P. But they are more constant with pistonic motion of the cone away from the rest position. So excursions means that the inductance is constant low for any particular excursion.
For more information on the Purifi drivers, if you are patient and hard core addict as some of us are, the video is nearly two hours long and the interesting things come in lumps here and there, so patience will be repaid. I think you putting in the time might be worth it to you:
This what the inductance look like in the Purifi driver at different frequencies.
SB takes a different tack. even lower inductance but not as flat, it looks more like this:
Inductance goes up as the cone goes in and inductance goes down as the cone goes out. That also indicates that BLi force is not linear with position.
But since we have four drivers, that means for any set SPL, the inductance gets flattened out to look effectively something like this:
So this makes all the Elsinores much superior to a single driver.
And: When we use Purifi drivers, even the Purifi's own results gets flattened even more again. But it is worth that all the versions of Elsinores do this and that even SB17 variations do well as they all use the same motor and has very low inductance.
In the video link above, this factor is linked to AM type distortions. But it is the current that is being effected since many of the imperfections of the driver (and we all know that there are no perfect drivers) and the Force - BLi indicates that the i - the current of the amplifier is being AM modulated.
What if we could make drivers without any inductance?
Then there would be no difference between voltage-drive (common) and current-drive (uncommon, but does lower i induced distortion). Also, there are questions as to how well the amplifier that is charged with keeping voltage controlled, is able to control i induced distortions. This is something that is begging to be looked at and has not been. I am hoping to do something about this, a paper that will discuss i induced distortions in amplifiers. But I digress.
Ultimately, what causes inductance is just as Faraday explained, that current through a coil produces a voltage and this voltage leads to distortions (I am glad to find an increasing number of people who now also understands this); it is also called back-EMF. It is this voltage that tries to oppose (back-EMF is an opposing force) current with rising frequency, which creates the final impedance of a driver (typically measured in free air)
Ei is controversially both a voltage (a back-EMF) and an impedance that can be quantified in Ohm (I have gotten attacked on this point), but the above proves that this is correct. Opposing current is also implied in that it is called back-EMF. It is current that is being opposed and the inverse of that is impedance and hence Ohm'age.
Anything above the DC resistance show up as added Ohm in the impedance plot. The DCR is very stable, but with music the Ei part of the impedance is fuzzy and not as shown a nice clean straight looking line. It is not stable with excursion for a start (the Purifi is much more stable here) where it will go up with the cone going in and down with the cone going out. The Em is the motional back-EMF and here is less of our concern here. But an interesting discussion can be had about that too.
So you have motional back-EMF Em and you have Ei and they are all in series with the Rc resistance of the voice coil. But here we are concerned with Ei being a source of i current induced distortion of the BLi force. The Ei fuzziness needs to be kept in mind and be designed for.
If only we have perfect drivers with zero inductance, but alas Mr Faraday tells us that we can't.
I am a bit reluctant to bring up too much of this stuff, but I seem to get drawn into it and it is fascinating (and maybe even addictive) to ponder over these things and understand mechanisms that lead to distortion.
But it does show that there is a lot of underlying theory behind the operation of the Elsinore designs.
Most if not all speaker designs, put the acoustic side of design as primary - and then take a glance at what happens on the current side. I have come to realise that my design have adopted the reverse position of that. I take the electriocal side of loudspeaker design as primary whilst also still keeping a keen eye on what happens on the acoustic side. I definitely put the electrical side first and the interface with the amplifier as primary. Many have said that my designs sound different. I take that as an encouragement.
The Purifi approach to the motor is very different to anybody else. I has quite low inductance, but not as ultra-low as SB variants including the MW16P. But they are more constant with pistonic motion of the cone away from the rest position. So excursions means that the inductance is constant low for any particular excursion.
For more information on the Purifi drivers, if you are patient and hard core addict as some of us are, the video is nearly two hours long and the interesting things come in lumps here and there, so patience will be repaid. I think you putting in the time might be worth it to you:
This what the inductance look like in the Purifi driver at different frequencies.
SB takes a different tack. even lower inductance but not as flat, it looks more like this:
Inductance goes up as the cone goes in and inductance goes down as the cone goes out. That also indicates that BLi force is not linear with position.
But since we have four drivers, that means for any set SPL, the inductance gets flattened out to look effectively something like this:
So this makes all the Elsinores much superior to a single driver.
And: When we use Purifi drivers, even the Purifi's own results gets flattened even more again. But it is worth that all the versions of Elsinores do this and that even SB17 variations do well as they all use the same motor and has very low inductance.
In the video link above, this factor is linked to AM type distortions. But it is the current that is being effected since many of the imperfections of the driver (and we all know that there are no perfect drivers) and the Force - BLi indicates that the i - the current of the amplifier is being AM modulated.
What if we could make drivers without any inductance?
Then there would be no difference between voltage-drive (common) and current-drive (uncommon, but does lower i induced distortion). Also, there are questions as to how well the amplifier that is charged with keeping voltage controlled, is able to control i induced distortions. This is something that is begging to be looked at and has not been. I am hoping to do something about this, a paper that will discuss i induced distortions in amplifiers. But I digress.
Ultimately, what causes inductance is just as Faraday explained, that current through a coil produces a voltage and this voltage leads to distortions (I am glad to find an increasing number of people who now also understands this); it is also called back-EMF. It is this voltage that tries to oppose (back-EMF is an opposing force) current with rising frequency, which creates the final impedance of a driver (typically measured in free air)
Ei is controversially both a voltage (a back-EMF) and an impedance that can be quantified in Ohm (I have gotten attacked on this point), but the above proves that this is correct. Opposing current is also implied in that it is called back-EMF. It is current that is being opposed and the inverse of that is impedance and hence Ohm'age.
Anything above the DC resistance show up as added Ohm in the impedance plot. The DCR is very stable, but with music the Ei part of the impedance is fuzzy and not as shown a nice clean straight looking line. It is not stable with excursion for a start (the Purifi is much more stable here) where it will go up with the cone going in and down with the cone going out. The Em is the motional back-EMF and here is less of our concern here. But an interesting discussion can be had about that too.
So you have motional back-EMF Em and you have Ei and they are all in series with the Rc resistance of the voice coil. But here we are concerned with Ei being a source of i current induced distortion of the BLi force. The Ei fuzziness needs to be kept in mind and be designed for.
If only we have perfect drivers with zero inductance, but alas Mr Faraday tells us that we can't.
I am a bit reluctant to bring up too much of this stuff, but I seem to get drawn into it and it is fascinating (and maybe even addictive) to ponder over these things and understand mechanisms that lead to distortion.
But it does show that there is a lot of underlying theory behind the operation of the Elsinore designs.
Most if not all speaker designs, put the acoustic side of design as primary - and then take a glance at what happens on the current side. I have come to realise that my design have adopted the reverse position of that. I take the electriocal side of loudspeaker design as primary whilst also still keeping a keen eye on what happens on the acoustic side. I definitely put the electrical side first and the interface with the amplifier as primary. Many have said that my designs sound different. I take that as an encouragement.
Last edited:
Thank you Joe for these very detailed explanations 🙂
I like a lot the way you explain thing, it reminded me my physic course on electromagnetism.
You have a very interesting approach because you take into account the amplifier. Your designs sound well with most of amplifiers. Not the case of all loudspeakers. Note I take into account now the electrical approach when I build my loudspeakers 🙂
Going to watch this video from Lars 🙂
I like a lot the way you explain thing, it reminded me my physic course on electromagnetism.
You have a very interesting approach because you take into account the amplifier. Your designs sound well with most of amplifiers. Not the case of all loudspeakers. Note I take into account now the electrical approach when I build my loudspeakers 🙂
Going to watch this video from Lars 🙂
Mind you, there will be people who disagree with me, so I am no oracle here, just sounding out.
Yes, do watch the video, even if you don't get everything, at least not a first, there is plenty to take in.
Yes, do watch the video, even if you don't get everything, at least not a first, there is plenty to take in.
No oracle here but discussion, knowledge and sharing 😉
To make the story long and explain my point of view which is an other paradigm ?
The context is when you write you have better result with the SBA NBAC versus Purifi NFA, I have a similar case at home between two 8" drivers SEAS W22NX003 and SICA 8H2CP in two loudspeakers two ways. But not the same tweeter and not the same box, not the same crossover but the overall result is the same and not the same price of driver 80€ versus 450€. I don't know really why but have the idea the motor of the SICA is very well designed, the bass is better. I don't have an enough good microphone to make a precise distortion measurement. But The distortion of the SICA doesn't vary a lot at high level according to Dibirama's test.
I consider three thing in a driver : electrical - mechanical - acoustical. Yes it is a very large subject with a lot of disciplines.
What I think reading your result is : it is a pure mechanical problem, the alu cone has a better behavior than the paper cone. I consider, make an approximation at low level : low displacement the motors performances are nearly the same. But things seem not so simple and you demonstrate. What I understand and I apply, use a driver in the area of flat impedance (flat phase and not a lot impedance variation).
I think Purifi aluminium has better performance than the paper because of the mechanical part, the cone has a better behavior. But I also think things are not so simple because we should consider the coherence of reproduction of the entire spectrum. Material cone has different behaviors in the sonic spectrum depend on stiffness, damping etc.
Second idea I see thing at the level of global acoustical response and fidelity of the loudspeaker, Is the acoustical response has all harmonics in the right place/level to a have the right sounding instrument ? you have a tweeter with a better energy transfert via the Cu cap, the loudspeaker sounds better.
Sure I miss a lot of things in the electrical-mechanical-acoustical aspects, this is why I read carefully and happy to learn new things. I take time to listen the interview 🙂
To make the story long and explain my point of view which is an other paradigm ?
The context is when you write you have better result with the SBA NBAC versus Purifi NFA, I have a similar case at home between two 8" drivers SEAS W22NX003 and SICA 8H2CP in two loudspeakers two ways. But not the same tweeter and not the same box, not the same crossover but the overall result is the same and not the same price of driver 80€ versus 450€. I don't know really why but have the idea the motor of the SICA is very well designed, the bass is better. I don't have an enough good microphone to make a precise distortion measurement. But The distortion of the SICA doesn't vary a lot at high level according to Dibirama's test.
I consider three thing in a driver : electrical - mechanical - acoustical. Yes it is a very large subject with a lot of disciplines.
What I think reading your result is : it is a pure mechanical problem, the alu cone has a better behavior than the paper cone. I consider, make an approximation at low level : low displacement the motors performances are nearly the same. But things seem not so simple and you demonstrate. What I understand and I apply, use a driver in the area of flat impedance (flat phase and not a lot impedance variation).
I think Purifi aluminium has better performance than the paper because of the mechanical part, the cone has a better behavior. But I also think things are not so simple because we should consider the coherence of reproduction of the entire spectrum. Material cone has different behaviors in the sonic spectrum depend on stiffness, damping etc.
Second idea I see thing at the level of global acoustical response and fidelity of the loudspeaker, Is the acoustical response has all harmonics in the right place/level to a have the right sounding instrument ? you have a tweeter with a better energy transfert via the Cu cap, the loudspeaker sounds better.
Sure I miss a lot of things in the electrical-mechanical-acoustical aspects, this is why I read carefully and happy to learn new things. I take time to listen the interview 🙂
Exactly, sharing is good.
Just listening to this, note how the Strat is strung, there isn't anybody quite like Uffe Steen:
The NBAC are growing on me, but now that I have the Miflex on both of them, they are both great. If I had left the Miflex in one or not the the other, then that is the one I would prefer. Re the NBAC, it reminds me of a ribbon speaker, but not with the room problems of a large fullrange ribbon. The NBAC also has a stronger bass, I call it more "fruity" there, but it seems to well with the rest of the picture it presents. It has character and I think others would like it to. Just don't use 6 Watt (I know people with MFC running them with 6W).
Maybe I am too close to all this to be totally objective on the NBAC. I would like to see what others would think, after building them. There is the fact it needs more amplifier power and of all the Elsinores, the ULD are the most efficient and requires less current to drive them. The MFC is in the middle.
Yes, maybe up to a certain frequency that is true. But I have not heard it with pink noise or an MLS signal through it, the raw driver on its own.
I suspect the Purifi will not pass this test. It will have modes between 4-5KHz and up to 10KHz. The NBAC has one that is around 7-8KHz and is very mild and relatively high up. It passes this test and of course the Purifi paper ones does as well, as they are in the 'soft' cones category along with polycones and most fibre cones (TexTreme cones are an interesting exception as they are 'fibre' and yet rated 'hard' by SB - they admirably pass the test too).
With most soft cones I can get away with a single inductor that needs to be much larger than the rated inductance of the driver. Ideally a ratio of 5:1 would be great, but 3:1 is not to be sneezed at. If the driver measures 0.15mH @2KHz and you can get 0.5mH, this is pretty good. In the NBAC I have reached a ratio of 5:1 plus and the ULD it is around 3:1, but they gain something from the motor and ultra-flat inductance. I think the MFC is around 4:1.
This means that at critical mid frequencies we get the the benefit [of so-called] current drive. That's right current drive from an inductor. The real advantage of current drive is low distortion when you have a rising inductance/impedance. The imperfections of the driver (and none are perfect) will show up there and modulate the current of the amplifier, and the lower the impedance the better it can control the voltage, but the lower impedance also means the amplifiers is virtually zero control over the current.
There is another key to make this all work, and here we are talking about the acoustic side, we must largely ignore the straight on axis response and look where the response is around 15° of axis and maybe even a bit wider than that. Let us say that 15° looks reasonable (it need not be perfect and what I am doing takes a bit of faith that it will work), now add the inductor and see what inductance is required to get it to -6dB at the crossover frequency. The ULD and MFC it is 3KHz and with the NBAC it is 2.5KHz. Now bring in your tweeter and tailor it for that fit. It should sum flat at the crossover frequency. The inductor may give you a bit of a droop just above 1KHz (there may even be a sign of the surround affecting response here) and a slight BBC dip effect. I don't really design for that, but I am simply going where the driver wants to go and I have never disliked the BBC dip effect anyway.
So I won't be able to do the above with the Purifi aluminium cone drivers. The fact that I have gotten away with it in the NBAC aluminium, that is quite amazing to me. I thank two people for this, hope they don't mind me naming them, Steve Morton and Troels Gravesen.
For sure. But I definite believe what must come first, the interface between the amplifier and speaker must be the most complex of all the electrical interfaces between the microphone to the speaker. It comes before the acoustic stage, so if we make a mess of it, the acoustic will be too late and the horse has already bolted. No doubt the acoustic side is very complex, and it has been well researched and well measured, but that electrical interface is still the most misunderstood stage of them all.
Some time ago I did some measurements and simulations of current affecting the dBSPL output in around the peak of the Em and I discovered something. The change in dB-SPL acoustic output was always proportional to the change in current. It supports what Richard H. Small told a youngster (me) in 1975, that the damping is entirely defined by the alignment and that the amplifier had no damping that it can magically apply to the speaker (at the time I did not agree with that view, but I came around later). A few weeks ago I was read an pro audio ad by a friend over the phone, for an amplifier used in recording studios. It talked about the fact that the miniscule output impedance had the ability to be a brake and somehow control or pull the cone back in the bass. This impolies that low impedance can control the position of the cone (Allen, I know you will be reading this and I welcome your thoughts) and that to me (and Small) is not supported by the physics.
My own measurements and simulations made it clear that something else was happening.. At all time the output change in level was always prerational to the current flowing through the voice coil irrespective of the output impedance of the amplifier. Measure the current and SPL, then change the current like doubling it, and you saw an exact same proportional 6dB increase in SPL. And pick any frequency and the driver always followed the current. This to me means there is no brake, rather the driver always tracked the current. Simulation with 1000 Ohm source impedance and it still responded to current alone.
OK, I am only sharing and saying my own piece: IMHO there is stuff being peddled out there that does not add up. It is mostly coming from the amplifier crowd and it has been the most incredible campaign that has sucked so many in. Speaking to the speaker guys, if they are knowledgeable and grounded, they hard ever speak of 'damping factor' and that has been my experience with them. They are more concerned how their speaker performs when tube amps are used and will the higher output impedance still be catered for. If you aim for something near a Bessel alignment, the speaker can be very tolerant of the amplifier, tubes and all. EQ the current and the output impedance doesn't matter.
So here I am just thinking aloud and typing. I am just sharing, that's all. And I don't want to correct everything out there, I just want to increase in knowledge.
Let me know your thoughts then.
Yes, that perspective trips up a lot of people who are trying to learn about this. The best advice I could give them is to initially just look at the response rolloff and that shows us the damping, which is simply the inverse of the resonance Q, like 0.7 Butterworth.
The 6 ohms of the voice coil is always there so low amplifier source impedance won't give total damping. A current source doesn't create an infinite resonance. The driver mechanical resonance has influence that will also always be part of the picture.
I listened to the interview, it is very didactic, very instructive, I saw the purify Laboratory and Lars has a little humour 🙂
What to drive with 1.2kW amplifier in a measurement system ? LOL
I understand what is modulation distortion. The motor of the Purifi is a low IMD motor.
But I have more questions after listening Lars.
What kind of distortion is really important in a driver ?
What make a good driver ?
Is resonance frequency vary with the voltage in Purifi driver ? Do they work on the spider ?
Suspension distortion ?
Note we also have videos from Kartesian
https://www.youtube.com/@kartesian5737/videos
Almost videos are in French but you can have translation with subtitles.
Inductance :
Suspension :
A big digression, I understand very well the importance of the amplifier. I designed class AB transistors amplifiers. The common problems are the power supply is no enough and impedance driving stage is insufficient to the power stage.
When you drive a load, the way you drive the final stage is important, it is the basic theory.
What I can say I designed in the past an amplifier which can really move a speaker with MOSFET but to drive the MOSFET I used a class A amplifier and 1KW power supply for 100W. MOSFET are insensitive to current load, just drive them correctly.
Tube is a little different because power supply is better, less current demand due to HT and better power for the speaker. This explains why a 10W tube amplifier delivers subjectively more power than a 50W BJT amplifier.
It is a large subject. Some people try to control the speaker driver with feedback but I don't know if it works.
And sure a bad baffle, like a bad vent tuning, can ruined the sound whatever the amplifier.
Kartesian has made this documentation About Spider in english on this page : TECHNICAL NOTES
The last thing is the membrane ie the cone and we have the main subjects on a driver unit : Electrical, suspension, membrane
Honestly very happy to dive/dig into these subjects.
🙂
The last thing is the membrane ie the cone and we have the main subjects on a driver unit : Electrical, suspension, membrane
Honestly very happy to dive/dig into these subjects.
🙂