kgrlee my curiosity and newbieness was peaked by your statements and entire thread so forgive my kindergardeness :
With current drive, one of the biggest faults is that certain variable resonances (eg the cabinet wall resonances that John Atkinson of Stereophile is so obsessed with) appear on the Impedance curve and are exacerbated by current drive.
If to use a DSP setup to process the music before it goes to the current amp- then one would want to profile the characteristics of the cabinets and then somehow have those measurements realign the DSP output prior to going to the current amp-
Variables are only those that are not defined-
There are additional sensor arrays that could be implemented in the cabinets- to measure EMF backforce and its varied actions and how they relate in the 3d inner-space of the volume and subsequent results in listening-space- and pressures across the entire surfaces and so on--- Is there perhaps a database of materials with regard to resonant frequencies and pressers at frequency?
This in my meager opinion would solve the microphone delay
With current drive, one of the biggest faults is that certain variable resonances (eg the cabinet wall resonances that John Atkinson of Stereophile is so obsessed with) appear on the Impedance curve and are exacerbated by current drive.
If to use a DSP setup to process the music before it goes to the current amp- then one would want to profile the characteristics of the cabinets and then somehow have those measurements realign the DSP output prior to going to the current amp-
Variables are only those that are not defined-
There are additional sensor arrays that could be implemented in the cabinets- to measure EMF backforce and its varied actions and how they relate in the 3d inner-space of the volume and subsequent results in listening-space- and pressures across the entire surfaces and so on--- Is there perhaps a database of materials with regard to resonant frequencies and pressers at frequency?
This in my meager opinion would solve the microphone delay
Non est tantum facile.
There's the small matter of making sure the model has some semblance to real life.
I did something similar in the 80's to get models to run on my own linear circuit analyser but I had the facility to check out real life stuff in dem days.
A big caveat with 3886 is that it appear a Zobel is essential and this mucks up the HiZ output of a Current amp.
I've said "Blameless" isn't my favourite topology but I've posted more detailed stuff & advice on it than any other on this forum. That's cos I have 'real life' experience of it from Jurassic days and Toni has been good enough to try stuff out for me in his 2stageef-high-performance-class-ab-power-amp-200w8r-400w4r.html
I like Self's & Cordell's stuff, not cos they support some pet theory of mine .. but cos for their pet theories, they present 'real life' verification.
Bob is in the process of revising some of his 'real life' (and SPICE world) examples .. for which I take my hat off to him.
From your forum name, do I understand that you've done some work with scanned laser interferometry?
Wharfedale & Celestion where the first to do serious work on this, both on drive units as well as cabinets.
But you have to translate this to the 3D output of the loudspeaker. You might be able to do this with an array of accelerometers for each panel but the computational power required would be immense. You need to do it with each individual speaker in situ as just changing the speaker stand has an effect. As does temperature & humidity.
I think Klippel's work is Bleeding Edge at the moment but my secret sources tell me he's stopped developing it cos he can't see how to protect his intellectual property.
If anyone knows better, please speak out.
So I've yet to see any (reported) evidence that his device results in better sound.
______________________
To put some of this into perspective, Richard Greene of Hawkesford's Essex group developed one of the first practical (linear) DSP EQs. They tried it out on a Celestion SL6 (or maybe SL600) and got better sound but not as much as they expected.
They then tried it out on one of my designs and got much better improvement cos that speaker had addressed many of the non-LTI evils. SL6 & 600 compressed badly at HF as the treble unit didn't have Ferrofluid.
And to round off the story, based on Listening Tests on the EQ'd speaker, I was able to modify my plain speaker to have a substantial part of the improvement brought about by the DSP EQ.
______________________
To come back to the cabinet stuff, there's a BBC Eng. Monograph on panel damping of LS3/5a. An independent Wharfedale Enginering Memo. confirmed their conclusions as to what was audible and what wasn't.
Richard, I hoped it was clear enough this is the way how I try to work with current driven speaker. DSP pre-equalization of the signal before it goes to the V/I input.
It is really very easy, these days. You generate the requested transfer function, make an impulse response from the transfer function and then you play music (or test signals) through a convolver with the impulse response. At 32-bits, there is no signal degradation and today's PCs have no problems to handle this. Quite an Utopia, 40 years ago, when I was 20
To build a V/I amplifier, one needs a soldering iron, as you have already stated correctly. So, I understand that the beach is probably not the best platform for the real work, though it may be inspiring, to some extent, for suggestion thoughts
Last edited:
I think that these two measurements may be quite interesting. They show distortion of tweeter current, when the tweeter is supplied from amplifier with current output, and corresponding voltage distortion at tweeter terminals. The distortion of tweeter voltage is a result of tweeter non-linear electrical impedance. Distortion of the current that feeds the tweeter is negligible, compared to its non-linearity.
Attachments
Pavel, which convolver do you use?Richard, I hoped it was clear enough this is the way how I try to work with current driven speaker. DSP pre-equalization of the signal before it goes to the V/I input.
It is really very easy, these days. You generate the requested transfer function, make an impulse response from the transfer function and then you play music (or test signals) through a convolver with the impulse response. At 32-bits, there is no signal degradation and today's PCs have no problems to handle this. Quite an Utopia, 40 years ago, when I was 20
Can it be integrated seamlessly with Windoze Media Player?
___________________
When I emerged from the bush after nearly a decade from civilisation, I found that computing power had increase out of all my expectations. Circa 1990, I developed the theory of measuring Response & Distortion using a log sweep in the theoretically shortest possible time for production. But the computing power and particularly good A/Ds & D/As were too expensive.
Today, I can do measurements in my beach shed mostly better than I could in da last Millenium with Anechoic chambers and $$$ .. using just a laptop with a good sound card.
Meanwhile Prof Angelo Farina, Parma U., had developed this independently and Angelo's method is now in the latest AP.
___________________
I'm presently involved with a number of DSP projects. eg the digital filters in VVmic for TetraMic are mine.
But the interface is clunky and only plays Ambisonic files. The ideal is to replace the Windoz mixer so all sound output can be processed conveniently.
___________________
And how do you measure the response and derive your filters?
Browsing sites like REW show a lot of naive filters generated which will often not result in better sound .. OK with dinosaur footsteps
___________________
Lastly, are you actually doing what I proposed .. combining the inverse of the Impedance curve and 'measured' response with a target function to get good sound from a Current fed speaker?
Richard, I make an impulse response with rePhase or REW5 and then I use foobar convolver plug-in.
foobar2000: Components Repository - Impulse Response Convolver
Pavel, have you measured Zo with frequency of this amp? Without the R23 C11 & R14 C5 Zobels of course.
Only simulated, I will measure it as well. This amp works even without those two Zobels for speaker testing, but I do not feel very well then
Audio does not yield signals that would make a HF trouble, it is rather a question of sudden RFI EMI impulses, without Zobels you may sometimes hear a click from the speaker, which does not make me feel good. Scope check of amplifier output is of course used.
So, I read the PDF. No way that any DIY person is going to do this level of DSP work. Anyone who can is not an amateur.
The approach is complicated. DIY people want a methodology that is actually attainable. AT HOME.
If you read his chronology he mentions estimation of velocity. I think that the solution is actually measuring cone velocity without the heroics of accelerometers, capacitive pickups, extra coil windings, etc.
laserpaddy,
Unless you are going to actually measure real resonance values off the panels of and enclosure I don't think any formulaic method will predict resonance values with any accuracy. There are just to many variable involved here as far as I understand. You would have to have a very stable material to start with and wood isn't one of those materials as there are just so many naturally occurring variations right there. Then you have to look at all the joints and how they are attached to each other. Add in some ribs and you change everything. I have seen cabinet designs that have also used both tension and compression on the walls to tune this and things start to look pretty hairy to just use a computational method to solve these problems.
I have taken some of this information that I understand and have worked with alternative materials that do not have some of the variation issues of natural materials. I actually work with polyurethane high density foam materials and this allows me to control the density of the final material and also can be used in proper design to eliminate many of the joints and adhesive that need to be used and also the ribbing or stiffeners become an integral part of the structure. Being able to control both density and flexural modulus is a great advantage. I have been working with this material since 1976 and have spent in the neighborhood of over one Million dollars to get where I am today. I actually built a manufacturing facility to do this work in 1985 and it was not something that many would attempt. I was young and stupid at the time and wasn't smart enough to understand the amount of time it would take to do all this and ended up doing product development and materials development to support this effort. Of course that took most of my time but I did learn a lot doing other peoples product development.
I have an old patent, long since expired using these materials in horns/waveguides and could have made some real trouble for others doing some of the things they were doing at the time but didn't feel the need as I knew that what I understood was not common knowledge or available to others to use. I basically owned the patent to make any audio acoustical horn of a foamed plastic material in the USA. I could have crimped Earl Geddes efforts early on if I had attack everyone else who was working in this area but why stop others from improving the art of acoustics was my feeling. I also beat Earl to the idea of the Oblate Spheroid design by a few years but didn't call it that or protect it with a patent so he won that one. His math skills are way beyond me but I did imagine and work with my brother the math wiz and some professor from Caltec on how to do the math to describe that type of waveguide design. The original JBL Eon speaker was a knockoff of some work I was doing at the time that was seen by their internal engineers as I was also manufacturing component parts for one of their high end consumer speakers at the time called the XPL series.
One reason I really like this site is the free exchange of information between people and some of these are real player in the industry who really don't have to do this. I appreciate all the help others have offered here and I try to do the same where I can.
Unless you are going to actually measure real resonance values off the panels of and enclosure I don't think any formulaic method will predict resonance values with any accuracy. There are just to many variable involved here as far as I understand. You would have to have a very stable material to start with and wood isn't one of those materials as there are just so many naturally occurring variations right there. Then you have to look at all the joints and how they are attached to each other. Add in some ribs and you change everything. I have seen cabinet designs that have also used both tension and compression on the walls to tune this and things start to look pretty hairy to just use a computational method to solve these problems.
I have taken some of this information that I understand and have worked with alternative materials that do not have some of the variation issues of natural materials. I actually work with polyurethane high density foam materials and this allows me to control the density of the final material and also can be used in proper design to eliminate many of the joints and adhesive that need to be used and also the ribbing or stiffeners become an integral part of the structure. Being able to control both density and flexural modulus is a great advantage. I have been working with this material since 1976 and have spent in the neighborhood of over one Million dollars to get where I am today. I actually built a manufacturing facility to do this work in 1985 and it was not something that many would attempt. I was young and stupid at the time and wasn't smart enough to understand the amount of time it would take to do all this and ended up doing product development and materials development to support this effort. Of course that took most of my time but I did learn a lot doing other peoples product development.
I have an old patent, long since expired using these materials in horns/waveguides and could have made some real trouble for others doing some of the things they were doing at the time but didn't feel the need as I knew that what I understood was not common knowledge or available to others to use. I basically owned the patent to make any audio acoustical horn of a foamed plastic material in the USA. I could have crimped Earl Geddes efforts early on if I had attack everyone else who was working in this area but why stop others from improving the art of acoustics was my feeling. I also beat Earl to the idea of the Oblate Spheroid design by a few years but didn't call it that or protect it with a patent so he won that one. His math skills are way beyond me but I did imagine and work with my brother the math wiz and some professor from Caltec on how to do the math to describe that type of waveguide design. The original JBL Eon speaker was a knockoff of some work I was doing at the time that was seen by their internal engineers as I was also manufacturing component parts for one of their high end consumer speakers at the time called the XPL series.
One reason I really like this site is the free exchange of information between people and some of these are real player in the industry who really don't have to do this. I appreciate all the help others have offered here and I try to do the same where I can.
This is the concept and a bit of the implementation.
If you need the complete circuit, PM me
George
Attachments
Current-drive should work wonders with easy to DIY moving magnet - large fixed coil drivers. Imagine a circular piece of foam (30 inch or so) (divinycell) covered in carbon fiber or thin layers of veneer, with a series of small lightweight Nd magnets along the outer rim (or a fixed electromagnet along the rim). Outside of this is a large fixed magnet that can be much much larger then any possible moving voice coil (say 1000 turns of 5 mm diameter cable). The added Le is not a problem as the current drive largely ignores added inductance. The fixed voicecoil can be so many turns with soo thick cable that the possible magnetic field is strong enough to overcome the efficiency loss of the comparably heavier moving mass and lack of magnetic field-strength from the Nd magnets.
With a large diameter surround it is much easier to create a very linear resistive suspension with a low Qms, which enables a much more pure current drive without the typical Fs-peak from modern "normal" high Qms drivers.
I want to DIY something along this line. It should make a very OB-suitable bass driver.
Cheers,
Johannes
With a large diameter surround it is much easier to create a very linear resistive suspension with a low Qms, which enables a much more pure current drive without the typical Fs-peak from modern "normal" high Qms drivers.
I want to DIY something along this line. It should make a very OB-suitable bass driver.
Cheers,
Johannes
Kindhornman,
What do you think about Magico metal structure loudspeaker enclosures? Is the final result worth it, because the sheer weight of these enclosures is fascinating. My friend recently bought one of these floor standing monsters (because Swiss company Solution told him they are using Magico and he owns these amps already).
What do you think about Magico metal structure loudspeaker enclosures? Is the final result worth it, because the sheer weight of these enclosures is fascinating. My friend recently bought one of these floor standing monsters (because Swiss company Solution told him they are using Magico and he owns these amps already).
Ivan,
For some reason people assume if you use a stiff material that it will magically remove resonance problems and everything is simple. That is far from the truth. Think about a metal bell, it is very dense and at the same time as the name implies it will resonate like a bell if not well damped. I can't say anything about the final properties of those Magico enclosures it is very engineering dependent. I can see by the website that these are very highly structured enclosures with many machined sections. That can very well be used to damp the natural ringing tendencies of a flat metallic panel. As you can well imagine that takes a very heavy enclosure to do that and it does not say that there is not also some internal stuffing to do with the metallic structure. I have sitting in front of me some very heavy granite enclosure I once had made for an enclosure. It is very dense and in the configuration it is machined in it is fairly well damped, if you knock on it it does not ring but that body is made of a single piece of granite machined down through the center, no joints at all, one solid block. At the same time the damned things weigh 500 lbs each or in your side of the world over 200 kilos. I did this as a design exercise, it was for the looks more than for the material properties. I could have done something in wood that would have sounded just as good but wouldn't have that look, it is black granite with blue colors running through the granite.
So my answer would be that though it could be very good sounding to use aluminum for an enclosure if well designed if not it could sound like a bell and ring for long periods of time. That is both expensive, aluminum plate of any size it not cheap and expensive to manufacture at the same time. There is nothing wrong with a wooden enclosure if you understand how to build it and how to use stiffeners in the correct fashion. Simple things like moving stiffening ribs off center from a panel so you have two different surface areas and such are some of the things required but very well understood.
For some reason people assume if you use a stiff material that it will magically remove resonance problems and everything is simple. That is far from the truth. Think about a metal bell, it is very dense and at the same time as the name implies it will resonate like a bell if not well damped. I can't say anything about the final properties of those Magico enclosures it is very engineering dependent. I can see by the website that these are very highly structured enclosures with many machined sections. That can very well be used to damp the natural ringing tendencies of a flat metallic panel. As you can well imagine that takes a very heavy enclosure to do that and it does not say that there is not also some internal stuffing to do with the metallic structure. I have sitting in front of me some very heavy granite enclosure I once had made for an enclosure. It is very dense and in the configuration it is machined in it is fairly well damped, if you knock on it it does not ring but that body is made of a single piece of granite machined down through the center, no joints at all, one solid block. At the same time the damned things weigh 500 lbs each or in your side of the world over 200 kilos. I did this as a design exercise, it was for the looks more than for the material properties. I could have done something in wood that would have sounded just as good but wouldn't have that look, it is black granite with blue colors running through the granite.
So my answer would be that though it could be very good sounding to use aluminum for an enclosure if well designed if not it could sound like a bell and ring for long periods of time. That is both expensive, aluminum plate of any size it not cheap and expensive to manufacture at the same time. There is nothing wrong with a wooden enclosure if you understand how to build it and how to use stiffeners in the correct fashion. Simple things like moving stiffening ribs off center from a panel so you have two different surface areas and such are some of the things required but very well understood.
Bl is composed of Tesla * meters. Less Tesla is easily compensated for by more meters.
A "normal" voice-coil is 3 - 4 inch in diameter and constrained by having to be light and fit into a very narrow gap. A fixed rim-mounted voice-coil can be huge! It is 1,44 meters around the circumference of an 18 inch driver, but only 0,32 meters circumference for a 4 inch voice-coil. 200 turns around the 4 inch coil is 64 meters. 1000 turns (no need to make it lightweight) around the the outer rim of a 18 inch driver is 1500 meter (the coil grows in diameter as one stacks several layers of wire). One can use less Tesla from the Nd magnets for any given value of Bl, since the large coil has 23 times the "meters" compared to the 4 inch coil.
Even though the magnetic losses is much larger for a rim mounted fixed coil, i think it is very easy to build extremely strong high Bl motor this way.
Current drive turns everything upside down. With current drive one does not have to care about reactive loads and high inductance in the coil. It is better to use this advantage in the design of the driver.
Cheers,
Johannes
Circlomanen,
I just try to imagine the mass increase you are talking about with moving magnets around that size diameter. The bandwidth would be seriously compromised with the increase in mass. Also a flat diaphragm that large, even one made of carbon fiber would have some serious breakup modes compared to a curvilinear cone.
I just try to imagine the mass increase you are talking about with moving magnets around that size diameter. The bandwidth would be seriously compromised with the increase in mass. Also a flat diaphragm that large, even one made of carbon fiber would have some serious breakup modes compared to a curvilinear cone.
S-06-10-N: Rod magnet Ø 6 mm, height 10 mm (Neodymium Magnets) - supermagnete.de
Each magnet weights 2,1 grams. A B&C 18IPAL driver has 330 grams mms. We can use 80 of these spaced around the rim of the "cone" for a total weight of 168 grams. A circular disk of 60g/dm2 divinycell or Airex foam, 20 mm thick in the middle and 10 mm at the rim, sanded to a smooth radius convex shape weights about 100 grams. Add one layer of 150 g/m2 carbon-fiber in epoxy which gives another 80 grams. Say a total of 350 grams. moving mass. I can´t model magnet systems, but i believe the strength of those magnets would be quite powerful. Add a huge extremely powerful fixed voice coil.
This is just a fast guesstimate. I have not modeled anything. But i think it could be fun.
I tested this idea by coiling 150 turns of 0,75 mm2 cable around the end of a pvc-pipe. Inside the pipe a placed a stack of small Nd magnets glued to a thin piece of paper. I used a 10 ohm power-resistor in series with the amp to save it from driving a very low impedance. It could fill my kitchen with sound. It was not pretty since it was just a fast and simple test to show my kids how a loudspeakers works. But i was impressed by the spl possible by this very crude and simple "loudspeaker". If scaled up and done with some serious effort (not a 10 minute fast demonstration for a 10 year old kid) i think it could work well.
Cheers,
Johannes
Each magnet weights 2,1 grams. A B&C 18IPAL driver has 330 grams mms. We can use 80 of these spaced around the rim of the "cone" for a total weight of 168 grams. A circular disk of 60g/dm2 divinycell or Airex foam, 20 mm thick in the middle and 10 mm at the rim, sanded to a smooth radius convex shape weights about 100 grams. Add one layer of 150 g/m2 carbon-fiber in epoxy which gives another 80 grams. Say a total of 350 grams. moving mass. I can´t model magnet systems, but i believe the strength of those magnets would be quite powerful. Add a huge extremely powerful fixed voice coil.
This is just a fast guesstimate. I have not modeled anything. But i think it could be fun.
I tested this idea by coiling 150 turns of 0,75 mm2 cable around the end of a pvc-pipe. Inside the pipe a placed a stack of small Nd magnets glued to a thin piece of paper. I used a 10 ohm power-resistor in series with the amp to save it from driving a very low impedance. It could fill my kitchen with sound. It was not pretty since it was just a fast and simple test to show my kids how a loudspeakers works. But i was impressed by the spl possible by this very crude and simple "loudspeaker". If scaled up and done with some serious effort (not a 10 minute fast demonstration for a 10 year old kid) i think it could work well.
Cheers,
Johannes