System7,
If you go look for some old posts of mine you will find that long ago I talked about using conjugate networks for correcting the impedance in a passive network. This is old news for some of us, that has nothing to do with passive network noise. I was taught some of what I am seeing here in these passive networks about 1993 when a friend and mentor of mine did some networks for me for my horn loaded speakers, the ones in my avatar and I didn't have any idea what he was doing, it just looked strange. I finally realized what he was doing and how he was adding a slope function to the passive network, so this is truly nothing new, just not something that you see in a classic passive network in a book. Those never work anyway as they never take into consideration the actual impedance at the crossover point, just plug in 4 or 8 ohms in the calculator and get whatever order filter you want. Never works correctly.
I actually have the book by Esa Merilainen, Current-Driving of Loudspeakers. It leave so much not answered, he always talks about modifying normal speakers for current drive but besides a servo feedback or something like that never explains how you would modify a normal speaker without adding a second voicecoil or any other practical changes like raising the impedance much higher than a normal speaker, say 100 ohms.
If you go look for some old posts of mine you will find that long ago I talked about using conjugate networks for correcting the impedance in a passive network. This is old news for some of us, that has nothing to do with passive network noise. I was taught some of what I am seeing here in these passive networks about 1993 when a friend and mentor of mine did some networks for me for my horn loaded speakers, the ones in my avatar and I didn't have any idea what he was doing, it just looked strange. I finally realized what he was doing and how he was adding a slope function to the passive network, so this is truly nothing new, just not something that you see in a classic passive network in a book. Those never work anyway as they never take into consideration the actual impedance at the crossover point, just plug in 4 or 8 ohms in the calculator and get whatever order filter you want. Never works correctly.
I actually have the book by Esa Merilainen, Current-Driving of Loudspeakers. It leave so much not answered, he always talks about modifying normal speakers for current drive but besides a servo feedback or something like that never explains how you would modify a normal speaker without adding a second voicecoil or any other practical changes like raising the impedance much higher than a normal speaker, say 100 ohms.
I do agree that Joe goes off the reservation sometimes.
But I have to admit he is in my top ten of audio people. 😀
And he answers my PM questions honestly. Power to you, Joe! 🙂
BTW, I have been investigating a top German loudspeaker man recently. Much as I love him, he makes it more difficult than it is. Sorry I can't be more specific. 😱
Last edited:
Hi Steve,
If you can't mention any details, why on earth would you even bring it up in the first place? Your mystery speaker man has become completely irrelevant.
Nothing wrong with liking Joe. Yes, he does his best to answer questions honestly as he knows. But there's the rub ... he is completely wrong about things that conflict with the known universe. Saying, "well, it could happen" doesn't cut it at all. You could be a guiding light to lead Joe back from the cold place he finds himself in right now. Help him.
-Chris
If you can't mention any details, why on earth would you even bring it up in the first place? Your mystery speaker man has become completely irrelevant.
Nothing wrong with liking Joe. Yes, he does his best to answer questions honestly as he knows. But there's the rub ... he is completely wrong about things that conflict with the known universe. Saying, "well, it could happen" doesn't cut it at all. You could be a guiding light to lead Joe back from the cold place he finds himself in right now. Help him.
-Chris
That's odd. This thread sort of started on flattening the impedance response. I posted about that with a reasoned opinion a few pages back. I am interested. So why don't you explain why it is important. Was I wrong? Why was I wrong?
I also have a question about your lyrical review of that smooth speaker. You mentioned that in your opinion the smooth and great sound was a result of it's flat impedance response. Can you explain why that should be, and not just a coincidence? After all, two things happening at the same time does not mean that one caused the other.
I am not trolling but am genuinely interested. The fact that you find a speaker smooth is in itself not of much interest for the rest of us. The fact that you think you have found out how to make it so certainly would be of interest!
Jan
I'm still wondering what 'buttery' sounds like in a speaker. I am not sure it's something I would want.
You know, much has been written about current drive; about reduction in distortion etc. Unfortunately most of it only applies to a driver connected directly to an current source amplifier. As soon as any parallel component or network is placed across the driver's terminals it no longer sees constant current drive.
The basic premise of current drive is that it eliminates the influence variations in impedance due to VC heating, the back emf, etc on the current, thus the driving force. The driver's impedance can be expressed as
Z = Re / (1 – BL * v / V)
where v = cone velocity and V is the voltage across the VC. In a perfect world Re, BL(x), and the ratio of v/V would be constant. But in the real world they are not thus Z will vary due to any number of reasons. With a driver connected directly to a current source none of this will effect the driving force, BL * I , since I will remain constant. But as soon as there is something in parallel with the Z, like a conjugate network, Z*, the component of current through the driver will vary as Z varies because Z in parallel with Z* forms a current divider and I through the driver will be
I = 1/( 1 + Z/Z*)
Now all the factors that influence Z are back in the equation and will influence the current actually flowing in the VC. Additionally, since the parallel impedance,
Z|| = Z x Z* /(Z + Z*)
will change due to the variation of Z, the voltage across the driver will also change when connected to a current source,
V = Is / Z|| where Is is the (constant) current from the source.
So, is this better or worse that a voltage source, or just different?
Adding a conjugate across a driver, or a conventional speaker, to flatten the impedance will make it suitable for use with a current source amplifier, but this is still a far cry the supposed benefits obtained by connecting a driver directly to a current source amplifier. And recognize that since Z varies, the conjugate network will not provide the necessary correction to flatten the system impedance under real world conditions.
Last edited:
Indeed. An other way to look at it is that the 'black box' seen from the driver, containing a current source and some // network is no longer a current source.
Jan
Jan
...so what is your point on topic? VS is also not ideal!
It's all a tradeoff, ESPECIALLY for EE. Or isn't it so?
I have lots of years of experience with colleagues in production or RnD or even former colleagues in academia, who, while forced to expand their formal expertise too fast it can happen that they might fully grasp what's happening (on intuitive//experience level) but if you put them to explain and write... oh better not!!! I always advise them to get "professional corrections" before they write something. Some listen, some not, either way this does not mean almost anything. The products are always good because they KNOW their job. They are also a reliable team of experts for critical support of RnD.
So, I read you take no lessons from people with lesser formal education than yours. OK, it's not nice, but I don't care. But when I offered an educated opinion, apparently is still your choice to pass. OK, not nice, again I don't care what you choose to do. My conclusion is that you are no better than Joe, at least in what you complain with him. Off topic of course.
The thread has gone a bit too far with this duplicity attitude: intransigent with CS but fully relaxed about VS. May I ask you WHY? I do not see any chance for a scientific answer. It's just.. stirring.
If you can, please let people be. It's a really really big world, and you know sometimes is very small as well. As I said, I personally indulge Joe's statements... based on his merits or on my hopes.
As I said, if Joe does not bring on the table the TS+ electrical parameters then I will simply ignore this X-over thread like a did with many others (about 98%???)... inclusive some of Nelson Pass... inclusive SY's web site beyond a point... etc.
Best to all,
Last edited:
Where has Jan said that? From all his posts here you can see he is very open to all well reasoned viewpoints. He has asked some very pertinent questions. When a theory is proposed that is not in line with generally accepted physics why should leading questions not be asked?
Maybe I have read that in your posts? Or maybe Sy? or.. some others?
What theory? Joe's?.. Mine couple of pages before? Mine was more correct.
Sorry... I really am.
JAN, sorry.
I have misdirected the source of that statement.
Bill: I have said that already above. Maybe you owe me an apology?
Ionmw
Accepted. But it is not that you only attributed it wrongly - the statement was never made by anyone.
Sometimes it is good to spend a bit more time reading and less typing.
Jan
John K,
The use of the conjugate network was always used in the context of a voltage source, so sorry for the confusion on that point. My point was that I don't have anything against using a conjugate network, I still think it is more precise than a simple Zobel, but I understand careful selection of a zobels components can be effective, but I think the LCR conjugate approach if more effectively correct. Don't ask me to put up the math such as you do, I only understand it intellectually.
Now my real question all along has been how you change the standard driver, no servo or extra magnetic motor circuit with a secondary coil? This is never a clear concept, what Esa would do or what changes would be made in a standard type dynamic driver to make it work with a current source. I think the speaker that I have developed would be a good device to try the theory as it has some of the properties you talked about. since the gap length is extremely long vs the coil height the coil never will see the flux density change that a normal speaker will see and the gap has a Faraday sleeve over its entire length. At full excursion the energy in the gap will not change. Now there is still the fact that the suspension is not perfectly linear due to the surround and spider but that is something I have some ideas about also, just not something I will talk about yet, not till I prove or disprove my concept of linearizing these two components to myself. May be a patent in there!
Now here is another question of the practically of current drive and having no components between the amplifier and the drive unit. Now if we are talking about a bass/mid cone driver I wouldn't have much hesitation to try this, but would anyone do this with a dome tweeter? What I mean would you take the chance of destroyed dome tweeters with no way to block any DC current leakage by accident or poor amplifier design, would you take the chance of not even having a single capacitor between the amplifier and the dome tweeter for protection purposes, wouldn't that simple protection defeat the idea of the current drive by your above explanation or would a series cap with no shunt element across the dome tweeters voicecoil still give a linear current drive and meet the requirements?
The use of the conjugate network was always used in the context of a voltage source, so sorry for the confusion on that point. My point was that I don't have anything against using a conjugate network, I still think it is more precise than a simple Zobel, but I understand careful selection of a zobels components can be effective, but I think the LCR conjugate approach if more effectively correct. Don't ask me to put up the math such as you do, I only understand it intellectually.
Now my real question all along has been how you change the standard driver, no servo or extra magnetic motor circuit with a secondary coil? This is never a clear concept, what Esa would do or what changes would be made in a standard type dynamic driver to make it work with a current source. I think the speaker that I have developed would be a good device to try the theory as it has some of the properties you talked about. since the gap length is extremely long vs the coil height the coil never will see the flux density change that a normal speaker will see and the gap has a Faraday sleeve over its entire length. At full excursion the energy in the gap will not change. Now there is still the fact that the suspension is not perfectly linear due to the surround and spider but that is something I have some ideas about also, just not something I will talk about yet, not till I prove or disprove my concept of linearizing these two components to myself. May be a patent in there!
Now here is another question of the practically of current drive and having no components between the amplifier and the drive unit. Now if we are talking about a bass/mid cone driver I wouldn't have much hesitation to try this, but would anyone do this with a dome tweeter? What I mean would you take the chance of destroyed dome tweeters with no way to block any DC current leakage by accident or poor amplifier design, would you take the chance of not even having a single capacitor between the amplifier and the dome tweeter for protection purposes, wouldn't that simple protection defeat the idea of the current drive by your above explanation or would a series cap with no shunt element across the dome tweeters voicecoil still give a linear current drive and meet the requirements?
DC blocking technology existed long time ago. People like to add parts instead of using a single component, and I'm not talking about cap.
First of all, a driver, any driver, will work fine with a current source. It will just have a different response because the driving force will vary differently with frequency. The only issue is really one of VC heating and burn out but I don't want to get into that, other to point out some very simple issues. I think most know that heating in the VC is due to heat generation which is I^2 x Re, or, equivalently, V^2 / Re. Just looking at this it should be apparent that as the VC heats up and Re increases, the heat generation for a current source increases where as for a voltage source it decreases. So, in a sense, with a voltage source the heat generation is self limiting but for a current source heating grows exponentially. Details in heat rejection to the surroundings will ultimately decide if the VC burns out in either case, but the difference is apparent.
Back to the difference in current and voltage drive and the driver response. Again, consider the driver's Zd. When connected to a voltage source the current through the VC goes like 1/Zd and around the resonance peak the current, thus driving force, drops and we seen the typical response. When connected to a current source, the current doesn't drop around resonance. Instead, the voltage across the driver's terminals goes like Zd, with a peak at resonance. In effect, connecting an ordinary driver to a current source yields a response that looks like what you would have if you placed a parametric EQ stage with a Q Peak filter before the input of the voltage source amplifier.
Now put a perfect conjugate network across the driver. The parallel impedance of the driver with conjugate, Z||, will be constant and purely resistive. For a voltage source this makes no difference, since the voltage across Z||, hence the driver, will not change. The current through the VC will still be V/Zd, but the amplifier will also have to deliver more current since there will be current though the conjugate will be V/Zc. So, other than the additional current load on the amplifier, nothing really changes. On the other hand, what happen with a current source is that the amplifier now sees a constant load, Z||. Since I is constant and Z|| is constant, the voltage across Z|| is V =I x Z|| and is also constant instead of having a peak around resonance, and the current through the VC is now V/Zd. There are always detail to consider, but in the ideal case when you place a conjugate network across a driver or a complete speaker, as far as the driver/speaker is concerned, it might as well be connected to a voltage source.
Anyway, that's my very simplistic view.
Seanny,
I'll assume your talking about an output zobel on the amp and any servo to counteract any dc offset. But even with that without knowing an amplifiers exact specifics which most people don't know and wouldn't think about when they connect an amplifier would you not put a cap on the tweeter just for a safety margin? Simple enough to do an not really affect the network if your just putting a first order cap in series with the dome tweeter, it doesn't have to be at the crossover frequency to offer protection. Whether that would cause a problem with current drive is the real question that I had.
ps. Saw your response after I posted this.
I'll assume your talking about an output zobel on the amp and any servo to counteract any dc offset. But even with that without knowing an amplifiers exact specifics which most people don't know and wouldn't think about when they connect an amplifier would you not put a cap on the tweeter just for a safety margin? Simple enough to do an not really affect the network if your just putting a first order cap in series with the dome tweeter, it doesn't have to be at the crossover frequency to offer protection. Whether that would cause a problem with current drive is the real question that I had.
ps. Saw your response after I posted this.
- Status
- Not open for further replies.