Thanks especially for confirming my observation wrt speaker drivers being local feedback systems at low frequencies were microphonic velocity voltage is large, depending on total termination impedance (as seen from an ideal zero-ohms voice coil), something many speaker, and even driver designers are not actually aware of as it is seldom stated explicitly.
Like Lars said, driver terminal voltage requirements are set by the driver as is, following the desired frequency response target, regardless of how it is driven. Plus, for most practical purposes, current drive is close to ideal when the output impedance is about 5 times the driver impedance within the frequency range of interest.
What is extremely hard to design is a universal laboratory-grade 4-quadrant current source that is high precision, high output compliance, large max output current, still high-Z at very high frequencies (MHz territory), and low distortion, all at the same time.
I think I'll continue with my little class-A gadget. Mixing DC servo feedback with current feedback at high frequencies is an easy enough starting point, but it's a gradual 1st order transition.
I may have to build a couple of switchable options for testing:
-Voltage feedback that tracks the speaker's cone excursion, crossing over to current feedback at the minimum impedance. For a woofer in a sealed box that looks like a 2nd order filter, so I'd have to wrap my head around avoiding instability.
-And/or enhance the voltage feedback around the bass impedance peak. Residual amounts of current feedback could be removed with a notch filter.
I may have to build a couple of switchable options for testing:
-Voltage feedback that tracks the speaker's cone excursion, crossing over to current feedback at the minimum impedance. For a woofer in a sealed box that looks like a 2nd order filter, so I'd have to wrap my head around avoiding instability.
-And/or enhance the voltage feedback around the bass impedance peak. Residual amounts of current feedback could be removed with a notch filter.
The thread looks to be dying...
It is interesting that any thread about current drive comes alive and then dies.
What happens?
The usual thing happens. A few will have their words of wisdom expressed, the usual links and websites are pointed to, and few hopefully friendly exchanges. Then nothing.
And I suspect it has happened again.
Let me show you guys something a little different, something a bit outside the box.
There a phrase used by Terry Demol on the old Blowtorch thread, where he said some were "textbook jockeys" and rather rigid if something is not said exactly as the textbooks do. The language is all wrong; so I have been told. I agree that language is important, but we should also be flexible.
With that in mind, take a look below, here is something curious:
What is happening here? Two FR graphs that look very similar and yet they are not exactly the same. Indeed multiple single measurements can yield these kind of differences, that is why it is a good thing to use many samples and average them out. But these are not samples of the same measurement, they are actually two entirely separate measurements.
Two plots, Red and Green.
One is Voltage Drive and the source impedance is 0.1 Ohm and since this is an 8 Ohm speaker, the nominal damping factor is supposed to be a healthy 80.
The other is Current Drive and the source impedance is 270 Ohm and as this is an 8 Ohm speaker, then nominal damping factor is... calculate... zero.
Which is which? Why do the look the same? How can that be?
Can you guess which is Current Drive and which is Voltage Drive?
I can't.
Looking closely at the FR (2Pi) we can see that this is around -3dB @ 40 Hertz and -12dB at 20 Hertz. What we see is a classic 2nd order Butterworth alignment tuned to 40 Hertz. Also I am using classic language.
What does this prove? It must prove something. This is what I said.
Note the critical phrase here is "varying current with frequency."
When I said that, the nameless KSTR starting to criticise pretty much anything said.
His answer in the form of two statements:
Mimics damping?
Good one! I don't know of anything alluding to that in any textbook.
But I completely understand him. He is not the only. But it is all a misunderstanding.
He has completely misunderstood something. He assumes that the frequency response has been EQ'd, and this is not so! He does not realise that it is current EQ. The frequency response then naturally responds to the EQ current. It becomes the result of the current being equalised.
I consider KSTR ghost with no name. My name is Joe Rasmussen, so what is his? He could at least PM me and introduce himself. If I offer criticism, then at least others know that I am not a ghost. I have a backbone and I should hope that others also so.
The alignment defines the damping of a speaker at low frequencies.
And what defines that alignment is the current that flows at:
Fl, Fc and Fh.
These three frequencies are not hard to calculate. If the right current turn up at the right frequency in both Red and Green examples above, then both have the same alignment and the damping is the same too. So simple.
KSTR's second statement, that Current Drive is equivalent to an open circuit. And he adds "by sheer definition."
But how that be? Open terminals of the speakers does not see a Current Source looking back "by sheer definition." Open terminals will only see air, and air cannot supply current. So sorry, the analogy does not suit the physics.
A current source is not air.
Air cannot produce current.
_______________________________________________________________________
Back to our two graphs. So I checked my records, Red is Voltage Drive and Green is Current Drive.
And guess what?
They sound the same!
I have demonstrated this to people coming through my front door. Play Joe Morello's "Take Five" drum solo, that is some test for any loudspeaker. Try organ music too. Anything you like.
I know what textbook jockeys think because I am well ahead of them. But they lack something, the ability to extend their thinking beyond the textbook. But a fine reading of those textbooks does allow us to say certain things. Loudspeakers are electrical motors and that is classic thinking. Hence there are already certain facts in place. Motors only respond to current. Our loudspeakers are both motors and transducers. Again I am speaking 100% classically. Being transducers they are also bandpass devices in acoustic terms. How they respond within that bandwidth, its fundamental behaviour, must be linked to the current of the amplifier, and we hope in a faithful way. That is the source of the force-factor in every sense of the word.
Everything the driver does comes back to the current, hence much of its behaviour can be shaped by current. Indeed we can manipulate and control that current in selective ways, and then we directly influence how the driver behaves. It is classic, and it is physics. It may not be fully spelled out like that in textbooks, but it is still in total harmony with what they say.
_______________________________________________________________________
The two graphs...
How was it done?
I will tell you all. It is what I have repeated many times: If you EQ the current under Voltage Drive, so that a voltage source amplifier produces the same current att all frequencies from DC and upwards, then when you connect a current source, the same LF alignment gets locked in, and as you can see above examples. Even the Crossover gets locked in. Now you have a speaker system that has the same damping no matter what the source impedance of the amplifier is. Even the Crossover get the correct 'damping' as well. And why is that? Because the real source of sound comes from the force-factor and that you listening to. As long as the right amount of current turns up at the right frequency, then both FR and damping will stay put with any amplifier you connect your current EQ'd loudspeaker.
You have effectively cancelled out the output impedance of the amplifier.
There a misunderstanding that damping always needs a low output impedance from the source. It does not. It just needs the right current.
Once you understand that principle of current, and where it shows up in the right amount, knowing that it is really the current versus frequency that is the key, then you can find ways of also making Voltage Drive behave better.
_______________________________________________________________________
And that last point, that is where I am focusing, not on Current Drive.
I am focused on getting Voltage Drive to sound better.
Unlike Esa Merilainen and this "Current Driving Loudspeakers" book (which I do consider an important book), this is where he and I are different. He feels that Current Drive is the only way, and that Voltage Drive should be dumped entirely. That is not my position.
Current Drive is not a practical solution. The real challenge is getting Voltage Drive to sound better and get similar sonic benefits (which are real), but without the downsides of Current Drive.
We need to make Voltage Drive to sound more like Current Drive.
Most practical current source amplifiers needs to have a minimum ratio of 5:1 that both Esa and I agreed upon. 40 Ohm min for 8 Ohm and 20 Ohm min for 4 Ohm speakers.
We can match that result with Voltage Drive. We just need to know how.
Once thing I have a firm opinion about: That the vast number of speaker designs out there are LR4 Crossovers. These can be designed almost by following a textbook recipe' and end up with a technically correct loudspeaker. Good FR, decent polar responses, claimed low distortion, mainly the tweeter tends to be well protected with 4th order and gross tweeter distortion can be avoided. Just follow the textbook and you can design a loudspeaker. But they just don't sound great.
John Curl said: "I know how to design a loudspeaker, I just don't know how to design a 'great' loudspeaker."
Designers like Troels Gravesen have gradually gravitated away from those Crossovers and heard a better result using low order and even first order. John Atkinson of Stereophile has pointed out that low 1st order Crossovers have a more "accessible" sound - and he is right.
Why are people like that, and a small growing number, hearing something better.
I have an opinion, this about getting the current right. And certain Crossovers do that better than others, and in some cases, when targeted, matches Current Drive.
OK, that it...
But let us be decent about it. If you don't disagree, that's fine. Maybe you think LR4 speakers sound great? I find them bland and boring, but that seems to satisfy those who don't know otherwise, and I get that.
But those here, please have an open mind because at a later date you may change your mind.
Bottom line?
So to me the answer is about getting the current right, it is not about Voltage Drive versus Current Drive.
As I said, agree or disagree, but please please be civil - and I will try to be as well.
Cheers, Joe
It is interesting that any thread about current drive comes alive and then dies.
What happens?
The usual thing happens. A few will have their words of wisdom expressed, the usual links and websites are pointed to, and few hopefully friendly exchanges. Then nothing.
And I suspect it has happened again.
Let me show you guys something a little different, something a bit outside the box.
There a phrase used by Terry Demol on the old Blowtorch thread, where he said some were "textbook jockeys" and rather rigid if something is not said exactly as the textbooks do. The language is all wrong; so I have been told. I agree that language is important, but we should also be flexible.
With that in mind, take a look below, here is something curious:
What is happening here? Two FR graphs that look very similar and yet they are not exactly the same. Indeed multiple single measurements can yield these kind of differences, that is why it is a good thing to use many samples and average them out. But these are not samples of the same measurement, they are actually two entirely separate measurements.
Two plots, Red and Green.
One is Voltage Drive and the source impedance is 0.1 Ohm and since this is an 8 Ohm speaker, the nominal damping factor is supposed to be a healthy 80.
The other is Current Drive and the source impedance is 270 Ohm and as this is an 8 Ohm speaker, then nominal damping factor is... calculate... zero.
Which is which? Why do the look the same? How can that be?
Can you guess which is Current Drive and which is Voltage Drive?
I can't.
Looking closely at the FR (2Pi) we can see that this is around -3dB @ 40 Hertz and -12dB at 20 Hertz. What we see is a classic 2nd order Butterworth alignment tuned to 40 Hertz. Also I am using classic language.
What does this prove? It must prove something. This is what I said.
Note the critical phrase here is "varying current with frequency."
When I said that, the nameless KSTR starting to criticise pretty much anything said.
His answer in the form of two statements:
Mimics damping?
Good one! I don't know of anything alluding to that in any textbook.
But I completely understand him. He is not the only. But it is all a misunderstanding.
He has completely misunderstood something. He assumes that the frequency response has been EQ'd, and this is not so! He does not realise that it is current EQ. The frequency response then naturally responds to the EQ current. It becomes the result of the current being equalised.
I consider KSTR ghost with no name. My name is Joe Rasmussen, so what is his? He could at least PM me and introduce himself. If I offer criticism, then at least others know that I am not a ghost. I have a backbone and I should hope that others also so.
The alignment defines the damping of a speaker at low frequencies.
And what defines that alignment is the current that flows at:
Fl, Fc and Fh.
These three frequencies are not hard to calculate. If the right current turn up at the right frequency in both Red and Green examples above, then both have the same alignment and the damping is the same too. So simple.
KSTR's second statement, that Current Drive is equivalent to an open circuit. And he adds "by sheer definition."
But how that be? Open terminals of the speakers does not see a Current Source looking back "by sheer definition." Open terminals will only see air, and air cannot supply current. So sorry, the analogy does not suit the physics.
A current source is not air.
Air cannot produce current.
_______________________________________________________________________
Back to our two graphs. So I checked my records, Red is Voltage Drive and Green is Current Drive.
And guess what?
They sound the same!
I have demonstrated this to people coming through my front door. Play Joe Morello's "Take Five" drum solo, that is some test for any loudspeaker. Try organ music too. Anything you like.
I know what textbook jockeys think because I am well ahead of them. But they lack something, the ability to extend their thinking beyond the textbook. But a fine reading of those textbooks does allow us to say certain things. Loudspeakers are electrical motors and that is classic thinking. Hence there are already certain facts in place. Motors only respond to current. Our loudspeakers are both motors and transducers. Again I am speaking 100% classically. Being transducers they are also bandpass devices in acoustic terms. How they respond within that bandwidth, its fundamental behaviour, must be linked to the current of the amplifier, and we hope in a faithful way. That is the source of the force-factor in every sense of the word.
Everything the driver does comes back to the current, hence much of its behaviour can be shaped by current. Indeed we can manipulate and control that current in selective ways, and then we directly influence how the driver behaves. It is classic, and it is physics. It may not be fully spelled out like that in textbooks, but it is still in total harmony with what they say.
_______________________________________________________________________
The two graphs...
How was it done?
I will tell you all. It is what I have repeated many times: If you EQ the current under Voltage Drive, so that a voltage source amplifier produces the same current att all frequencies from DC and upwards, then when you connect a current source, the same LF alignment gets locked in, and as you can see above examples. Even the Crossover gets locked in. Now you have a speaker system that has the same damping no matter what the source impedance of the amplifier is. Even the Crossover get the correct 'damping' as well. And why is that? Because the real source of sound comes from the force-factor and that you listening to. As long as the right amount of current turns up at the right frequency, then both FR and damping will stay put with any amplifier you connect your current EQ'd loudspeaker.
You have effectively cancelled out the output impedance of the amplifier.
There a misunderstanding that damping always needs a low output impedance from the source. It does not. It just needs the right current.
Once you understand that principle of current, and where it shows up in the right amount, knowing that it is really the current versus frequency that is the key, then you can find ways of also making Voltage Drive behave better.
_______________________________________________________________________
And that last point, that is where I am focusing, not on Current Drive.
I am focused on getting Voltage Drive to sound better.
Unlike Esa Merilainen and this "Current Driving Loudspeakers" book (which I do consider an important book), this is where he and I are different. He feels that Current Drive is the only way, and that Voltage Drive should be dumped entirely. That is not my position.
Current Drive is not a practical solution. The real challenge is getting Voltage Drive to sound better and get similar sonic benefits (which are real), but without the downsides of Current Drive.
We need to make Voltage Drive to sound more like Current Drive.
Most practical current source amplifiers needs to have a minimum ratio of 5:1 that both Esa and I agreed upon. 40 Ohm min for 8 Ohm and 20 Ohm min for 4 Ohm speakers.
We can match that result with Voltage Drive. We just need to know how.
Once thing I have a firm opinion about: That the vast number of speaker designs out there are LR4 Crossovers. These can be designed almost by following a textbook recipe' and end up with a technically correct loudspeaker. Good FR, decent polar responses, claimed low distortion, mainly the tweeter tends to be well protected with 4th order and gross tweeter distortion can be avoided. Just follow the textbook and you can design a loudspeaker. But they just don't sound great.
John Curl said: "I know how to design a loudspeaker, I just don't know how to design a 'great' loudspeaker."
Designers like Troels Gravesen have gradually gravitated away from those Crossovers and heard a better result using low order and even first order. John Atkinson of Stereophile has pointed out that low 1st order Crossovers have a more "accessible" sound - and he is right.
Why are people like that, and a small growing number, hearing something better.
I have an opinion, this about getting the current right. And certain Crossovers do that better than others, and in some cases, when targeted, matches Current Drive.
OK, that it...
But let us be decent about it. If you don't disagree, that's fine. Maybe you think LR4 speakers sound great? I find them bland and boring, but that seems to satisfy those who don't know otherwise, and I get that.
But those here, please have an open mind because at a later date you may change your mind.
Bottom line?
So to me the answer is about getting the current right, it is not about Voltage Drive versus Current Drive.
As I said, agree or disagree, but please please be civil - and I will try to be as well.
Cheers, Joe
Just an added thought, I am very well aware that Current Drive can't fix everything, so please don't think I am making claims that I am not. But I do think there is a primary problem that makes the most difference, if the driver is well-designed. I have always said that the driver needs to be well designed and that there is no substitute. Indeed for what I have in mind (and what I do), the driver has to be very good to start with.
Yeah, when Zmax can reach 30-40-50 Ohm. But I don't think that it needed, I say, stay with Low Z amplifiers. I have a suggestion: We should view all amplifiers as "current delivery systems" and High Z is one and Low Z is another. And there are a variant(s) in-between.
I am convinced that Current Drive is not the way to go and that there are other techniques that does the job effectively with good drivers.
Yeah, when Zmax can reach 30-40-50 Ohm. But I don't think that it needed, I say, stay with Low Z amplifiers. I have a suggestion: We should view all amplifiers as "current delivery systems" and High Z is one and Low Z is another. And there are a variant(s) in-between.
I am convinced that Current Drive is not the way to go and that there are other techniques that does the job effectively with good drivers.
Re: 2 frequency response curves that look very similar. Here's a quick thought experiment. Consider a signal with 1% distortion. What is the ratio of signal to 'noise'? 100 to 1, or about 40dB. So far, so good. But what if we add 2 signals with a 100:1 ratio together on a logarithmic scale? That 1% only adds (or subtracts) a small fraction of a dB to the total output, rendering it practically invisible. Yet 1% is potentially quite a lot of distortion.
Similarly, you could add a resistor in series with a dynamic speaker and adjust the volume and Eq so that the response 'looks' the same as the raw response with no resistor. Then perform a listening test. They will not sound the same at all.
I am prepared to accept that voltage control may indeed be superior in the bass, as it helps ignore big swings in the driver impedance, and also because at low frequencies those box-related reactive elements are relatively linear. But at higher frequencies, voice coil related nonlinearity begins to dominate, adding many ohms in series, and therefore acting a bit like an unwanted voltage divider between the amplifier and the speaker. Except that it doesn't divide voltages linearly like resistors or taps on an inductor, but the ratios are dynamically changing. And that's the crux of the problem. A voltage amplifier by its nature can't control voltage in the middle of some elements in series. But a current amplifier can control the current in the middle of the same system.
Similarly, you could add a resistor in series with a dynamic speaker and adjust the volume and Eq so that the response 'looks' the same as the raw response with no resistor. Then perform a listening test. They will not sound the same at all.
I am prepared to accept that voltage control may indeed be superior in the bass, as it helps ignore big swings in the driver impedance, and also because at low frequencies those box-related reactive elements are relatively linear. But at higher frequencies, voice coil related nonlinearity begins to dominate, adding many ohms in series, and therefore acting a bit like an unwanted voltage divider between the amplifier and the speaker. Except that it doesn't divide voltages linearly like resistors or taps on an inductor, but the ratios are dynamically changing. And that's the crux of the problem. A voltage amplifier by its nature can't control voltage in the middle of some elements in series. But a current amplifier can control the current in the middle of the same system.
What you seem to be proposing is some form of feedback, or feed-forward or dead reckoning to de-facto convert one system into another. So you sample the current across a range of frequencies and manually adjust the voltage to get the same result. This may solve the gross impedances and damping factors.
But if voice coil Le at 1kHz changes dynamically +/-5% (just guessing, it may be less or a lot more) because of modulation by a 100Hz signal, a static change in the Eq of a voltage amplifier will not fix this. Signals around 1kHz will be directly modulated in proportion to the total impedance (say 6 ohms + inductance) On the other hand, if the amplifier that already has 100 ohms output impedance, that source of IMD is greatly reduced.
But if voice coil Le at 1kHz changes dynamically +/-5% (just guessing, it may be less or a lot more) because of modulation by a 100Hz signal, a static change in the Eq of a voltage amplifier will not fix this. Signals around 1kHz will be directly modulated in proportion to the total impedance (say 6 ohms + inductance) On the other hand, if the amplifier that already has 100 ohms output impedance, that source of IMD is greatly reduced.
I've been building and demonstrating speakers for a long time and I have noticed a trend in what some people like better in a loudspeaker. It came into focus while helping my friend, who only builds 1st order crossovers, design a crossover for a 4 driver system he was not happy with. His new speaker had a rough shouty response when he used his typical minimum part count first order crossover. This was because he had not leveled out the impedance of the different drivers before he started and some drivers had peaks in the response. So I added parts to level the impedance, notched a breakup mode and then implemented 2nd order crossovers. To verify the midrange crossover was working, we flipped the phase and looked to see if there were notches in the response at the crossover points. It worked and the notches were there. After measurements were done we played music and flipped the phase of the midrange, normal and reversed. Listening to just one speaker as you do when building, the sound was good and very localized with normal phase. Inverting the phase of the midrange, the sound became wider and took on a pleasant and entertaining sort of simulated stereo effect around the single speaker. Of course this was due to the cancelation of the direct response at the crossover frequencies and the emphasis of the reflected sound instead. In a similar way low order crossovers with the drivers overlapping produce wild variations in the response at different off axis angles and bring the room acoustics into play likely producing a wide image from a single speaker. I liken it to the simulated stereo button on old receivers for creating a pleasant stereo effect from a mono source material. I suspect this is what people are enjoying when listening to these speakers with low order crossovers. These people when listening to a very accurate, low distortion speaker with a uniform on and off axis response as one finds in high quality professional studio monitors, will find the sound to be some what boring as it is missing this wide effect sound.
Last edited by a moderator:
Yep. It's these sort of discussions that give me "aha!!" moments when something just clicks. With even crossover orders you're always shorting the speaker terminals with the last filter component, like a capacitor in the case of a low-pass. So, going from 1st to 2nd order, you ought to gain a steeper roll-off slope, and the unwanted out-of-band signals get quieter more quickly, but you also lose the benefits of the inductance in series and break-up modes may seem harsher.
You will use prefilter anyway to get flat response. And the voltage will become the same as with voltage drive.
@Joe Rasmussen ,
A screen name is just a handle, nothing more and nothing less.
Your screen handle may or may not be your real name (absolutely no way to prove it), and it's not important at all.
My screen name may or may not be an abbreviaton of my real name, and it's not important at all (but a bit of google-fu could reveal).
The only thing that is important -- to me -- is consistency (use same handle everywhere) and uniqueness (no sock-puppet accounts).
Judge people by their posted content, not by any superficialities like whether a screen handle sounds like a typical real name (or actually even is one) to you or not.
Assume my screen handle were John Bloggs (John Doe in the US, or Max Mustermann in Germany) would that make me any more credible? Probably not, as people may internationally be aware of such simple placeholder names. But if I chose a more subtle pseudonym, just some very common German name like Thomas Müller? Well, maybe that's my real name, after all? Who knows, and more importantly, who cares?
-------------
Now, as for actual content, you seem to always like to put words into people's mouths and re-interpret things, preferably out of context, so that it fits your agenda, for example:
Another one:
There is a number of ways to realize this but it is absolutely clear that pure current drive cannot give the same result. What you have done is a using current drive amplifier and modify (severly reduce) its output impedance with parallel cells so that the source impedance, as seen from the speaker, is not inifinite (or 270Ohms) anymore, so it clearly is not current drive anymore and therefore is not undamped anymore. Otherwise, the only way to get that result is to pre-EQ the current drive as is (not degenerated) so that the same voltage is generated at the driver terminals, but this time there is no electrical damping at all, regardless of whether the SPL transfer looks identical to voltage drive.
We've been at that point a few pages back, haven't we? The thread title is "Current Drive for Loudspeakers".
Typical example of a re-definition of things to your likings. Sorry mate, that's quite lame...
A screen name is just a handle, nothing more and nothing less.
Your screen handle may or may not be your real name (absolutely no way to prove it), and it's not important at all.
My screen name may or may not be an abbreviaton of my real name, and it's not important at all (but a bit of google-fu could reveal).
The only thing that is important -- to me -- is consistency (use same handle everywhere) and uniqueness (no sock-puppet accounts).
Judge people by their posted content, not by any superficialities like whether a screen handle sounds like a typical real name (or actually even is one) to you or not.
Assume my screen handle were John Bloggs (John Doe in the US, or Max Mustermann in Germany) would that make me any more credible? Probably not, as people may internationally be aware of such simple placeholder names. But if I chose a more subtle pseudonym, just some very common German name like Thomas Müller? Well, maybe that's my real name, after all? Who knows, and more importantly, who cares?
-------------
Now, as for actual content, you seem to always like to put words into people's mouths and re-interpret things, preferably out of context, so that it fits your agenda, for example:
What I said is, very obviously, that with regard to damping current drive is equivalent to open terminals, infinite impedance is infinite impedance, no matter how it is realized.
Another one:
It is very obvious from very first principles that the output can only be the same when the terminal voltage at the driver is also the same (or terminal current, doesn't matter). Fine with that, so far.With that in mind, take a look below, here is something curious:
What is happening here? Two FR graphs that look very similar and yet they are not exactly the same. Indeed multiple single measurements can yield these kind of differences, that is why it is a good thing to use many samples and average them out. But these are not samples of the same measurement, they are actually two entirely separate measurements.
Two plots, Red and Green.
One is Voltage Drive and the source impedance is 0.1 Ohm and since this is an 8 Ohm speaker, the nominal damping factor is supposed to be a healthy 80.
The other is Current Drive and the source impedance is 270 Ohm and as this is an 8 Ohm speaker, then nominal damping factor is... calculate... zero.
Which is which? Why do the look the same? How can that be?
Can you guess which is Current Drive and which is Voltage Drive?
There is a number of ways to realize this but it is absolutely clear that pure current drive cannot give the same result. What you have done is a using current drive amplifier and modify (severly reduce) its output impedance with parallel cells so that the source impedance, as seen from the speaker, is not inifinite (or 270Ohms) anymore, so it clearly is not current drive anymore and therefore is not undamped anymore. Otherwise, the only way to get that result is to pre-EQ the current drive as is (not degenerated) so that the same voltage is generated at the driver terminals, but this time there is no electrical damping at all, regardless of whether the SPL transfer looks identical to voltage drive.
We've been at that point a few pages back, haven't we? The thread title is "Current Drive for Loudspeakers".
Typical example of a re-definition of things to your likings. Sorry mate, that's quite lame...
I am not a ghost:
Bjørn Ingolf (aka Joe) Rasmussen
133 Dalmeny Drive
Prestons 2170, NSW
(Sydney Metro)
Australia
Mobile: +61-412-203382
Landline: +612-96074650
Website: www.customanalogue.com
Bushiness Name: "Custom Analogue Audio"
Australian Business Number: 60632798606
I am definitely not a ghost. All of the above have been public knowledge for well over twenty years.
If you want to be a ghost, then expect to be treated accordingly, or it is not fair criticism. I put greater store on what people say when I know their name. Even just a PM to introduce themselves. Man, even a first name would be nice, right?
Bjørn Ingolf (aka Joe) Rasmussen
133 Dalmeny Drive
Prestons 2170, NSW
(Sydney Metro)
Australia
Mobile: +61-412-203382
Landline: +612-96074650
Website: www.customanalogue.com
Bushiness Name: "Custom Analogue Audio"
Australian Business Number: 60632798606
I am definitely not a ghost. All of the above have been public knowledge for well over twenty years.
If you want to be a ghost, then expect to be treated accordingly, or it is not fair criticism. I put greater store on what people say when I know their name. Even just a PM to introduce themselves. Man, even a first name would be nice, right?
Interesting. Maybe not a bad trend.
Then the quality of driver must bear some of the blame. But correcting faults of drivers in a Crossover, and there is a danger of throwing the baby out with the bathwater. I would rather use less or no corrections, which can become over-corrections. Driver selection is paramount. And often it is not price, but good drivers do tend to be expensive.
Not so sure. If a sound does not sound wide and open on tracks that should sound that way, then something is not right. I believe that those who have tasted 1st order done well are in fact listening to less distorted current because not only can less harm done with a single inductor in line with the driver VC, that it can in fact improve the situation by lessening current distort. But of course, many will not understand what I mean by that. Not in the textbook. In that case I have been misunderstood, at least for now and with them.
But thank you for your comment. Food for thought.
Often it is no one such moment, but a series of them.
Yes, there is an intrinsic advantage of a single series inductor to a Midbass driver. But again, driver selection then becomes very important. When I look at using a driver, there are a number of things I look for. One is as low inductance as possible, but since the cone shape intrinsically becomes a horn above the knee, where the driver ceases to be a piston and now have to break up in a very controlled way. But low inducdance will mean a rising response becomes exxagerated. But it is a mixed blessing bewcause it will allow you to use a greater value of series inductance.
Guys, take a look below. The driver has to be just right to get away with it, but it need not be perfect above the knee, but it susally means a nice soft cone will do the job, such as fibre or poly sones. Not hard ones like aluminium or Magnesium, Ceramic, and so one.
I know one polycone driver that comes very close to the ideal and where you can do this:
Inductive Current Drive with a Voltage Source?
Maybe I am giving away secrets, but I don't care. Analyse the above. We have gotten rid of most of the driver's inductance, but this is not something you can do. I have said it many times. 1) There is no substitute for a quality driver and 2) this may be considered controversial, but Current Drive cannot reduce distortion in the driver. I know that will cause me strife, but note the 0R1 current sense resistor. This allows us to check for distortion on the current side of the amplifier.
I said the current side of the amplifier and not the speaker. The speaker does not generate current. The current sense resistor tells us what the amplifier is doing on the current side.
Note the inductance ratio is 5:1
That is the same minimum that Esa and I agreed would be the minimum ratio of source impedance to nominal impedance of the speaker.
Inductive Current Drive?
Yes, you will have all the same advantages of 5:1 ratio current drive, and yet be using a voltage source amplifier.
The above inductor is likely to be -6dB or in some cases a bit lower, at 3KHz. Now carefully build a high-pass filter to the tweeter. Yes, it is a challenge and once again, driver selection is crucial.
Yes, it could be done that way, sort of.
But that means that you would end up with a speaker that is not fully contained as it would require a another "box" in the line level of the system. The solution should be entirely in the speaker box, after the power amplifier.
We want a self-contained speaker that we can connect to any voltage or current source - and get the right current where we need it. So we are looking at passive parallel networks, sometimes also called conjugate filters. These we can use to control the current going to the VC. Remember I said the key is to make the amplifier supply the same current at all frequencies.
No, that clearly makes you less credible. It could lead to questions like: What have you got to hide?
I tried Google...
But if you want to remain a ghost... then I will acct accordingly.
parallel notch in series to rescue, or DSP for the extra orders past 1st
That's been done before. But that is not how these two graphs looks similar. Once understood, then you can't unlearn it.
Here is the Crossover (no values, but you can make rough guesses)
Please note that with this Crossover, the current drawn by the amplifier will be the same at all frequencies.
This can be confirmed by the impedance plot:
That's right.
When the amplifier current is the same at all frequencies, it effectivley cancels out the output impedance of the amplifier.
Now a 2nd order Butterworth alignment, -3dB @ 40Hz and -12dB at 20 Hertz always stays that way.
It proves that controlling the current is what provides the correct damping. Then the series/output impedance has no influence on the damping of the speaker.
The coirrect current will show up at the three critical frequencies of Fl, Fc and Fh. If you know how to calculate these frequencies based on equations based on Thiele-Small Parameters, the you have the correct damping regardsless of the amplifier.
It does not mimic a 2nd order Butterworth alignment, it is a 2nd order Butterworth alignment.
Use dominant single series element to filter' the drivers. Ideally a single inductor on the MidBass and ideally a single capacitor on the tweeter. Yes, it is a challenge, but can largely be done. Study the basic Crossover above. The conjugates does only one thing, EQ the current of the amplifier. What this means is that those conjugates always draw the same current, then a by-product of that is that it also means that the correct current turns up in the VC of the driver.
Whatever deviation you use in the Crossover or corrections, be careful not to throw out the baby with the bathwater. Note the null filter on the tweeter, this will typically tuned to the Fs of a well-chosen Tweeter and around 500-700 Hertz. Also note that the Tweeter's phase in inverted. Why is another topic again.
The "open terminals" has gone down in folklore, it's part of the damping factor myth.
Perhaps reading what I have just posted above, properly understood, will make that clear. And there is more to it than presented here.
Open the floor... any comments? But please keep them civil. I promise that I am trying likewise.
Joe
Keeping the speakers independent and compatible with a range of amplifiers is one design philosophy. Another would be to fine-tune each amplifier channel to the speaker that it's paired with. Too bad there's no "active speaker" forum category, but it's probably not needed anyway.
To flatten the bass impedance, quite large component values would be needed, so there's no free lunch as they say. Active filters may be cheap in parts, but take a long time as well.
To flatten the bass impedance, quite large component values would be needed, so there's no free lunch as they say. Active filters may be cheap in parts, but take a long time as well.
Active filters are now very inexpensive. For example a miniDSP 2x4HD gives you 60 biquad filters for $200 if you get the PCB version, 10 for each of two input channels and 10 for each of four output channels. If you are using a PC as a source, there is freeware like equalizerAPO that will give you all the filters you would want. Amplifier channels are also very inexpensive. So I see no reason to have any passive crossover components in a system anymore. I guess if you really want to you can build a passive series crossover and use it successfully with a transconductance amplifier. I prefer to just use a separate amp channel for each driver and then use a Linkwitz transform in the active crossover to produce what ever low frequency response I desire. No need for anything else. As I mentioned before, low order crossovers simply produce wide off axis response variations and I think that is the dominant source of the difference in sound that people seem to like. Its all about drivers playing the same range of frequencies and adding or subtracting a at different angles. That is very audible. The tiny changes in distortion are not going to be audible. Since Joe doesn't believe that transconductance amplifiers are a good idea for audio, maybe he could stop commenting on the only thread dedicated to the topic.
I guess I'm too much of a romantic to use off-the-shelf IIR style filters in the "ultimate system". Maybe for prototyping, but building analogue circuits is where it's at for me.
I'm on a quest for some simple-yet-good alternatives to op-amps as buffer stages. Op-amps are nice and reliable, at least if you use something like the NE5532x (avoiding many potential pitfalls with faster chips). MOSFET compound pairs (a la Sziklai) seem to reduce distortion compared to single FETs, but push-pull might be another good option, or using depletion fets as simple CCSs. JFET cascodes seem to have very weak drive capability for the highly capacitive loads that are typical for some filters, but I'm not ruling them out. Even if the triple-0 distortion of op-amps is unbeatable with just 2-3 transistors, I may be able to beat the noise performance.
The filters can be directly built into a special-purpose power amp, whose output impedance is shaped as discussed (voltage control at low frequency, current control at high frequency).
I'm on a quest for some simple-yet-good alternatives to op-amps as buffer stages. Op-amps are nice and reliable, at least if you use something like the NE5532x (avoiding many potential pitfalls with faster chips). MOSFET compound pairs (a la Sziklai) seem to reduce distortion compared to single FETs, but push-pull might be another good option, or using depletion fets as simple CCSs. JFET cascodes seem to have very weak drive capability for the highly capacitive loads that are typical for some filters, but I'm not ruling them out. Even if the triple-0 distortion of op-amps is unbeatable with just 2-3 transistors, I may be able to beat the noise performance.
The filters can be directly built into a special-purpose power amp, whose output impedance is shaped as discussed (voltage control at low frequency, current control at high frequency).