Yes. Impedance curve flatness does not correlate with effectiveness of current drive. AMT tweeters have flat impedance and yet are a type of driver were often huge improvements (20dB) can be found, IME.
No. It only mimics damping (in that it can yield a frequency response as with actual electrical damping by means of EQ). But the cone is still not damped in in its movement.
Current drive is equivalent to open terminals (by sheer definition), so where would the damping come from?
Electrical damping is a control loop mechanism, an inner feedback in the driver itself when terminated with some non-infinite impedance. With current drive you can only steer the current to counteract the expected movement, but you have no actual control loop to counteract any unexpected movement. Which means you cannot force the driver to make only the movement you've requested, the driver is completely free to move arbitrarily on top of that steering.
A simple test (besides the simple knuckle test) is to drive the speaker into over-excursion for a brief period -- something that will always happen in practice. No matter what alignment (and thus apparent damping, not actual damping) you have dialed in. the driver will recover and go back to its rest position following its undamped natural response. If the undamped Q is, say, 5, it will ring with that Q during the settling time. And that's what you will hear... the dreaded "one note bass" as the driver will react to any "unplanned" excitation, the tiniest amount of nonlinearity completely undamped. With a low Qms driver, resulting in a box Qtc of below 2 or so, you might get away with no damping but the bass sound will never be as tight as from a well-damped speaker with the same frequency response. Some people like a somewhat "spongy/springy" bass sound but I don't.
This is the reason why I tend to design the speaker-amplifier interface so that critical damping is established as the natural response of the combo (around resonance), and from there I apply EQ to dial in the target frequency response (aka alignment). Critical damping gives the fastest and cleanest error recovery without ringing.
For ported speakers, the requirement is a bit different, here the goal is to have a reasonable Q for the Helmholtz resonator.
Different natural damping response can also have a slight effect on room mode excitation. Brickwall damping (cone movement not affected in any way by different acoustic load) will excite room modes fully. A fully undamped cone will have some give with resonant air load and thus will excite some frequencies a bit less but the higher inherent ringing will often nullify any benefit from that. Best case is when the natural ringing frequency coincides with a room mode frequency.
Definitely my last post addressed to you, Joe. It's futile.
Last edited:
Then you think you know, but it may come as a shock, you have got this wrong. Yes, I know that in your mind what you are saying is 100% clear. I get that. But you have not understood it as well as you think.
Again, I repeat, I am not in the current drive camp. But I do have such an amplifier and I can demonstrate it. Not theory but demonstrable fact. I can take you through it step-by-step and watch the amazement on your face.
Your position is a common one. The 'open circuit' is not a proper definition. I understand very well the point and indeed going back 15-20 years ago, I would have said the same thing.
If you could walk through my front door, I can actually demonstrate it for you. Or if you know somebody here in Sydney who can act as your surrogate.
Or indeed anybody here?
You are all welcome.... come and see it, and then come back to this thread and report it. I won't bribe you, I promise. Besides I can't afford it.
I am game!
[BTW, I designed a Critical Sub and published it, Peerless XLS 12". Was capable of clean 20 Hertz picked up by a nearfield mic.]
Last edited:
Hi Guys
Just to sum up, this thread was dead for several years. At least I got it back to life.
But the way things are shaping up, it going to die again.
The thread is about current drive, but the issue of current as a subject in loudspeakers, it is an incredibly difficult subject.
Ironically, I am happy that KSTR turned up, because (without knowing it) brought down to a single issue:
When the impedance of the amplifier goes high, all electrical damping is lost and cannot be recovered!
NOT TRUE!
Can I be any more emphatic than that?
Of course, KSTR will go away, he thinks the impossible (to him) is futile. But at least he highlighted the 'problem' and not wanting the 'solution.'
This gets to the heart of the problem about current in loudspeakers. It is the damnedest subject in all audio. It makes grown people almost go crazy.
I keep coming back to Menno Vanderveen's words: "Thinking current is hard." It is like a foreign language.
Because electrical damping can be recovered despite any high source impedance. For some this will seem totally impossible. It will sound like snake oil. Except that is not snake oil. In fact it is an incredibly useful thing to know and understand how it works.
With a current source amplifier, I can design any sealed box alignment. I am not lying. You want a Bessel alignment with a Qc of 0.56, I can do it. To KSTR, since you like a critically damped alignment of 0.5, I can do that too. But most I think would be happy to get a classic 2nd order Butterworth with a Qc of 0.707 as that would considered fairly normal. It is not theory, I can do this. But here is a clue, it only works when the source impedance is high, it does not work with a low impedance voltage source.
If you think it can't be done?
Get over it (please), send somebody here for me to show it and be a witness.
Just to sum up, this thread was dead for several years. At least I got it back to life.
But the way things are shaping up, it going to die again.
The thread is about current drive, but the issue of current as a subject in loudspeakers, it is an incredibly difficult subject.
Ironically, I am happy that KSTR turned up, because (without knowing it) brought down to a single issue:
When the impedance of the amplifier goes high, all electrical damping is lost and cannot be recovered!
NOT TRUE!
Can I be any more emphatic than that?
Of course, KSTR will go away, he thinks the impossible (to him) is futile. But at least he highlighted the 'problem' and not wanting the 'solution.'
This gets to the heart of the problem about current in loudspeakers. It is the damnedest subject in all audio. It makes grown people almost go crazy.
I keep coming back to Menno Vanderveen's words: "Thinking current is hard." It is like a foreign language.
Because electrical damping can be recovered despite any high source impedance. For some this will seem totally impossible. It will sound like snake oil. Except that is not snake oil. In fact it is an incredibly useful thing to know and understand how it works.
With a current source amplifier, I can design any sealed box alignment. I am not lying. You want a Bessel alignment with a Qc of 0.56, I can do it. To KSTR, since you like a critically damped alignment of 0.5, I can do that too. But most I think would be happy to get a classic 2nd order Butterworth with a Qc of 0.707 as that would considered fairly normal. It is not theory, I can do this. But here is a clue, it only works when the source impedance is high, it does not work with a low impedance voltage source.
If you think it can't be done?
Get over it (please), send somebody here for me to show it and be a witness.
@Rallyfinnen , in case you are addressing me in above post...
no, I'm talking about an intrinsic feedback mechanism in the driver which affords the electrical damping. This is somewhat similar to linearizing local degenerative feedback in a simple emitter follower output stage of an amplfiier.
The driver's terminal voltage is comprised of the sum of two parts:
Part one is the normal voltage drop along the static VC impedance (which just Re, simplified) when some current is flowing, so we have have Ve = I * Re.
The second part is the microphonic voltage (aka "back-EMF" but I hate the term) which is proportional to cone velocity, Vm = k*BL*dx/dt (x is cone displacement). It does not matter why the cone is moving in the first place.
So we have Vtotal = Vm + Ve.
Now assume we have standard voltage drive (Ramp = 0) and the output voltage of the amp is zero but the cone is moving (for whatever reason).
This means Vtotal = 0 and thus 0 = Vm + Ve or Vm = -Ve = -I*Re.
Solving for I gives I = -Vm/Re and what this means is that an opposing current is injected, counteracting the movement of the cone, damping it. This in turn directly affects Vm itself, reducing it, and this is where the loop is closing.
Re is the effective transfer impedance in this case -- creating the counteracting current from the microphonic voltage -- for damping in voltage drive, as Ramp = 0.
With current drive, we have Ramp approaching infinity and the effective transfer impedance now becomes Re + Ramp = infinity, too.
Therefore, corrective current I = -Vm/(Re + Ramp) is zero. No damping at all, the cone can move freely, Vm simply goes into the voids, literally. Cone motion is excited by generating a force, F = BL * I, this is mentioned force steered mode. We inject a force and the forget about it.
Let's now look at the other extreme, negative impedance drive with Ramp approaching -Re. Let's assume we have reduced the effective transfer impedance to one milliohm. This means when the microphonic voltage of the driver is off by one millivolt from the applied voltage (as seen from before the negative resistance) this already produces one Ampere of corrective current. This is really quite a lot of local feedback, obviously, compared to standard voltage drive. This the mentioned velocity controlled mode, when the velocity doesn't match a large corrective current is produced to make it match, the control loop's action. Around resonance only little current is needed to keep the cone moving as desired, in other words the ratio of Vm/Ve is is very high and thus we have the largest amount of feedback there.
Standard voltage drive is somewhere in between these extremes.
Hope this helps.
no, I'm talking about an intrinsic feedback mechanism in the driver which affords the electrical damping. This is somewhat similar to linearizing local degenerative feedback in a simple emitter follower output stage of an amplfiier.
The driver's terminal voltage is comprised of the sum of two parts:
Part one is the normal voltage drop along the static VC impedance (which just Re, simplified) when some current is flowing, so we have have Ve = I * Re.
The second part is the microphonic voltage (aka "back-EMF" but I hate the term) which is proportional to cone velocity, Vm = k*BL*dx/dt (x is cone displacement). It does not matter why the cone is moving in the first place.
So we have Vtotal = Vm + Ve.
Now assume we have standard voltage drive (Ramp = 0) and the output voltage of the amp is zero but the cone is moving (for whatever reason).
This means Vtotal = 0 and thus 0 = Vm + Ve or Vm = -Ve = -I*Re.
Solving for I gives I = -Vm/Re and what this means is that an opposing current is injected, counteracting the movement of the cone, damping it. This in turn directly affects Vm itself, reducing it, and this is where the loop is closing.
Re is the effective transfer impedance in this case -- creating the counteracting current from the microphonic voltage -- for damping in voltage drive, as Ramp = 0.
With current drive, we have Ramp approaching infinity and the effective transfer impedance now becomes Re + Ramp = infinity, too.
Therefore, corrective current I = -Vm/(Re + Ramp) is zero. No damping at all, the cone can move freely, Vm simply goes into the voids, literally. Cone motion is excited by generating a force, F = BL * I, this is mentioned force steered mode. We inject a force and the forget about it.
Let's now look at the other extreme, negative impedance drive with Ramp approaching -Re. Let's assume we have reduced the effective transfer impedance to one milliohm. This means when the microphonic voltage of the driver is off by one millivolt from the applied voltage (as seen from before the negative resistance) this already produces one Ampere of corrective current. This is really quite a lot of local feedback, obviously, compared to standard voltage drive. This the mentioned velocity controlled mode, when the velocity doesn't match a large corrective current is produced to make it match, the control loop's action. Around resonance only little current is needed to keep the cone moving as desired, in other words the ratio of Vm/Ve is is very high and thus we have the largest amount of feedback there.
Standard voltage drive is somewhere in between these extremes.
Hope this helps.
@KTSR Thank you for the long post, I think I have grasped the basic concept, even if I try to think of it in simpler terms of 'lookback impedance' seen from the driver. My guess would also be that in case of 'negative amplifier output impedance', the distortion would also increase from the driver in the same manner it would decrease from a higher amp impedance?
I'm staying home with a flu (or covid, who knows) so my brain is really slow at the moment.
Actually my 'motional feedback' was a response to Joe, based on his above claim being able to alter Q to his liking while using current drive (as I understood it). That was the only option I could think of with a very high amp output impedance, and I find it an interesting concept, even if it gets complicated, most likely giving little 'bang for the buck' in the end. It should minimize distortion (both the current drive and feedback), and give good control of the cone movement though.
I should have replied to his message to make it clear what I was commenting, sorry.
I'm staying home with a flu (or covid, who knows) so my brain is really slow at the moment.
Actually my 'motional feedback' was a response to Joe, based on his above claim being able to alter Q to his liking while using current drive (as I understood it). That was the only option I could think of with a very high amp output impedance, and I find it an interesting concept, even if it gets complicated, most likely giving little 'bang for the buck' in the end. It should minimize distortion (both the current drive and feedback), and give good control of the cone movement though.
I should have replied to his message to make it clear what I was commenting, sorry.
KSTR, this seems nice in theory, but does it sound good? This litmus test should help determine overall correctness.
I don't have a system with negative output resistance to verify, but my guess is that it would tend to sound bad, or at least worse than something very similar but without a zero or higher output resistance. Reasoning: it is better to let the cone ring freely in response to spurious microphonic signals (such as the bass resonance, which can be EQ'd anyway), rather than attempt to oppose those forces.
For one thing, we already know that the conversion process between acoustic and electrical is slightly non-linear, so the amplifier by necessity would be forced to inject harmonics and inter-modulation products to counteract what it "thinks" is the error signal. 'Hardening' the voice coil against unwanted vibrations seems like a particularly bad idea in the cone break-up region, as the voice coil would act like an immovable surface that strongly reflects transverse waves on the surface of the cone. As an example, the amplifier generates an impulse:
0-0-0-1-0-0-0-...
which then travels across the surface of the cone. Some of it is progressively absorbed by the air and the rubber surround (which is also a source of non-linearity because of hysteresis), and a residual echo returns to the voice coil and applies a force to it. With this in mind, the electrical impedance should probably be tuned to either ignore it altogether, or maximise the rate at which the energy is absorbed.
I don't have a system with negative output resistance to verify, but my guess is that it would tend to sound bad, or at least worse than something very similar but without a zero or higher output resistance. Reasoning: it is better to let the cone ring freely in response to spurious microphonic signals (such as the bass resonance, which can be EQ'd anyway), rather than attempt to oppose those forces.
For one thing, we already know that the conversion process between acoustic and electrical is slightly non-linear, so the amplifier by necessity would be forced to inject harmonics and inter-modulation products to counteract what it "thinks" is the error signal. 'Hardening' the voice coil against unwanted vibrations seems like a particularly bad idea in the cone break-up region, as the voice coil would act like an immovable surface that strongly reflects transverse waves on the surface of the cone. As an example, the amplifier generates an impulse:
0-0-0-1-0-0-0-...
which then travels across the surface of the cone. Some of it is progressively absorbed by the air and the rubber surround (which is also a source of non-linearity because of hysteresis), and a residual echo returns to the voice coil and applies a force to it. With this in mind, the electrical impedance should probably be tuned to either ignore it altogether, or maximise the rate at which the energy is absorbed.
The above sentence drives my crazy 😮 If you drive speaker into over-excursion it’s disaster already as it is not intended to use like this in any case. At least if quality is important.
Yes agree that critical damping is best. Yes no damping for external forces with current drive. Yes agree that voltage drive is better than current at resonance. But mainly due to THD reduction. Ringing issue with current drive i don’t see so important. You have left with some mechanical damping if some errors. I work mostly with open baffles so maybe box adds some problems, not sure. But theoreticaly don’t see big problems here. I will not do current drive at resonance next time, but i have open baffles with current drive 20hz-20khz and don’t hear any problems. I have done box speaker with voltage at resonance and current upwards and that bass is muddy compared to open baffle. But it is box vs baffle thing. Another topic
I was thinking of using dual feedback: an integrator for "servo" NFB to stabilize DC, and connecting the capacitor (from the integrator RC) to the sense resistor. So you say it turned out muddy?
The other possibility I was thinking about was:
1) varying the gain with an in-line filter. This could either be a notch in a gain stage, or a band-boost in the NFB. Either way, the net effect on gain should be the same, but there may be some differences in the noise and distortion.
The other possibility I was thinking about was:
1) varying the gain with an in-line filter. This could either be a notch in a gain stage, or a band-boost in the NFB. Either way, the net effect on gain should be the same, but there may be some differences in the noise and distortion.
In my case, I think it must be abundantly clear that I am talking voltage sources (near zero Ohm) or current source (very high Ohm) and no motional feedback and no negative impedance trick. Just plain vanilla voltage sources and current sources.
A voltage source amplifier is what you can walk into a store and buy.
A current source amplifier is something that you will never be able to buy in a store.
The first part of that sentence, about open terminals, I keep coming back to this because KSTR openly declares that he DOES know about the physics and he knows it all.
Yet what he says is highly inaccurate.
A current source applied to the terminals is not open terminals at all. The open terminals implies that there is air between them. Well, air cannot supply current. So how can it be the same.
What is a current source? Simple:
A SOURCE OF CURRENT.
It's not air and the terminals are not open, they are connected to something that supplies current.
This is the kind of flawed logic I see every time when the behaviour of current is discussed. It is also why so many make little progress in understanding it. Thinking in terms of current is different than thinking in terms of voltage. But many who start talking about current behaviour, soon without knowing they go back to think voltage because that is their comfort zone.
Now KSTR will take this as some kind of 'attack' and it is not. Only if we make this an argument where egos come in. That should not be the case, or else that just becomes a more persistent obstacle.
And it is not as if I haven't explained it. Measure the current, not the voltage, at the height of the impedance (Zmax) and then at a frequency below that and also at a frequency above that where the current is the same. Because a current source supplies a fixed amount of current, now set it up to match the current levels that you saw when using a voltage source. Match the current at those three frequencies. It sound more difficult than it really is. But you must have something that can measure impedance to do that. I use Clio.
This is an instance where social media is problematic. I offered to Allen to call me and offered to explain it. Or if a practical demonstration, just walk through the front door.
But if some put their heads in the sand, that is their decision. Once you poor concrete it is hard to undo, and people get set in their ways and would rather not know. Fragile egos? We should all be aware.
Not from the amplifier. This must surely be the biggest myth in all audio. It is what feeds the damping factor myth (and I am not the only one calling it that, please read on).
Here is a tidbit from the Purifi blog, strangely in paratheses:
"(Wait, isn’t the amplifier supposed to stop cone movement? Eventually, yes, but not nearly as fast as the damping factor myth would have it. We’ll devote another blog post to that.)"
https://purifi-audio.com/2019/12/10/a-fast-driver-needs-a-light-cone-or-does-it
I sure hope he writes that blog, it should be very interesting. I believe we can ascribe the above words to Bruno Putzeys.
Note what he said?
"...the damping factor myth..."
There! This time he said it, I have been using that phrase for decades.
So where does the damping come from? Ironically from the driver itself and the current that feeds it. So you need to check what the current does in the voice coil. It does not necessarily rely on the source impedance. If you do nothing, then increasing the impedance will make the damping worse, no argument about that. Just that is does not provide the complete picture.
This will be reflected in the alignment of the speaker. But it can be corrected by the designer by using current equalisation, current EQ.
There, I have said it.
Do I need help in this area, or are you looking for someone to help explain it?
Social media has limits. I am trying to explain something that many (maybe even most) think is snake oil. Like KSTR put his finger on it, that a current source would take away all electrical damping because of the high impedance. Then what I am saying cannot be true. The electrical damping is lost forever, never to be recovered. It is nonsense to think it can be recovered.
Social media, no. Doesn't work. A two-way conversation is different. That works because I have explained to others on the phone or face-to-face, and they got it.
I will tell you what I need here: A witness.
Then you will appear here on this thread and say that you understand it too, and that it is not snake oil once you understood it.
Also, I think you will enjoy the explanation, because it leads to an even greater understanding, such as Bruno declaring that the idea of damping factor is myth. Why? I do hope he writes a blog about it. Maybe we should present a petition to Purifi. I have Claus Neesgaard's email address, we could send it to him: "Please tell Bruno to write a new blog on damping factor and why he calls it a myth."
My phone number is publicly known everywhere, so I might as well put it here: 0412-203382 (+61-412-203382, as I take international calls all the time)
Social media, no. Doesn't work. A two-way conversation is different. That works because I have explained to others on the phone or face-to-face, and they got it.
I will tell you what I need here: A witness.
Then you will appear here on this thread and say that you understand it too, and that it is not snake oil once you understood it.
Also, I think you will enjoy the explanation, because it leads to an even greater understanding, such as Bruno declaring that the idea of damping factor is myth. Why? I do hope he writes a blog about it. Maybe we should present a petition to Purifi. I have Claus Neesgaard's email address, we could send it to him: "Please tell Bruno to write a new blog on damping factor and why he calls it a myth."
My phone number is publicly known everywhere, so I might as well put it here: 0412-203382 (+61-412-203382, as I take international calls all the time)
I've already demonstrated that you can manipulate damping both up and down, in low and high impedance environments. The theory is in my textbooks.
What is it that you want?