I am curious as to from where the notion comes that we have to control the cones displacement. Acoustic output above resonance is directly proportional to cone acceleration, and since the driver is mass controlled above resonace, then acoustic output is also directly proportional to drive force.
However, it should be clarified that displacement, velocity and acceleration are all linearly related, and that they differ only in proportionality to frequency. If you linearize one with respect to proportionality to input signal, you have linearised them all.
Andrew
However, it should be clarified that displacement, velocity and acceleration are all linearly related, and that they differ only in proportionality to frequency. If you linearize one with respect to proportionality to input signal, you have linearised them all.
Andrew
Bill F. said:
I have a system exactly like this:
18" ELF woofer 10-35Hz
2*18" dipole (Qm=1.98) 35-250Hz
Goodmans Axiom 80 in a 200Hz straight spherical front horn
RatShack supertweeter above 16k
four way EV digital crossover
transformer output 20 bit no oversampling multibit dacs
separate current drive chip amp for each speaker.
Works extremely well in my small room (16*12*7) where I always had problems with standing waves.
I've tried to replace the chip amp for the Axiom with a good 300B amp, but in this setup the tube distortion was simply too evident.
I am beginning to realize the flaw in my argument.
a) In an amp in which the feedback loop tries to keep the current at some value that is determined by the input signal, if you were to push the cone, you would induce a current that the feedback loop would try to compensate by injecting an extra current of opposite polarity. I guess this picture is still right?
b) At resonance, the impedance of the speaker increases. With voltage drive and a Q=0.71 alignment, frequency response remains flat down to resonance, but this is because much less current flows at resonance because of the higher impedance. In other words, power sensitivity is greater around resonance and frequency response is kept flat by drawing less power around resonance.
In current drive, the current and hence power drawn are flat over frequency, but this means the frequency response goes up around resonance. Makes perfect sense, doesn't it?
In both cases, the magnet/VC acts as a transformer that converts mechanical movement into current. So what is the difference between a) (force acting on cone, inducing current that is countered by the feedback loop and causes a counterforce) and b) (mechanical forces of the mass/spring oscillator coupling to the electric side and hence raising impendance around resonance)?
a) In an amp in which the feedback loop tries to keep the current at some value that is determined by the input signal, if you were to push the cone, you would induce a current that the feedback loop would try to compensate by injecting an extra current of opposite polarity. I guess this picture is still right?
b) At resonance, the impedance of the speaker increases. With voltage drive and a Q=0.71 alignment, frequency response remains flat down to resonance, but this is because much less current flows at resonance because of the higher impedance. In other words, power sensitivity is greater around resonance and frequency response is kept flat by drawing less power around resonance.
In current drive, the current and hence power drawn are flat over frequency, but this means the frequency response goes up around resonance. Makes perfect sense, doesn't it?
In both cases, the magnet/VC acts as a transformer that converts mechanical movement into current. So what is the difference between a) (force acting on cone, inducing current that is countered by the feedback loop and causes a counterforce) and b) (mechanical forces of the mass/spring oscillator coupling to the electric side and hence raising impendance around resonance)?
Hi,
It never was right. If a current source supplies 0.0000A into a given load you can do with that load anything you wish, the current supplied by the (ideal) current source WILL NOT CHANGE, Except of being adkusted to a different value by the input signal.
Yup. Basiocally, the mecanical system RINGS LIKE A BELL.
Assuming no measures are taken to naturally damp the resonance of course....
The German Eckmiller Coax from the late 1930's used a combination of an "apperiodic" (fully covered with holes) basket and a grease damping elemet to make a driver with an inherently critically damped resonance, a feat of engineering sadly unrepeated since....
The basic problem with your view is that speakers are usually operating (excep the bass in extremely badly designed systems) otside the resonant range. Hence the whole view of what happens at the low frequency resonance (which is usually more or less the limit of response of the system) is irelevant EXCEP at the resonance and most of the music happen way above the LF cone resonance. Hence concentrating on the resonant operation mode of the driver misses 99.9% of the music.
Moreover, making a more or less "non resonant" mechanical system is not terribly difficult. It is just that in speakers it is rarely (if ever) done.
I have seen moving coil meters (the moving coil drives a needle indicating with good precision a given value) using a strip of aluminum in the same magnet gap to "damp" the inherent mechanical resonance of the Coil/Needle arrangement. It is simple engineering. Nothing that cannot be applied to a speaker driver.
Sayonara
It never was right. If a current source supplies 0.0000A into a given load you can do with that load anything you wish, the current supplied by the (ideal) current source WILL NOT CHANGE, Except of being adkusted to a different value by the input signal.
Yup. Basiocally, the mecanical system RINGS LIKE A BELL.
Assuming no measures are taken to naturally damp the resonance of course....
The German Eckmiller Coax from the late 1930's used a combination of an "apperiodic" (fully covered with holes) basket and a grease damping elemet to make a driver with an inherently critically damped resonance, a feat of engineering sadly unrepeated since....
The basic problem with your view is that speakers are usually operating (excep the bass in extremely badly designed systems) otside the resonant range. Hence the whole view of what happens at the low frequency resonance (which is usually more or less the limit of response of the system) is irelevant EXCEP at the resonance and most of the music happen way above the LF cone resonance. Hence concentrating on the resonant operation mode of the driver misses 99.9% of the music.
Moreover, making a more or less "non resonant" mechanical system is not terribly difficult. It is just that in speakers it is rarely (if ever) done.
I have seen moving coil meters (the moving coil drives a needle indicating with good precision a given value) using a strip of aluminum in the same magnet gap to "damp" the inherent mechanical resonance of the Coil/Needle arrangement. It is simple engineering. Nothing that cannot be applied to a speaker driver.
Sayonara
Meter damping.
The coil former on which meter movements have their coil wound are often aluminium and form a shorted turn in the same area as the coil. You mightn't have realised what a good job these do until you look at an undamped meter that may take 30 seconds or so to come to rest.
That the aluminium coil former is a shorted turn would also affect the flux risetime presumably, and if this was principle applied overzealously to a loudspeaker I think the efficecieny would be very low with the voicecoil running even hotter than usual. All the speakers that I have seen that do have an aluminium voice coil former have it made out of a strip with the ends insulated from one another to avoid a shorted turn. Maybe if they were joined with a little bit of resistance?
The coil former on which meter movements have their coil wound are often aluminium and form a shorted turn in the same area as the coil. You mightn't have realised what a good job these do until you look at an undamped meter that may take 30 seconds or so to come to rest.
That the aluminium coil former is a shorted turn would also affect the flux risetime presumably, and if this was principle applied overzealously to a loudspeaker I think the efficecieny would be very low with the voicecoil running even hotter than usual. All the speakers that I have seen that do have an aluminium voice coil former have it made out of a strip with the ends insulated from one another to avoid a shorted turn. Maybe if they were joined with a little bit of resistance?
Current drive
Hi
For al the guys that have been reading this thread and are wanting make a idiea of the sound of a current drive speaker and a voltage drive here is a litlle experience that anybody can do:
Add a 8 Ohms resistor in series with one of yours speakers leads and listen!
The damping factor will be one...and your amp have gone a step in the direction of a current amp...
Listen an see if the sound also have gone a step in the right directio(better sound)...if yes,maybe current drive could be more investigated..if the step is for worst ,well forget and have a good sleep.
I prefer the practical side of the things!
Arriverderci!
Jorge Santos
Pslease repport yours experiences and opinions!
Hi
For al the guys that have been reading this thread and are wanting make a idiea of the sound of a current drive speaker and a voltage drive here is a litlle experience that anybody can do:
Add a 8 Ohms resistor in series with one of yours speakers leads and listen!
The damping factor will be one...and your amp have gone a step in the direction of a current amp...
Listen an see if the sound also have gone a step in the right directio(better sound)...if yes,maybe current drive could be more investigated..if the step is for worst ,well forget and have a good sleep.
I prefer the practical side of the things!
Arriverderci!
Jorge Santos
Ps
lease repport yours experiences and opinions!
While, from a first glance, it looks as if it doesn't matter whether damping is done mechanically or electrically - things might be much more complicated in practice.
There are sources claiming that drivers with low mechanical losses (i.e. high Qms) seem to deliver more precise bass than those with high mechanical losses AND same Qts.
Regards
Charles
P.S. Graham did you get my photocopies ?
There are sources claiming that drivers with low mechanical losses (i.e. high Qms) seem to deliver more precise bass than those with high mechanical losses AND same Qts.
Regards
Charles
P.S. Graham did you get my photocopies ?
Kuei Yang Wang said:
Assuming you have a bipolar current source with feedback (either from a sense resistor or by using the loudspeaker as part of the feedback loop) that is set to 0 A, I bet that the moment the load injects a current, the feedback will act by injecting a current of opposite sign so that the sum is 0 again.
Hi Capslock!
If you have a loudspeaker conected to a transconductance amplifier with a near infinite output impedance (a constant current source) and you disturbe the loudspeaker mecanicly .
the loudspeker will produce a voltage, but as the output of the amp as a near infinite output impedance there will not be any current in the voice coil only a voltage because a current to existe need a closed circuit.
So the amp c'ant send any correcting current because it d'ont see the disturbance in the load even if it have overall feedback!
Regards
Jorge Santos
If you have a loudspeaker conected to a transconductance amplifier with a near infinite output impedance (a constant current source) and you disturbe the loudspeaker mecanicly .
the loudspeker will produce a voltage, but as the output of the amp as a near infinite output impedance there will not be any current in the voice coil only a voltage because a current to existe need a closed circuit.
So the amp c'ant send any correcting current because it d'ont see the disturbance in the load even if it have overall feedback!
Regards
Jorge Santos
Yes Charles, they came last Monday. Because your real name was on the envelope rather than "phase_accurate" I couldn't remember who it was that sent it to me, otherwise I would have thanked you straight away. Very interesting in fact. Have put it in my stash of stuff to read over and over on rainy days, and one day I might just astound myself and build something like it. Thanks again Charles!phase_accurate said:
How would you go about this with an inverting (voltage feedback)amp? Your input current would have to equal the current through the feedback impedance = load. So you need another power amp and a resistor to generate the input currentphase_accurate said:The current-drive subject has also been discussed within a damping-factor thread : http://www.diyaudio.com/forums/showthread.php?s=&threadid=4536&highlight=info+on+damping+factor
The main advantages of current drive are (in no particular order):
1.) Compensation of sensitivity changes due to VC temperature rise.
2.) Elimination of the effects of the voice-coil inductance.
3.) Added linearity.
The main disadvantages:
1.) Loss of damping around fs.
2.)Some additional circuit complexity
A workaround for the damping loss can either be a mixed mode (i.e. current drive above fs and voltage drive through the fs range) or MFB. Though MFB is also possible with voltage drive, it would be easier to do it with current drive because the aforementioned advantage # 2.) removes one lowpass pole. This is exactly what the German manufacturer Silbersand does.
The disadvantage can best be tackled using an inverting amp and current feedback (one more reson to use inverting amplifiers
Regards
Charles
The paper written jointly by Hawksford and Mills was a research paper for Tannoy's 625ALF subwoofer. The Tannoy was the inspiration for my MFB speaker described in the "Simple MFB...." thread.
They started out with the intention of using a (complicated) current output amplifier, and used velocity feedback from a second coil to control the damping at resonance.
Would it be OK to post the paper, or would copyright make it naughty?
So I approached it from the opposite direction to them, in that I used current feedback to control the damping of the coil feedback... Hmmm It does work, but the Tannoy 625ALF was not a commercial success, so maybe it's just a nice idea.
They started out with the intention of using a (complicated) current output amplifier, and used velocity feedback from a second coil to control the damping at resonance.
Would it be OK to post the paper, or would copyright make it naughty?
So I approached it from the opposite direction to them, in that I used current feedback to control the damping of the coil feedback... Hmmm It does work, but the Tannoy 625ALF was not a commercial success, so maybe it's just a nice idea.
Konnichiwa,
Hmm.
The dreaded "loss of damping".
Actually, there are many methodes to damp a mechanical resonance system. it would seem to me that the best solutions would be mechanical (obviously).
Indeed, I would go as far as to say that a speaker that requires substantial electrical damping is badly designed to start with, as it will in most cases result in maximised distortion and compression.
So, if we actually take the Horse and put it before the cart (instead of the other way around), we find that the linear variable controling the force on the drivers cone is the current, so a current output amplifier is used.
Secondly, we find now that we need to account for a suddenly sky-high Drive unit Qt. We find this of course easy to do, as all we need to do is to make the resonant system less effective, by using a leaky box, acoustic labyrinth etc. We do not need to boost bass by adding rsonance (Horn, Transmission line or Reflex).
So, suddenly we have freedom from compression, low distortion in certain areas and with a suitably adjusted enclosure even low frequency response reaching down to the build in resonance of the driver.
So, I fail to see the problem with current source output amplifiers, probably because I am aware of the huge slew of problems introduced by attempting to use electrical damping.... ;-)
Sayonara
BTW, given that this threads came back from the undead, this little article can be interesting.....
Current Source Amplifiers and Sensitive / Full-Range Drivers by Nelson Pass
capslock said:
Hmm.
The dreaded "loss of damping".
Actually, there are many methodes to damp a mechanical resonance system. it would seem to me that the best solutions would be mechanical (obviously).
Indeed, I would go as far as to say that a speaker that requires substantial electrical damping is badly designed to start with, as it will in most cases result in maximised distortion and compression.
So, if we actually take the Horse and put it before the cart (instead of the other way around), we find that the linear variable controling the force on the drivers cone is the current, so a current output amplifier is used.
Secondly, we find now that we need to account for a suddenly sky-high Drive unit Qt. We find this of course easy to do, as all we need to do is to make the resonant system less effective, by using a leaky box, acoustic labyrinth etc. We do not need to boost bass by adding rsonance (Horn, Transmission line or Reflex).
So, suddenly we have freedom from compression, low distortion in certain areas and with a suitably adjusted enclosure even low frequency response reaching down to the build in resonance of the driver.
So, I fail to see the problem with current source output amplifiers, probably because I am aware of the huge slew of problems introduced by attempting to use electrical damping.... ;-)
Sayonara
BTW, given that this threads came back from the undead, this little article can be interesting.....
Current Source Amplifiers and Sensitive / Full-Range Drivers by Nelson Pass
Kuei Yang Wang said:
That would describe pretty much every dynamic loudspeaker driver out there.
I don't see how a leaky box or a labarynth would lower a sky-high drive unit Q.
Far as I'm aware, a box can only result in a system Q higher than that of the driver Q. A leaky box simply wouldn't raise the Q as high as a non leaky box of the same volume. But you're still left with a system Q that's at best that of the sky high driver Q.
If you want to reduce the driver's mecahnical Q, you'd need to increase the mechanical losses of the driver. Which would mean using lossier materials for the spider and surround.
se