Kuei Yang Wang said:
Maybe I'm out of school here, but damping is only required when there is a spring, or its equivalent, to store energy. Now, I'm not sure how to get rid of the springs (spider and surround, in this case) cause somethings gotta keep the cone (or its equivalent) from flyin out of the frame. But maybe it doesn't have to be a mechanical spring. I don't guess electrostatic panels need a lot of damping.
Sheldon
No, as a matter of fact, because of their very small mass, the air-to-membrane friction serves to damp the membrane, as I recall (something I read in Stereophile, I think). But you can hardly do that with an electrodynamic driver.Sheldon said:
Konnichiwa,
I never suggested to remove centering and restoring spring. The front surround mainly exists to keep the cone moving straight and to seal the drive unit and is usually needed allways.
The spider can be replaced by making the voicecoil former from a moderatly magnetic material, whih invariably will keep the voicecoil centered in the gap and will also provide the restoring spring force. If the voicecoil former is then conductive too we have damping of the spring action.
This principle was invented by Hartley and is still found in Hartley drive units. BTW, Hartley is one of the few people ever who has build domestic cpeakers that actually make acoustical sense (Siegfried Linkwitz is another, as are the guys at Gradient).
Sayonara
Sheldon said:
I never suggested to remove centering and restoring spring. The front surround mainly exists to keep the cone moving straight and to seal the drive unit and is usually needed allways.
The spider can be replaced by making the voicecoil former from a moderatly magnetic material, whih invariably will keep the voicecoil centered in the gap and will also provide the restoring spring force. If the voicecoil former is then conductive too we have damping of the spring action.
This principle was invented by Hartley and is still found in Hartley drive units. BTW, Hartley is one of the few people ever who has build domestic cpeakers that actually make acoustical sense (Siegfried Linkwitz is another, as are the guys at Gradient).
Sayonara
Kuei Yang Wang said:
I didn't mean to imply that you had said that. Actually I was suggesting, in a very oblique way, getting rid of the springs or more precisely the spring function of the spider and surround, as much as possible anyway. Reducto etc.,: if there is no DC, the waveform should be mostly symmetrical and the electrical input should bring the assembly back to center (in a dipole configuration, at least), and assuming a symmetrical moving assembly (for symnmetrical air resistance). Of course things are not that perfect but it's fun to contemplate.
In the real world it sounds like the idea of low mass, strong magnets, etc., IOW, high eff. drivers . Consistent with Pass's conclusion that these benefitted most from current drive. Don't know about the spring rates in those though.
Sheldon
Onto another topic: what do you think is the most efficient method of providing flux in a motor with (a) field coil(s)? With the coil wrapped around the pole piece, as shown here? Or something like the Parthenon motor, except with the NdFeB ring magnets as part of the columns, the columns themselves would have wire wrapped around them and powered? (see here) Or perhaps with the coil wrapped around the outside of the entire motor structure, as seen here? This last method seems like it would dissipate heat the most readily...
A little off topic, but I believe you can do a field-coil motor a number of ways and see little difference in performance, although some architectures will provide better heat sinking than others. Otherwise, IMO, simpler/cheaper is generally better. Just simulate to get your gap geometry right and make sure you're staying well away from saturation elsewhere.
My interest/experience runs more toward loudspeakers than electronics, so I have a couple semi-ignorant questions:
1. Nelson mentions in his SOZ writeup that Ro can be lowered somewhat, but only at an efficiency penalty. Does that mean the opposite would hold true--that one could raise the Ro of an SOZ and gain the side benefit of improved efficiency?
I've calculated that around a 20-ohm Ro would be just about right for my poor homeless AuraSound 18-8 drivers in NaO-style U-baffles with perhaps some additional stuffing. I think that could be a very nice alignment for for ~ 30-200 Hz to augment a wide-ranger.
2. Is there a simple/easy gainclone-type circuit that would allow me to tailor Ro? Perhaps something along the lines of Rod Elliott's description of a mixed-mode-feedback circuit?
1. Nelson mentions in his SOZ writeup that Ro can be lowered somewhat, but only at an efficiency penalty. Does that mean the opposite would hold true--that one could raise the Ro of an SOZ and gain the side benefit of improved efficiency?
I've calculated that around a 20-ohm Ro would be just about right for my poor homeless AuraSound 18-8 drivers in NaO-style U-baffles with perhaps some additional stuffing. I think that could be a very nice alignment for for ~ 30-200 Hz to augment a wide-ranger.
2. Is there a simple/easy gainclone-type circuit that would allow me to tailor Ro? Perhaps something along the lines of Rod Elliott's description of a mixed-mode-feedback circuit?
An externally hosted image should be here but it was not working when we last tested it.
Bill F. said:
No. As you raise the bias resistor values of the SOZ the
efficiency, what little there is, will decline. On the other hand,
an efficient loudspeaker won't require much, and you're already
at 16 ohms - going to 10 ohm resistors won't change much.
optimal driver values for current amp.
What would be the ranges in an optimal driver for a current amp then? ie. Qes, Qms, Qts, BL, etc.
I want to start looking for these qualities as I am driver shopping.
David
What would be the ranges in an optimal driver for a current amp then? ie. Qes, Qms, Qts, BL, etc.
I want to start looking for these qualities as I am driver shopping.
David
Qes is the important parameter affected by amp output impedence (which I've been calling Ro but should probably call Zo).
To figure out how a driver's Qes will be changed by a given Zo, use this formula:
Qes2 = [Qes1 x (Re + Zo)] / Re
Where:
Qes1 is the driver's original Qes
Qes2 is the new Qes
Re is the driver's VC resistance
Zo is the amplifer's output impedance
(Note: speaker cable resistance is not factored in, but could be added to Zo)
And of course, Qts will reflect the change:
Qts = (Qes x Qms) / (Qes + Qms)
The upshot is the behavior of just about any dynamic driver driven by a current amp (which I'll qualify somewhat arbitrarily as Zo > 40 ohms) will be dominated by Qms.
So, if you want to use a dynamic driver anywere near its resonance with a current amp, pick one with a Qms roughly equal to the Qts you need for your desired alignment.
Otherwise, parallel resonance circuits, active EQ, flow resistance, etc. could be used to deal with an underdamped driver.
To figure out how a driver's Qes will be changed by a given Zo, use this formula:
Qes2 = [Qes1 x (Re + Zo)] / Re
Where:
Qes1 is the driver's original Qes
Qes2 is the new Qes
Re is the driver's VC resistance
Zo is the amplifer's output impedance
(Note: speaker cable resistance is not factored in, but could be added to Zo)
And of course, Qts will reflect the change:
Qts = (Qes x Qms) / (Qes + Qms)
The upshot is the behavior of just about any dynamic driver driven by a current amp (which I'll qualify somewhat arbitrarily as Zo > 40 ohms) will be dominated by Qms.
So, if you want to use a dynamic driver anywere near its resonance with a current amp, pick one with a Qms roughly equal to the Qts you need for your desired alignment.
Otherwise, parallel resonance circuits, active EQ, flow resistance, etc. could be used to deal with an underdamped driver.
Re: optimal driver values for current amp.
Konnichiwa,
Well, the answer depends entierly upon application. I did write up the reasoning behind a simple and basic active speaker using current feed in a different thread, but pretty much everyone posting was so eager to explain why it would not work that the thread is basically dead.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=39338
Sayonara
Konnichiwa,
EternaLightWith said:
Well, the answer depends entierly upon application. I did write up the reasoning behind a simple and basic active speaker using current feed in a different thread, but pretty much everyone posting was so eager to explain why it would not work that the thread is basically dead.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=39338
Sayonara
According to a certain site found using a Google search ("Hartley magnetic suspension"), Hartley's magnetic suspension uses a layer of powdered iron behind the voice coil. The problem I see is the returning force gets weaker and weaker as excursion increases, just as BL vs excursion does in most drivers. Unless, of course, the driver motor was built using an XBL^2 topology, in which the returning force would be roughly flat across linear excursion... no?
Beyond a certain point in the excursion, probably about where an edge of the ferromagnetic ring enters the gap, I believe you would be correct.
However, (I haven't simulated it, but I believe) the return force would rise to that point from the zero point. So the ring's contribution to the compliance curve would be a smooth rise, then a somewhat steeper fall with excursion.
IIRC, Hartley combined the magnetic suspension with a fabric spider. So the spider might have been engineered to assert itself as the ring's contribution fell.
Obviously, the height of the ring would have a lot to do with its behavior. Maybe it was tall enough that no edge would enter the gap in its normal operating range.
I haven't given it much thought yet, but my impression is this particular type of magnetic suspension wouldn't be very well suited to an XBL topology.
Take an outward excursion for example. Soon after the ring leaves its neutral zero point between the gaps and loses touch with the lower gap, its midline will be *attracted* to that of the upper gap, so it will contribute a *negative* return force until its midline is center-gap. Only thereafter will it begin to provide positive return force.
Mmm... you're probably right. To get a constant Cms you'd probably need a progressive spider in addition to the layer of iron.Bill F. said:
Perhaps 2 discrete rings behind the coil, symmetrically displaced outward (away from the coil)?
Two things to avoid that would play havoc with your compliance curve:
1. either ring leaving a gap
2. either ring entering another gap
Right now, the only way I can think of to make a passive magnetic suspension work in an XBL motor over the coil's full Xmag would be to use a single magnetic ring >3 times the height of the VC. That way it can stay in both gaps all the time. (Heavy!)
(Sorry if we're getting too far off topic here, folks...)
1. either ring leaving a gap
2. either ring entering another gap
Right now, the only way I can think of to make a passive magnetic suspension work in an XBL motor over the coil's full Xmag would be to use a single magnetic ring >3 times the height of the VC. That way it can stay in both gaps all the time. (Heavy!)
(Sorry if we're getting too far off topic here, folks...)
Mmm... if we can find a low-density magnetically-permeable alloy to make the former out of...Bill F. said:
I checked www.matweb.com for materials with density <= 4g/cc, and a magnetic permeability of >= 1000. No luck.
Hi,
Appended an electro-mechanical model of an electro dynamic loudspeaker. This model shows more direct what is going on in a loudspeaker than the model of T&S. What can be seen is that damping is a force proportional to cone velocity. Cone velocity generates a back EMF that is subtracted from the driving voltage and as such acting as an electrical damping mechanism. The back EMF relies on the magnetic field in the gap and the place of the voice coil in the gap and is as such a major source of distortion. When driving the speaker with a current source this damping is not effective anymore and damping relies only on mechanical friction (Ds). So far nothing new.
Now is raising the mechanical loss a viable solution the get proper damping? IMHO we are trading bad things for other bad things then. It is very hard to get large mechanical damping ( = friction) with low distortion.
A better approach looks sensing cone velocity and feed back that to the input of the current amp. By adjusting the feedback factor you can adjust the damping at wish. Sensing cone velocity can be done with very low distortion. A second sense coil (on the existing voice coil) is not the way to go IMHO due to mutual coupling and still relative high distortion. Better use an acceleration sensor mounted on the dust cap and integrate the output with say, a 1 Hz low pass filter to get a representation of cone velocity. Piezo disks senses acceleration with pretty low distortion and are cheap.
Cheers 😉
Appended an electro-mechanical model of an electro dynamic loudspeaker. This model shows more direct what is going on in a loudspeaker than the model of T&S. What can be seen is that damping is a force proportional to cone velocity. Cone velocity generates a back EMF that is subtracted from the driving voltage and as such acting as an electrical damping mechanism. The back EMF relies on the magnetic field in the gap and the place of the voice coil in the gap and is as such a major source of distortion. When driving the speaker with a current source this damping is not effective anymore and damping relies only on mechanical friction (Ds). So far nothing new.
Now is raising the mechanical loss a viable solution the get proper damping? IMHO we are trading bad things for other bad things then. It is very hard to get large mechanical damping ( = friction) with low distortion.
A better approach looks sensing cone velocity and feed back that to the input of the current amp. By adjusting the feedback factor you can adjust the damping at wish. Sensing cone velocity can be done with very low distortion. A second sense coil (on the existing voice coil) is not the way to go IMHO due to mutual coupling and still relative high distortion. Better use an acceleration sensor mounted on the dust cap and integrate the output with say, a 1 Hz low pass filter to get a representation of cone velocity. Piezo disks senses acceleration with pretty low distortion and are cheap.
Cheers 😉