I don't know if anyone here has tried this, but I'd like some feedback on this idea..
As I understand it, power compression with dynamic drivers results primarily from voice coil heating and the positive temperature coefficient of the wire resistance. The same effect also alters other T/S parameters, like Qes and, hence, Qts.
Using a dual coil driver in a similar fashion to the output transformer of a push-pull valve amplifier, except with MOSFET transistors and a highish DC current, I'm thinking this should cancel.
Essentially, the dissipation in the coil remains constant, meaning that heat-derived power compression is eliminated. Also, since the coils are in counterphase to each other, the standing current may not displace the cone, depending on the winding technique. The AC is in-phase, of course.
Combine this with a biquad transform and a sealed enclosure, so as to match room gain, and I'd guess one would be getting very decent and natural bass
As I understand it, power compression with dynamic drivers results primarily from voice coil heating and the positive temperature coefficient of the wire resistance. The same effect also alters other T/S parameters, like Qes and, hence, Qts.
Using a dual coil driver in a similar fashion to the output transformer of a push-pull valve amplifier, except with MOSFET transistors and a highish DC current, I'm thinking this should cancel.
Essentially, the dissipation in the coil remains constant, meaning that heat-derived power compression is eliminated. Also, since the coils are in counterphase to each other, the standing current may not displace the cone, depending on the winding technique. The AC is in-phase, of course.
Combine this with a biquad transform and a sealed enclosure, so as to match room gain, and I'd guess one would be getting very decent and natural bass
Since this would be resemble a "class A device" efficiency would be quite low. And that counts for both - driver and amplifier !
In a "normal case" the peak to average ratio would make dissipation in the voice-coild significantly lower than your proposal.
For midrange and tweeters current-drive would be one solution to compensate for the thermal resistance-increase.
Meyersound has a solution that seems to work at the lower end as well, but I don't know how they do it.
Maybe they use the poweramp offset-voltage to measure the DC resistance of the coil and apply the result to some control circuitry.
Regards
Charles
In a "normal case" the peak to average ratio would make dissipation in the voice-coild significantly lower than your proposal.
For midrange and tweeters current-drive would be one solution to compensate for the thermal resistance-increase.
Meyersound has a solution that seems to work at the lower end as well, but I don't know how they do it.
Maybe they use the poweramp offset-voltage to measure the DC resistance of the coil and apply the result to some control circuitry.
Regards
Charles
phase_accurate said:
Yes, unless you employ feedback, the amplifier has to be operated with a balanced single-ended output stage, hence low efficiency. For those of us who enjoy the SoZ, for example, it should be quite bearable, though.
However, I fail to see how the efficiency (by which I assume you mean efficiency and maximum SPL) of the driver is reduced. Essentially, the DC cancels out, and produces no force, while the AC, being in phase, acts as if the voice coils were paralelled?
Yes. And this variation between peak and average is the main source of dynamic compression in the speaker.
Heavy-duty speakers should have no problem with this dissipation; for example, the JBL2226 can dissipate 600Wrms continous. (Although that driver is not, to the best of my knowledge, available in a dual-coil version.)
For less rugged speakers, just keep below the rated continous dissipation.
Current drive is indeed an option, regardless of which type of driver you are using, but few, if any, drivers are designed for current drive, meaning that you will need to have quite a lot of compensation circuitry.
Sounds doable, but hardly high fidelity? They would, I guess, be using an automatic gain control circuit that aims to keep the DC current constant. Such a circuit can be difficult to implement without loss of fidelity, and it would only respond to changes in the DC characteristic of the load, whereas my approach, admittedly at a very high efficiency penalty, would cancel changes in the AC characteristic without the use of an AGC circuit or similar complexities.
I wouldn't think that the speaker efficiency would be reduced, but rather that the power handling available for music would be reduced. although the forces produced by the opposing coils would cancel, the resistive heating would add.
For normal home use, it seems to me that this is a non problem. Since an average of a watt or two isn't going to heat the voice coils much anyway. Short peaks of a few cycles aren't likely to produce much of a temperature rise due to the mass of the voice coil assembly.
I'd be curious to hear the results of your testing, when you get to it. Adire has high power dual voice coil drivers that may be suitable.
For normal home use, it seems to me that this is a non problem. Since an average of a watt or two isn't going to heat the voice coils much anyway. Short peaks of a few cycles aren't likely to produce much of a temperature rise due to the mass of the voice coil assembly.
I'd be curious to hear the results of your testing, when you get to it. Adire has high power dual voice coil drivers that may be suitable.
BobEllis said:
Yes, you will pretty much limit yourself to the long term max power. I don't see this as much of a problem, since short term power handling is rarely more than some 6dB or so above long term, and most woofers are excursion limited before they hit dissipation limits.
Since you are essentially using each coil as the load for one leg of a balanced single ended output stage, you do not reduce the power handling available for music: the total power remains constant, regardless of signal level. You are just alternating the same net current between two coils.
First off, it depends on your definition of home use. At realistic listening levels and full bass extension, you are going to peak more than a couple of watts into anything but the most efficient of woofers.
Secondly, discussing averages is not really interesting. We are talking here about dynamic compression, not static conditions.
Thirdly, due to the time it takes for the heat to distribute itself into the voice coil assembly (thermal time constant?), the voice coil wire is going to be heated quite a lot during a peak, causing dynamic compression. After the peak has passed, this heat has time to dissipate.
Thanks for the tip.
phase_accurate said:
Ahh. Then we agree. The arrangement *does* waste a lot of energy to minimize nonlinearities by keeping conditions constant.
Hmm... A few thoughts here.
First, isn't the motion of the coil an important component of the cooling of a modern speaker? If that is case, the coil would end up warmer at standstill, and there would be a reverse compression effect.
Second, if the speaker is continously loaded with max power, the resistance would be higher, and this would lead to a higher Qts, which would have to be taken into account during the design, and this would also mean that the speaker would need some warmup time until it would sound as intended.
Third, I think that the lifetime of the driver would be seriously shortened if it was run at full power continously. But of course, I don't know.
I think it is a thought-provoking idea, but I think there is a reason why it is not done. Compression is a too small problem to go through all of these efforts, IMO.
First, isn't the motion of the coil an important component of the cooling of a modern speaker? If that is case, the coil would end up warmer at standstill, and there would be a reverse compression effect.
Second, if the speaker is continously loaded with max power, the resistance would be higher, and this would lead to a higher Qts, which would have to be taken into account during the design, and this would also mean that the speaker would need some warmup time until it would sound as intended.
Third, I think that the lifetime of the driver would be seriously shortened if it was run at full power continously. But of course, I don't know.
I think it is a thought-provoking idea, but I think there is a reason why it is not done. Compression is a too small problem to go through all of these efforts, IMO.
Good thoughts.
The warmup time of the coil can be controlled however, since it is possible to measure DC resistance of the driver and control the heating current accordingly.
But I still don't like the idea.
And there's another drawback: Not only the DC resistance influences Qes, also the magnet's temperature, which has a very large time-constant compared to the voice-coil. And this effect would be more difficult to control than the coil's Re.
The two best things against thermal power compression is the use of generously dimensioned voice-coils (both drivers of my two-way system use 3" coils !) or the use of efficient drivers with light coils.
BTW: The intentional heating of voice-coils is indeed used for some outdoor installations (using low power ultrasonic noise) in order to keep humidity out of the air-gaps.
Regards
Charles
Compression is mostly mechanical
Before you worry too much about the electrical aspects consider that power compression is mostly a mechanical problem. When the cone/voice coil travel approaches Xmax the 6dB/octave excursion equation means that at lower frequencies the suspension puts finite limits on output capability while at higher frequencies the cone is able to travel freely with much higher input levels. If the driver is mounted in an enclosure that adds acoustic resistance, be it a bandpass or a horn enclosure, that acoustic resistance is also non-linear, being another source of limiting cone movement at lower frequencies but not high frequencies and thus being another source of non-linear response.
Before you worry too much about the electrical aspects consider that power compression is mostly a mechanical problem. When the cone/voice coil travel approaches Xmax the 6dB/octave excursion equation means that at lower frequencies the suspension puts finite limits on output capability while at higher frequencies the cone is able to travel freely with much higher input levels. If the driver is mounted in an enclosure that adds acoustic resistance, be it a bandpass or a horn enclosure, that acoustic resistance is also non-linear, being another source of limiting cone movement at lower frequencies but not high frequencies and thus being another source of non-linear response.
Re: Compression is mostly mechanical
Yes. I forgot about that. Unless the airflow can be raised, either by using infrasound during idle, or by using fans, you will indeed see a greater heating of the coil.
I'm not sure what you mean by a reverse compression effect, though?
Actually, Qts always changes during use, and in normal speaker design you have to take into account where you think it will be most of the time, while with this design, you know where it will be after a little while. Also note that heating will raise Qes, not lower it, so you'll be hearing a slightly tight bass until the warmup is complete, which will happen quite quickly.
Also, the compression-cancelling effect will act immediately, even though low end extension will be slightly off for a few secs.
This depends on the driver. The JBL2226 can take 600Wrms indefinitely, and the 2242 can take 800Wrms indefinitely. If they are given an infrasound signal at idle, to avoid standstill overheating, they should not see a reduction in lifespan.
However, I agree that the method would impose some seriously heavy demands on the driver.
Hmm... I wonder if it might be sensible to use a metal cone driver with thermal coupling between the voice coil and cone for this.
When you've been spoiled growing up listening to electrostatic headphones and loudspeakers, dynamic compression is a real issue in most systems. Just desperately looking for a way to "fix" it.
Even with the low compression of the 2226, you're seeing 4dB power compression at rated power. Remember that this is a figure which bases itself on compression of the steady-state signal when averaging the rated power over time. A short peak of rated power is likely to see another digit or so of compression losses due to the small amount of time available to dissipate the heat.
And, as to the benefit, ISTR that Steen Duelund has experimented with NTC resistance in series with the woofer, and found it to do quite a lot to improve dynamics, both macro and micro.
It would not even work at low frequencies, I think? There is the issue of magnetic coupling between the coils that should by far dwarf any back-Emf, I think. Hence, you're really only getting time-delayed negative feedback.
To do that, I think you might want to use a transconductance-output into a speaker that has a very precise Zobel network attached to it. Then measure the deviation from ideal voltage on the output, and amplify the error, before feeding it back to the input. If you set the right error gain, you should see some interesting results, I imagine.
Actually, this argues my case, I think. One thing I had not even considered, is that you don't need to idle at full power, just the amount of power necessary to be able to move up to excursion bounds without increasing dissipation.
Yes, excursion doubles as frequency is halved. Yes, this limits the maximum output. No, that does not in itself contribute to limiting the ability of the speaker to reach said maximum output for a given input. High vs low frequencies are not an issue.
In fact, I'd think tweeters and midranges would benefit far more than woofers, but I haven't found any dual coil tweeters or midranges yet. If financially capable, one could of course order custom units.
Oh, and I was primarily thinking about sealed enclosures. Their 12dB/oct rolloff is ideal for matching room boundary gain.
Subtractive nonlinearities (manifesting themselves as odd order harmonic distortion, IIRC) in the suspension and air load will contribute compression. In a sealed enclosure, the air load should only contribute additive nonlinearities, again IIRC.
The suspension and spider remain obstacles, though. Replacing the suspension with foam, and the spider with a thread suspension, and letting the enclosure and amplifier deal with the damping, should reduce this compression even further.
For ideal dynamics, matching the air load to both front and back of the cone, using a dipole horn, would seem like a nice approach, provided you're single, and rich. I'm neither, though, so I guess I'll leave that option to others who are at liberty to build architectural horns
Svante said:
Yes. I forgot about that. Unless the airflow can be raised, either by using infrasound during idle, or by using fans, you will indeed see a greater heating of the coil.
I'm not sure what you mean by a reverse compression effect, though?
Actually, Qts always changes during use, and in normal speaker design you have to take into account where you think it will be most of the time, while with this design, you know where it will be after a little while. Also note that heating will raise Qes, not lower it, so you'll be hearing a slightly tight bass until the warmup is complete, which will happen quite quickly.
Also, the compression-cancelling effect will act immediately, even though low end extension will be slightly off for a few secs.
This depends on the driver. The JBL2226 can take 600Wrms indefinitely, and the 2242 can take 800Wrms indefinitely. If they are given an infrasound signal at idle, to avoid standstill overheating, they should not see a reduction in lifespan.
However, I agree that the method would impose some seriously heavy demands on the driver.
Hmm... I wonder if it might be sensible to use a metal cone driver with thermal coupling between the voice coil and cone for this.
When you've been spoiled growing up listening to electrostatic headphones and loudspeakers, dynamic compression is a real issue in most systems. Just desperately looking for a way to "fix" it.
Even with the low compression of the 2226, you're seeing 4dB power compression at rated power. Remember that this is a figure which bases itself on compression of the steady-state signal when averaging the rated power over time. A short peak of rated power is likely to see another digit or so of compression losses due to the small amount of time available to dissipate the heat.
And, as to the benefit, ISTR that Steen Duelund has experimented with NTC resistance in series with the woofer, and found it to do quite a lot to improve dynamics, both macro and micro.
arniel said:
It would not even work at low frequencies, I think? There is the issue of magnetic coupling between the coils that should by far dwarf any back-Emf, I think. Hence, you're really only getting time-delayed negative feedback.
To do that, I think you might want to use a transconductance-output into a speaker that has a very precise Zobel network attached to it. Then measure the deviation from ideal voltage on the output, and amplify the error, before feeding it back to the input. If you set the right error gain, you should see some interesting results, I imagine.
BillFitzmaurice said:
Actually, this argues my case, I think. One thing I had not even considered, is that you don't need to idle at full power, just the amount of power necessary to be able to move up to excursion bounds without increasing dissipation.
Yes, excursion doubles as frequency is halved. Yes, this limits the maximum output. No, that does not in itself contribute to limiting the ability of the speaker to reach said maximum output for a given input. High vs low frequencies are not an issue.
In fact, I'd think tweeters and midranges would benefit far more than woofers, but I haven't found any dual coil tweeters or midranges yet. If financially capable, one could of course order custom units.
Oh, and I was primarily thinking about sealed enclosures. Their 12dB/oct rolloff is ideal for matching room boundary gain.
Subtractive nonlinearities (manifesting themselves as odd order harmonic distortion, IIRC) in the suspension and air load will contribute compression. In a sealed enclosure, the air load should only contribute additive nonlinearities, again IIRC.
The suspension and spider remain obstacles, though. Replacing the suspension with foam, and the spider with a thread suspension, and letting the enclosure and amplifier deal with the damping, should reduce this compression even further.
For ideal dynamics, matching the air load to both front and back of the cone, using a dipole horn, would seem like a nice approach, provided you're single, and rich. I'm neither, though, so I guess I'll leave that option to others who are at liberty to build architectural horns
If you are listening to real music and not your function generator then you'd have to feed it insanely high peak power (in the multi-kilowatt range) in order to achieve 600 Watts average, except you use a compressor in front of the poweramp. In this case any discussion regarding compression would be obsolete !!!
Edit: The only solution that deals with power compression, mechanical compression and all the problems with non-linearity is one that controls the radiated sound pressure like this one:
http://www.meyersound.com/products/studioseries/x-10/pdfs/x-10_ds.pdf
Regards
Charles
The predominant source of compression in home audio systems is not from thermal compression (as Charles alludes to, the average power delivered to home speakers is incredibly low, typically below 1W), but BL compression.
As the driver moves in and out, the BL value can change. With typical overhung drivers, the BL begins falling as soon as the driver moves at all. Dropping BL means a loss in efficiency (efficiency is BL and Mms). Which means that BL compression is an instantaneous effect with most drivers - happens on every stroke of the driver, and is independent of what happened previously (unlike power compression).
Drivers with flat BL curves (underhungs operated well within the "underhang", and drivers with new motor topologies like XBL^2) do not exhibit this effect nearly as much, because the BL is constant over a much wider range of motion. BL compression is greatly reduced.
Dan Wiggins
Adire Audio
As the driver moves in and out, the BL value can change. With typical overhung drivers, the BL begins falling as soon as the driver moves at all. Dropping BL means a loss in efficiency (efficiency is BL and Mms). Which means that BL compression is an instantaneous effect with most drivers - happens on every stroke of the driver, and is independent of what happened previously (unlike power compression).
Drivers with flat BL curves (underhungs operated well within the "underhang", and drivers with new motor topologies like XBL^2) do not exhibit this effect nearly as much, because the BL is constant over a much wider range of motion. BL compression is greatly reduced.
Dan Wiggins
Adire Audio
I'll see what I can find... It's pretty basic/simple though - as you traverse the BL curve, the BL of the driver is constantly changing. Efficiency is directly proportional to BL, so if the BL drops 10%, the efficiency drops 10%.
There's some information at http://www.klippel.de and they have a LOT of BL curve measurements, too. You can see how the dynamic efficiency of the driver changes with excursion. This would be, essentially, compression. The higher the stroke, the higher the SPL, but if it's not linear, then that is compression (doubling the current does not double the force on the driver, meaning that SPL does not increase the predicted amount).
Dan Wiggins
Adire Audio
There's some information at http://www.klippel.de and they have a LOT of BL curve measurements, too. You can see how the dynamic efficiency of the driver changes with excursion. This would be, essentially, compression. The higher the stroke, the higher the SPL, but if it's not linear, then that is compression (doubling the current does not double the force on the driver, meaning that SPL does not increase the predicted amount).
Dan Wiggins
Adire Audio
Not really... overhung drivers rarely, if ever, exhibit flat BL curves. A few highly optimized designs have gotten pretty good, but even then there's some significant tradeoffs in terms of BL, flatness, excursion, and inductance.
If you want a low BL compression driver, it's best to look at either an underhung or an XBL^2 type approach (JBL's dual drive is another good one). Straight overhung doesn't give you a flat BL curve at all.
Dan Wiggins
Adire Audio
If you want a low BL compression driver, it's best to look at either an underhung or an XBL^2 type approach (JBL's dual drive is another good one). Straight overhung doesn't give you a flat BL curve at all.
Dan Wiggins
Adire Audio
There's an interesting article by Eugene Czerwinski et al in the latest JAES.
There they present the BL product as a function of coil travel. The underhung version was distinctively better.
When I saw it, I thought that it was a pity that I didn't order my (otherwise excellent) woofers with underhung voice-coils (as the manufacturer proposed !). The downside of his proposed version was only a little less x-max and a slightly higher price. Now that I have experience with the excursions I usually encounter the way I usually listen, I'd definitely order them with underhung voice-coil !
BTW: Convection cooling of the coil is increased with underhung construction also and thereby reducing thermal power compression.
Regards
Charles
There they present the BL product as a function of coil travel. The underhung version was distinctively better.
When I saw it, I thought that it was a pity that I didn't order my (otherwise excellent) woofers with underhung voice-coils (as the manufacturer proposed !). The downside of his proposed version was only a little less x-max and a slightly higher price. Now that I have experience with the excursions I usually encounter the way I usually listen, I'd definitely order them with underhung voice-coil !
BTW: Convection cooling of the coil is increased with underhung construction also and thereby reducing thermal power compression.
Regards
Charles
Bl
You're on the right track with Bl. As to thermal compression it's really not that much of an issue because it's only significant at very high power levels and even then mechanical compression effect dominates.
But as to Bl it's far more important than even Gene and his cohorts suspect. The great bulk of research on power compression has been in direct radiator systems; the phenomena is greatly magnified in horn loaded enclosures, most especially those with very high throat impedances. Since my research into horns lately has concentrated on LF horns with very high throat impedances I've had to deal with LF power compression on the order of 6 to 8dB.
Simply put what happens at high power levels (100 watts or more) and low frequencies (less than 80 Hz), with average Bl figures (less than 12) the motor is unable to overcome the high acoustic pressure of the throat and there is a loss of cone travel linearity; relative response at 50 Hz, for instance, can be 8 dB lower than that at 100 Hz under high power than with small signal inputs. However, when the same horn is loaded with a high Bl driver (16 or more) linearity is maintained to higher power levels at lower frequencies as Bl is increased. The downside to the equation is that as Bl is increased Qts is lowered, leading to a loss of sensitivity and LF extension at low power levels, but once you get past 100 watts or so the lessened LF power compression results in a net increase in sensitivity below 80 hz, with the advantage in total ouput capability for the high Bl driver increasing with additional power input.
This effect is quite obvious with a high throat impedance horn, and would be far less noticeable with a direct radiator, but it would certainly still be there.
You're on the right track with Bl. As to thermal compression it's really not that much of an issue because it's only significant at very high power levels and even then mechanical compression effect dominates.
But as to Bl it's far more important than even Gene and his cohorts suspect. The great bulk of research on power compression has been in direct radiator systems; the phenomena is greatly magnified in horn loaded enclosures, most especially those with very high throat impedances. Since my research into horns lately has concentrated on LF horns with very high throat impedances I've had to deal with LF power compression on the order of 6 to 8dB.
Simply put what happens at high power levels (100 watts or more) and low frequencies (less than 80 Hz), with average Bl figures (less than 12) the motor is unable to overcome the high acoustic pressure of the throat and there is a loss of cone travel linearity; relative response at 50 Hz, for instance, can be 8 dB lower than that at 100 Hz under high power than with small signal inputs. However, when the same horn is loaded with a high Bl driver (16 or more) linearity is maintained to higher power levels at lower frequencies as Bl is increased. The downside to the equation is that as Bl is increased Qts is lowered, leading to a loss of sensitivity and LF extension at low power levels, but once you get past 100 watts or so the lessened LF power compression results in a net increase in sensitivity below 80 hz, with the advantage in total ouput capability for the high Bl driver increasing with additional power input.
This effect is quite obvious with a high throat impedance horn, and would be far less noticeable with a direct radiator, but it would certainly still be there.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.