Hi
Here is an idea that mixes class D with ESLs.
One of the main difficulties associated with ESLs is the step-up transformer: the magnetizing inductance and core saturation limit the low frequency response, and the leakage inductance causes problems at the high end of the spectrum, especially considering the highly capacitive nature of the load.
Direct drive is the obvious solution, but is difficult to achieve and probably requires tubes, with problems of their own.
I propose the following arrangement:
A conductive membrane is suspended between two metallic grids; the grids are driven by high frequency, high voltage sinusoids having +90° and -90° phases; the center membrane is driven by a phase modulable sinusoid of similar voltage and 0° quiescent phase. Under those conditions, the net force exerted on the membrane will be zero: the force at 2*Fo on one side will be exactly compensated by an opposite force on the other side.
Now, if the phase shifts from 0°, the average absolute potential difference between the membrane and each of the grids will become unbalanced, resulting in a net force on the membrane.
Because the electrostatic force is independent of the sign of the potential difference, the membrane acts as a "phase comparator".
In summary, all we have to do in order to generate an audio signal is to phase-modulate the membrane drive wrt. the reference grids signals.
We need a center-tap transformer with its driver for the grids, and another transformer and driver, plus a phase modulator for the signal channel.
The big advantage is the fact that the transformers only need to operate at a high frequency, making them small, light and non-critical; their amplitude linearity is irrelevant, as only the phase information is critical.
The output drivers can operate in class D/E, the speaker capacitance being part of the output filter.
I see some potential difficulties: with kilovolt signals in the 100~200KHz range, great care would be needed to avoid corona and arcing; the drivers would need to supply a huge reactive power to the system.
For the signal channel, a compromise would have to be found between the Q of the output network and the bandwidth needed to reproduce the PM sidebands.
I think these are practical, solvable issues, what's your opinion about it?
Note that this outline is mainly conceptual, made to ease the comprehension of the principles, but practical implementations could significantly differ.
Enhancements are also possible: for instance, the linearity could be improved by means of a feed-forward analogue signal for correction purpose. This way, the missing parts of the Bessel function due to band-limiting could be compensated for.
Etc, etc...
Here is an idea that mixes class D with ESLs.
One of the main difficulties associated with ESLs is the step-up transformer: the magnetizing inductance and core saturation limit the low frequency response, and the leakage inductance causes problems at the high end of the spectrum, especially considering the highly capacitive nature of the load.
Direct drive is the obvious solution, but is difficult to achieve and probably requires tubes, with problems of their own.
I propose the following arrangement:
A conductive membrane is suspended between two metallic grids; the grids are driven by high frequency, high voltage sinusoids having +90° and -90° phases; the center membrane is driven by a phase modulable sinusoid of similar voltage and 0° quiescent phase. Under those conditions, the net force exerted on the membrane will be zero: the force at 2*Fo on one side will be exactly compensated by an opposite force on the other side.
Now, if the phase shifts from 0°, the average absolute potential difference between the membrane and each of the grids will become unbalanced, resulting in a net force on the membrane.
Because the electrostatic force is independent of the sign of the potential difference, the membrane acts as a "phase comparator".
In summary, all we have to do in order to generate an audio signal is to phase-modulate the membrane drive wrt. the reference grids signals.
We need a center-tap transformer with its driver for the grids, and another transformer and driver, plus a phase modulator for the signal channel.
The big advantage is the fact that the transformers only need to operate at a high frequency, making them small, light and non-critical; their amplitude linearity is irrelevant, as only the phase information is critical.
The output drivers can operate in class D/E, the speaker capacitance being part of the output filter.
I see some potential difficulties: with kilovolt signals in the 100~200KHz range, great care would be needed to avoid corona and arcing; the drivers would need to supply a huge reactive power to the system.
For the signal channel, a compromise would have to be found between the Q of the output network and the bandwidth needed to reproduce the PM sidebands.
I think these are practical, solvable issues, what's your opinion about it?
Note that this outline is mainly conceptual, made to ease the comprehension of the principles, but practical implementations could significantly differ.
Enhancements are also possible: for instance, the linearity could be improved by means of a feed-forward analogue signal for correction purpose. This way, the missing parts of the Bessel function due to band-limiting could be compensated for.
Etc, etc...
This is quite a clever idea and I could imagine that it would actually work. Phase modulation like that is also used as a power-efficient solution to generate AM broadcast signals BTW.
The problem arises from somewhere else: If you want to drive an ESL diaphragm homogenuously over the whole are you have to keep the charge from wandering on the diaphragm. This is usually done by using a diaphragm of high resistance but this would be in the way of efficiently driving it the way you want. So you'd have to come up with a solution for this problem as well in order to make it work. Maybe driving it in isolated segments would help.
Regards
Charles
The problem arises from somewhere else: If you want to drive an ESL diaphragm homogenuously over the whole are you have to keep the charge from wandering on the diaphragm. This is usually done by using a diaphragm of high resistance but this would be in the way of efficiently driving it the way you want. So you'd have to come up with a solution for this problem as well in order to make it work. Maybe driving it in isolated segments would help.
Regards
Charles
phase_accurate said:
Final claim they have solved this in their 'inverted ESL', they have a patent that describes how they drive the stators. I think there is another problem with a conducting diaphragm that the force is depending on the excursion so there will be a lot of distortion. But the idea is a good one. The rest is engineering, thus solvable.
Jan Didden
phase_accurate said:
FINAL
http://www.final.nl/electro.html#advantage
http://patft.uspto.gov/netacgi/nph-...L&OS=AN/FINAL+AND+FINAL&RS=AN/FINAL+AND+FINAL
Jan Didden
I'm aware that voltage drive will cause a reduced efficiency and increased non-linearities compared to charge drive, but there are mitigating factors: the push-pull construction will cancel this effect to the first order, and the effect will increase the sensitivity for high excursions, where mechanical effects will tend to cause a compression.phase_accurate said:
In addition, this topology allows for a simple and accurate method of feedback: if a small DC bias is superimposed on the HF drive, the speaker becomes its own, high accuracy microphone.
The low frequency transfer function can then be corrected by a MFB loop at low frequencies, where the non-linearities are susceptible to happen.
LV
el`Ol said:Could this also be a solution:
Asymmetric PWM with Square wave = max. AC voltage, narrow spike = min. AC voltage, also high switching rate with small core-less transformer plus a rectifier after the transformer for decoding.
The big problem with this type of circuit is discharging the panel capacitance after a peak. If full power bandwidth up to 20KHz is required, this means a low value resistance, and a huge amount of power to dissipate; not quite what I try to achieve, just the opposite in fact.
LV
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.