As far as I know from opening mine '63 the complete chassis of te 988 and 989 are different. Allso I have a doubt on the highvoltage supply sinceit is not too overbuild.
Well, I guess you'll just have to try. Stacking a pair of ESL63 on top of each other has been done with a good result. But I do not know how they were wired. I'm actually tempted too...
Be careful with the high voltages involved (5,25 kV)
Well, I guess you'll just have to try. Stacking a pair of ESL63 on top of each other has been done with a good result. But I do not know how they were wired. I'm actually tempted too...
Be careful with the high voltages involved (5,25 kV)
A more sensible option would be to rebuild the Quads and
add dipole subs/stands , there's plently of info on the net.
http://www.euronet.nl/users/temagm/audio/dipolesub.htm
sreten.
add dipole subs/stands , there's plently of info on the net.
http://www.euronet.nl/users/temagm/audio/dipolesub.htm
sreten.Oh, and when soldering:
Be oh so carefull with the diaphragm: when using onely 20 W with a very small tip I managed to fry one of the panels without even touching. The diaphragm is very sensitive for heat...
After thinking, the basspanels are integrated in the delayline. This would mean that stacking is possible with only a litthe problem with the basspanels. The supply does only have to maintain the electrostatic field. It works like a conductor, so I guess one could stack them and wire them in parallel with the other two basepannels (no. 8 wire)...
Be oh so carefull with the diaphragm: when using onely 20 W with a very small tip I managed to fry one of the panels without even touching. The diaphragm is very sensitive for heat...
After thinking, the basspanels are integrated in the delayline. This would mean that stacking is possible with only a litthe problem with the basspanels. The supply does only have to maintain the electrostatic field. It works like a conductor, so I guess one could stack them and wire them in parallel with the other two basepannels (no. 8 wire)...
If you want to keep a flat(tish) bass and lower midrange response, you'll definitely have to modify the circuitry at the end of the transmission line, and it is not obvious at all what modifications will be necessary. If you want to try it, the first things to do are to get a copy of the 989 schematic and of
Peter J. Baxandall, “Electrostatic loudspeakers”, chapter 3 of John Borwick (editor), Loudspeaker and headphone handbook, Focal Press, Oxford, 1997, ISBN 0-240-51371-1
This chapter contains a very detailed description of the design of the ESL63 and the reasons why it is designed like that, including things like the resonance frequency and quality factor of the resonance of the bass panels and how the response peak can be removed for by a good choice of the series resistors.
Peter J. Baxandall, “Electrostatic loudspeakers”, chapter 3 of John Borwick (editor), Loudspeaker and headphone handbook, Focal Press, Oxford, 1997, ISBN 0-240-51371-1
This chapter contains a very detailed description of the design of the ESL63 and the reasons why it is designed like that, including things like the resonance frequency and quality factor of the resonance of the bass panels and how the response peak can be removed for by a good choice of the series resistors.
Summarising what Peter Baxandall has written about the ESL-63 and its bass response:
The ESL-63 is designed in such a way that the response at practical listening distances (2 metres or more) is essentially the far-field response. (The same can not be said of electrostatics with a large uniformly driven diaphragm, at least not when the uniform drive is maintained up to the highest audio frequencies. Those require a listening distance much greater than the square of the largest dimension divided by the wavelength to obtain far-field characteristics, which may be hundreds of metres at high audio frequencies.)
Anyway, according to an equation derived by Peter Walker (Walker's equation), the far-field axial response of a flat panel electrostatic dipole loudspeaker with a diaphragm with negligible mass and negligible stiffness and with acoustically transparent stators is determined by the vector sum of the currents through the stator segments.
Ensuring that the sum of the stator segment currents is frequency-independent is therefore a necessary and sufficient condition to obtain a flat far-field midrange response. Some errors occur for the highest treble and deep bass, though:
A. At high frequencies, the response drops somewhat due to the fact that the diaphragm is not entirely massless and that the stators (and dust covers) are not perfectly acoustically transparent. This is compensated for electrically; I don't believe Baxandall discusses how.
B. At low frequencies, there is a resonance of the stiffness of the diaphragm with the reactive part of the air load (air load mass). This causes an additional second-order high-pass effect. In the QUAD-63, the diaphragm resonance frequency is in the order of 45Hz (if I remember it correctly) and the Q is acoustically damped to about 2 by means of thin cloths glued to the stator plates.
C. The transformers aren't entirely perfect either. Baxandall doesn't discuss the required corrections in much detail.
The resonance with Q=2 would give a response peak in the order of 6dB if the drive current would be kept frequency-independent down to subsonic frequencies. However, by rolling off the current with a first-order slope below a properly chosen frequency, the total response can be kept constant to within +/-0.5dB. (This is rather similar to the way in which a first-order section and a second-order section with a Q of about 2 can together become a third-order Chebyshev filter with a response with only 1dB peak to peak ripple.)
Because of the capacitance of the stators, there is a simple way to obtain a drive current that drops with frequency below a certain frequency and stays constant above that frequency. Simply connect series resistors between a signal voltage source (the transformer) and the stators. The cut-off frequency is then set by the stator capacitance and the resistors.
The series resistors R5 and R7 (360 kohm each) in the ESL-63 have this purpose; at the same time they are terminating resistors for the LC delay line. I don't remember why the bass part is split into two parts with R6 and R8 in between.
If the (damped) inductors in the delay line, the wiring and the stators had no parasitic capacitance, the 10pF...22pF cross capacitors C1...C12 would have been unnecessary, as would R3, R4 and C13. In this case, the delay line would have a nearly frequency-independent input impedance above the frequency where current drive starts, and all current flowing into the delay line would have to flow through the stator segments, since it wouldn't have any other place to go.
However, in reality C1...C12 are necessary to compensate for the influence of parasitic capacitances on the delay line. At midrange and high frequencies, some of the input current flows through C1...C12 instead of through the stator segments. R3, R4 and C13 ensure that the current through the bass segments is equally attenuated, to maintain a flat response.
So if you add bass panels, you may have to increase C13 as well. R3, R4, R5 and R7 have to remain the same, otherwise the delay line isn't properly terminated. It isn't clear to me what you should do with R6 and R8. Maybe looking at the ESL-989 schematic can give you a clue.
The ESL-63 is designed in such a way that the response at practical listening distances (2 metres or more) is essentially the far-field response. (The same can not be said of electrostatics with a large uniformly driven diaphragm, at least not when the uniform drive is maintained up to the highest audio frequencies. Those require a listening distance much greater than the square of the largest dimension divided by the wavelength to obtain far-field characteristics, which may be hundreds of metres at high audio frequencies.)
Anyway, according to an equation derived by Peter Walker (Walker's equation), the far-field axial response of a flat panel electrostatic dipole loudspeaker with a diaphragm with negligible mass and negligible stiffness and with acoustically transparent stators is determined by the vector sum of the currents through the stator segments.
Ensuring that the sum of the stator segment currents is frequency-independent is therefore a necessary and sufficient condition to obtain a flat far-field midrange response. Some errors occur for the highest treble and deep bass, though:
A. At high frequencies, the response drops somewhat due to the fact that the diaphragm is not entirely massless and that the stators (and dust covers) are not perfectly acoustically transparent. This is compensated for electrically; I don't believe Baxandall discusses how.
B. At low frequencies, there is a resonance of the stiffness of the diaphragm with the reactive part of the air load (air load mass). This causes an additional second-order high-pass effect. In the QUAD-63, the diaphragm resonance frequency is in the order of 45Hz (if I remember it correctly) and the Q is acoustically damped to about 2 by means of thin cloths glued to the stator plates.
C. The transformers aren't entirely perfect either. Baxandall doesn't discuss the required corrections in much detail.
The resonance with Q=2 would give a response peak in the order of 6dB if the drive current would be kept frequency-independent down to subsonic frequencies. However, by rolling off the current with a first-order slope below a properly chosen frequency, the total response can be kept constant to within +/-0.5dB. (This is rather similar to the way in which a first-order section and a second-order section with a Q of about 2 can together become a third-order Chebyshev filter with a response with only 1dB peak to peak ripple.)
Because of the capacitance of the stators, there is a simple way to obtain a drive current that drops with frequency below a certain frequency and stays constant above that frequency. Simply connect series resistors between a signal voltage source (the transformer) and the stators. The cut-off frequency is then set by the stator capacitance and the resistors.
The series resistors R5 and R7 (360 kohm each) in the ESL-63 have this purpose; at the same time they are terminating resistors for the LC delay line. I don't remember why the bass part is split into two parts with R6 and R8 in between.
If the (damped) inductors in the delay line, the wiring and the stators had no parasitic capacitance, the 10pF...22pF cross capacitors C1...C12 would have been unnecessary, as would R3, R4 and C13. In this case, the delay line would have a nearly frequency-independent input impedance above the frequency where current drive starts, and all current flowing into the delay line would have to flow through the stator segments, since it wouldn't have any other place to go.
However, in reality C1...C12 are necessary to compensate for the influence of parasitic capacitances on the delay line. At midrange and high frequencies, some of the input current flows through C1...C12 instead of through the stator segments. R3, R4 and C13 ensure that the current through the bass segments is equally attenuated, to maintain a flat response.
So if you add bass panels, you may have to increase C13 as well. R3, R4, R5 and R7 have to remain the same, otherwise the delay line isn't properly terminated. It isn't clear to me what you should do with R6 and R8. Maybe looking at the ESL-989 schematic can give you a clue.
This seems to neglect the fact a Q of around 3* is
deliberately chosen to counteract the 6dB/octave open
baffle roll-off, bass is thus maintained down to diaphragm
resonance and falls off sharply below this.
(*according to the original hi-fi choice review)
A flat nearfield response will give a 6dB/octave bass rollof farfield.
This explains the Quad's poor bass power handling performance
as it rapidly runs out of excursion capability at low frequencies.
sreten.I wonder if it is correct to call a normal dipole ESL an "open-baffle" design. Neglecting the frame, there is basically no baffle because the loudspeaker is acoustically transparent for most frequencies. However, this is not true for the deepest bass frequencies, where diaphragm stiffness comes into play.
Anyway, according to Baxandall, who has had discussions with Walker about it, Q is about 2 for the ESL-63 and about 2.5 for the original QUAD ESL (also known as ESL-55 or ESL-57). I don't know where the discrepancy between 2 and 3 comes from.
Apart from that, I don't see any discrepancy between what Sreten has written and what Baxandall has written.
When current driven, a flat-panel electrostatic loudspeaker with negligible diaphragm mass and stiffness has a flat far-field response, but the near-field response drops with 6dB per octave with increasing frequency.
With voltage drive, the near-field response is flat to a good approximation, but the far-field response increases with 6dB per octave with increasing frequency. This is the proper way to drive electrostatic headphones, but not loudspeakers. It is probably also the reason why many do-it-yourself and some commercial ESL's produce insufficient bass.
When the current drive in a loudspeaker is realised with some form of series resistor (or a termination resistor, in the ESL-63), the loudspeaker necessarily goes from current drive to voltage drive below a certain frequency determined by the resistance and stator capacitance. Below this frequency, the far-field response drops with 6dB per octave. When you want to make this frequency very low, you need a very large resistor and, hence, a very large drive voltage for a given stator current. It is better for the sensitivity to use a somewhat smaller resistor and to compensate for the resulting bass drop by using somewhat less damping for the diaphragm resonance. I presume that this also reduces the influence of the damping cloth on the treble response, although I don't know how serious that is.
By the way, what has the Q got to do with the excursion capability?
Anyway, according to Baxandall, who has had discussions with Walker about it, Q is about 2 for the ESL-63 and about 2.5 for the original QUAD ESL (also known as ESL-55 or ESL-57). I don't know where the discrepancy between 2 and 3 comes from.
Apart from that, I don't see any discrepancy between what Sreten has written and what Baxandall has written.
When current driven, a flat-panel electrostatic loudspeaker with negligible diaphragm mass and stiffness has a flat far-field response, but the near-field response drops with 6dB per octave with increasing frequency.
With voltage drive, the near-field response is flat to a good approximation, but the far-field response increases with 6dB per octave with increasing frequency. This is the proper way to drive electrostatic headphones, but not loudspeakers. It is probably also the reason why many do-it-yourself and some commercial ESL's produce insufficient bass.
When the current drive in a loudspeaker is realised with some form of series resistor (or a termination resistor, in the ESL-63), the loudspeaker necessarily goes from current drive to voltage drive below a certain frequency determined by the resistance and stator capacitance. Below this frequency, the far-field response drops with 6dB per octave. When you want to make this frequency very low, you need a very large resistor and, hence, a very large drive voltage for a given stator current. It is better for the sensitivity to use a somewhat smaller resistor and to compensate for the resulting bass drop by using somewhat less damping for the diaphragm resonance. I presume that this also reduces the influence of the damping cloth on the treble response, although I don't know how serious that is.
By the way, what has the Q got to do with the excursion capability?
A great stacking Quad's link
Far down on this page the original question seems answered with a practical implementaion and pictures + sonics
http://www.audiocircuit.com/9041-esl-circuit/9041IMAI-CO.htm
Petter
Far down on this page the original question seems answered with a practical implementaion and pictures + sonics
http://www.audiocircuit.com/9041-esl-circuit/9041IMAI-CO.htm
Petter
It would also seem more optimal in my eyes to change the "ripple effect" stator setup so that there is one big version rather than two smaller ones.
If one were to use a digital device such as the Behringer DCX2496 one could do digital delay rather than analog delay. This would surely be preferable. Now, this would require multiple amps/transformers. This is the way we should go, particularly if one would be able to build direct drive high voltage amplifiers.
Petter
If one were to use a digital device such as the Behringer DCX2496 one could do digital delay rather than analog delay. This would surely be preferable. Now, this would require multiple amps/transformers. This is the way we should go, particularly if one would be able to build direct drive high voltage amplifiers.
Petter
Because digital is "perfect". Any analog delay lines are as has been said dependent on a number of things we don't have control over all the time: termination being one of them.
I am using a digital crossover for a dynamic speaker today.
Of course if you are using analogue sources the above arguement may not hold water for you
Petter
I am using a digital crossover for a dynamic speaker today.
Of course if you are using analogue sources the above arguement may not hold water for you
Petter
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.