Hi all,
I am new to this forum and I have been doing some research on building DIY electrostatic speakers. I have a questions that may be a stupid one but please enlighten me.
Electrostatic panels work by having the normal amp output of a few tens of volts stepped up to a few kV at the stator. My question is, if I'm driving them with my tube amps, which already put out ~400V from the output tubes, can I just step that up with a 10X transformer directly driving th stators?
It just seems a little "silly" for the voltage to be steped down through the amp OPT and then back up again through the ESL transformer.
???
Thanks...
I am new to this forum and I have been doing some research on building DIY electrostatic speakers. I have a questions that may be a stupid one but please enlighten me.
Electrostatic panels work by having the normal amp output of a few tens of volts stepped up to a few kV at the stator. My question is, if I'm driving them with my tube amps, which already put out ~400V from the output tubes, can I just step that up with a 10X transformer directly driving th stators?
It just seems a little "silly" for the voltage to be steped down through the amp OPT and then back up again through the ESL transformer.
???
Thanks...
Certainly possible and ultimately better than the conventional step-down-then-step-up. But as a practical matter, it will require a custom transformer wound by an expert and may not be trivial to stabilize. For all that effort, intrepid experimenters will go all the way to direct drive, but that's no slam-dunk, either.
If you decide that this would be a fun project to keep you occupied all winter, one thing I might suggest is a tertiary winding that you can use in conjunction with an impedance network (perhaps just a resistor) to add a bit of damping.
If you decide that this would be a fun project to keep you occupied all winter, one thing I might suggest is a tertiary winding that you can use in conjunction with an impedance network (perhaps just a resistor) to add a bit of damping.
Thanks for the clarification, SY.
Are there any articles or existing projects of this sort that you know of?
I am concerned about the frequency response of using a single transformer for the full freq range as well as the impedance on the input and output end of the tranny.
I'm not good at calculations so any help on this area will be much appreciated.
Thanks.
Are there any articles or existing projects of this sort that you know of?
I am concerned about the frequency response of using a single transformer for the full freq range as well as the impedance on the input and output end of the tranny.
I'm not good at calculations so any help on this area will be much appreciated.
Thanks.
I don't know if someone has documented an actual project. It's certainly worth doing some searching in a library (it would probably be from the pre-'Net days).
I think there's a few papers floating around from the old Acoustat company on transformers and bandwidth- they ended up using two different step-ups for bass and treble to drive the same panel. There's probably a few tricks in there that you could use.
Good luck and if you try this, please post about it!
I think there's a few papers floating around from the old Acoustat company on transformers and bandwidth- they ended up using two different step-ups for bass and treble to drive the same panel. There's probably a few tricks in there that you could use.
Good luck and if you try this, please post about it!
Hi Folks,
the direct coupled OTL tube amplifier is best matching to the high impedance of an ESL.
There were some systems avaiable in the past (Beveridge, acoustat, Pütz..), the sound has been amazing, but those systems had been not very relaiable.
In our days there is one interesting concept of an OTL-ESL-tube-amp. Currentyl i have two amps for testing. These amps are capable to deliver 4000 Vpp, which enables quite satisfying sound pressures. Former concepts were limited to room lisening levels.
Take a look: -innoxx audio- High voltage tube amplifier for esl (electrostatic loudspeakers)
I will present my reference ESL E2.5 on a show, driven by these amps. I intend to make a video presentation at the show. If this comes true you will be able to watch the amps "in action" at youtube.
Capaciti
the direct coupled OTL tube amplifier is best matching to the high impedance of an ESL.
There were some systems avaiable in the past (Beveridge, acoustat, Pütz..), the sound has been amazing, but those systems had been not very relaiable.
In our days there is one interesting concept of an OTL-ESL-tube-amp. Currentyl i have two amps for testing. These amps are capable to deliver 4000 Vpp, which enables quite satisfying sound pressures. Former concepts were limited to room lisening levels.
Take a look: -innoxx audio- High voltage tube amplifier for esl (electrostatic loudspeakers)
I will present my reference ESL E2.5 on a show, driven by these amps. I intend to make a video presentation at the show. If this comes true you will be able to watch the amps "in action" at youtube.
Capaciti
www.ultranalog.com built an amplifier employing 845's and a centertapped choke to drive Martin Logans. The site is down, but luckily we have www.archive.org. Look here
Go to schematics (bottom of the page). Look at power amplifiers: 845 OTL
An interesting variation I saw once consisted of operating the whole output stage between 'ground' and a negative supply. Using a CT choke the plates are at about ground potential, and one doesn't need capacitors at the output. On the other hand, one still needs them at the input, or a transformer... See the attached schematic (I can't remember the author, maybe member forum revintage)
Still, I have no practical experience at all with these amplifiers.
Go to schematics (bottom of the page). Look at power amplifiers: 845 OTL
An interesting variation I saw once consisted of operating the whole output stage between 'ground' and a negative supply. Using a CT choke the plates are at about ground potential, and one doesn't need capacitors at the output. On the other hand, one still needs them at the input, or a transformer... See the attached schematic (I can't remember the author, maybe member forum revintage)
Still, I have no practical experience at all with these amplifiers.
Last edited:
Hi Erik,
the concept of ultranalog is limited by the rate of high voltage.
1200 Volts is OK for running a mid-high-range panel like the martin logans, but for an fullrange-ESL it is far away to be sufficient. Even the 4000 Volts do not drive my fullrange-ESL to its sound pressure limit. It needs about 7000 Vpp, which is true for most fullrange-ESL with extended stator-stator spacing
Capaciti
the concept of ultranalog is limited by the rate of high voltage.
1200 Volts is OK for running a mid-high-range panel like the martin logans, but for an fullrange-ESL it is far away to be sufficient. Even the 4000 Volts do not drive my fullrange-ESL to its sound pressure limit. It needs about 7000 Vpp, which is true for most fullrange-ESL with extended stator-stator spacing
Capaciti
Hi,
the problems with very high voltage ESL-amplifiers are high losses and bandwidth. The falling impedance of the ESL -remember its a quality cap- asks for increasing currents with increasing frequency. While at low and midle frequencies the power need of an ESL is very modest (efficiency can be as high as 30%!!) the power need at high frequencies can be enormous.
To reach a power bandwidth of >20kHz the amp must be able to supply several dozend to hundreds of mA. As far as I know no commercial amp can do this (the linked Innoxx shows measurements of a 1nF-panel just for 1kV@20kHz sine against 20kHz with a 240pF panel.....guess why).
They can only supply currents that allow for lower power bandwidths and rely on the fact that the typical music signal doesn´t content high levels above 5kHz. Powerwise its the same, 1500V/40mA or 3000V/20mA, but while 1500V is enough to drive a hybrid panel with high capacitance, 3000V is not enough to drive a low capacitance FR-panel and 20mA not enough to drive a high capacitance panel.
1200-1500V is the maximum Voltage modern power transistors (standard types, not the unobtainable expensive special thingies) and familiar tubes like the 845 can withstand. And that´s the voltage non segmented hybrid-panels like ML´s will happily sing with (max. 1.5mm d/s). Segmented panels (FullRanger) need higher drive voltages but lower currents because of greater d/s (hence smaller capacitance). With transistor amps the output transistors need to be cascaded to allow for those high rail voltages which rises problems of its own (reliability, reduced bandwidth, etc.).
Such limited amps wouldn´t be considered hifi, let alone highend if they were ´normal´ amps, because they don´t even fulfill the basic demand of >20kHz power bandwidth. They rather perform like a lowpass filter with varying and power determined crossover frequency.
There´s a silvery shine at the horizon.....Power-JFETs based on SIC-Technology are just making their way into the market. These could be the devices of tomorrows amps since they could allow for higher voltage rails, larger currents, higher ´speed´ and less heat probs.
They could replace high voltage triodes with the advantage of no heaters needed.
But at the time tubes are still better signalwise but need heating and may be difficult to obtain.
The afore mentioned tubed commercial amps use TV-line deflection pentodes which were never designed to withstand continous high anode voltages but just very short pulses of several kV. The 6HB5 used in the Acoustat amp and the Audio Exclusiv were not very reliable. The PL519 is hairyer chested (well, 35W of limiting anode dissipation aint that much anyway). 3kV or 4kV supply voltage still means a bias of 1.5-2kV at the anode of a tube specced for 700V (the 6HB5 rated at only 18W). No wonder the tube life is shorted, rated around just 2000hrs.
For high capacitance hybrid panels I would choose the supply voltage around 2-2.5kV and currents >50mA. For low capacitance and fullrange 7-8kV and >20mA could be fine. Anyone knows of an commercial amp fulfilling one of these demands?
jauu
Calvin
the problems with very high voltage ESL-amplifiers are high losses and bandwidth. The falling impedance of the ESL -remember its a quality cap- asks for increasing currents with increasing frequency. While at low and midle frequencies the power need of an ESL is very modest (efficiency can be as high as 30%!!) the power need at high frequencies can be enormous.
To reach a power bandwidth of >20kHz the amp must be able to supply several dozend to hundreds of mA. As far as I know no commercial amp can do this (the linked Innoxx shows measurements of a 1nF-panel just for 1kV@20kHz sine against 20kHz with a 240pF panel.....guess why).
They can only supply currents that allow for lower power bandwidths and rely on the fact that the typical music signal doesn´t content high levels above 5kHz. Powerwise its the same, 1500V/40mA or 3000V/20mA, but while 1500V is enough to drive a hybrid panel with high capacitance, 3000V is not enough to drive a low capacitance FR-panel and 20mA not enough to drive a high capacitance panel.
1200-1500V is the maximum Voltage modern power transistors (standard types, not the unobtainable expensive special thingies) and familiar tubes like the 845 can withstand. And that´s the voltage non segmented hybrid-panels like ML´s will happily sing with (max. 1.5mm d/s). Segmented panels (FullRanger) need higher drive voltages but lower currents because of greater d/s (hence smaller capacitance). With transistor amps the output transistors need to be cascaded to allow for those high rail voltages which rises problems of its own (reliability, reduced bandwidth, etc.).
Such limited amps wouldn´t be considered hifi, let alone highend if they were ´normal´ amps, because they don´t even fulfill the basic demand of >20kHz power bandwidth. They rather perform like a lowpass filter with varying and power determined crossover frequency.
There´s a silvery shine at the horizon.....Power-JFETs based on SIC-Technology are just making their way into the market. These could be the devices of tomorrows amps since they could allow for higher voltage rails, larger currents, higher ´speed´ and less heat probs.
They could replace high voltage triodes with the advantage of no heaters needed.But at the time tubes are still better signalwise but need heating and may be difficult to obtain.
The afore mentioned tubed commercial amps use TV-line deflection pentodes which were never designed to withstand continous high anode voltages but just very short pulses of several kV. The 6HB5 used in the Acoustat amp and the Audio Exclusiv were not very reliable. The PL519 is hairyer chested (well, 35W of limiting anode dissipation aint that much anyway). 3kV or 4kV supply voltage still means a bias of 1.5-2kV at the anode of a tube specced for 700V (the 6HB5 rated at only 18W). No wonder the tube life is shorted, rated around just 2000hrs.
For high capacitance hybrid panels I would choose the supply voltage around 2-2.5kV and currents >50mA. For low capacitance and fullrange 7-8kV and >20mA could be fine. Anyone knows of an commercial amp fulfilling one of these demands?
jauu
Calvin
Last edited:
Although an ESL panel forms a capacitive load, this does not automatically mean that increased current is required for constant sound pressure. In fact for an unsegmented panel in the far field, 'Walkers equation' tells us that we need constant current drive at all frequencies to get a flat response. Clearly a compromise can be made between directionality and current requirements. Another solution is to segment the panels and not have a constant d/s spacing, but lower the spacing for high frequency elements hence raising the sensitivity and lowering voltage requirements. A small step-up transformer for bass only would significantly lower the requirements for the transformer.
In short what I'm considering is a 1200V RMS direct drive amp to a panel with small spacing for 400Hz and up, and a step-up from 1500V to 6000V for the bass. Bass panels of course will have significant higher d/s.
In short what I'm considering is a 1200V RMS direct drive amp to a panel with small spacing for 400Hz and up, and a step-up from 1500V to 6000V for the bass. Bass panels of course will have significant higher d/s.
Hi,
any concept using high voltage amps require an appropriate design of the ESL-panel.
Curved panels (mostly non-segmented) and non-segmented flat panels are and not the requested choice, since Capacity might be 1nF and higher.
In my ESL-design there is just one small stripe for frequencies higher than 5000 Hz. Its 69" (175cm) long, 0,8" (2,1cm) wide, having a capacity of 140 pf.
The tube amp shows no band width limitation in the audible range (see frequency response taken at 40cm and 1 metere) and distortions at 97 dB soundpressure (1m) are extremely low. No step-up design enables such low K3 values. There is just an increase of distortions for frequncies heigher than 5 000 Hz indicating the current limit of the amp. In Music such frequencies are not present with this level.
any concept using high voltage amps require an appropriate design of the ESL-panel.
Curved panels (mostly non-segmented) and non-segmented flat panels are and not the requested choice, since Capacity might be 1nF and higher.
In my ESL-design there is just one small stripe for frequencies higher than 5000 Hz. Its 69" (175cm) long, 0,8" (2,1cm) wide, having a capacity of 140 pf.
The tube amp shows no band width limitation in the audible range (see frequency response taken at 40cm and 1 metere) and distortions at 97 dB soundpressure (1m) are extremely low. No step-up design enables such low K3 values. There is just an increase of distortions for frequncies heigher than 5 000 Hz indicating the current limit of the amp. In Music such frequencies are not present with this level.
Attachments
Last edited:
Hi,
1200Vrms means 1700V-1800V of supply voltage. Add losses and some overhead and You end up with a 2kV supply line. This is too much for single MoSFETs. You need cascaded stages or tubes for this job or maybe some HV-IGBT like the YXIS BIMOS-series (up to 4kV). http://www.ixys.com/news/HV_BiMOSFET_2009-10-13.pdf and IXYS Product Portfolio | Power Devices. YXIS manufactures some very interesting HV-devices like HV-depletion mode MOSFETs up to 1kV, that might be useful as current source load in an HV-amp.
Panels working from 400Hz on upwards need less voltage than 1200Vrms, since You can design the d/s around 0.5mm. This would allow for peak-voltages ~1kV. With 1kV of supply voltage You might reach 1.9kVp-p (950Vp/670Vrms), which could be enough already. Shematics for such amps can be found in this forum.
jauu
Calvin
1200Vrms means 1700V-1800V of supply voltage. Add losses and some overhead and You end up with a 2kV supply line. This is too much for single MoSFETs. You need cascaded stages or tubes for this job or maybe some HV-IGBT like the YXIS BIMOS-series (up to 4kV). http://www.ixys.com/news/HV_BiMOSFET_2009-10-13.pdf and IXYS Product Portfolio | Power Devices. YXIS manufactures some very interesting HV-devices like HV-depletion mode MOSFETs up to 1kV, that might be useful as current source load in an HV-amp.
Panels working from 400Hz on upwards need less voltage than 1200Vrms, since You can design the d/s around 0.5mm. This would allow for peak-voltages ~1kV. With 1kV of supply voltage You might reach 1.9kVp-p (950Vp/670Vrms), which could be enough already. Shematics for such amps can be found in this forum.
jauu
Calvin
Hi,
AJ is right with his link to Walker´s equation, but the equation has its limits. It holds only true for on-axis measurements, because it relies on the increasing directivity of a panel to ´beef up´ amplitude response on axis.
The equation strictly applies only to non-segmented flat panels of round or square shape. And it applies only over a freq-range of 5 octaves.
When designing a panel one should take care of directivity issues anyway and well designed commercial ESLs all feature constructional measurements regarding this point. Walker´s equation implies that -at least on axis- neither directivity measurements nor equing should be needed. In fact current drive could be considered as kind of equalization in itself. Funny though that all Quads care very much about directivity and include massive electrical equalization too.
So if we need the same design features to construct a linear amplitude response, why should we use current drive?
Is there any HV-amp that works as transconductance amp, hence current source? All I know of work as voltage source (more or less). The Quads ´simulated´ a current source drive by using large valued resistances. As a downside, the driving voltages must be very high and as such efficiency would become low (so high voltages, so low efficiency that Walker omitted with current drive below app. 300-400Hz). This asks for very high transformation factors for the trannies with all the inherent probs of high-U or amps with very high supply lines.
The problem of high losses at idle especially rely to those amps that have to be run in class-A, e.g. single ended amps. Pushpull amps can be run in class-AB (SRPP can´t be considered as true pushpull with an ESL as load) and could supply larger currents on demand.
jauu
Calvin
AJ is right with his link to Walker´s equation, but the equation has its limits. It holds only true for on-axis measurements, because it relies on the increasing directivity of a panel to ´beef up´ amplitude response on axis.
The equation strictly applies only to non-segmented flat panels of round or square shape. And it applies only over a freq-range of 5 octaves.
When designing a panel one should take care of directivity issues anyway and well designed commercial ESLs all feature constructional measurements regarding this point. Walker´s equation implies that -at least on axis- neither directivity measurements nor equing should be needed. In fact current drive could be considered as kind of equalization in itself. Funny though that all Quads care very much about directivity and include massive electrical equalization too.
So if we need the same design features to construct a linear amplitude response, why should we use current drive?
Is there any HV-amp that works as transconductance amp, hence current source? All I know of work as voltage source (more or less). The Quads ´simulated´ a current source drive by using large valued resistances. As a downside, the driving voltages must be very high and as such efficiency would become low (so high voltages, so low efficiency that Walker omitted with current drive below app. 300-400Hz). This asks for very high transformation factors for the trannies with all the inherent probs of high-U or amps with very high supply lines.
The problem of high losses at idle especially rely to those amps that have to be run in class-A, e.g. single ended amps. Pushpull amps can be run in class-AB (SRPP can´t be considered as true pushpull with an ESL as load) and could supply larger currents on demand.
jauu
Calvin
Special ESL transformer
Thought about this a few years back and did contact a few of the transformer manufacturers like Sowter. Consensus was that it wouldn't work because of something to do with the large inductance affecting the frequency response somehow - don't remember the details. Also the toroidal company in Canada said they could probably do it but they just weren't interested if only 2 pieces were involved.
I have a friend in CT who powers the stators of his panels directly from the plates of his Heathkit W6's; gets plenty of sound output - AND SOUNDS GREAT. Doesn't work with the Quads as the stator spacing is too great.
Charles
Thought about this a few years back and did contact a few of the transformer manufacturers like Sowter. Consensus was that it wouldn't work because of something to do with the large inductance affecting the frequency response somehow - don't remember the details. Also the toroidal company in Canada said they could probably do it but they just weren't interested if only 2 pieces were involved.
I have a friend in CT who powers the stators of his panels directly from the plates of his Heathkit W6's; gets plenty of sound output - AND SOUNDS GREAT. Doesn't work with the Quads as the stator spacing is too great.
Charles