Anyone seen this class-D for ESL design?
High-voltage class-D direct-drive audio amplifier for electrostatic loudspeakers
I can't get into the paper itself, but this might be helpful in ironing out some of the issues.
High-voltage class-D direct-drive audio amplifier for electrostatic loudspeakers
I can't get into the paper itself, but this might be helpful in ironing out some of the issues.
DIRECT-DRIVE DIGITAL AUDIO AMPLIFIER FOR ELECTROSTATIC LOUDSPEAKERS
https://www.linkedin.com/in/john-kota-07156619
quote
Electrostatic loudspeakers (ESLs) are capable of low distortion and have desirably flat frequency response within the audio frequency range. In this paper, a novel approach is proposed for audio amplification using a high-voltage direct-digital solid-state class-D amplifier as a method that replaces impedance matching ESL audio transformers. Optical coupling is used to achieve isolation from the high-voltage side. The proposed stacked MOSFET output stage provides a solution to the voltage swing limitations of the available switching power MOSFETs. An example digital amplifier having 14-bit ADC precision PWM with fc = 50 kHz was implemented at ± 375 V delivering 7W into Ra =20kΩ. The digital modulator was realized on a Xilinx Virtex4 Sx35-10ff668 at Fs = 100 MHz clock.
end quote
https://www.apexanalog.com/resources/products/pa99u.pdf
https://www.linkedin.com/in/john-kota-07156619
quote
Electrostatic loudspeakers (ESLs) are capable of low distortion and have desirably flat frequency response within the audio frequency range. In this paper, a novel approach is proposed for audio amplification using a high-voltage direct-digital solid-state class-D amplifier as a method that replaces impedance matching ESL audio transformers. Optical coupling is used to achieve isolation from the high-voltage side. The proposed stacked MOSFET output stage provides a solution to the voltage swing limitations of the available switching power MOSFETs. An example digital amplifier having 14-bit ADC precision PWM with fc = 50 kHz was implemented at ± 375 V delivering 7W into Ra =20kΩ. The digital modulator was realized on a Xilinx Virtex4 Sx35-10ff668 at Fs = 100 MHz clock.
end quote
https://www.apexanalog.com/resources/products/pa99u.pdf
Last edited:
What would you expect from student project anyway, no pun intended.
Amp uses four FETs with rather awkward driver topology which includes fiber optics for isolation with small signal BJTs as boosters. No Bandwidth or other audio charts are given. Membrane looks graphite coated: very dark - could be publishing artifacts.
Cheers
Amp uses four FETs with rather awkward driver topology which includes fiber optics for isolation with small signal BJTs as boosters. No Bandwidth or other audio charts are given. Membrane looks graphite coated: very dark - could be publishing artifacts.
Cheers
I have thought about a HV Class D system every once in a while.
But, Being an analog purist I am not ready to go down that road yet.
Besides such a system gets way more complicated than a simple Class A P-P setup would be and for mediocre performance just doesn't weigh in very well compared to an analog setup.
jer
But, Being an analog purist I am not ready to go down that road yet.
Besides such a system gets way more complicated than a simple Class A P-P setup would be and for mediocre performance just doesn't weigh in very well compared to an analog setup.
jer
As said before: there is nothing particular in the amp per se. Q=C*U => 1500*1e-9=1.5 uC => 50k*1.5e-6=75 mA. 75mA*1500V=112.5W of idle dissipation making D-class action rather questionable.
Transition time is around 4 us at +/- 750 V.
old thread but the contents are still very valid
http://www.diyaudio.com/forums/planars-exotics/80714-another-direct-drive-thread.html
http://www.diyaudio.com/forums/planars-exotics/80714-another-direct-drive-thread.html
I got an email today about some new SIC FET's at Digikey and I really like there specs!!
They have a very low Coss (output capacitnace) in the range of 15pf-30pf !!
These may work very nice for my Desktop models and slightly large Bookshelf type of ESL's.
1200V, 85watt,
http://rohmfs.rohm.com/en/products/databook/datasheet/discrete/sic/mosfet/sct2450ke-e.pdf
SCT2450KEC Rohm Semiconductor | Discrete Semiconductor Products | DigiKey
1700V, 69watt,
http://www.wolfspeed.com/media/downloads/173/C2M1000170D.pdf
C2M1000170D Cree Inc. | Discrete Semiconductor Products | DigiKey
1700v, 35watt,
http://rohmfs.rohm.com/en/products/databook/datasheet/discrete/sic/mosfet/sct2h12nz.pdf
SCT2H12NZGC11 Rohm Semiconductor | Discrete Semiconductor Products | DigiKey
1200v, 108watt,
http://rohmfs.rohm.com/en/products/databook/datasheet/discrete/sic/mosfet/sct2280ke-e.pdf
http://www.digikey.com/product-detail/en/rohm-semiconductor/SCT2280KEC/SCT2280KEC-ND/4171677
These are about $5 or $6 or so, and are just a few of what is available.
I haven't gone through everything they have yet.
Others are in the $20 to $40 each, but must of those also have a higher Vdds and power dissipation as well.
jer
They have a very low Coss (output capacitnace) in the range of 15pf-30pf !!
These may work very nice for my Desktop models and slightly large Bookshelf type of ESL's.
1200V, 85watt,
http://rohmfs.rohm.com/en/products/databook/datasheet/discrete/sic/mosfet/sct2450ke-e.pdf
SCT2450KEC Rohm Semiconductor | Discrete Semiconductor Products | DigiKey
1700V, 69watt,
http://www.wolfspeed.com/media/downloads/173/C2M1000170D.pdf
C2M1000170D Cree Inc. | Discrete Semiconductor Products | DigiKey
1700v, 35watt,
http://rohmfs.rohm.com/en/products/databook/datasheet/discrete/sic/mosfet/sct2h12nz.pdf
SCT2H12NZGC11 Rohm Semiconductor | Discrete Semiconductor Products | DigiKey
1200v, 108watt,
http://rohmfs.rohm.com/en/products/databook/datasheet/discrete/sic/mosfet/sct2280ke-e.pdf
http://www.digikey.com/product-detail/en/rohm-semiconductor/SCT2280KEC/SCT2280KEC-ND/4171677
These are about $5 or $6 or so, and are just a few of what is available.
I haven't gone through everything they have yet.
Others are in the $20 to $40 each, but must of those also have a higher Vdds and power dissipation as well.
jer
Last edited:
Not at audio frequnecy's.
Just need a good opamp that can deliver some current and it still isn't that much at probably less then 50ma and much less considering the voltage swing on the gate is less than 1-2Vp-p.
The drive imedance into the gate for say 667pf is only about 2Kohms at 100Khz, so that comes out to be 1ma, any decent opamp should do it.
jer
Just need a good opamp that can deliver some current and it still isn't that much at probably less then 50ma and much less considering the voltage swing on the gate is less than 1-2Vp-p.
The drive imedance into the gate for say 667pf is only about 2Kohms at 100Khz, so that comes out to be 1ma, any decent opamp should do it.
jer
Last edited:
The article inspired many people. The symmetric amp design is a pretty suitable concept although susceptible to high frequency oscillation.That is the article that got me started in ESL's back in 1986 !!!!
jer
My amp was pretty close to Sanders inspiration but using big resistors as a load works better than having capacitative cells as the load (even though that's the relevant load). ESLs need so little horsepower, that they can be just parasites on the resistors load and the amp hardly knows the cells are there.
Final design, at least in my case, depends on your local surplus electronics store, esp for the costly items used in the B+ supply. You might luck-into some components that allow you to make a superior power supply (or not).
Ben
Direct is direct. I never had such effortlessly clean and luminous sound any other way.
Ben
Hi,
with resistor loaded amps the upper bandwidth limit should be kept an eye on as with typical values seen for the R the bandwidth doesn't even reach 20kHz.
It can only be recommended -strictly technical speaking- for low capacitance low voltage panels, e.g tweeters or headphones.
Besides, Alex's Re already touched another problem, the terribly low efficiency of such an amp.
Theoretically the maximum efficiency would reach a meager 6%.
In practical use its less than 1% for most of the time.
In the light of dramatically changing climate I think everybody should ask himself if he/she could really recommend such a energy wasting technology.
jauu
Calvin
with resistor loaded amps the upper bandwidth limit should be kept an eye on as with typical values seen for the R the bandwidth doesn't even reach 20kHz.
It can only be recommended -strictly technical speaking- for low capacitance low voltage panels, e.g tweeters or headphones.
Besides, Alex's Re already touched another problem, the terribly low efficiency of such an amp.
Theoretically the maximum efficiency would reach a meager 6%.
In practical use its less than 1% for most of the time.
In the light of dramatically changing climate I think everybody should ask himself if he/she could really recommend such a energy wasting technology.
jauu
Calvin
Well, nice to see "green" becoming an issue at this forum. That could lead to some very novel thinking esp in relation to tube amps.
Disregardless of Calvin's insights into theoretical matters, the amp worked just great for me and needed no attention for decades.
But talking about efficiency, to find the basic name for what we are talking about, is really odd in making sound. I hope everybody knows what one watt of acoustic energy sounds like in your music room at home?* Our music systems produce flea-power acoustic output. But, at least for tube gear, we can heat our room in the winter with the waste heat from the amps and need air-conditioning for tri-amped systems consuming maybe 1500 watts.
Can anybody post the efficiencies of cone drivers and ESL panels? My recollection is that a boxed cone speaker might be like 2% efficient. So you'f multiply the sound by 50X to estimate the amp output needed.
But a raw ESL panel is substantially more efficient, although the hardware to drive an ESL is cumbersome and inefficient. With a DD amp, the gear is especially cumbersome and inefficient, as Calvin says, but the ESL play just fine symbiotically off the load resistors because they really require very little power to vibrate.
Ben
*OK, I admit that's a trick question since nobody could possibly stand to be in a room with one watt of sound blaring at them.
Disregardless of Calvin's insights into theoretical matters, the amp worked just great for me and needed no attention for decades.
But talking about efficiency, to find the basic name for what we are talking about, is really odd in making sound. I hope everybody knows what one watt of acoustic energy sounds like in your music room at home?* Our music systems produce flea-power acoustic output. But, at least for tube gear, we can heat our room in the winter with the waste heat from the amps and need air-conditioning for tri-amped systems consuming maybe 1500 watts.
Can anybody post the efficiencies of cone drivers and ESL panels? My recollection is that a boxed cone speaker might be like 2% efficient. So you'f multiply the sound by 50X to estimate the amp output needed.
But a raw ESL panel is substantially more efficient, although the hardware to drive an ESL is cumbersome and inefficient. With a DD amp, the gear is especially cumbersome and inefficient, as Calvin says, but the ESL play just fine symbiotically off the load resistors because they really require very little power to vibrate.
Ben
*OK, I admit that's a trick question since nobody could possibly stand to be in a room with one watt of sound blaring at them.
Last edited: