After much searching I finaly found a large list of dielectric constants and also data on 3m vhb tapes you might find to be very interesting. jer
Dielectric Constants of Materials
http://multimedia.3m.com/mws/mediawebserver?mwsId=66666UuZjcFSLXTtnxMtLXs6EVuQEcuZgVs6EVs6E666666--
Dielectric Constants of Materials
http://multimedia.3m.com/mws/mediawebserver?mwsId=66666UuZjcFSLXTtnxMtLXs6EVuQEcuZgVs6EVs6E666666--
Thank you all for your answers. I feel much better about my knowledge of this kind of speaker. Unfortuneately, I'm still stupid about several issues.
I wanted to make both as easy to drive speaker, and as efficient as possible.
My questions are: as speaker size goes up efficiency goes up, but is it an easier to drive speaker, or harder.
With a smaller d/s spacing efficiency goes up, but is it an easier speaker to drive.
Assuming that I wanted to run diaphram bias at as high a level as possible,
I would also have the choice of different bias voltages for different spacing all having the same efficiency, but
Which one would be easier to drive, a higher bias/ larger d/s panel ,or a "mid" bias smaller d/s panel.
Assuming for now that we arent worried about a speakers ability to perform bass frequencies.
I take it this tendency for esl's to be difficult on wimpy amplifiers is all about impedance in the top octave.
I wanted to make both as easy to drive speaker, and as efficient as possible.
My questions are: as speaker size goes up efficiency goes up, but is it an easier to drive speaker, or harder.
With a smaller d/s spacing efficiency goes up, but is it an easier speaker to drive.
Assuming that I wanted to run diaphram bias at as high a level as possible,
I would also have the choice of different bias voltages for different spacing all having the same efficiency, but
Which one would be easier to drive, a higher bias/ larger d/s panel ,or a "mid" bias smaller d/s panel.
Assuming for now that we arent worried about a speakers ability to perform bass frequencies.
I take it this tendency for esl's to be difficult on wimpy amplifiers is all about impedance in the top octave.
Forgot to add that if I was to do similar to Acoustat style (yes Moray- making every improvement I can) with either 2 panels or 4 panels. Are the panels connected in parallel to the transformer. I guess series connection is not possible? Would this raise the impedance of the speaker in the top octave?
Thanks Paul
Thanks Paul
With Acoustats, the more panels you have, the easier the load on the amp. The 2-panel 1+1 being more brutal than the 4-panel 2+2, for example. Acoustat panels are hooked up in parallel. Never considered trying a series connection. No way to series the connections since there is only one per stator/membrane, but I guess additional connections could be made. Interesting thought; I have NO idea what the results would be, though.
It bears mention that the bias voltage on the membranes of the Acoustat are always run off of one wire open (thus static) circuit. That connection to the high voltage bias is by definition parallel. I think we're talking about the amplifier connections and that would mean plus(+) minus(-) connections that are analogous to front stator (+) back stator (-) of each panel. Acoustat did indeed always connect them front to front back to back (wires) in parallel simply adding panels as you go. There is no way to put them in series. Think about it. They would be front back front back on the amplifier. That could not work could it?
Yeah, the thought of putting the bias connections in series doesn't seem to make sense. Somehow connecting the end of front stator wire #1 to the beginning of front stator wire #2 (assuming two panels) would, I suppose, electrically create a single front stator, spread over two panels. This I imagine would function, although I don't know what the sonic or electrical outcome would be. Actually, perhaps the bias could be done this way as well, as long as the last panel connection was left open? I'm no engineer so this is merely conjecture on my part. 
Hella,
if I imagine this right if indeed there were an input and output wire on the diaphragm say top and bottom that would essentially make the conductive coating a "wire" in the series circuit. If you connected 4 such wired diaphragms in the "daisy chain" series fashion and the last panel in the circuit "open" thus the static nature of the high voltage, then I'm not sure how the charge would carry through the panels but given resistivity there would be some odd voltage drops and charge distribution I'd surmise. That intuitively does not sound like a good combo for the DC charge. On to the stators then there is no closed connection there either. Generally for the original Acoustat panel there are 3 wire panels and 5 wire. Each of these reserve one wire for the DC charge voltage leaving a front and back stator wire for the 3 wire version and thus 2 front and 2 back stator wires for the 5 wire version. The 5 wire panel has two wires to feed the wire loop from two point but still is not a closed loop if connected plus to minus. Simply put 3 or 5 wire Acoustat stator wires are meant to be connected front to plus(+) back to minus(-) or visa versa and maintaining that phase of connection as panels are added. That is in my mind one of the most electrically interesting things about E-stats in general. Current flows but the panels themselves are all open circuits. Only the interface transformers that the amp sees are the closed loop the amps sees. Quite astounding when you think of it that such great sound comes out of such an unlikely device.
To sum it up though it indeed looks like no series connection per se of panels is possible.
Oh and yes in a sense you and all of us are engineers. Untrained "armchair" and way behind the curve "engineers" however our interest in this stuff belies the little engineer inside us all.
if I imagine this right if indeed there were an input and output wire on the diaphragm say top and bottom that would essentially make the conductive coating a "wire" in the series circuit. If you connected 4 such wired diaphragms in the "daisy chain" series fashion and the last panel in the circuit "open" thus the static nature of the high voltage, then I'm not sure how the charge would carry through the panels but given resistivity there would be some odd voltage drops and charge distribution I'd surmise. That intuitively does not sound like a good combo for the DC charge. On to the stators then there is no closed connection there either. Generally for the original Acoustat panel there are 3 wire panels and 5 wire. Each of these reserve one wire for the DC charge voltage leaving a front and back stator wire for the 3 wire version and thus 2 front and 2 back stator wires for the 5 wire version. The 5 wire panel has two wires to feed the wire loop from two point but still is not a closed loop if connected plus to minus. Simply put 3 or 5 wire Acoustat stator wires are meant to be connected front to plus(+) back to minus(-) or visa versa and maintaining that phase of connection as panels are added. That is in my mind one of the most electrically interesting things about E-stats in general. Current flows but the panels themselves are all open circuits. Only the interface transformers that the amp sees are the closed loop the amps sees. Quite astounding when you think of it that such great sound comes out of such an unlikely device.
To sum it up though it indeed looks like no series connection per se of panels is possible.
Oh and yes in a sense you and all of us are engineers. Untrained "armchair" and way behind the curve "engineers" however our interest in this stuff belies the little engineer inside us all.
Last edited:
Hi guys, wondering if you have checked out my most recent tests? http://www.diyaudio.com/forums/plan...t-panel-angles-8-vs-9-panels.html#post2170405
By increasing the the surface area from 1 panel to 3 panels increases the spl output with the same drive level however this may not mean that the power requirements have not gone up due to the lowering of impedence by adding more capacitance.
But it does allow you to run a lower drive level using say 3 panels or more to get the same spl as for one panel.
so I guess it is safe to say,yes, it is easier on the amp as the higher the surface the less drive voltage level for a determined spl level,but not necessarily the power requirement as you can't get more for less.
However increasing capacitance lowers the impedence
and raises the current demand on the ampilfier,which is where my issues are at the moment.
I am currently using small d/s of around .060" and a considerably high bias voltage of 5kv to 7kv and a high transformation ratio of 1:216.
Trying to contain a voltage that high has proven to be a challenge but has been acomplished and my goal is to get to 10kv.
The high transformation ratio has issues of its own as well as one of my panels is only around 70pf this is a reflected impedence of around 2ohms at 20khz and with three panels it is well below 1ohm.
I have not yet measured the internal capacitance of the transformer (and will do so shortly) as I am sure it is quite high and causing most all of my amplifier issues with the higher frequency's.
So a ground up build with a (very)large core is in order to achieve my particular goals in order to keep the winding count down, thus reducing the internal transformer capacitance to a minimum inorder to keep the high transformation ratio.
Also a large core will allow me to use thicker insulation techniques to further increase the breakdown factor and also reduce capacitance at the same time.
In my last test I also discovered that the white panel seemed a bit louder than the other two even at a lower bias of around 2kv to 3kv as it was reduced because that particular panel can't contain as high of a bias voltage as well as the black ones do.
It was a noticeable difference as it has a compleatly different stator coating and may have a higher capacitance than the other two of the same size.
So I am going to build a few more panels of the same size with different stator coatings with different dielectric constants to determine how much of a factor this may be.
I did try connecting the stators in series and it did work but it also reduced the efficiency as drive voltage was divided equally across the panels.
I will experiment more with this technique more at a later time. jer
By increasing the the surface area from 1 panel to 3 panels increases the spl output with the same drive level however this may not mean that the power requirements have not gone up due to the lowering of impedence by adding more capacitance.
But it does allow you to run a lower drive level using say 3 panels or more to get the same spl as for one panel.
so I guess it is safe to say,yes, it is easier on the amp as the higher the surface the less drive voltage level for a determined spl level,but not necessarily the power requirement as you can't get more for less.
However increasing capacitance lowers the impedence
and raises the current demand on the ampilfier,which is where my issues are at the moment.
I am currently using small d/s of around .060" and a considerably high bias voltage of 5kv to 7kv and a high transformation ratio of 1:216.
Trying to contain a voltage that high has proven to be a challenge but has been acomplished and my goal is to get to 10kv.
The high transformation ratio has issues of its own as well as one of my panels is only around 70pf this is a reflected impedence of around 2ohms at 20khz and with three panels it is well below 1ohm.
I have not yet measured the internal capacitance of the transformer (and will do so shortly) as I am sure it is quite high and causing most all of my amplifier issues with the higher frequency's.
So a ground up build with a (very)large core is in order to achieve my particular goals in order to keep the winding count down, thus reducing the internal transformer capacitance to a minimum inorder to keep the high transformation ratio.
Also a large core will allow me to use thicker insulation techniques to further increase the breakdown factor and also reduce capacitance at the same time.
In my last test I also discovered that the white panel seemed a bit louder than the other two even at a lower bias of around 2kv to 3kv as it was reduced because that particular panel can't contain as high of a bias voltage as well as the black ones do.
It was a noticeable difference as it has a compleatly different stator coating and may have a higher capacitance than the other two of the same size.
So I am going to build a few more panels of the same size with different stator coatings with different dielectric constants to determine how much of a factor this may be.
I did try connecting the stators in series and it did work but it also reduced the efficiency as drive voltage was divided equally across the panels.
I will experiment more with this technique more at a later time. jer
hella356;With Acoustats, the more panels you have, the easier the load on the amp. The 2-panel 1+1 being more brutal than the 4-panel 2+2, for example.
Is it the size of the combined panels that makes for the easier load?
In other words two speakers one with a giant panel and one with 4 separately wired panels (Acoustat 2+2) the latter would be easier to drive?
Is it the size of the combined panels that makes for the easier load?
In other words two speakers one with a giant panel and one with 4 separately wired panels (Acoustat 2+2) the latter would be easier to drive?
That I don't know. The increase in the number of panels being easier on the amp is something I have gleaned from others who know more about this stuff than I do, and is specific to Acoustats, although it may apply to 'stats in general. As I recall, it has to do with the change in the capacitance being easier on the amps and an easier impedance curve. Efficiency also increases, which helps.
However I did use trasparent double sided tape on my micro esl and plan to use it to mount my diagphram frames on my mini esl's.
So that I can take them apart easily to do away with the bolt as Iam tired of got shocked by them sometimes when grab them.
You just have to acount for the extra thinkness as two layers is about 4.5mil. jer
So that I can take them apart easily to do away with the bolt as Iam tired of got shocked by them sometimes when grab them.
You just have to acount for the extra thinkness as two layers is about 4.5mil. jer
The more panels you have, the lower the impedance seen by the amplifier..thus a more difficult load. That is assuming the transformer is the same for both configurations.
But, it is more complicated than that. The transformer has stray capacitance and leakage inductance. The capacitance of the transformer is in parallel with the speakers and the inductance is in series. This causes a rise in voltage peaking at the resonant frequency. This resonance offers a series resonant circuit as seen by the amplifier. At this resonance, the amplifier sees a very low impedance minimized only by the series equivalent resistance of the transformer windings and any resistance added to the low impedance (drive side) of the transformer.
It is desirable that this resonance occur well above the audio range so that a series resistance can be added to the input side of the transformer to flatten the response.
This forms a low pass filter to roll of the highs at the same rate that the response is climbing due to the resonance. It also helps the amplifier. This resistance is now an absolute minimum impedance that the amplifier would ever see. It also helps that it adds a resistive component to the impedance. A purely reactive load is tough on amplifiers. Because of the phase difference with voltage and current, the amplifier has to dissipate more power in the output devices. This is an entirely different subject, but suffice it to say that the amplifier can self destruct, oscillate or both. The protection circuitry in the amplfier can also cause distortion because of the load that it does not understand.
The more panels you have, the higher the capacitance and the LOWER the high frequency resonant frequency. If the resonant frequency is too low, a maximally flat response cause by the proper input resistance would mean that the frequency response is now limited at high frequencies.
There are several ways to handle this problem. One, is to use a separate high frequency transformer (As Strickland did) with much lower leakage inductance and stray capacitance. Another way is to break up the panels with RC filters. The transformer secondary only sees one section as a pure capacitance, and the other sections (at varying degrees) are seen as a resistance in series with the panel capacitance. This means that these panels do not operate full range, but are rolled off on the high end. (Spectra) That is, one section operates full range (with highs), the next section operates up to upper mid range, the next section operates up to lower mid range and so on. This means that all sections operate at low frequencies.
If you are interested, I can post a PSPICE model showing the electrical equivalent of a real transformer (with leakage inductance and stray capacitance) and the result of frequency response and impedance with a capacitive load (esl).
Jim
But, it is more complicated than that. The transformer has stray capacitance and leakage inductance. The capacitance of the transformer is in parallel with the speakers and the inductance is in series. This causes a rise in voltage peaking at the resonant frequency. This resonance offers a series resonant circuit as seen by the amplifier. At this resonance, the amplifier sees a very low impedance minimized only by the series equivalent resistance of the transformer windings and any resistance added to the low impedance (drive side) of the transformer.
It is desirable that this resonance occur well above the audio range so that a series resistance can be added to the input side of the transformer to flatten the response.
This forms a low pass filter to roll of the highs at the same rate that the response is climbing due to the resonance. It also helps the amplifier. This resistance is now an absolute minimum impedance that the amplifier would ever see. It also helps that it adds a resistive component to the impedance. A purely reactive load is tough on amplifiers. Because of the phase difference with voltage and current, the amplifier has to dissipate more power in the output devices. This is an entirely different subject, but suffice it to say that the amplifier can self destruct, oscillate or both. The protection circuitry in the amplfier can also cause distortion because of the load that it does not understand.
The more panels you have, the higher the capacitance and the LOWER the high frequency resonant frequency. If the resonant frequency is too low, a maximally flat response cause by the proper input resistance would mean that the frequency response is now limited at high frequencies.
There are several ways to handle this problem. One, is to use a separate high frequency transformer (As Strickland did) with much lower leakage inductance and stray capacitance. Another way is to break up the panels with RC filters. The transformer secondary only sees one section as a pure capacitance, and the other sections (at varying degrees) are seen as a resistance in series with the panel capacitance. This means that these panels do not operate full range, but are rolled off on the high end. (Spectra) That is, one section operates full range (with highs), the next section operates up to upper mid range, the next section operates up to lower mid range and so on. This means that all sections operate at low frequencies.
If you are interested, I can post a PSPICE model showing the electrical equivalent of a real transformer (with leakage inductance and stray capacitance) and the result of frequency response and impedance with a capacitive load (esl).
Jim
One thing I failed to mention:
If you double the number of panels, the impedance as seen by the amplifier is half. Therefore it is a more difficult load to drive. If you want the load on the amplifier to be the same, then the transformer ratio must be changed.
As an example, if you had one panel of 300pf, it has a reactance of 26526 ohms at 20khz. If the transformer had a voltage ratio of 100:1, then the amplifier sees 2.6526 ohms reactive at 20khz.
If you have 2 panels each of 300pf, then the reactance at 20khz is 13263 ohms. In order to reflect 2.6526 ohms to the amplifier, the voltage ratio of the transformer must be 70.7:1.
Each panel now sees 3.01db less drive, but the panel area is doubled.
Jim
If you double the number of panels, the impedance as seen by the amplifier is half. Therefore it is a more difficult load to drive. If you want the load on the amplifier to be the same, then the transformer ratio must be changed.
As an example, if you had one panel of 300pf, it has a reactance of 26526 ohms at 20khz. If the transformer had a voltage ratio of 100:1, then the amplifier sees 2.6526 ohms reactive at 20khz.
If you have 2 panels each of 300pf, then the reactance at 20khz is 13263 ohms. In order to reflect 2.6526 ohms to the amplifier, the voltage ratio of the transformer must be 70.7:1.
Each panel now sees 3.01db less drive, but the panel area is doubled.
Jim
Thanks,Jim, for the detailed explanation.
Now all of the bits and pieces I have learned about transformers have now been put into a perspective of how and why.
It has made me understand much more clearly now.
Thanks again!
"Each panel now sees 3.01db less drive, but the panel area is doubled."
Does this mean that the efficiency gain is equal (0db gain) or 3db more?
Now all of the bits and pieces I have learned about transformers have now been put into a perspective of how and why.
It has made me understand much more clearly now.
Thanks again!
"Each panel now sees 3.01db less drive, but the panel area is doubled."
Does this mean that the efficiency gain is equal (0db gain) or 3db more?
For those who hadn't been following the transformer design thread, there was a post with some plots showing graphically the 3 things jelanier is describing.
1) Leakage inductance & (transformer winding capacitance + panel capacitance) form an under-damped LP filter which defines the upper bandwidth limit.
2) Adding a series resistance damps this LP filter.
3) Adding a series resistance has the added benefit of raising the minimum load impedance and making the impedance more resistive.
Here is a link to the post.
http://www.diyaudio.com/forums/planars-exotics/161485-step-up-transformer-design-20.html#post2147513