ESL electronics 101 for the electronics challenged

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
This is a spin-off from the "DIY esl for dummies" thread.
I realized there are a lot of basic theory that eludes me and I'm hoping some of you smart and capable people can help me filling in the blanks.

It would be great if this thread could turn into a tutorial in electronics basics for the DIY ESL n0ob. If there is already such a thing please point the way because I need to do some reading.

The electrostatic loudspeaker or esl as they are often called is basically a huge plate capacitor.
The metaphorical driving force is the electric field generated between the two stators.
To increase the field strength you can either increase the voltage or decrease the D/S distance.

Since the ESL is a parallel-plate capacitor of sorts, decreasing the D/S will increase the capacitance. The other way of increasing the capacitance is to increase the stator area.

When dealing with AC such as a music signal the usually friendly and easily accessible electronics turn complicated really fast.
Instead of the easy to comprehend resistance R we have to deal with impedance Z.

The panels are almost purely reactive which lead to Z=R-j/WC [ohm], W (omega) being the angular velocity [radians/s]. W=2Pi*f, f being the frequency [Hz].

What this means is just that the impedance will drop as the frequency goes up and -j/WC is the imaginary/complex component of Z.

This is the point where I'm starting to get confused.
Purely resistive loads are ideal for your average amplifier while a purely reactive load is about as far from ideal as you come.

At this point there are two roads to chose from.
1) The common solution of using a step up transformer and a regular amplifier.
2) Get yourself a monster of a Direct drive amplifier. This is clearly not for the faint of heart. It's a big project and you're quite possibly going to be handling lethal currents.

Going with option #1 you attach a transformer. This sounds easy enough but the already hairy Z suddenly gets an additional component, the inductive part jWL.
At this point Z=R+jX where X=WL-1/WC.

To understand the inner mechanics of a esl I'd like to start with option #2, using a DD amp. Not really building one but learning how everything fits together and how it works.

Some amplifiers are prone to oscillating when confronted with difficult loads while others are fine.

How does the preferred load look? (Speaking of ESL's)
A large capacitance will result in a low impedance and a relatively small phase angle.
A small capacitance will be the exact opposite, high impedance and a relatively large phase angle.

A large C will have a high demand for current but be an easier load?
A small C will need less current but it'll be more difficult on the amp?
Personally I always thought this to be a no-brainer, the small C was obvious to me.
After reading a few posts in another thread it suddenly didn't seem as clear any more.

How do you estimate how much current you'll need?
How do you know if the amplifier is in danger of oscillating?
How does an amplifier look that is immune to oscillating while working with audio frequencies?
How can I make the panel an easier load?
When will the frequency response suffer?
What are the demands on the amp that I need to meet?

As long as I looked upon Z as a resistive part everything was dandy. Now it's all confusing?

Can someone please explain to me how to handle the complex load?

I used to approximate the current demand using f= 20kHz.
In a practical application you could probably get away with designing for half that current since there really isn't all that much information in music at 20kHz.
Was my thinking correct?

This is probably enough questions to fill a book on it's own but figured I'd give it a shot anyway. :)

This is a free for all, if you have any questions about the electronics involved with ESL's go ahead and ask them.
Hopefully we can find answers to at least a few of our questions.
 
Not to complicate matters but it has also been discussed that, It is a posibility that the strain on the amplifier can be reduced from using a larger panel becuase of the added gain of having a larger surface area.

Which by the way is why I built a panel with 4 times the surface area of my little panel, that is too explore this.

Only I haven't got as far in testing this becuase #1, I haven't rebuilt new frames for the bigger panels yet, as the old one got broke, and #2 I don't have suitable measurement microphone or SPL meter at the moment.
And, I would have done so when I built them in 2003 but I didn't have the proper electronics to drive them until a year ago.

Increasing the surface area does increase the capacitance but because of the added gain of the larger area this allows you to use a lesser transformation ratio which inturn reduces the load on the amplifier for the same SPL level.

This is where compromization comes into play unless you are trying to go out with all balls to the walls type perfromance like I have been trying to do to get every last drop of efficiency out of them.

Apparently (and it has been said) that the laws of physic's due to the inoization of air will only allow you so much.

This is why the Wright speakers are in a sealed bag of sulfer hexaflorine in order to run higher voltages in order to increase the efficiency.

I have pushed my panels to this level of ionization of the air with the common purple glow as an indicator and could not get any more sound preasure out of them.

So all that extra voltage was a waste and didn't do to much for loudness except stress the materials to their limits aswell.

The good thing was that the resistance of the stator coating and the diagphram coating was high enough so that the currents in the arcs were low enough not to burn holes in the mylar diagphram.

I found this to be true when I found holes burned in the diagphram of the panels that were coated with the white spray paint when subjected to the same conditions. jer
 
I had the same questions as you do as well as many others.

And it has takin alot of arm twisting and teeth pulling, Aswell as my on experimentations to find such answers.

Even through all the the thread digging I could only find bits and pieces to the answers here and there and many without solid definition, It is good to have them all in one spot.

As you can see how I keep having to back track to find the shortcuts to the best threads and they get buried very easily and quickly.

It would be nice if they would split this part of the forum into there own enities of the different technologies like they did with the amplifier section.

As I tried to state in a funny way but with all seriousness is that, my goal is to "take the mystery out" of such devices in order to get more diy'ers intrested in the technology.

As it is truley amazing the kind of sound you get with a fairly simple (said lightly) form of construction. jer
 
To understand the inner mechanics of a esl I'd like to start with option #2, using a DD amp. Not really building one but learning how everything fits together and how it works.

Some amplifiers are prone to oscillating when confronted with difficult loads while others are fine.

How does the preferred load look? (Speaking of ESL's)
A large capacitance will result in a low impedance and a relatively small phase angle.
A small capacitance will be the exact opposite, high impedance and a relatively large phase angle.

A large C will have a high demand for current but be an easier load?
A small C will need less current but it'll be more difficult on the amp?
Hello MarkusA,

Remember no matter if you are dealing with a large capacitance(low impedance) or small capacitance(high impedance), the phase angle will be the same, close to -90 degrees.

The only way to make the phase angle smaller is to add resistance to the load, or reactance of the opposite sign(ie inductance). The resulting phase angle will be dependent on the relative magnitudes of the resistance, inductance, and capacitance. Vector math is involved, but nothing too difficult. I can post a few mathematical & graphical examples if it would help. But, a few wiki searches may help just as well.

A simple example of how to reduce the phase angle is to add a high power shunt resistance of low enough value that will swamp the capacitive reactance.
 
Thanks but this time it's just me expressing myself poorly.
I was thinking of "phi" and |Z| when I wrote that.
Then again, I made a mess of it and it's always good to have a proper explanation. If you have the time, everyone will gain from it.

My problem is rather understanding how the reactive part of the load is affecting the amplifier?
 
If you have the time, everyone will gain from it.

Attached is a *.zip file containing an Excel 2004 compatible and Excel 2007+ compatible impedance visualization spreadsheet for helping understand 4 simple circuit combinations.

1) R and C in series: See pic#1
The spread sheet lets you enter values for R and C and plots the impedance magnitude and phase. There is also a slider bar at the bottom of the main plot that lets you move the green marker to select a particular frequency of interest. The values for resistance, reactance, impedance magnitude and phase are displayed in the colored table above the plot. Also, the vector diagram is plotted off to the right, showing resistance in blue along the X-axis, reactance in red along the Y-axis, and the resulting impedance in green. The phase is the angle between the blue and green lines.

It is quite interesting to scroll the marker along and watch the vector diagram change with frequency.

Note that for R and C in series, R dominates and sets the minimum impedance at high frequency; the phase approaches zero. At lower frequencies the reactance of C dominates and the phase angle approaches -90 degrees.

See pic#2 for the vector diagram at a lower frequency(set by moving the marker)

2) R and C in parallel: See pic#3
With this type of circuit, impedance at lower frequency is dominated by R which sets a maximum value for the impedance; phase approaches zero. At high frequencies, C dominates and the impedance approaches zero; phase approaches -90.

Combinations of resistance in series and parallel with a capacitor can be used to make the impedance more resistive(low phase angle) over the whole audio range.


3) R, L, and C in series: See pic#4
This is a series resonant circuit. L and C determine the frequency of resonance. R determines the minimum value of impedance at resonance and the sharpness of the resonance. Impedance is capacitive below resonance(phase ~ -90deg), and inductive above resonance(phase ~ +90 deg).

4) R, L, and C in parallel: See pic#5
This is a parallel resonant circuit. L and C determine the frequency of resonance. R determines the maximum value of impedance at resonance and the sharpness of the resonance. Impedance is inductive below resonance(phase ~ +90deg), and capacitive above resonance(phase ~ -90 deg).




If people find this type of spreadsheet of interest, I also have one for a complete model of a transformer driven ESL taking into account transformer parasitics.


I almost forgot to mention that the spreadsheet does use complex variable functions, so you will need to load the Analysis Toolpak Addin before it will work properly.
 

Attachments

  • Zvis_01.jpg
    Zvis_01.jpg
    141 KB · Views: 276
  • Zvis_02.jpg
    Zvis_02.jpg
    142.8 KB · Views: 274
  • Zvis_03.jpg
    Zvis_03.jpg
    142.3 KB · Views: 260
  • Zvis_04.jpg
    Zvis_04.jpg
    146.5 KB · Views: 263
  • Zvis_05.jpg
    Zvis_05.jpg
    144.4 KB · Views: 261
  • Z_visualizer.zip
    209.1 KB · Views: 80
Last edited:
Very cool spreadsheet.
The model for a transformer driven ESL sounds interesting. :)

Attached is a *.zip file containing an Excel 2004 compatible and Excel 2007+ compatible spreadsheet for modeling impedance and response of an ESL connected to a step-up transformer. Also attached is the specification sheet for the Plitron 4314 transformer which I will post a couple sample calculations for.

Lower left plot shows impedance magnitude and phase.
Lower right plot shows normalized output response of voltage applied to ESL stators.
Upper right plot shows the impedance vector diagram for marker frequency.

Note that Rc(core losses) and Lp(primary inductance) are in reality a function of input voltage magnitude and frequency. In this spreadsheet they are constants. So not perfect modeling, but pretty good. Rc is not specified in the Plitron data. I estimated it at 500 ohm. This sets the peak magnitude of the impedance peak for the parallel resonance between Lp and (Cw + Cesl)*N^2.

Pic #1: Plitron 4314 with 1000pF ESL and recommended Rdamp = 0.8 ohm
You can see that the impedance and response match well with page 2 of the Plitron data sheet.

Pic #2: Plitron 4314 with 1000pF ESL and Rdamp = 0.2 ohm
The recommend value of 0.8 ohm gives you a maximally flat HF electrical response(stator voltages). If a lower value is chosen for Rdamp, the under-damped series resonance between LL and (Cw + Cesl)*N^2 causes a peak in the HF response. Also note that HF impedance drops and the phase angle gets even more negative (ie capacitive) in the top two octaves.

Pic #3: Plitron 4314 with 1000pF ESL and Rdamp = 1.5 ohm
If a higher value is chosen (Rdamp = 1.5 ohm) the HF electrical response droops a bit. But, since the native response of an ESL is a rising response (+3dB/oct for line source, +6dB/oct for point source) this is actually closer to the value needed for a flat acoustic response. Not that the minimum impedance has risen and the phase angle is reduced.

Pic #4: Plitron 4314 with 1000pF ESL and Rdamp = 1.5 ohm, Rshunt = 16 ohm
Adding a shunt resistance before Rdamp does not change the response. But, it does make the impedance load the amplifier sees more resistive. If used, this resistor would need to be able to dissipate considerable power/heat.

Like before, the spreadsheet does use complex variable functions, so you will need to load the Analysis Toolpak Addin before it will work properly.
 

Attachments

  • ESL_4314_pic_03.jpg
    ESL_4314_pic_03.jpg
    467.8 KB · Views: 231
  • ESL_4314_pic_02.jpg
    ESL_4314_pic_02.jpg
    470 KB · Views: 231
  • ESL_4314_pic_01.jpg
    ESL_4314_pic_01.jpg
    465.5 KB · Views: 236
  • ESL_4314_pic_04.jpg
    ESL_4314_pic_04.jpg
    465.8 KB · Views: 224
  • ESL_Z_dB_model.zip
    204.7 KB · Views: 112
  • plitron_pat4134-ES.zip
    410.2 KB · Views: 94
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.