An attempt at a constant dipolar directivity esl panel

The following graphs result from running the segments each as second order rc-filter with segment voltage gain set to 0.1 at the frequency where lambda/2 is the panel diameter. This improves directivity. Whether those curves are viable regarding sensitivity, tweeter excursion capability and phase behavior I'm not able to model yet... At 180hz the outer panel is already down to a quarter of the input voltage for example; so the tweeter would do four times the excursion...
 

Attachments

  • 2ndOrderRC.png
    2ndOrderRC.png
    80.9 KB · Views: 253
  • 2ndOrderRCPhase.png
    2ndOrderRCPhase.png
    97.2 KB · Views: 257
  • 2ndOrderRCImpedance.png
    2ndOrderRCImpedance.png
    22.2 KB · Views: 256
The new curves translate to way improved directivity (example sims again at 5kHz left, 10kHz right). I still don't know about sensitivity and max spl regarding excursion...
 

Attachments

  • 5kHzDirectivity.jpg
    5kHzDirectivity.jpg
    129.9 KB · Views: 221
  • 10kHzDirectivity.jpg
    10kHzDirectivity.jpg
    129.9 KB · Views: 216
… May I ask, how did you generate that directivity pattern? (expert software, analytical insight?)
The directivity patterns were generated using an Excel spreadsheet calculation of the generalized Walker Equation. I use a VBA macro to step thru different angles and output impulse response files that can be read into ARTA for plotting directivity. If you are not familiar with the Walker Equation and/or ESL operation, you might find this document useful. http://home.kpn.nl/verwa255/esl/ESL_English_2011.pdf

Could you make a directivity pattern of a simple rectangular panel of same proportion for comparison?
Attached is horizontal directivity pattern for your 34cm wide rectangular panel if it was driven uniformly.

… I can't think of an easy way to steepen the low-pass passively after the transformer.
The easiest way, is to use a resistive ladder network(ie RC transmission line). As you work outward in segments, you retain the roll-off from the previous segment and add an additional 1st order slope. I’ll post an example shortly.
 

Attachments

  • Polar_34cm_width.png
    Polar_34cm_width.png
    336.4 KB · Views: 108
The following graphs result from running the segments each as second order rc-filter with segment voltage gain set to 0.1 at the frequency where lambda/2 is the panel diameter. This improves directivity. Whether those curves are viable regarding sensitivity, tweeter excursion capability and phase behavior I'm not able to model yet... At 180hz the outer panel is already down to a quarter of the input voltage for example; so the tweeter would do four times the excursion...

The Walker Equation for ESLs can relate pressure in the far field to panel dimensions and applied voltages:
P = e0 * Vsig * Vpol * A * f / (2 * d^2 * c * r)
Where:
P = sound pressure (N/m^2)
Vsig = signal voltage applied to stators
Vpol = bias voltage applied to diaphragm
A = area of ESL (m^2)
f = frequency (hz)
d = stator to diaphragm spacing (m)
c = speed of sound (m/s)
r = listening distance (m)
e0= electric constant

Note that:
- with constant stator voltage, far field pressure is proportional to frequency (ie increases +6dB/oct)
- there is a limit to the voltage that can be applied before the air in the gap will ionize and conduct and/or arc. This puts a ceiling on the max SPL that can be generated for a given panel area. Increasing the gap size does not increase the max SPL, it just increased the voltages required to achieve that SPL. If interested, more details can be found in the Baxandal ESL chapter.

P. J. Baxandall, “Electrostatic Loudspeakers,” in
Loudspeaker and Headphone Handbook, J. Borwick, Ed.
(Reed Educational and Professional Pub., Woburn, MA,
1998), pp. 106–195.

Also here is a link showing what happens when you exceed the ionization voltage gradient in the gap,
and a link on how to set optimum bias voltage.
First time ESL builder
Measured my ESL attempt


Attachment #1 Directivity Plot and individual segment contributions to SPL @ 1m on-axis for the filtering shown in Post #5

Attachment #2 Directivity Plot and individual segment contributions to SPL @ 1m on-axis for the filtering shown in Post #21

Attachment #3 Directivity Plot and individual segment contributions to SPL @ 1m on-axis for an example ladder resistor network. This is by no means optimized, but you can see that usefully broad polar can be achieved. In this case, I actually combined the inner 2 segments since the output from the first segment was so small on it’s own there wasn’t any benefit to separating it.

*** In attached plots, an estimate of the max possible SPL @ 1m on-axis for unsegmented panel is shown as a dashed black line.
Also note that I set a notional diaphragm resonance of 30Hz with Q=4
 

Attachments

  • Polar_KalleMan_Post5.png
    Polar_KalleMan_Post5.png
    131.2 KB · Views: 116
  • Polar_KalleMan_Post21.png
    Polar_KalleMan_Post21.png
    125.3 KB · Views: 107
  • Polar_KalleMan_Ladder_Example.png
    Polar_KalleMan_Ladder_Example.png
    127.2 KB · Views: 99
Last edited: