experiences with ESL directivity?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I should have time to post an updated spreadsheet in a few days which includes the symmetric layout.

Here is Version 2 of the spreadsheet.
Changes are:

1) Added formulas for SPL calcualtions to Notes TAB.
2) Added Reference for line termination to bottom of Directions TAB.
3) Added 3 TABs showing schematics for asymmetric and 2 possible symmetric configurations.

Each of the configuration TABs show R & C values for single ended and symmetric drive of the ESL sections.
Most people use the symmetric drive layout when actually constructing ESLs.
However, for SPICE simulation, the single ended version is more convenient.

When simulating with SPICE, SPL is proportional to:
20*LOG[(total current) / sqrt(frequency)]


Oh, almost forgot to mention... I had protected each TAB just so you didn't accidently mess up one of the formulas. There isn't a password though, so you can unprotect each TAB by simply selecting [Unprotect Sheet] from the [Tools] menu. Then you can change or rearrange things as you see fit.
 

Attachments

  • ESL_line_sectioned_DIY_v2.zip
    31 KB · Views: 926
Making ESL's better for people not in the hot seat.

Having now owned Quad ESL 57's and Quad ESL 63's I have to say quads combination of filter and delay line in the ESL 63 is very impressive, for improving the treble dispersion. Unless you play with a pair of ESL 63's for a while you would not realize how good they are at treble dispersion because the bass is still very directive.

Having had the Quad ESL 63's for longer now, and experimented with active cross overs its worth giving you all an update.

I have now wired up a cheap bought active cross over at about 100Hz, and now the Quad ESL 63's with the sub woofer are more tolerable off axis. With a cross over of 400Hz the off axis sound is quiet good, improving on conventional speakers with smoothness of an ESL midrange tweeter and suitable for a dinner party, though not ESL perfect. The cross over at 100Hz in my opinion enhances the sound when sitting at the hot seat for best sound, the cross over at 400Hz sounds better when in a Dinner Party situation and is nicer enough that I am not tempted to use conventional speakers.

Before these discoveries I was considering having the ESL63's on Sculpture rotation plates:

 
Here is Version 3 of the spreadsheet.
The following functionality was added based on feedback from users:

1) Added ability to choose units for the dimensions
2) Added inputs for transformer step-up ratio, input voltage, and bias voltage
- among other things this will allow you to easily determine sensitivity @ 2.83Vrms.
3) Added selection for line source type(ideal, finite, finite on floor)
4) Added ability to model diaphragm resonance effect on response
5) Added calculations for parameters to model response with calculator esl_seg_ui
- see attachment below for description of how to copy and paste segment parameters

As with previous version, each TAB is protected so you don't accidently mess up one of the formulas. There isn't a password though, so you can easily unprotect each TAB and change or rearrange things as you see fit.

Oh, I almost forgot, this version uses a short macro code to populate the esl_seg_ui parameters. So when you open the spreadsheet, be sure to enable Macro content when prompted.
 

Attachments

  • Copy_to_esl_seg_ui.png
    Copy_to_esl_seg_ui.png
    112.4 KB · Views: 714
  • ESL_line_sectioned_DIY_v3.zip
    284.4 KB · Views: 940
Last edited:
Having had the Quad ESL 63's for longer now, and experimented with active cross overs its worth giving you all an update.

I have now wired up a cheap bought active cross over at about 100Hz, and now the Quad ESL 63's with the sub woofer are more tolerable off axis. With a cross over of 400Hz the off axis sound is quiet good, improving on conventional speakers with smoothness of an ESL midrange tweeter and suitable for a dinner party, though not ESL perfect. The cross over at 100Hz in my opinion enhances the sound when sitting at the hot seat for best sound, the cross over at 400Hz sounds better when in a Dinner Party situation and is nicer enough that I am not tempted to use conventional speakers.

Before these discoveries I was considering having the ESL63's on Sculpture rotation plates:


? should the of axis be pretty ok for this low frequencies ?? even for a quad. talking meters of wavelength here when talking about 100hz. the panels are only 60cm. only gain in my opinion is getting higher spl. And maybe increase dynamics. and maybe just maybe your sub is not open bafle and act as a omni directional at these low frequency's. but the quad itself wont change in response. it stay figure of 8 at these low frequency's.
 
Here is Version 3 of the spreadsheet.
The following functionality was added based on feedback from users:

1) Added ability to choose units for the dimensions
2) Added inputs for transformer step-up ratio, input voltage, and bias voltage
- among other things this will allow you to easily determine sensitivity @ 2.83Vrms.
3) Added selection for line source type(ideal, finite, finite on floor)
4) Added ability to model diaphragm resonance effect on response
5) Added calculations for parameters to model response with calculator esl_seg_ui
- see attachment below for description of how to copy and paste segment parameters

As with previous version, each TAB is protected so you don't accidently mess up one of the formulas. There isn't a password though, so you can easily unprotect each TAB and change or rearrange things as you see fit.

Oh, I almost forgot, this version uses a short macro code to populate the esl_seg_ui parameters. So when you open the spreadsheet, be sure to enable Macro content when prompted.

Niceeeeeee. an update! i dont get thbe 5 5 5 5 5 you wont get an error anymore when leaving the resistor values blank ?

EDIT aha i see its a excel sheet. thought it was the esl gui :) but great work anyhow! i always have problems having excel sheets to work proper somehow. remember the stacked sheet you gave me ? i could open it but it never calculated new values or plotted :( might be my excel is to old ? oh i got excel 2003 ...... that might be a problem i guess haha
 
Last edited:
I have Excel 2000 installed on one of my old test computers. I checked and both spreadsheets work fine, so should work with your Excel 2003. Remember the stacked sheet (ie multi-diaphragm) calculator requires the use of higher math functions so you need to have the Analysis Toolpak Add-In installed and active. If you are unsure about it, a quick web search for "excel 2003 analysis toolpak" will find you plenty of details on how to install and select/activate. The spreadsheet posted here does not require the Analsysis Toolpack. When you open it you should get a prompt to enable the macro. Once you click "Enable Macros", you should see the parameter input section and plot.


For those interested, the multi-diaphragm calculator mentioned was posted here for new and older version of Excel:
http://www.diyaudio.com/forums/planars-exotics/272742-sandwich-esl-yummie-3.html#post4292045
http://www.diyaudio.com/forums/planars-exotics/272742-sandwich-esl-yummie-3.html#post4292147
 

Attachments

  • macro_prompt.jpg
    macro_prompt.jpg
    145.6 KB · Views: 662
  • open-screen.png
    open-screen.png
    76.3 KB · Views: 668
I have been using the ESL-line-sectioned spreadsheet (v.3) but am unsure whether the D/S spacing to be entered includes or excludes the thickness of the stator wire insulation. I have searched but couldn't find a clarification of this point. Of course it makes slight difference to the result of the calculations.
 
The D/S spacing is between the diaphragm and the outer surface of the wire insulation.
I will put that on the list of things to add/clarify in v4 of the spreadsheet. ;)

Here is a post showing pics of two different sized wires and insulation thickness, but with the same D/S = 0.059" (1.5mm).
http://www.diyaudio.com/forums/planars-exotics/281348-sanity-check-wire-stator.html#post4492349

Hmm so it is only used for max excursion ? I mean the spacing between membrane and conductive stator will dictates the efficiency , that's kind of weird to take insulation since there is no set value for thickness for the insulation . So efficiency is random ?
 
I mean the spacing between membrane and conductive stator will dictates the efficiency , that's kind of weird to take insulation since there is no set value for thickness for the insulation . So efficiency is random ?
Remember that even though PVC is considered an insulator, it is MUCH more conductive than air both for DC bias voltage and AC stator signal voltage. The PVC bulk resistitivity and dielectric constant K are the determining properties. With this in mind when calculating SPL for wire ESLs built with PVC insulation, it is most appropriate to use the distance from the diaphragm to the outside of the insulation as D/S.

You can easily verify this by measuring the capacitance between the stators. You will find with varying insulation thickness, the capacitance measurement will correspond well with the flat plate capacitor calculation when you use the dimension between the outer diameter of the insulation, not the outer diameter of the copper conductor inside the insulation.
http://www.diyaudio.com/forums/planars-exotics/246846-first-time-esl-builder-12.html#post3881204

More discussion and examples on this topic can be found here:
http://www.diyaudio.com/forums/plan...truct-cube-louver-acoustat-7.html#post2154621
http://www.diyaudio.com/forums/planars-exotics/281348-sanity-check-wire-stator.html#post4492011
 
Remember that even though PVC is considered an insulator, it is MUCH more conductive than air both for DC bias voltage and AC stator signal voltage. The PVC bulk resistitivity and dielectric constant K are the determining properties. With this in mind when calculating SPL for wire ESLs built with PVC insulation, it is most appropriate to use the distance from the diaphragm to the outside of the insulation as D/S.

You can easily verify this by measuring the capacitance between the stators. You will find with varying insulation thickness, the capacitance measurement will correspond well with the flat plate capacitor calculation when you use the dimension between the outer diameter of the insulation, not the outer diameter of the copper conductor inside the insulation.
http://www.diyaudio.com/forums/planars-exotics/246846-first-time-esl-builder-12.html#post3881204

More discussion and examples on this topic can be found here:
http://www.diyaudio.com/forums/plan...truct-cube-louver-acoustat-7.html#post2154621
http://www.diyaudio.com/forums/planars-exotics/281348-sanity-check-wire-stator.html#post4492011

oh wow, thats a nice feature. since it still protects against arcing and such, without losing the spacing to insulation? hmmm very nice !! does it also holds with perf sheets that have some laquer? or only for PVC ?
 
oh wow, thats a nice feature. since it still protects against arcing and such, without losing the spacing to insulation? hmmm very nice !! does it also holds with perf sheets that have some laquer? or only for PVC ?
Indeed, PVC insulation has many desirable features for ESL stators, as outlined in the original JansZen patent that first promoted its use.
As far as coatings used on perf sheets, you would need to look at the electrical properties of the coating and compare them to PVC. In general, most coatings have as high a dielectric constant(good) as PVC, but not as low a bulk resistivity(bad) as PVC. Calvin mentioned in one thread that Nylon 66 is one particular coating that is as good if not better than PVC. Use of polyamide coatings such as nylon were also suggested in one of the Beveridge patents because of their low bulk resistivity.
 
Hi,

a coating functions like a second capacitor in series with the 'air-capacitor'.
As such they also form a voltage divider.
Thats the reason why low-k high-R materials cost on efficiency ... voltage loss over the dielectric,
When the k is high, the insulating layer can be made thicker, as the material acts like a field conductor, increasing field strength in the air capacitance, but still achieving high flashover tresholds due to its thickness.

Fluidized bed coating won't work with metal sheet stators ... at least not as single procedure.
The heat capacitance of a metal sheet is too low.
It cools down too fast for the powder to melt completely.
At least a additional heating/melting processes would be required with the problem of the coating beginning to flow again.
Its a process to produce thick coatings on big parts (offshore, oil rigs) that once heated keep the temperature over a longer time.

jauu
Calvin
 
Here is Version 4 of the spreadsheet.
The following functionality was added based on feedback from users:

  • Parameter values now auto convert when you switch units.
  • Damping section has been expanded to allow specifying either resonance Q or acoustic resistance of damping mesh.
    Whichever parameter you specify, the other is calculated and displayed.
  • An estimate of peak diaphragm excursion is plotted for frequencies below fL where the entire diaphragm area is uniformly driven.
  • Response curves are now continuous rather than straight line approximations connecting the break points.
    Many thanks to golfnut for his math wizardry in deriving and simplifying the equations.
  • Added input to specify number of wires per section to use for the esl_seg_ui import data.
As with previous versions, each TAB is protected so you don't accidently mess up one of the formulas. There isn't a password though, so you can easily unprotect each TAB and change or rearrange things as you see fit. Changing any blue input cell, will re-protect the main input TAB.

Let me know if you have any questions or if something doesn’t seem to be working properly.
 

Attachments

  • ESL_line_sectioned_DIY_v4.png
    ESL_line_sectioned_DIY_v4.png
    98.6 KB · Views: 907
  • ESL_line_sectioned_DIY_v4.zip
    334.1 KB · Views: 500
Last edited:
Here is Version 4 of the spreadsheet.
The following functionality was added based on feedback from users:

  • Parameter values now auto convert when you switch units.
  • Damping section has been expanded to allow specifying either resonance Q or acoustic resistance of damping mesh.
    Whichever parameter you specify, the other is calculated and displayed.
  • An estimate of peak diaphragm excursion is plotted for frequencies below fL where the entire diaphragm area is uniformly driven.
  • Response curves are now continuous rather than straight line approximations connecting the break points.
    Many thanks to golfnut for his math wizardry in deriving and simplifying the equations.
  • Added input to specify number of wires per section to use for the esl_seg_ui import data.
As with previous versions, each TAB is protected so you don't accidently mess up one of the formulas. There isn't a password though, so you can easily unprotect each TAB and change or rearrange things as you see fit. Changing any blue input cell, will re-protect the main input TAB.

Let me know if you have any questions or if something doesn’t seem to be working properly.

GREAT! thats a keeper ! might come in handy, thanks allot for all the hard works once again !
 
Also opens and looks beautiful in Mac OSX both in "Excel" and in "Numbers" (which I almost always prefer). But I haven't confirmed all the functionality is functioning.

Of course, there are the usual warnings from Microsoft, buggy and otherwise. Excel for example warns "File error: Data may have been lost".

B.
 
Last edited:
Here is Version 4 of the spreadsheet.
Let me know if you have any questions or if something doesn’t seem to be working properly.

I need a sanity check on my interpretation of the LF break point and it's correlation to the crossover frequency in a hybrid application:

Is it correct to say that the LF break point is effectively the upper frequency limit of the last segment in the transmission line (i.e... the last wire group(s) in the series)?

If so; the crossover frequency should be lower than the LF break point; else the crossover would chop off the signal above the upper limit of the latter wire group(s) in the series.

Did I get that right?

Lastly; can you recommend some general guidance as to where the crossover frequency needs to be in relation to the LF break point (or visa versa)?
 
Last edited:
Also opens and looks beautiful in Mac OSX both in "Excel" and in "Numbers" (which I almost always prefer). But I haven't confirmed all the functionality is functioning.
Unfortunately you will find "Numbers" does not handle the VBA code properly.
You should be able to use Version 2 of the spreadsheet just fine though as it has no VBA code.

Of course, there are the usual warnings from Microsoft, buggy and otherwise. Excel for example warns "File error: Data may have been lost".
As much as I like to complain about Microsoft and Apple software, I believe that error is due to uninstalled service packs. The spreadsheet may work just fine even with the error...can't tell you for sure though. If you let me what version of Excel you are running, I can email you a copy saved for that version that should open up fine.
 
…the crossover frequency should be lower than the LF break point; else the crossover would chop off the signal above the upper limit of the latter wire group(s) in the series. Did I get that right? …can you recommend some general guidance as to where the crossover frequency needs to be in relation to the LF break point (or visa versa)?
Sounds like you got it right. Basically fL is the frequency at which all segments or groups of wires have achieved their maximum SPL contribution. The last segment is the last to have its output rolled in, but the final octave shows all segments reaching there maximum output pretty much together. The upper plot in the esl_seg_ui window shows the magnitude of the voltages applied to each of the individual segments by the ladder resistor network. You can see that at fL, all segments are receiving full voltage, and the response has started to follow the -3dB slope of a dipole line source ESL. So, if you set fL lower than fX(crossover frequency) you will just be throwing away sensitivity.

General guidance is to set fL = fX, or at most fL = fX * 1.15.
If your diaphragm resonance is undamped and within 1.5 octaves of fX, you will see some response lift at fX from the resonance. So, you could move fL up in frequency by as much as a factor of 1.5 and gain 1 - 2dB of efficiency while retaining the response level at fX. Of course with resonance so close to fX you would need to use a notch filter in combination with the crossover to avoid over driving the diaphragm at resonance. I seem to recall you already having experience with this.

***
Found what I was thinking of: Thinking about a segmented wire stator ESL - Post#186 & #187
 

Attachments

  • fL_for_Fx.png
    fL_for_Fx.png
    208.9 KB · Views: 791
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.