
Home  Forums  Rules  Articles  The diyAudio Store  Gallery  Blogs  Register  Donations  FAQ  Calendar  Search  Today's Posts  Mark Forums Read  Search 
Planars & Exotics ESL's, planars, and alternative technologies 

Please consider donating to help us continue to serve you.
Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving 

Thread Tools  Search this Thread 
29th May 2010, 06:11 AM  #51 
diyAudio Member
Join Date: Oct 2009


30th May 2010, 08:20 AM  #52 
diyAudio Member

Hi, real life got a little bit in the way. Here's a quick summary I made of the paper, let me know if something is unclear, then I'll try to elaborate!
Kenneth * Walker's equation for onaxis SPL is linear with the signal current and has a 1/sqrt(f) dependency * To use voltage drive and get a flat frequency response, we therefore need an impedance which is inverse proportional to the square root of the frequency. An RC transmission line satisfies this requirement. * Since we use a finite number of sections, there are 3 operating areas: 1. Low frequencies: unattenuated by the transmission line, whole surface area used. Here the impedance becomes capacitive (1/f), and SPL becomes proport. to sqrt(f) so it rolls off at 10dB/decade. 2. Mid frequencies: desired flat response. 3. High frequencies: only the first RC section is active, the impedance is predominantly resistive and we get SPL proport. to 1/sqrt(f) so it rolls off at 10dB/decade. * If N is the number of RC sections:  the LF transition frequency is 1/(2 Pi N^2 RC)  the HF transition frequency is 2/(Pi RC)  the number of sections needed to have flat behavior from fL to fH is given by fH/fL = 4N^2 * For a good zerofree polar response, the segments should be no more than 1012mm wide. This was determined by simulation. Hence, for a fullrange ESL, 30 or more sections may be needed. * Asymmetric ESL (RC line driven from one side of the speaker) has a very poor feature: the location of the peak in the polar response is a little offaxis and this is dependent on the frequency. Symmetric ESLs are much better: they obviously don't have this error. Also, the author shows that they have an inherent Butterworth (optimally flat) 2ndorder lowpass response in their polar response, giving a better sweet spot and very linear response near onaxis. Of course, symmetric construction doubles the number of sections in practice. * With a segment width of w, a flat and linear polar response is obtained up to this frequency: (2RC c^2)/(Pi w^2) with c=speed of sound in air * Terminating the transmission line with its Zc helps in reducing a "bump" in the response near fL but at the cost of a slightly higher fL. It is better to terminate the line with an impedance simulating between 0.1N and 0.15N additional segments.
__________________
Never send a human to do a machine's job. Agent Smith Last edited by kavermei; 30th May 2010 at 08:25 AM. 
30th May 2010, 01:55 PM  #53 
diyAudio Member
Join Date: Oct 2009

Kenneth,
if it's all impedance wise 20 sections per panel half is relatively good. Below are calculated plots magnitude of impedance vs frequency @ 20,40,80 sections. It really unfortunate that the resistance per cell has to be quite high at low ladder length. Alex Last edited by alexberg; 30th May 2010 at 02:01 PM. 
30th May 2010, 09:34 PM  #54  
diyAudio Member
Join Date: Apr 2004
Location: Maine, USA

Kenneth,
Thanks for the very useful contribution to this (old) thread. I've only done quick calculations based on bits and pieces of your summary because I don't have capacitance values to work with yet. If I start with this: Quote:
Quote:
Few 

30th May 2010, 10:16 PM  #55  
diyAudio Member

Hi,
Quote:
Quote:
The 'ideal' sectioning would be an infinite number of infinitely narrow sections Think stators with a resistive coating, fed from a central strip To continue with your example: in an asymmetrical ESL with 4 sections, yes (according to this paper) you would end up with 48mm max if you want a zerofree polar response. Whether or not such an ESL could generate sound down to 300Hz is an additional requirement which needs to be checked on a casetocase basis. Personally I think it is very possible to make such an ESL. If the ESL is made symmetrical (recommended), we get 7 sections (centerfed) and a width of 84mm. Other important remark: the paper discusses the low frequency 3dB point fL due to sectioning, but in practice you'll have another fL due to dipole baffle step (frontback cancellation). The former is determined by the number of sections, and the fact that the RC transmission line no longer acts like a transmission line below that frequency. The latter is determined by the width of the speaker which is quite unrelated from N. Then there is a third unrelated effect that the SPL output drops down below the membrane's mechanical resonance frequency. Whew! HTH Kenneth
__________________
Never send a human to do a machine's job. Agent Smith Last edited by kavermei; 30th May 2010 at 10:18 PM. 

31st May 2010, 04:39 AM  #56  
diyAudio Member
Join Date: Jun 2005
Location: Fort Worth, Texas

Quote:
As you pointed out, the effect of diaphragm resonance on the response is not considered by the paper. Neither is the matter of HF resonance between leakage inductance and panel capacitance which is a dominate factor in defining the HF rolloff behavior. The matter of capacitive coupling between segments is also ignored. In practice more high frequency energy couples to the outer sections than the theory assumes, bypassing the transmission line resistors. However, the dipole rolloff behavior of a line source is contained in the linesource version of the Walker equation. For floor to ceiling line sources, the native response falls 3dB per octave with decreasing frequency. You do not have to account for it separately 1) If your line source is not fully floor to ceiling, you will get an additional 3dB rolloff below the frequency where the height of the line source is half wavelength. Thus 6dB/octave decreasing below this breakpoint frequency. See the AES paper "The Acoustic Radiation of Line Sources of Finite Length" by Lipshitz & Vanderkooy for more details. 2) If you add baffle wings to the sides of the a line source ESL panel you will halt the 3dB rolloff slope from frequency f1 where half wavelength equals the width of the ESL panel until the f2 where half wavelength equals the width of the (ESL panel + baffle). Below f2 response reverts back to the 3dB/octave slope. 

31st May 2010, 04:57 AM  #57 
diyAudio Member
Join Date: Oct 2009

Thanks for sharing such an important info.
I am still confused... What are we trying to obtain? 1. 1/Sqrt(f) to equalize the response? According to simulations 4 sections are not even near 2. Time delay to enhance off axis response (eliminate beaming)? Walker defines the purpose as a formation of a particular wavefront driving concentric rings through the dealy line. Then the ratio of phase/frequency have to rize propotionally to the frequency applied. Again by simulation 4 sections have decent accuracy for 1 decade, let's say 200Hz2000Hz 3. Time dealy of the line does not depends on frequency. Or is it? Let's assume we applied step to the input (central strip). This step will appear on each tap (more or less distorted) with the dealy depending on propagation speed of a given line. I have to read the original  that's my verdict. Thanks, Alex P.S. It's definetely a grey area: http://en.wikipedia.org/wiki/Phased_array_ultrasonics http://www.explorelearning.com/index.cfm?method=cResource.dspDetail&ResourceID=37 Last edited by alexberg; 31st May 2010 at 05:00 AM. 
31st May 2010, 05:05 AM  #58  
diyAudio Member
Join Date: Jun 2005
Location: Fort Worth, Texas

Quote:
When designing a panel, I'd suggest starting out by: 1) defining the largest panel(height & width) you can live with. This will give you the highest sensitivity and maxSPL. 2) Then, pick your LF breakpoint. With area already defined in 1), this also sets the max SPL for your design. 3) Figure how many sections you need to reach beyond 20kHz. 4) If your sections are already 1/2" or whatever you deem "small enough" then you are done. Otherwise, increase the number of sections until you get there. Your high frequency extension will be much higher then needed, but the leakage inductance of your transformer will most likely cut of the HF response at not much higher than 20kHz. Another option is to only use the "small enough" section width on the first few sections of the transmission line and then increase the size of the section after that. If you look at the frequencies fed to each section you can see that as you travel down the line you pretty quickly lose the highs that would cause beaming problems. You just need to adjust the size of the transmission line resistors when you reach these sections. I had thrown together an XL spreadsheet a few months back to allow fiddling with panel size, LF breakpoint, and number of sections to see how they affected efficiency and max SPL. It also calculated sections sizes, capacitance, and feed resistor sizes. Unfortunately looks I left it at work. I do have a screenshot I'd saved at one time shown here so you can get an idea what I am talking about. Inputs 1 thru 4 define panel size and listening distance. This defines the max obtainable SPL for the line source seen in blue on the plot. Note the 3dB/octave slope as defined by the linesource version of the Walker Equation. Inputs 5 & 6 define the LF breakpoint and number of sections. The resulting response is shown in red on the plot. I'll be back to work on Tuesday, and should be able to post the spreadsheet for you to experiment with then. 

31st May 2010, 05:34 AM  #59  
diyAudio Member

Hi,
Quote:
P = I(signal) * V(pol) / (2 Pi d h sqrt(c r f) ) Quote:
Kenneth
__________________
Never send a human to do a machine's job. Agent Smith 

31st May 2010, 05:58 AM  #60  
diyAudio Member

Quote:
Quote:
The paper contains a formula for polar response based on 3 effects: 1. cos(theta) offaxis attenuation because of dipole 2. sinc(A f sin(theta) ) attenuation (A is a constant) because output from different segments arrives outofphase at the offaxis listener position 3. complicated factor containing sin(theta)^(2) because the output of different segments has different time delay due to the RC line. Kenneth
__________________
Never send a human to do a machine's job. Agent Smith 

Thread Tools  Search this Thread 


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Constant Directivity Horn EQ  Horizons  MultiWay  4  19th December 2008 11:41 AM 
Constant directivity EQ  Saurav  MultiWay  11  19th May 2008 02:41 AM 
Constant directivity horn  or waveguide?  hasselbaink  MultiWay  6  1st April 2008 01:24 PM 
A question on directivity  swak  MultiWay  18  31st July 2005 03:03 PM 
Open baffles and directivity  Yuihb  MultiWay  18  27th February 2004 08:22 PM 
New To Site?  Need Help? 