
Home  Forums  Rules  Articles  diyAudio Store  Blogs  Gallery  Wiki  Register  Donations  FAQ  Calendar  Search  Today's Posts  Mark Forums Read  Search 
Software Tools SPICE, PCB CAD, speaker design and measurement software, calculators 

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 
3rd May 2019, 07:55 PM  #1181  
diyAudio Member

Quote:
 your coefficient for 0°/0° is (approximately) 0.0004758 / 2 = 0.0002379  your coefficient for 180°/180° is (approximately) 0.0004758 / 2 / 2 = 0.00011895 This means that the sum in your code includes the 0°/0° point twice, and the 180°/180° point is included four times for some reason. Correct?
__________________
 Group buy for augerpro waveguides for Scan Speak tweeters  elliptical version  circular version 

3rd May 2019, 08:35 PM  #1182 
diyAudio Member
Join Date: Oct 2005
Location: Kuopio

Correct.

3rd May 2019, 10:03 PM  #1183 
diyAudio Member

Ok, so I tried the Tylka calculation using your coefficient values from post 1177. To this end I first tried with 0.000237... for the last coefficient for the 180° point, and then I also repeated with 0.000118... . Both results were very close to the one with the original Tylka calculation (less than 0.1 dB difference everywhere). This is not surprising, as your coefficient values are essentially identical to the Tylka values.
This means that although my Tylka code and Vituix use the same coefficients, the result is not the same. Therefore Vituix and my Tylka code therefore don't implement the same calculation. I attached the formula I used for the Tylka calculation. How is this different from the Vituix calculation? Note: the w_n correspond to the coefficients as discussed in post 1177.
__________________
 Group buy for augerpro waveguides for Scan Speak tweeters  elliptical version  circular version 
4th May 2019, 05:24 AM  #1184 
diyAudio Member
Join Date: Oct 2005
Location: Kuopio


4th May 2019, 05:35 AM  #1185 
diyAudio Member
Join Date: Oct 2005
Location: Kuopio

PS1. Just want to make sure that you calculate VCAD results with the main program  not with Calculator tool which uses the simplest weighting with sin() function i.e. 0 deg and 180 are totally ignored.
PS2. The latest built of 2.0.15.3 (downloadable since yesterday evening) uses Archimedes' hatbox theorem cos(a)cos(b) for area calculation. Difference to old versions such as obsolete 1.1 is insignificant, visually undetectable with default dB scale. 
4th May 2019, 06:21 AM  #1186 
diyAudio Member
Join Date: Oct 2005
Location: Kuopio

Difference between simple sin() and area (or integral of sin) weighting is quite small if angle interval is only 5 deg:
In practice, amount of offaxis angle measurements is the most significant factor in addition to limit of dual planes only. Curve fitting would work better than weighting by sector factors but that would reduce calculation performance. 
4th May 2019, 05:33 PM  #1187  
diyAudio Member

Alright, so now at least I know the formula used in Vituix. So the "How does it work?" question is now settled.
With respect to the question "Is the result correct?", here are some updates: Quote:
There is some progress on my side. In my code, I screwed up the allocation of the data files to the corresponding H_m,n terms where the angles wrapped around the hemispheres (i.e., at the rear of the speaker). The Vituix convention is slightly different to what Tylka uses, and I got confused by this. While the w_n coefficients were right all the time, they were not always multiplied with the correct SPL curves (H_m,n terms). I fixed this now. I have also tested my code by artificially setting all H_m,n to zero with the exception of one (n=k), which was set to 1:  H_m,n = 1 for n = k  H_m,n = 0 for all other n By looking at the summation (see post 1183) you will find that whatever the choice of k is, the resulting power response (before conversion to dB) will be equal to 2 x w_k. Implementing this in the code is easy, and I have done this test for all possible k values (with a 5° angle step, there are 72 possibilities). The code always gave the correct result. I am therefore quite confident that it works correctly now. The power response curve from the fixed code is now closer to the Vituix curve than before, but there is still some discrepancy of almost 1.5 dB above 10 kHz (see attached plot). I don't know if this is just a difference related to the numerical techniques, or if it's something more conceptual.
__________________
 Group buy for augerpro waveguides for Scan Speak tweeters  elliptical version  circular version 

4th May 2019, 06:26 PM  #1188 
diyAudio Member
Join Date: Oct 2005
Location: Kuopio

That basic part was already explained 3 days ago and exception logic for weighting factors 2 days ago. But I guess Greek hieroglyphs suit better than my bad English flavored with minor amnesia. I'm done with this item.

5th May 2019, 05:34 AM  #1189 
diyAudio Member
Join Date: Oct 2005
Location: Kuopio

Rev. 2.0.15.4 (20190505)
Diffraction * Effect of Floor and Wall boundary reflections reprogrammed. New calculation method enables easy adding of new boundaries such as front wall and ceiling, which might be added later. Enclosure * Added name and Vd[cm3] readonly text boxes to Driver configuration group. Name saved to user settings and enclosure project file. * Added name and Vd[cm3] readonly text boxes to Passive group. Name saved to user settings and enclosure project file. 
5th May 2019, 02:43 PM  #1190 
diyAudio Member
Join Date: Apr 2011

Hi Kimmosto:
Regarding front wall and ceiling reflections, I would love to see them. I immediately tried out the new revision but the changes apparently aren't apparent to the user. As you may recall I tried simulating them for a line array via image method in the main program but there were multiple issues that you noted causing me to abandon that approach. But floor and ceiling reflections are critical to line array performance in a home environment so I have gone as far as I can with them in the current program and would appreciate your help going further. The goals are not just to understand line array behavior but to be able to develop shading coefficients without endless cut and try or audible inlivingroom tweaking. Any improvement to the accuracy of these sims is welcome but especially if it would shed light on why the comb filtering seen in simulations of line arrays is usually worse than seen in measurements. Looking at geometry of a 2m tall line array, one needs high resolution vertical directivity only within 0 to 27 degrees for the main array and its ground reflection or out to 37 degrees if including a ceiling reflection. Diffraction tool only goes down to 5 degrees but I am able to import piston directivity from HornResp down to 1 degree. It would be nice if diffraction tool would do this for me as that takes some time. Its not practical for me to do 1 degree measurements but next time out I will get down to 5 degrees in this critical region. I wonder if there is a way to interpolate between 5 or 10 degree measurement data in some external tool then import or if that feature could be added to calculator. Now I'm adding ground reflection in main program by the simple expedient of adding 2x number of drivers. This puts the vertical axis on the floor, which is not ideal, but it does give me answers. I can't add ceiling reflection this way because baffle height seems to be limited to 2m. If you would raise that limit (3m would be sufficient) then I could add the ceiling reflection. Simulation performance is not yet an issue. There is a lag to recompute but only a few seconds. Looking at reflections in diffraction tool is certainly easier. In the main program there is more flexibility to add frequency dependent reflection coefficients in an attempt to correlate with measurements. I will shortly be able to do measurements of a full line array so this will be interesting. Thanks, Jack
__________________
My Synergy Corner Horns and Bass Bins http://www.diyaudio.com/forums/multi...bassbins.html 
Thread Tools  Search this Thread 


New To Site?  Need Help? 