Acoustic Horn Design – The Easy Way (Ath4)

mabat · 2021-09-16 7:02 pm

To follow the OS coordinates the wavefronts as a whole should form ellipses. I'm not sure that can happen with the vanes in place...

- Hmm, but the points on the ellipses of the OS coordinates are not equidistant from the origin, are they. Maybe that's the problem.

mabat · 2021-09-16 7:32 pm

These are the OS coordinates, and after all, I think this is exactly what we see - by the time a wave in the blue vane reaches the ellipse at u=0.6, a wave in the yellow vane will be already slightly beyond that line.

(I could as well choose neighbouring vanes but the path length difference would be too small to be obvious.)

gedlee · 2021-09-17 1:00 am

Seems to me that you are proving my point. For the zeroth mode the red lines are wavefronts - each channel is in-phase. Those channels are equal distance at .6. One does not "lead" the other as I see in the sims.

gedlee · 2021-09-17 1:03 am

mabat said:
To follow the OS coordinates the wavefronts as a whole should form ellipses. I'm not sure that can happen with the vanes in place...

- Hmm, but the points on the ellipses of the OS coordinates are not equidistant from the origin, are they. Maybe that's the problem.

The average distance from the origin, which is a line, not a point, to each point on the ellipse is equidistant. That's how one draws an ellipse - the old pencil and string technique.

mabat · 2021-09-17 5:37 am

No. Imagine the vanes starting at 0 with separated sources. The average path lengths inside the channels are obviously different and that's what is effectively happening here, IMO.

AllenB · 2021-09-17 7:00 am

This should appear as an animated gif showing with and without the vanes.

mabat · 2021-09-17 7:16 am

I think we can leave the notion that the OS contours alone provide such feature but otherwise I still like the vanes principle. All (or "all") that seems to be required is to adjust the channels to be more aligned length-wise at the summing points. Shapes of the individual channels seem to matter much less as long as they stay narrow.

mabat · 2021-09-17 9:19 am

I was thinking about making the channels simply "meandering", separating them - I would keep the center channel as it is and add more and more of a smooth meander towards the outer one. Should not be hard to implement nor make.

miero · 2021-09-17 9:28 am

Strange behavior is simulated on the waveguide outside.

BTW, Paul has replied on the Github:

The first one just seems like there are different standing waves in the various sections. The second one certainly seems incorrect. I don't think I'm simulating angled walls properly with acoustic waves, but I'm not sure how I would fix it.

Source: sound wave propagation speed issue * Issue #4 * pfalstad/ripplegl * GitHub

mabat · 2021-09-17 9:32 am

That's because of reflections at the screen border, that's just the behviour of the applet - you should always stop the sim before the waves reach screen boundaries.

Well I think the simulations are actually correct in the big picture, however strange it may seem.

miero · 2021-09-17 9:53 am

No, this applet does not reflect from the borders.

Here is a earlier screenshot of the same situation.

It nicely shows how a waveguide converts planar wave to the spherical one, because the wavefront in the center of the waveguide is aligned with the wave outside of waveguide. I have never realized that earlier

To me, it also looks that the simulation tool works. Maybe vanes could be adjusted to generate a nicer wavefront.

mabat · 2021-09-17 9:56 am

It definitely reflects off the screen boundaries - try placing a sole point source in the middle of a blank screen and keep it running... no waves should ever return back. See what you get.

miero · 2021-09-17 9:59 am

I cannot confirm this in the current version of Ripple Tank Simulation

mabat · 2021-09-17 10:03 am

OK, that's nice! It seems it has been fixed in the meantime, which would be great.
I certainly remember it behaved differently.

Could you post the code of the WG sim above?

miero · 2021-09-17 10:10 am

I just extended source from your second example by 400 units to both sides:
"S 2 251 701 251 1299 0 0.08186844405594404 0 10.000178395549446 99.9998652719957 1 0"
changed to
"S 2 251 301 251 1699 0 0.08186844405594404 0 10.000178395549446 99.9998652719957 1 0"
but you can use more for longer source:
"S 2 251 0 251 2000 0 0.08186844405594404 0 10.000178395549446 99.9998652719957 1 0"

miero · 2021-09-17 10:17 am

And try to remove also the back wall of the source, it seems to help to generate nicer wavefront.

miero · 2021-09-17 10:43 am

Just a single wall, no reflections that would boost the fading wave.

Looking on the patterns on the other side, I'd guess that the width of a vane would be important too.

$ 3 2000 125 0 0 604 0.00037463
S 2 0 0 0 2000 0 0.08186844405594404 0 10.000178395549446 99.9998652719957 1 0
w 0 250 700 254 700
w 0 254 700 258 700
w 0 258 700 263 700
w 0 263 700 268 699
w 0 268 699 274 699
w 0 274 699 281 698
w 0 281 698 287 698
w 0 287 698 294 697
w 0 294 697 302 696
w 0 302 696 311 694
w 0 311 694 320 692
w 0 320 692 329 690
w 0 329 690 339 687
w 0 339 687 349 684
w 0 349 684 360 680
w 0 360 680 373 676
w 0 373 676 385 671
w 0 385 671 398 665
w 0 398 665 412 659
w 0 412 659 427 652
w 0 427 652 442 644
w 0 442 644 459 634
w 0 459 634 475 625
w 0 475 625 493 614
w 0 493 614 512 602
w 0 512 602 531 589
w 0 531 589 550 576
w 0 550 576 572 560
w 0 572 560 593 544
w 0 593 544 616 527
w 0 616 527 639 508
w 0 639 508 663 489
w 0 663 489 690 468
w 0 690 468 715 447
w 0 715 447 742 424
w 0 742 424 770 400
w 0 770 400 798 375
w 0 798 375 827 350
w 0 827 350 858 322
w 0 858 322 887 296
w 0 887 296 919 267
w 0 919 267 952 237
w 0 952 237 984 207
w 0 984 207 1017 177
w 0 1017 177 1049 146
w 0 1049 146 1085 113
w 0 1085 113 1118 82
w 0 1118 82 1151 51
w 0 1151 51 1186 17
w 0 1186 17 1219 -14

tmuikku · 2021-09-17 11:13 am

mabat said:
OK, that's nice! It seems it has been fixed in the meantime, which would be great.
I certainly remember it behaved differently.

Could you post the code of the WG sim above?

It sure did bounce when last time tried it few days ago!

This is nice, adds to the usability.

miero · 2021-09-17 1:31 pm

miero said:
Looking on the patterns on the other side, I'd guess that the width of a vane would be important too.

It seems so.

Code:

$ 3 2000 125 0 0 456 0.00037463
S 2 249 -1000 249 3000 0 0.08186844405594404 0 10.000178395549446 99.9998652719957 1 0
w 0 250 650 250 -1000
w 0 250 1500 250 3000
w 0 250 1300 254 1300
w 0 250 700 254 700
w 0 254 1300 258 1300
w 0 254 700 258 700
w 0 258 1300 263 1300
w 0 258 700 263 700
w 0 263 1300 268 1301
w 0 263 700 268 699
w 0 268 1301 274 1301
w 0 268 699 274 699
w 0 274 1301 281 1302
w 0 274 699 281 698
w 0 281 1302 287 1302
w 0 281 698 287 698
w 0 287 1302 294 1303
w 0 287 698 294 697
w 0 294 1303 302 1304
w 0 294 697 302 696
w 0 302 1304 311 1306
w 0 302 696 311 694
w 0 311 1306 320 1308
w 0 311 694 320 692
w 0 320 1308 329 1310
w 0 320 692 329 690
w 0 329 1310 339 1313
w 0 329 690 339 687
w 0 339 1313 349 1316
w 0 339 687 349 684
w 0 349 1316 360 1320
w 0 349 684 360 680
w 0 360 1320 373 1324
w 0 360 680 373 676
w 0 373 1324 385 1329
w 0 373 676 385 671
w 0 385 1329 398 1335
w 0 385 671 398 665
w 0 398 1335 412 1341
w 0 398 665 412 659
w 0 412 1341 427 1348
w 0 412 659 427 652
w 0 427 1348 442 1356
w 0 427 652 442 644
w 0 442 1356 459 1366
w 0 442 644 459 634
w 0 459 1366 475 1375
w 0 459 634 475 625
w 0 475 1375 493 1386
w 0 475 625 493 614
w 0 493 1386 512 1398
w 0 493 614 512 602
w 0 512 1398 531 1411
w 0 512 602 531 589
w 0 531 1411 550 1424
w 0 531 589 550 576
w 0 550 1424 572 1440
w 0 550 576 572 560
w 0 572 1440 593 1456
w 0 572 560 593 544
w 0 593 1456 616 1473
w 0 593 544 616 527
w 0 616 1473 639 1492
w 0 616 527 639 508
w 0 639 1492 663 1511
w 0 639 508 663 489
w 0 663 1511 690 1532
w 0 663 489 690 468
w 0 690 1532 715 1553
w 0 690 468 715 447
w 0 715 1553 742 1576
w 0 715 447 742 424
w 0 742 1576 770 1600
w 0 742 424 770 400
w 0 770 1600 798 1625
w 0 770 400 798 375
w 0 798 1625 827 1650
w 0 798 375 827 350
w 0 827 1650 858 1678
w 0 827 350 858 322
w 0 858 1678 887 1704
w 0 858 322 887 296
w 0 887 1704 919 1733
w 0 887 296 919 267
w 0 919 1733 952 1763
w 0 919 267 952 237
w 0 952 1763 984 1793
w 0 952 237 984 207
w 0 984 1793 1017 1823
w 0 984 207 1017 177
w 0 1017 1823 1049 1854
w 0 1017 177 1049 146
w 0 1049 1854 1085 1887
w 0 1049 146 1085 113
w 0 1085 1887 1118 1918
w 0 1085 113 1118 82
w 0 1118 1918 1151 1949
w 0 1118 82 1151 51
w 0 1151 1949 1186 1983
w 0 1151 51 1186 17
w 0 1186 1983 1219 2014
w 0 1186 17 1219 -14

mabat · 2021-09-17 1:39 pm

Sorry, I don't follow what you are trying to show. What's the point of all the waves outside the waveguide?

Acoustic Horn Design – The Easy Way (Ath4)

Member

Member

Attachments

Member

Member

Member

diyAudio Moderator

Attachments

Member

Member

Member

Attachments

Member

Member

Attachments

Member

Member

Attachments

Member

Member

Member

Attachments

Member

Attachments

Member

Member

Attachments

Member