An Improved Transmission Line Alignment

Neil,
And here are the HR inputs I interpreted from the drawing above. Note, the 89 cm length for the length of the second segment comes from adding the dimensions 26.1+4.4+4.5 on the drawing. I think the rest should be clear.

This version results in a volume of 64.4 L and a TL length of 188 cm which I think are consistent with your enclosure dimensions.

Eric
 

Attachments

  • neil4.jpg
    neil4.jpg
    123.7 KB · Views: 305
Neil,
I believe that by using this "diagonal" dimension as the cross-sectional area for S2, you have created volume in the model that simply isn't there in reality. Using your inputs, HR calculates a "System Volume" of nearly 74 liters. But if I calculate the volume of your actual enclosure, I only get around 65 liters, and that even includes the dead air in the corners.
But if I use the step function, and set S2 in segment 1 to 411 (i.e. 7.5"), and set S2S in segment 2 to 343 (i.e. 6.25"), and use all your segment lengths, then the HR System volume comes out at 64.8 liters, which is much more consistent with your actual enclosure volume.
Eric

Eric, you make an excellent point. The balance between the centerline length, and the overall volume - these will be different when the system is folded vs straight; kind of by definition. If you hold the volume from straight to folded, you'll get something different than if you hold the centerline length, from straight to folded.

I have a feeling that either holding the length or holding the volume - are both a fudge from what the real world performance would be. And holding both isn't really possible.

I would differ a bit from you, though - the air is resonating in all directions, and especially in the closed end, the diagonals are going to "count" in terms of the harmonics. A pressure wave front, like ripples on the surface are affected by edges / surfaces no matter what incident angle they hit at. If the lower left corner is a transitional tapered space - then why isn't the upper right corner where the woofer is, also a tapered space?

The dimensions of the sections used in Hornresp are essentially "samples" - or more accurately "key samples" of the volume. As far as I can tell, the program is using a model of 3D space that would look like the schematic diagram they show.

And it is simply not possible to hold both centerline lengths and volume, when folding a 3 dimensional space. I will see if I can reinterpret the design to get a volume closer to what it it really is - and then see how much different the response is.

Do we know how Hornresp "builds" the model it uses? I wonder how much would be involved in getting a "front end" that would load a profile of a shape, and give it a depth - and then use the existing programming engine to generate the response for the folded shape?|


Is the schematic profile that Hornresp shows ROUND - or is it SQUARE? If I had to guess, I'd say it was square.
 
Last edited:
I have a feeling that either holding the length or holding the volume - are both a fudge from what the real world performance would be. And holding both isn't really possible.

I have the same feeling. But I think it's likely important to come pretty close to holding both.

If the lower left corner is a transitional tapered space - then why isn't the upper right corner where the woofer is, also a tapered space?

I agree, both should be tapered spaces. But your enclosure basically has 5 segments, and HR only allows 4. So you have to compromise (fudge) and decide to model one of the tapered sections as a step, instead of a taper. I actually tried it both ways and the differences seemed subtle to me.

Do we know how Hornresp "builds" the model it uses?
Is the schematic profile that Hornresp shows ROUND - or is it SQUARE? If I had to guess, I'd say it was square.

I don't. But David McBean (it's creator), frequently post here, so maybe he will reply. But my best guess is that it's actually neither round nor square. That is to say, I suspect the model is basically one dimensional (a line) and that each point along the line has an associated differential volume of air (A x dx), with the associated properties of that air (density, spring constant, viscosity, etc?). I would further guess that viscous drag at the walls is neglected. So since a round cross section would minimize wall drag compared to a rectangular cross section, I suppose the results for a cyclindrical cross section might be slightly more accurate than for a rectangular cross section, but they are probably approximations in both cases. All just my guesses, however.

Concerning how you are choosing how to represent the position of the driver: You seem compelled to draw the "centerline" of your TL so that it passes through the acoustical center of the woofer. Rather, I wonder if the better approach isn't to draw the best geometric centerline of the enclosure, then find the spot along that line that comes physically closest to that centerline, and consider that spot to be the effective location of the driver?

Eric
 
Neil,
And here are the HR inputs I interpreted from the drawing above. Note, the 89 cm length for the length of the second segment comes from adding the dimensions 26.1+4.4+4.5 on the drawing. I think the rest should be clear.

This version results in a volume of 64.4 L and a TL length of 188 cm which I think are consistent with your enclosure dimensions.

Eric

Well, the good news is, this still has pretty good response. It has a ~1dB ripple at about 125Hz, but other than that, it is pretty darn close to the response I get.

The thing is, this design is not a stepped design. A stepped design is something like this.


And I measured it a third way that doesn't have the diagonal S2 - and the response is also very similar to the one you just did, and the one I did before. If the lower left corner is a tapered section, then so is the the upper right corner, I would think.

There is no perfect way to slice this. This design is not unusual, and the dividing line has to be somewhere - and the one thing that has to be right for Hornresp to be anywhere near correct - is the location of the woofer.
 

Attachments

  • Stepped TL.png
    Stepped TL.png
    6.2 KB · Views: 325
Last edited:
The thing is, this design is not a stepped design. A stepped design is something like this.

Agreed, but at the same time I don't think that modelling with a step is much of a compromise.

If you really want to avoid the step in the model, there are two easy things you could do:
1) split the enclosure in half at 6.875 in both sides. Then S1, S2 and S3 are all 6.875 (377 cm2). I tried it and it seemed pretty good. No real difference.
2) Leave the center divider where it is at the bottom end of the enclosure, but slide it over to 6.875" at the top. Again then S2 can be considered to be 377 without requiring a step. I tried that too, and likewise no real difference.

To be clear, I'm not suggesting that either of these changes would improve the final result. I'm only saying that they would free you of the need to use the step function and perhaps then feel more confident of the model.

There is no perfect way to slice this. This design is not unusual, and the dividing line has to be somewhere - and the one thing that has to be right for Hornresp to be anywhere near correct - is the location of the woofer.

No, there's no perfect way. It's only a model. I think you are pretty close. Probably we are splitting hairs at this point.

Regarding woofer location. I do think you want to get it as close to correct as possible, like the volume and line length. But the real result might not be quite as sensitive to the woofer location as the model seems to suggest. Martin King noted to me once that his model (and I assume HR too) assume that the woofer is effectively a point source, whereas in reality, the woofer has a significant area. So effectively, the woofer location is really spread out a bit along the line, and not concentrated at a single point.

Eric
 
The reason that I shifted the vertical baffle to the back and made the closed end a bit larger volume than the rear - is it extended the flatter response a bit deeper, but kept that shelf all the way up to about 200Hz. And it allows there to be a brace, if needed in the closed end; which where the highest pressures are likely to be.

Your simplified model with all right angles is what got this so close to where it is now. But the slight asymmetry extended the shelf down quite a bit farther - on the 8 ohm woofer in particular.

This now is an inch less deep - 16" is still chunky but I like the package.
 
Okay - shows what I know! Thanks for making this program.

Knowing this is helpful, I think. Fluid dynamics are better in round tubes (round chimneys and HVAC ducts are preferred) - and so close(r) to square is the next best behaved. Square ducts / tubes act more or less like round, as the corners are close to still pockets. And the slimmer a rectangle gets the worse it performs for its given area.

An 8 inch square acts essentially like an 8" diameter round, even though it has 64 square inches of area vs 50.27 sq in. But a 2" x 32" rectangle will have much worse flow, even though it is the same area as the 8" square.

The same goes for resonances with these shapes, I think.
 
Last edited:
I am wondering something - should the section areas from the driver location up through the terminus opening be counted as the area of the best circle / ellipse that fits in the rectangle? Since this is the area that the air can actually flow in?

I think the closed end to the driver should have the full dimensions used for the volume, since this end is acting as a resonant chamber - the air basically stays in place, an transmits the pressure waves moving through it.

But maybe the portion from the opening to the driver, need to have the section areas reduced, to more accurately capture what is happening?
 
I discovered I had an error for the 4 ohm Dayton woofer - I had used a Vas value that I had arrived at with a calculator - but realized that Dayton Audio lists it. I had a value that was way too high, and now the response is very, very similar to the 8 ohm.

I also widened the last section from 3" to 3 3/4", and this results in the virtually flat shelf from 50Hz up to 200Hz, for both Dayton woofers. The extension frame at the top is for the Linaeum TLS tweeter. I have updated the images - I will attach them to this post.

In fact, I plugged in the SEAS Curv Cone U22REX/P-SL (H1659-08) woofer that I was considering - it is even lower Mms and lower Fs - but also lower Qts, and more than twice the cost. And the response with the Curv Cone woofer is virtually the same, in the same cabinet. The only functional difference - is the SPL level, based on the differing efficiencies between the three woofers.

They use different fill values - the Dayton 4 ohm and the SEAS Curv Cone uses 33 in the first two sections. And the Dayton 8 ohm uses 100 in the first section and 30 in the second.
 

Attachments

  • SEAS Curv Cone U22REX P-SL Response.png
    SEAS Curv Cone U22REX P-SL Response.png
    30.5 KB · Views: 317
  • Final Dayton Audio RS225P-8A Transmission Line Design.png
    Final Dayton Audio RS225P-8A Transmission Line Design.png
    30.5 KB · Views: 288
  • Final Dayton Audio RS225P-4A Transmission Line Design.png
    Final Dayton Audio RS225P-4A Transmission Line Design.png
    26.7 KB · Views: 322
  • Final Dayton Audio RS225P Transmission Line Design.png
    Final Dayton Audio RS225P Transmission Line Design.png
    28 KB · Views: 336
Last edited:
Neil,
I like the change to a bigger port.

Probably you are tired of my questions/suggestions for the enclosure inputs but I'll share them anyway :)

I understand now where you are getting 493 from for S2 but I still think it's not representative of your enclosure. With your inputs I still get over 71 liters, when the real enclosure is a bit under 65. And as you have noted there are likely dead spots that may make it behave as if it's even a bit smaller than that. And even if you assume that having the driver up in that upper right corner helps better utilize the volume there, it's still not reasonable (to me) to model the enclosure as anything larger than 65 L.

And I still don't understand how you chose 375 for S1. Sorry if you explained that before and I missed it. But it seems to me it pretty much has to be 411 (7.5x8.5x2.54^2) or darn near. What am I missing?

So overall it still seems to me that you'd have a more representative sim if you use 411 for S1 and S2. That gives a volume of 66.6 L, which is still a tad high but by a lot less than 71 L.

I'd tend to go even farther and either apply a step at S2 and call it 411 in the first segment and 343 in the second (63.8 L). Or if I was really adamant to avoid using the step, then I'd just fudge S2 to whatever input creates whatever I thought the most realistic volume was. For example, calling S2 390 gives 65 L which seems good to me.

All that said, I tried all the above and the results are not really so different from yours to be worth worrying about. But I would still be interested to know your reasoning for S1 = 375 and for justifying the 71 L volume.

Being new to HR and folded geometries myself, it would be interesting to get input from someone else more experienced, but lately you seem to be stuck with me!

Eric
 
Hi Eric,


On S1, I was reducing it for the 3/4" wide brace, that sits between the front baffle and the vertical inside baffle. Since this is the area that will see higher pressures, the brace is best there.

On the volume issue - most of that comes (I think?) from having a "quick taper" rather than a long even taper. I could try to represent this in Hornresp - it might be possible by going to a stepped mode, but then using numbers that show as short taper followed by longer straights?

I am surprised by the consistency of this design - as I posted, it works virtually the same for 3 different woofers. An it has so little fill required to get there. The few small peaks are all high enough for my active crossover to "shave" them right off with the custom EQ - sweep with REW and export the PEQ file. Load that into the crossover, and that's it.
 
If the net volume of the Hornresp sim is larger than the net volume of the box that you're trying to sim, that's usually a pretty good sign that the sim might be off a bit.


Hi Brian, I will try to figure a better way to model it in Hornresp. I want to do this right.

The closest place I can get Baltic birch plywood, is not making it easy to buy from them, so all the more reason to not screw up in wood.

My quandary comes down to this: to get the location of the woofer accurately from the closed end of the TL - AND to have the L12 and L23 volumes accurate, at the same time.

* Can I have an S2 section located independently from the woofer?

* Or, if the S2 is located at the top edge of the vertical baffle - and this is at the EDGE of the woofer (and not at its centerline) - is that better?

* OR - a third possibility is that the woofer frame and magnet structure, and the corner pocket above the woofer - reduce the S2 to an extent that S2 is roughly equal to S1?
 
Here's what I did - I tweaked the filling a bit more after the screen capture, so the peaks are slightly better in the export file.
 

Attachments

  • RS225P-4A Stepped Hornresp Model Drawing.png
    RS225P-4A Stepped Hornresp Model Drawing.png
    34.4 KB · Views: 147
  • RS225P-4A Stepped Hornresp Model Response.png
    RS225P-4A Stepped Hornresp Model Response.png
    25.3 KB · Views: 133
  • RS225P-4Astepped.txt
    1 KB · Views: 40
If that box is rectangular and that internal panel is parallel to the front and back panels, then S2a should be equal to S1

TBH, if I was doing this, I'd redesign the box a bit, rather than try to approximate it as-is in a Hornresp sim. For example, I'd lengthen that internal panel so it was equidistant from the top and the front, essentially making S2=S2a=S2b, and then adjust that chamfer in the back so that the CSA from the internal panel to the chamfer is equal to (S2+S3)/2. This are small changes that shouldn't change the box too much, but will make the sim of its response a bit more accurate.

BTW, make sure that you've set the value of "Path" in Hornresp to the distance between the center of the driver and the center of the vent.