Hornresp

[David McBean]

Originally Posted by mwmkravchenko

As a followup on the update notification:

Works very well.

Still have to figure out what the editor will do.

But very interesting.

Hi Mark,

Thanks for the feedback. To better understand how the Hornresp Data File Editor works, try the following:

1. Create a new data file using the File > New menu command.
2. Open the Data File Editor using the File > Editor menu command.
3. Click on the Hornresp.dat file name in the top left or right box.
4. Select a number of Hornresp.dat records to move or copy to the new file, or to delete from Hornresp.dat.

The Editor allows those Hornresp users with lots of records * to organise things better by storing their data in various different files if they wish.

Kind regards,

David

* A Hornresp "Grand Master" is a user with more than 1000 records :-).

[David McBean]

Originally Posted by schmeet

The equation for a Parabola is Y = a.X² + C

We have our two Areas: S1 and S2. From these we can calculate Y1 and Y2.

Y1 = SQRT[S1 / PI]
Y2 = SQRT[S2 / PI]

if we assume that the element starts at 0 and ends at L12 then:

X1 = 0
X2 = L12 (Length)

a and c can be calculated from:

a = [Y2 - Y1] / [X2² - X1²]
c = Y1 - aX1²

We can then calculate the area at any point along the segment by

Y = a.X² + C
S = PI.Y²

Hi schmeet,

You have assumed that X1 = 0. Isn't X1 the axial distance from the parabola vertex to the point at which Y = Y1?

Kind regards,

David

[Don Hills]

schmeet appears to have his x's and y's crossed...
"wires" crossed, get it? Oh, never mind...

I think it should read:
X1 = SQRT[S1 / PI]
X2 = SQRT[S2 / PI]
Y1 = 0
Y2 = L12 (Length)

S for any value of Y can be found as follows:
s = ((y - c) / a) * pi

His method is the same as I have been using. I also like your method because it avoids converting from area to radius and back.

[schmeet]

I guess so, but it makes more sense to me that x is the axis along the length of the pipe and y is its radius.

[just a guy]

Originally Posted by David McBean

Hi just a guy,

Further to my earlier post about keeping things in perspective, I think you might find in practice that it really doesn't make much of a difference if multiple parabolic segments or multiple conical segments are used in the simulation model. To illustrate, the attached screenprint compares the SPL response of a tapped horn having three conical segments against the response of the same horn, but with three parabolic segments. You can see that the traces are effectively identical. Greater inaccuracies are already inherent in the predictions due to the assumption that the driver diaphragm acts as a rigid plane piston at all frequencies, and that there are no losses due to panel vibrations, etc.

Kind regards,

David

Hi, I agree with you in most cases it won't have much effect. But it only took me about 3 minutes to create this mess. I'm pretty sure I could show a much worse example if I spent more time on it.



The only difference between these two lines is PAR vs CON segments and my lines don't overlay nearly as nicely as yours. I noticed this problem almost immediately when Hornresp allowed PAR segments. I don't remember which model caused me the most trouble but I've seen examples much worse than I've posted here.

As I pointed out originally this isn't a fatal flaw, it doesn't seem to shift the peaks right or left (which would be really bad) it just changes the amplitude a bit. Even though this isn't bad I would prefer to avoid surprises like this. In very large horns, 2 db at tuning can be a difference of hundreds of liters of enclosure size. I'd really like to be as accurate as possible when I fold.

Although I haven't studied in depth what factor is causing this, I think it's one or more very long segments that will cause this discrepancy.

(I got the mail, thanks!)

[mwmkravchenko]

* A Hornresp "Grand Master" is a user with more than 1000 records :-).

Aew man!

lost 400+ a few years ago and have 440

Still not a grand master!

But I'm working on it!

[Don Hills]

Originally Posted by schmeet

I guess so, but it makes more sense to me that x is the axis along the length of the pipe and y is its radius.

True, it makes sense to "lay it on its side". The trouble is that you mixed your axes in your example - some of your values were for a "horizontal" parabola, some were for an "upright" parabola.

[David McBean]

Hi Don and schmeet,

Thanks for the clarifications – everything makes sense now :-).

In summary:

Assuming that X is axial length and Y is radius, the equation used to define the parabolic curve should have been shown as:

X = a.Y² + c

Rather than:

Y = a.X² + c

This then gives:

a = [X2 – X1] / [Y2² – Y1²]
c = X1 – a.Y1²

S = ((X – c) / a) * Pi

Which provides the correct result.

Kind regards,

David

[Don Hills]

just a guy:

A very small spreadsheet is attached. All it does is calculate the csa at any point you specify along a conical segment.

[just a guy]

Originally Posted by Don Hills

just a guy:

A very small spreadsheet is attached. All it does is calculate the csa at any point you specify along a conical segment.

Thanks Don. I can't look at it at the moment (no excel on this laptop) but I'm sure it's just what I need to accurately fold my Akabak horns.