Don Snyder said:
Don, that layout you have posted - is it for the Kappalite 3015LF? I have 4 Kappalite 3015LF lying around and would love to try a TH with a lower cut-off of about 35Hz. I've been asking for quite a while but have not got much help.
Do you think you can help with a layout? Thanks a lot.
iand said:
Thanks to David's quick work I've been trying out a version of Hornresp where the extra section in a 4-section tapped horn is between the taps, not at the mouth.
This helps reduce ripple even further -- I've attached a plot comparing a two-section exponential (13Hz/26HZ flare cutoffs for first/last 50% of length) with a single exponential (18.5Hz flare cutoff throughout) -- conical is not shown, ripple is >3dB.
By playing around with the percentage lengths I can even get a response more like the TH-115 -- but I still can't get the higher efficiency and/or lower excursion that Tom is predicting :-(
Cheers
Ian
iand said:
Thanks to David's quick work I've been trying out a version of Hornresp where the extra section in a 4-section tapped horn is between the taps, not at the mouth.
This helps reduce ripple even further -- I've attached a plot comparing a two-section exponential (13Hz/26HZ flare cutoffs for first/last 50% of length) with a single exponential (18.5Hz flare cutoff throughout) -- conical is not shown, ripple is >3dB.
By playing around with the percentage lengths I can even get a response more like the TH-115 -- but I still can't get the higher efficiency and/or lower excursion that Tom is predicting :-(
Cheers
Ian
Attachments
Ian: Using EXP in hornresp will smooth the curve, but you're just kidding yourself if you don't build the real horn with exponential
curves.
Samuel: See the Hornresp screencapture below. That's everything you should need.
Sabbelbacke: Exponential, hyperbolic and conic horns relate areas. Only the tractrix horn (used in midrange horns and single driver setups) is defined by radii.
Conic horns have a linear change in area from throat to mouth. The LabHorn was constructed of several conic sections, and one area near the throat that defies description. All considered, it was roughly hyperbolic with an "M" of 0.7
The hyperbolic / exponential formula:
Area at x = Throat Area*[cosh( x*2*Pi*f/c)+M*sinh(x*2*Pi*f/c)]^2
where:
x = distance from throat
f = the cutoff frequency of the horn
M = the flare constant - M = 1 is exponential, 0 < M < 1 is hyperbolic
c = the speed of sound, approximately 344 m/s
curves.
Samuel: See the Hornresp screencapture below. That's everything you should need.
Sabbelbacke: Exponential, hyperbolic and conic horns relate areas. Only the tractrix horn (used in midrange horns and single driver setups) is defined by radii.
Conic horns have a linear change in area from throat to mouth. The LabHorn was constructed of several conic sections, and one area near the throat that defies description. All considered, it was roughly hyperbolic with an "M" of 0.7
The hyperbolic / exponential formula:
Area at x = Throat Area*[cosh( x*2*Pi*f/c)+M*sinh(x*2*Pi*f/c)]^2
where:
x = distance from throat
f = the cutoff frequency of the horn
M = the flare constant - M = 1 is exponential, 0 < M < 1 is hyperbolic
c = the speed of sound, approximately 344 m/s
Attachments
@Don
Hm.. Twisting my mind.....
Given a simple horn in hornresp with a CON Flare:
Exporting dimensions to csv:
Results in:
Radius follows a conical expansion. In order to use a straight peace of wood for the height-part, height should be of conical expansion, too (having the same increment per distance), but it´s not.
Same in AjHorn. Taking a look in the manual of AjHorn also confirms that radius not area is concerned by the flare rate:
rough Translation (I try...):
Parabolic horn follows a function where the radius is expanded by square root hence area is proportional to length (=using straight peaces of wood).
Conical horns follow a linear increase of radius resulting in an squarish increase of area per length increment.
.....
Am I in a state of coherent conclusion or are my thoughts plain rubbish?
Sorry for the long post...
Hm.. Twisting my mind.....
Given a simple horn in hornresp with a CON Flare:
Exporting dimensions to csv:
Results in:
Radius follows a conical expansion. In order to use a straight peace of wood for the height-part, height should be of conical expansion, too (having the same increment per distance), but it´s not.
Same in AjHorn. Taking a look in the manual of AjHorn also confirms that radius not area is concerned by the flare rate:
rough Translation (I try...):
Parabolic horn follows a function where the radius is expanded by square root hence area is proportional to length (=using straight peaces of wood).
Conical horns follow a linear increase of radius resulting in an squarish increase of area per length increment.
.....
Am I in a state of coherent conclusion or are my thoughts plain rubbish?
Sorry for the long post...
Hi Sabbelbacke: Ya got me, fair & square. I knew that I had it wrong just after I posted it, but for some reason the forum software wouldn't let me edit it.
I said "Conic horns have a linear change in area from throat to mouth." I should have said Conic horn areas vary as the square of horn length, or:
AREAx = CX^2
I had never thought about the radius thing.
I said "Conic horns have a linear change in area from throat to mouth." I should have said Conic horn areas vary as the square of horn length, or:
AREAx = CX^2
I had never thought about the radius thing.
Don Snyder said:
Hi Don
I agree, the "fine detail" of exponential vs. conical is irrelevant for a bass horn, in practice it will always be made from conical sections.
But it looks as if a single conical tapped horn expands too quickly at the beginning and too slow towards the mouth; a lot of box volume is wasted near the throat, leaving not enough near the mouth so the area is too small.
Having done a lot of playing with the "beta version" hornresp and multiple conical and exponential sections, it looks like the best solution overall (for the "TH-115" case) is actually an exponential (or an approximate one) throughout -- and having done some playing around with drawings of what could be inside the TH-115 I get the following (all results are attached)
Exponential profile, 20/280/20cm sections
Area 258cm2 at driver (3.3:1 CR), 2800cm2 at mouth (22Hz flare)
This gives an impedance curve which is almost an exact match (at least for the upper peaks), SPL averages about 101.5dB from 40-105Hz.
What is very interesting is that I've found that ripple can be considerably reduced by adding an extra series inductor -- the black line is with this, the grey line is without.
Another interesting point is that to get the lowest impedance peak to match up with the TH-115 (result not shown) I have to reduce the compliance of the 15TBX100 by about 1/3, which also removes the remaining ripple -- so maybe the driver Tom is using is modified, not the standard?
The discrepancies over SPL and displacement still remain. Simulated SPL is 101.5dB average which is about 2dB short of the TH-115, excursion maximum is 0.4mm at 50Hz with 1W/8ohms which would give 12.5mm at 1000W which is still over the 15TBX100 Xmax limit.
So getting closer, but still significant unexplained differences :-(
Cheers
Ian
Im thinking of building a small tapped horn. It doesnt have to dig very deep, and size is a lot more important than maximum spl. I would like something like 120 dB out of this thing in its passband.
Could anyone give me some feedback on this??? What can I change? Any better drivers?
The driver I choose is two B&C 8NDL51.
Cheers
Johannes.
Could anyone give me some feedback on this??? What can I change? Any better drivers?
The driver I choose is two B&C 8NDL51.
An externally hosted image should be here but it was not working when we last tested it.
Cheers
Johannes.
pinkmouse said:
I'm not familiar with the software (but am with the general MLS approach). The look of the FFT leads me to question what the resolution is. It looks as if no feature is narrower than ~50Hz. Is there an option to take a longer sweep and/or get higher resolution? (Or do you know already that the resolution is much better?)
Ken
It's measured with Soundeasy's default settings with the maximum length MLS, but even with the resolution the response is so far off the simulated one something has gone badly wrong. I was at my mum's for lunch and used the opportunity to use her big back garden to do a ground plane measurement at 1m. I'll fire up the laptop after I've had my tea and post that graph, but it's not hugely different.
For the specs, it's big/long enough to tune it 20 Hz in a T/S max flat 4th order band-pass and the simmed tapped pipe version rolls off the low end and boosts the dynamic range above Fb just as TD claims, so I guess my question is does it sound like it measures or it sims?
I mean I don't know anything about the measurement system, but the 200 - 1 kHz response looks about like I expected the 20 - 100 Hz BW to look.
GM
I mean I don't know anything about the measurement system, but the 200 - 1 kHz response looks about like I expected the 20 - 100 Hz BW to look.
GM
Well, if you compare the published results of the Audax HM100Z0 with my measured ones they correspond pretty well, so I have no reason to doubt the hardware, and my ears tell me much the same thing. The factor of ten thing you point out is fascinating though.
Anyone else got measured responses they can compare to simulations?
Anyone else got measured responses they can compare to simulations?