Thursday, January 8, 2015, 21:24

Posted by Anders

The idea I had was to sum two horn paths of different lengths over a single driver in order to expand the usable passband of the design in total.Posted by Anders

The "Dual Summation" part of the principle stems from (SUM 1) the summation of the different horn segments, paths 1 & 2 and (2) the summation of the paths (SUM 1) over the rear of the driver giving rise to the second summation (SUM 2).

In order to obtain a more equal contribution of the two segments making up SUM1 they have been compensated in cross section areas according to their respective difference in lengths so that the shorter segment does not dominate (the law of least resistance).

In principle the SUM1 can be regarded as a first taper, and the SUM2 as a second taper, this also means a number of out of passband cancellations, balancing these summations with regards to both contribution (levels) and distance (timing) is perhaps not easily done.

I made a baseline balancing of these summations according to below, but I have no illusions that they will not need to be tweaked further, perhaps, or rather probably, significantly so.

Overview :

In principle :

Unfortunately this is a very difficult concept to mathematically describe and simulate with any grater accuracy using tools I presently have at my disposal, and as such it remains a paper product with no mathematical proofs of concept.

Somewhat more developed it might look a bit like this, again this is only in concept form, the brace at the driver provides both more path length(s) and will hopefully make the summation a bit more gradual.

Or perhaps even something like this :

We can and should also alter the path length ratio in order to try and find the point just before the path dependent cancellations becomes to severe, here is a non optimized example using the same size enclosure and principle folding, just showing it's possible to tune it :

An attmept to describe it (a start) :

The dual summation idea might even be applied in a Cyclops style folded TP (constant area) format :

This is simply an idea, no claims, no proof, and perhaps even a fair bit "out there", I just thought I'd share it with you for the fun of it in this our common interest :)

view entry
( 58 views )
| permalink
| print article
| ( 3.1 / 40 )

Wednesday, December 31, 2014, 12:49

Posted by Anders

Posted by Anders

**Happy new year all you brilliant DIY'ers out there!**

Thanks for making 2014 one of the most interesting and rewarding years so far for me in this our common field of interest.

Tuesday, November 11, 2014, 14:02

Posted by Anders

It may sound crazy... I will not know for sure until I build one, if I build one that is.Posted by Anders

This is just a concept sketch, no plan, there is no plan, just a thought centered around the Fostex FX120.

The choise to go with a constant area (pipe) rather then a horn, is to achive a more efficient low end response with a shorter pathlength (smaller box).

This along with the folding will hopefully provide a more prenounced pass band with a reduced risk of upper band resonances making it to the summation area (filter effect).

I have not yet seen anything like this out there to date, but the internet is big so chances are I might have simply missed it, anyway, here it is :

Tuesday, April 8, 2014, 16:46

Posted by Anders

Some details and rough RTA measurements of the unholy union.Posted by Anders

Driver:

2" Tangband W2-803SM full-range.

Horn:

JBL 2386.

Circuit:

Amp:

LAB IP450

XO:

DBX 223 XL

Source:

Audacity on Linux (generating pink noise)

RTA:

Phonic PAA2 measuring non weighted.

Distance:

1 Meter

The measurement setup :

The driver and circuit complement :

Measurement with HP set at 45Hz 24dB/Oct 1M:

Measurement with HP set at 250Hz 24dB/Oct 1M:

Measurement with HP set at 350Hz 24dB/Oct 1M (prefered):

It sounds a lot better then the graphs would have you believe, and the SPL capacity is greater the the drivers would have you belive.

Here's an approximated summary SPL graph result :

**| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | Next> Last>>**