THAM15 - a compact 15" tapped horn

[martinsson]

Hello all fellow DIY'ers

I wanted to give this design a thread of it's own since it has attracted a certain amount of attention out there.

Sofar it has only been linked to in threads here and there and often with different names (martinsson TH, andershorn etc.) and since my 6,5" micro TH got it's own thread i felt it only fair to present this to you in the same manner.

This is a compact tapped horn messuring 550x680x500mm (wxhxd), the internal volume summs up at around 167 liters, and it is made from 18mm plywood, it was initially designed for homeuse but it soon found it's way into the proffesional realm where it has also proven most usefull.

The driver of choise is a B&C 15TBX100, but other happy customers has reported in that it works wery well with other drivers aswell, amongst these BMS, Eighteensound, eiminence and RCF variants can be found.

The design seems forgiving in this respect, but it should be said that I can only speak for the B&C 15TXBX100 loaded one, since this is the only one I have heard, sofar.

The base design, shown here, is the THAM15, and although there are proposals for improvement in the pipeline these are thought to add only nuances of improvement.

So here we go, this is the THAM15 tapped horn :





These are only a few pictures to help give you a quick view of what this DIY proposal contains and look like.

Detailed drawings ready for the "productionfloor" can be found here :

http://www.martinsson.cc/blog/index....y090721-180410

Best regards // Anders Martinsson

[Rademakers]

Hi Anders,

i think that is a nice design you made.

However the sim does not correspond with the construction. Hornresp simulates a nett volume of 173 liter, that would result in a construction of ~220 liter, whilst the construction drawning is 187 liter in total. Increasing the outer width up to 60 cm would level with the simulation.

Best regards Johan

[martinsson]

Hello Johan

If i do the math on the dims only i arraive at your 187dm3 on the external dimensions, so far so good, and the S4 value tells me that the internal width is 464mm also a check, and the lengths L12,L23 & L34 are all correct in the sim.

My only naggin doubt is that the expansion may not be corretly simulated as the front chamber takes a step up in volume that is not shown in the system volume calculations.

Since this is a linear expansion (apart from the last step mentioned above) there maight be a deviation in there somewhere, the material part of the volume should be about 50dm3 (approx 12x24x0.18dm).

The differance between the volumes 187-173 = 14dm3 (given that the 173dm3 sim figure is correct) does not account for the material only, there is still a diff of approx 50-14 = 36dm3 left somwhere.

Now 36dm3 is not to be taken lightly, this is a significant amount of volume, a part of which can depend on the linear simulation of a non linear expansion, and the bends has been approximated, this maty also contribute, the sum of all this may give us the difference you describe.

Where and how these deviations are and lokk like, aswell as how great a apart they play remains to be figured out.

I will not change the design to fit the simulation, the simulations however may need another go to figure out exactly how this design simulates in detail, but the real world result is still, and has allways been, a very good one, the comments i have recieved sofar from users has been very positive.

Thanks for bringing this to my attention, I will as soon as possible try to iron this out, meanwhile me advice to all that are thinking about building this box is still "go for it", it works wery well, despite the suspected simulation errors.

If anyone out there wants to lend a helping hand in figuring this out and perhaps speed thing up please do so and let us know of your findings, i would be very grateful for any help i can get.

Best regards // Anders Martinsson

[Don Snyder]

Hello Anders - Thanks for your excellent design. HornResp is a wonderful tool, and McBean deserves all our thanks for his devotion to the DIY community.

There is one change in version 28 that I wish to consider, that of the addition of the Parabolic expansion type (PAR). When we make a horn segment
with plywood, we almost always use constant width and linear taper. While the total horn can approximate Hyperbolic/Exponential or LeCleach, etc,
this segment, taken alone, is parabolic.

Your input to HornResp uses EXP for all segments. Selecting EXP or even CON will result in errors, most notable in the "system volume" being over
reported.

[Rademakers]

I will not change the design to fit the simulation, the simulations however may need another go to figure out exactly how this design simulates in detail, but the real world result is still, and has allways been, a very good one, the comments i have recieved sofar from users has been very positive.

I understand, measurements often show better results for the low end extension then the simulation does. So a smaller version would result in less efficiency in the lower department but due to the simulation being pessimistic, the actual results might agree with the simulation for the larger version.

If anyone out there wants to lend a helping hand in figuring this out and perhaps speed thing up please do so and let us know of your findings, i would be very grateful for any help i can get.

One note: Using exp instead of con segments reduced the simulated volume by 6 liter (exp 167 liter). That's just 30 instead of 36 liter

To get an idea of where they went I made my own simulation based upon the drawning and measurents given/recalculated. I actually got a longer horn length myself suggesting, for a fixed volume, lower partial horn area's.

The second drawning/graph shows how this influences the flare rate through-out the entire horn. The second graph I adjusted for the horn area taken up by the driver itself. Showing in comparisson with the third picture were the difference should result from.

The third picture being the horn flare predicted by HR 28.00.


Regards Johan

[epa]

i designed a simular model for an 18" a wile ago,just for fun







it was something like 60 wide*90 high*70 deep cm

[Xoc1]

I drew up the THAM 15 using Anders dimensions in Solidworks.
I measured the internal volume as 144.5 Litres.
I then plotted the internal schematic shape. The volume of the straightened shape was 146.5 litres which is just over a 1% error.
So what would be the best way to sim this in Hornresp?
Should I compensate for the volume of the driver?
BTW I seem to get a more linear taper by increasing the 135.1mm dimension by another 10mm.

[soho54]

Originally Posted by Xoc1

I drew up the THAM 15 using Anders dimensions in Solidworks.
I measured the internal volume as 144.5 Litres.
I then plotted the internal schematic shape. The volume of the straightened shape was 146.5 litres which is just over a 1% error.

I thought I was the only one to do things like that. Wish I had a copy of Solidworks though. Heck, even an old version of AutoCAD would be nice. I have Sketchup which make you do everything manually, Maya, or 3DMax, and the later two just make no sense.

Really though, nice work. If you are getting this crazy with it, I would skip HR, and use AkAbak. That way you could model the 27 sections needed, and forget trying to account for/estimate everything in HR.

[Xoc1]

The advantage of using Solidworks is that once a parametric model is set up it is easy to adjust the sizes and and see the results. In this instance I was just trying to help Anders get a closer sim to what he has already built.
The method I am using to get the horn length is based on the method that was developed in the TH spreadsheet thread. I could easily fully dimension the schematic sketch if that helps. Then the width dimensions could be entered in a spreadsheet to get the area data.
Using AkAbak sounds good, but I am still strugling with geting a TH model to work in AkAbak. I would like to learn more about getting AkAbak to work with a TH sim.
I am determined to build my own TH horns one day.......

[martinsson]

Thanks alot for the help, really, to all of you !

In Rademakers & Xoc1's excellent descriptions we see partly where the "missing" volumes are located (bend contributions) and even better, here we can easily see the distances between them and perhaps try to match those to potential resonaces/cancellations in the frequency responce, although i'm no longer shure where they occour and how significant they really are.

I'm also thinking that the approximation done by honrresp due to the odd shape of the expansion is a contributor to the "errors" we see.

Big thanks, now i understand my own design alot better (I've allways thought there was an element of luck involved in my designprocess somewhere

The 135.1 dimension should perhaps be reviewed, in light of your findings Xoc1, my thinking was to get as much L23 as possible and keep to the expansion up tp that point, but maybe there is a improvment possibillity here, thanks for telling us your thoughts.

This brings back an "issue" initially mentioned by Mr Cowan when he built his pair, would adding reflectors be of any grater benifit to this design ?

+ more even expansion (as shown above, less bumps on the way)
- less bend filter effect, may increase upper range resonances

What are your thoughts ?

sidenote:
I'm a CATA V4/V5 nerd myself (automotive comcept engineer), wich is not far from solidworks wich is a really nice software, i only wich i had CATA V5 at home, but it's rather expensive...