Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Collaborative Tapped horn project
Collaborative Tapped horn project
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 1st March 2008, 06:33 PM   #1301
Tom Danley is offline Tom Danley  United States
diyAudio Member
 
Join Date: Apr 2005
Location: Sitting behind the 'puter screen, in Illinois, USA, planet earth
Hi guys.

I guess I should check in here more often haha.

A few thoughts, Iand, a properly done measurement trumps a computer prediction every time, those measurements are what one uses to tweak the computer model so that it predicts more like what you measure.
Computer models in acoustics are rarely exact, at best they come close which makes them a powerful tool for those who make sawdust.
For a good chunk of my life, I would have given my arm for David’s horn program for example.

But, if you want to KNOW the actual excursion, that requires you observe it directly at a known level and frequency.
You want to know how much port loss there is at a given power level, measure the change in response with a given box / port / drive level etc.
You have to measure, models only include things to a given depth, for example you can represent an inductor as an L, or you can represent it as an L in series with an R paralleled by a small C which is in series with an R, and so on and on.

In the case you are investigating, in particular you should actually build / measure a high power ported box and tweak your computer model accordingly. These systems depend on having low losses and as the level increases and / or frequency decreases, so do the losses and the greater the impact that has.

You ask me to prove things, to design a 320L Tapped horn box for you, I would say step back and think about it.
Proof is a proper measurement; a proper execution is one that works “like intended”.
I occasionally have to build 3 or 4 different prototypes if the first ones measurements weren’t close enough.
A computer model has to be interpreted into an air and wood shape and that is not always a simple or obvious thing either.

Also, while I am a DIY’r and grew up doing that, I have kids and am trying to make a living doing this so there is a limit to how far I can go so far as “showing secret sauce” (if and when it exists) as you say. I try to make the simplest things which do the job and in general I don’t try to hide what I’m doing hence the white paper explains the nub of how it works but doesn’t show how to design one.
Actually you guys have made very fast progress figuring these out with only a few clues, which is pretty cool (thinking back to how long it took to figure out how to make one that could be a product).

Lastly, keep in mind that our products at work are mostly used in commercial sound, in that area the weight and cost of the speakers and power amps are also usually significant concerns.
Best,

Tom
__________________
Bring back mst3k and futurama
  Reply With Quote
Old 1st March 2008, 06:50 PM   #1302
tb46 is offline tb46  United States
diyAudio Member
 
Join Date: Jan 2006
Location: Texas
Default Tapped horn modeling

Ian:

Would you, please, post the Hornresp Input Parameter windows for your designs? It would make communication a lot easier and clearer. E.g.: In Post #1298 you tell us: "S1=285 (3:1 CR) L=20/280/20cm (all exponential) S3=2400", and I'm almost certain that's supposed to be S4=2400; or, in my last post you were immediately able to see that I used "Ang=0.5xPi" (fumble fingers, and I'm just not interested in using a tapped horn for outdoor concerts). This would also remove any doubt as to the driver parameters used (e.g: data sheet impedance v. measured impedance).

It is my belief, that the simulation tools (that have been made available for no cost and practically no outside contribution except by the original programers) are quite adequate to, quote: "come up with decent tapped horn designs (Post #1298)". This has been proven by quite a number of builders. If I understand you correctly, you are trying to compare data derived from your Hornresp model with data you extracted from Tom Danley's publications, and it looks as if you have gone about as far as one can considering the guesses and vagaries involved, as always, the devil seems to be in the details.

I agree, there is room for improvement in the simulation tools, and I am thankful for the continuing effort being put forward. You may want to try to extend your modeling efforts to AkAbak, and develop a more detailed model including exact driver front chamber dimensions, throat area, individual waveguide sections and individual corners, exact S3 area (minus the area displaced by the driver), exact mouth area with respective end correction and not to forget the resonators and their acoustic and physical contributions, and the impact of acoustic treatment; and then it would be just great if you would share the new found model and knowledge in this thread.

Regards,

P.S.: I just run a spell check on the above and WordPerfect’s spell checker wanted to replace AkAbak with Icebox, so much for that.
__________________
Oliver
  Reply With Quote
Old 1st March 2008, 10:16 PM   #1303
kstrain is offline kstrain  Scotland
diyAudio Member
 
Join Date: Apr 2006
Location: Glasgow
Default Re: AkAbaK vs. Hornresp.

Quote:
Originally posted by kstrain

The results are not identical but the excursion and response are both within 20% (and are both larger in Hornresp so the SPL for a given excursion is rather similar).
After reading subsequent posts I realise I should have made clear that the models were not identical (as David stated, anything other than identical output would be unexpected otherwise). So my post title was misleading. "A model that happens to be in Hornresp vs. a model, with some extra bits, in AkAbaK" would have been more accurate.

The point was that separately developed models were within 20% in SPL and excursion over a wide frequency band. This band includes regions where the rear (and front for that matter) load on the driver is high impedance and others where it is low impedance. I would be surprised to see reality much different over the *whole* band (unless the driver parameters are wrong), but I'd not be surprised at all to see differences at certain frequencies. Time will tell (laser vibrometer available in a couple of months for some tests on a model speaker, I think).

Ken
  Reply With Quote
Old 2nd March 2008, 07:43 AM   #1304
David McBean is offline David McBean  Australia
diyAudio Member
 
Join Date: Jun 2007
Quote:
Originally posted by iand
After all, I expect the same basic equations which relate impedance, radiation resistance, excursion and SPL are used in all cases (is this true?)
Hi Ian,

Just confirming that Hornresp uses a universal simulation model, applicable to any loudspeaker system.

Kind regards,

David
  Reply With Quote
Old 2nd March 2008, 05:21 PM   #1305
iand is offline iand  United Kingdom
diyAudio Member
 
Join Date: Feb 2007
Location: London
Quote:
Originally posted by Tom Danley
Hi guys.

I guess I should check in here more often haha.

A few thoughts, Iand, a properly done measurement trumps a computer prediction every time, those measurements are what one uses to tweak the computer model so that it predicts more like what you measure.
Computer models in acoustics are rarely exact, at best they come close which makes them a powerful tool for those who make sawdust.
For a good chunk of my life, I would have given my arm for David’s horn program for example.

But, if you want to KNOW the actual excursion, that requires you observe it directly at a known level and frequency.
You want to know how much port loss there is at a given power level, measure the change in response with a given box / port / drive level etc.
You have to measure, models only include things to a given depth, for example you can represent an inductor as an L, or you can represent it as an L in series with an R paralleled by a small C which is in series with an R, and so on and on.

In the case you are investigating, in particular you should actually build / measure a high power ported box and tweak your computer model accordingly. These systems depend on having low losses and as the level increases and / or frequency decreases, so do the losses and the greater the impact that has.

You ask me to prove things, to design a 320L Tapped horn box for you, I would say step back and think about it.
Proof is a proper measurement; a proper execution is one that works “like intended”.
I occasionally have to build 3 or 4 different prototypes if the first ones measurements weren’t close enough.
A computer model has to be interpreted into an air and wood shape and that is not always a simple or obvious thing either.

Also, while I am a DIY’r and grew up doing that, I have kids and am trying to make a living doing this so there is a limit to how far I can go so far as “showing secret sauce” (if and when it exists) as you say. I try to make the simplest things which do the job and in general I don’t try to hide what I’m doing hence the white paper explains the nub of how it works but doesn’t show how to design one.
Actually you guys have made very fast progress figuring these out with only a few clues, which is pretty cool (thinking back to how long it took to figure out how to make one that could be a product).

Lastly, keep in mind that our products at work are mostly used in commercial sound, in that area the weight and cost of the speakers and power amps are also usually significant concerns.
Best,

Tom
Hi Tom

I absolutely agree that an accurate real life measurement is the best thing, since it includes effects which might not be simulated such as nonlinearity. But you still need a starting point for a design (derived from simulation) which is close enough that you make the right basic design choices in the first place...

I wasn't asking or expecting you to give up the precise details of what you do inside your boxes (because it's your living!), I was asking if you could shed any light on the fact that your measured (or simulated using your own program?) excursion and SPL are considerably different to the best that it seems possible to do in Hornresp.

As was said in another post, the predicted dips and peaks won't agree exactly with the real thing, and they move around anyway with small tweaks to the design. So I doubt that the design I came up with is exactly the same as the TH-115, but any small detail differences can't explain the average measured SPL over the band being 3dB higher than simulated, or the peak excursion being 40% lower.

Errors this big are very likely to mean that people make wrong design decisions -- for example, deciding to use a reflex box because the tapped horn doesn't seem to give higher SPL or lower excursion (!!!), or to try and find a driver for a tapped horn with much higher Xmax (price, weight) than is actually needed (like the TC PA-5100 -- there are much cheaper drivers with similar BL/Mmd but smaller Xmax).

So it would be really helpful if you could give a few hints on how different the simulations are to real life, for example:

-- is the design I came up with close enough to the TH-115 that the SPL results should be *broadly* similar -- in other words, Hornresp consistently predicts SPL 3dB lower than a real-life tapped horn?

-- does a real-life tapped horn seem to have excursion about 60% or so of that predicted in Hornresp?

If we can establish rules-of-thumb like this (e.g. simulate in Hoirnresp, add 3dB to SPL and multiply excursion by 0.6) then it will enable people to take a much better shot at home-brewed tapped horns (not commercial obviously!).

Or if there's some missing factor in the simulations that can easily be fixed (and that you could tell us about) then maybe David could put this into Hornresp to improve the accuracy of the simulations.

Of course I realise size and weight matter for PA, that's exactly why I want the highest SPL I can get out of a given sized box (all our PA has to go in one Transit van) -- and it now *does* seem that the best answer is likely to be a tapped horn after all... :-)

Cheers

Ian

P.S. Hornresp input parameters attached as requested
Attached Images
File Type: gif tapped_input.gif (24.3 KB, 1119 views)
  Reply With Quote
Old 3rd March 2008, 02:16 AM   #1306
tb46 is offline tb46  United States
diyAudio Member
 
Join Date: Jan 2006
Location: Texas
Default B&C 15TBX100 designs

Hi Ian,
Thanks for posting the input parameters for your model. I'll attach what I came up with for the TH-115 for comparison, it is also based on a guess at the internals.
I'll also attach input and SPL for a B&C 15TBX100 design that trades off SPL for low frequency extension and box size.
Attached Images
File Type: jpg 2008_mar02_b&c_15tbx100_2_designs_small.jpg (70.7 KB, 1138 views)
__________________
Oliver
  Reply With Quote
Old 3rd March 2008, 08:55 AM   #1307
iand is offline iand  United Kingdom
diyAudio Member
 
Join Date: Feb 2007
Location: London
Default Re: B&C 15TBX100 designs

Quote:
Originally posted by tb46
Hi Ian,
Thanks for posting the input parameters for your model. I'll attach what I came up with for the TH-115 for comparison, it is also based on a guess at the internals.
I'll also attach input and SPL for a B&C 15TBX100 design that trades off SPL for low frequency extension and box size.
Hi Oliver

Thanks, I'll try your "TH-115" design tonight -- am I right in guessing that the average efficiency across the band is still close to what I got?

One way to resolve the possible "real-vs-simulated" discrepancy would be for Tom to simulate the exact TH-115 design in Hornresp (or AkAbak) and post just the results (SPL, excursion) for comparison with measurements -- this way we could see what the difference is, without Tom giving anything away about what his "104dB/W secret sauce" is -- how about it, Tom?

The second result is interesting -- you seem to be using the tapped horn to *reduce* the driver sensitivity from its direct-radiator value in order to get extended LF response, which is certainly a novel use of a horn, tapped or not :-)

Cheers

Ian

P.S. I'm not trying to get Tom to reveal *how* he gets such good results, just establish whether the differences are because he's so much better at tapped horn design than we are (I should hope so, but 3dB and 60% excursion is a hell of a lot!) or because the simulations are wrong -- or a bit of both...
  Reply With Quote
Old 3rd March 2008, 10:42 AM   #1308
iand is offline iand  United Kingdom
diyAudio Member
 
Join Date: Feb 2007
Location: London
Default Interesting new driver for tapped horn

Having been looking for something which might be better than the 15TBX100, I found that BMS released this since last time I looked:

http://www.bmspro.info/index.php?sho...282&id=5065912

http://tinyurl.com/2nf2tw

Details are sketchy so far, but going by the press releases it looks like a pretty damn impressive driver which could be a killer in a tapped horn:

-- carbon fibre cone (nice and strong for high CR), triple demodulating rings
-- 38mm coil in 12mm gap (15TBX100 has 25mm coil in 12mm gap)
-- Xmax 13mm linear, 16mm mathematical (15TBX100 is 6.5mm/9mm)
-- 1200W AES -- should be OK with 2.5kW amp...

If this simulates well, just think what the *real* performance could be... :-)

Cheers

Ian
  Reply With Quote
Old 3rd March 2008, 03:05 PM   #1309
tb46 is offline tb46  United States
diyAudio Member
 
Join Date: Jan 2006
Location: Texas
Default TH-115 guess models

Ian: You are correct, there is little difference between those two TH-115 models. Some of the real life differences may result from all the tight corners and bends that are not modeled in Hornresp. I'll try to model the new BMS driver tonight(?). Well, back to work.
__________________
Oliver
  Reply With Quote
Old 3rd March 2008, 06:40 PM   #1310
cowanaudio is offline cowanaudio  Australia
diyAudio Member
 
cowanaudio's Avatar
 
Join Date: Jul 2004
Location: Canberra
Default I get it!

G'day all

Iand said>P.S. I'm not trying to get Tom to reveal *how* he gets such good results, just establish whether the differences are because he's so much better at tapped horn design than we are (I should hope so, but 3dB and 60% excursion is a hell of a lot!) or because the simulations are wrong -- or a bit of both...

We have all been looking in the wrong place for the Tapped Horns hidden advantage.

This came to me while lying in bed at 2:00 this morning (as I am often inclined to do when faced with a challenge).

A loudspeaker radiating into half space has an output that decreases at -6dB with every doubling of distance. With a vented box, the origin is the driver or the port, depending on the freq the box is being driven at. Now a front loaded horn does the same thing, but the origin is still the driver, not the mouth of the horn. That means if you took a measurement at 10M out from a 4M long horn and compared it with a measurement 10M out from a vented box, the results would be very different, even if the horn output matched the VB output at 1M. I've measured this effect with front loaded horns and would have to assume it holds in a tapped horn. Tom measures his speakers at 10M for a VERY GOOD REASON. He could argue, and I'd have to agree, it more closely matches real life use. It accounts for the excellent output and efficiency seen in all the tapped horns that have been discussed here.

Ian, you are right in that a Tapped Horn has no advantage at one meter from the mouth when compared to a well designed vented box. As the distance increases, the advantage increases.

The plot thickens.

Cheers

William Cowan
  Reply With Quote

Reply


Collaborative Tapped horn projectHide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


New To Site? Need Help?

All times are GMT. The time now is 05:53 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.79%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio
Wiki