Hi Ian
Your trying pretty hard to make the case the Tapped horn does nothing unusual. Normally, when investigating something new one usually has better results with more of an open mind approach, as opposed to deciding on little information, but to each his own.
Also, so far, you are only comparing a loss-less ideal model where the port losses and size is not included in your model which is not the same as measured results in the real world.
As others have suggested, why not look at it from the other direction, take the measured results and see what it takes to equal them even if its only a prediction?
Take the measured response of the Mini, take its volume and see if you can make a vented box with the same sensitivity, low corner and include cabinet walls and a realistic port size which is taking up some of the space.
Same for the 4Xth115's below.
Also, a few Vented box Pro-sound subwoofers DO have real measured responses, to get more of a feel for that , you might want to examine these curves carefully relative to the TH-115 and TH mini.
http://www.jblpro.com/srx700/SRX700_Flash_Presentation/JBL.SRX718S.pdf
http://www.jblpro.com/srx700/SRX700_Flash_Presentation/JBL.SRX728S.pdf
Your assessment of how the horns work is curious, at least from my understanding.
If your assertion is right, it should be a snap for you to find a vented box combination which gives the same or greater efficiency shown below for four TH-115’s with the same physical size and low corner.. Don’t worry too much about if you can find / build the actual driver, for now anything goes.
Again, as with our other curves, this is in half space with 100W drive (25W each) and at 10M so it is a conservative “1w1m” equivalent.
Best,
Tom
Your trying pretty hard to make the case the Tapped horn does nothing unusual. Normally, when investigating something new one usually has better results with more of an open mind approach, as opposed to deciding on little information, but to each his own.
Also, so far, you are only comparing a loss-less ideal model where the port losses and size is not included in your model which is not the same as measured results in the real world.
As others have suggested, why not look at it from the other direction, take the measured results and see what it takes to equal them even if its only a prediction?
Take the measured response of the Mini, take its volume and see if you can make a vented box with the same sensitivity, low corner and include cabinet walls and a realistic port size which is taking up some of the space.
Same for the 4Xth115's below.
Also, a few Vented box Pro-sound subwoofers DO have real measured responses, to get more of a feel for that , you might want to examine these curves carefully relative to the TH-115 and TH mini.
http://www.jblpro.com/srx700/SRX700_Flash_Presentation/JBL.SRX718S.pdf
http://www.jblpro.com/srx700/SRX700_Flash_Presentation/JBL.SRX728S.pdf
Your assessment of how the horns work is curious, at least from my understanding.
If your assertion is right, it should be a snap for you to find a vented box combination which gives the same or greater efficiency shown below for four TH-115’s with the same physical size and low corner.. Don’t worry too much about if you can find / build the actual driver, for now anything goes.
Again, as with our other curves, this is in half space with 100W drive (25W each) and at 10M so it is a conservative “1w1m” equivalent.
Best,
Tom
Attachments
G'day again Ian
I'm thinking that the problem you might be having is that you don't appear to have measured any BR boxes at power. There are very good reasons why my "A" system at home uses a sealed bass section (with 10L displacement) and my "B" system uses a Tapped Horn for bass duties. You've seen my effort attempting to build a vented subwoofer that didn't suffer from port compression related effects, but the reality is that vented boxes do not perform when you start to pour in the power. Build one and measure it. I've built and measured a lot of speakers and I believe I have a pretty good idea what can be done and what can't be done with different loading techniques. You have scoffed about how far from optimal the 18LW1400 sub was, but I think you'll find it gets very close to what Hoffman's Iron Law says is possible, at least at 1W. The bottom line is I can get more output at low frequencies from a tapped horn than I can get from a vented box of the same total size. In it's simplest form these Tapped Horns are easier to build and often use cheaper drivers.
I encourage you to buy one of those BMS drivers and test it at 1KW in a vented box. I think you'll learn much. I also think the process will raise as many questions as it answers.
Cheers
William Cowan
I'm thinking that the problem you might be having is that you don't appear to have measured any BR boxes at power. There are very good reasons why my "A" system at home uses a sealed bass section (with 10L displacement) and my "B" system uses a Tapped Horn for bass duties. You've seen my effort attempting to build a vented subwoofer that didn't suffer from port compression related effects, but the reality is that vented boxes do not perform when you start to pour in the power. Build one and measure it. I've built and measured a lot of speakers and I believe I have a pretty good idea what can be done and what can't be done with different loading techniques. You have scoffed about how far from optimal the 18LW1400 sub was, but I think you'll find it gets very close to what Hoffman's Iron Law says is possible, at least at 1W. The bottom line is I can get more output at low frequencies from a tapped horn than I can get from a vented box of the same total size. In it's simplest form these Tapped Horns are easier to build and often use cheaper drivers.
I encourage you to buy one of those BMS drivers and test it at 1KW in a vented box. I think you'll learn much. I also think the process will raise as many questions as it answers.
Cheers
William Cowan
I hope Tom will answer these questions below:
-with horn type enclosures the max efficiency is around 50% while with direct radiators (like a vented cabinet) it is just around 25% (Keele)
All this without directivity!
How it is with the tapped horns?
- For me it seems like in the case of tapped horns there is a tradeoff - as with other types as well - between low cutoff frequency and efficiency. I mean with the same driver, and outer dimensions, you can have bigger efficiency with a higher low cutoff, or lower efficiency with lower cutoff frequency. Is it right?
What parameters dictates higher efficiency in a tapped Horn?
Thanks,
Tamas Tako
-with horn type enclosures the max efficiency is around 50% while with direct radiators (like a vented cabinet) it is just around 25% (Keele)
All this without directivity!
How it is with the tapped horns?
- For me it seems like in the case of tapped horns there is a tradeoff - as with other types as well - between low cutoff frequency and efficiency. I mean with the same driver, and outer dimensions, you can have bigger efficiency with a higher low cutoff, or lower efficiency with lower cutoff frequency. Is it right?
What parameters dictates higher efficiency in a tapped Horn?
Thanks,
Tamas Tako
A 'real' horn, a tapped horn, and a reflex box all have very different requirements for the drivers.
I don't see how they can be compared.
The tapped horn seems to model best with the driver Fs about a half octave above what you want the Fc of the horn to be. It also works well with Qts values that would seem too high for optimum reflex or a 'real' horn. The main disadvantage I see to the higher Qts is a larger throat and a higher x-max is needed.
Sometimes the gain over a reflex looks kind of low, but the driver is tuned much lower than you would normally try in a reflex.
I don't see how they can be compared.
The tapped horn seems to model best with the driver Fs about a half octave above what you want the Fc of the horn to be. It also works well with Qts values that would seem too high for optimum reflex or a 'real' horn. The main disadvantage I see to the higher Qts is a larger throat and a higher x-max is needed.
Sometimes the gain over a reflex looks kind of low, but the driver is tuned much lower than you would normally try in a reflex.
MaVo said:
I want to understand how the tapped horn compared to other solutions -- where it wins, where it's the same, where it loses, what its advantages and disadvantages are.
All enclosures have different advantages and disadvantages, that's engineering -- for example, reflex (and bandpass) boxes suffer from port noise and losses, are easy to design and build but need more drivers, tapped horns have resonant peaks above the passband and are hard to design and build but need fewer drivers which may (or may not) have parameters more suitable for manufacture.
I don't have an anti-tapped-horn axe to grind, I've been interested in tapped horns since before Tom invented/publicised the name, and certainly nothing that Tom has said is incorrect -- for example, with a single 15" driver the tapped horn will have much higher maximum output and a more buildable driver (just like it says on his website), and it works very well as a single box (unlike non-tapped horns) with high efficiency and output, and in big arrays it can get a higher limiting efficiency than a direct radiator.
So why can't I get a single tapped horn to exceed the performance of a reflex box of the same size and cutoff (with double the number of drivers, but probably the same cost)?
I'd really like this to be possible, otherwise I wouldn't have spent the last couple of years on and off going back to tapped horn designs to try and achieve this.
Cheers
Ian
P.S. Yes I'm ignoring port choking, the same way that tapped horn aficionados ignore the high only-just-out-of-band resonances which can colour the sound -- as I said, nothing's perfect
cowanaudio said:
I meant that the 18LW1400 didn't have anywhere near enough Xmax for a box tuned to 22Hz to handle anywhere close to Pmax, no pro driver does (though the Seismic 8196 I've got gets a lot closer) -- which is why your tapped horns will beat it at low frequencies. I also thought it's T-S parameters weren't close to optimum for that box, I could be wrong here so I'll try that out when I get home tonight.
I have measured the Seismic 8196 in a 400l box tuned to 25Hz, and with 1200W input the main problem was things falling off shelves and pans and cupboards rattling in the house 20 yards away. Not living in the middle of a field I didn't do frequency response measurements vs. level.
However I did note that the cone excursion at the box resonance was only a couple of mm pk-pk at 1200W, and the frequency for minimum motion didn't seem to vary significantly with drive level. This means the box/port Q must be reasonably high and the port losses fairly low, otherwise cone motion would be higher or resonant frequency would shift with drive level.
Am I missing something here?
And I've said all along that the driver requirements for reflex and tapped horns are very different, which means sometimes one will be easier/cheaper to build and other times the other one will be. What I'm questioning is the belief -- that seems to be widely held -- that tapped horns always have higher performance and lower cost.
Cheers
Ian
P.S. I don't imagine the BMS drivers will be so different given what I know about their design.
P.P.S. I'm not saying that reflex boxes don't suffer from port choking, this is a problem which tapped horns don't have.
Tom Danley said:
Hi Tom
I certainly don't have a closed mind on this otherwise I wouldn't still be trying to design a better tapped horn after a couple of years of trying!
I'm certainly *not* saying that that tapped horn is nothing special, or that a reflex can always equal a tapped horn, even ignoring port compression issues -- which I'm not saying can or should be ignored, and are a big advantage for the tapped horn, and I agree completely about this.
It's quite possible that the TH-Mini can't be beat, at least with drivers that can be built (though I'll have a go, but I'm sure it won't be easy). It's also very likely that 4 TH-115's can't be beat given the extremely high efficiency that results (and the large size that goes with it).
What this shows is that you've come up with designs which suit the tapped horn and show off its advantages, which is what I'd expect given your knowledge otherwise you should be sacked!
It seems most likely (from what I've found, and what you've said) that the size/cutoff/efficiency tradeoffs are different to a "normal" box, and that a tapped horn wins in the high efficiency/large array case (see TH-Mini and 4xTH-115). Certainly as soon as efficiency goes well over 100dB/W a direct radiator would need a lot of drivers to keep up which would put the cost through the roof -- but the case I found only needs 2, which seems quite acceptable.
Maybe the case I picked is one of those (lower efficiency <100dB/W, relatively small box for the cutoff) where the performance is the same, which is why I can't come up with a better tapped horn, or maybe I just haven't found the optimum design (though I've spent a lot of time with Hornresp trying). Or maybe this case has found a chink in the tapped horn armour...
Most pro-sound "subwoofers" -- name any manufacturer you want, including many famous ones (JBL, EAW...) -- are frankly rubbish compared to the boxes a few more enlightened manufacturers produce. Danley are probably at the top of this tree in my opinion and I'm not trying to put down the tapped horns that you've produced in *any* way, they are obviously very carefully optimised designs with excellent performance.
I did think my understanding of how horns -- and tapped horns -- worked was not only based on acoustics, but what you've said in the past about how they work -- certainly a single very-small-mouth tapped horn doesn't work anything like a classical horn does, especially regarding ripple sensitivity to driver parameters and acoustic impedance presented to the driver. I'm sure everyone who's tried designing a tapped horn has found this out for themselves when they've tried finding suitable drivers (or even working out what *is* suitable).
If you can explain this better than I did please feel free...
Cheers
Ian
P.S. I'll try beating the TH-Mini with a reflex but I suspect I'll lose if the design parameter set suits a tapped horn. How about you try and beat the "challenge" specification (99dB/W, 33Hz, 320l, 133dB continuous) which I suspect is well suited to a reflex?
Tom Danley said:
Hi Tom
I've done some modelling in Hornresp of what I think is something close to the TH-115 (but see comments about simulated vs. measured results) and a dual 18" reflex box with the same external size and cutoff frequency -- all results are for a single box in half space (but both boxes gain the same 10dB in a corner or a block of 4 according to Hornresp).
I know this is simulation only, but it's the only way I can see of doing a comparison of the two designs with a level playing field -- if there is any systematic error in Hornresp's calculations it should at least apply to both designs equally, so the differences should still be correct.
Box designs
---------------
For the TH-115 I'm using the B&C 15TBX100 in a 3m long conical tapped horn (tap points 20cm from each end) with 3:1 CR (best result I could get) and 2000cm2 mouth -- this gives just over 300l net volume which I calculate is what is inside the TH-115 allowing for all the internal partitions and the bottom rear corner chamfer.
For the reflex box I'm using 2 B&C 18PS76 with 320l net volume, (allowing 30l for ports which is a 10" diameter port per driver) tuned to 37Hz. Simulations for box this are with 2V applied to allow for 4ohm nominal impedance (2.82V for tapped horn).
Response curves
---------------------
The tapped horn has an impedance curve which is very close to the TH-115, but in spite of lots of tweaking I can't get either a flat response or one that's quite as efficient as the one in the data sheet -- from 37Hz to 100Hz it varies from 99dB/W to 102dB/W (3dB ripple), average is about 100.5dB/W, -6dB point from this average is 34Hz.
==> Tom, can you explain why the measured results in the data sheet look better than this? (I have noticed the poor frequency resolution, also the differences between the current data sheet response and the one I have saved from a couple of years ago
The reflex is not quite maximally flat, from 38Hz to 100Hz it varies from 100dB/W to 101dB/W (1dB ripple), average is 100.5dB/W, -6dB point from this average is 32Hz.
Maximum SPL
-----------------
Maximum cone travel for the tapped horn with 2.82V applied is 0.49mm at 47Hz (also at 33Hz). Mathematical Xmax for the 15TBX100 is 9.5mm, so worst-case power handling is 380W without exceeding Xmax, which is over 4dB less than the 1000W continuous power handling of the driver. This limits peak output to 126dB at 50Hz and 119dB at 33Hz (average would be 130.5dB at 1000W without Xmax limit).
Maximum cone travel for the reflex with 2V applied is 0.29mm at 51Hz (also 32Hz). Mathematical Xmax for the 18PS76 is 9mm, so worst-case power handling is 980W without exceeding Xmax, which is 1dB less than the 1200W continuous rating for the drivers. This limits peak output to 130.5dB at 50Hz and 125dB at 32Hz (average inband output would be 131.5dB at 1200W without Xmax limit).
Comparison
---------------
The 15PZB100 costs 225 euros+VAT, the 18PS75 cost 167 euros each + VAT which is 50% more for a pair than the single 15PZB100, but I would expect the extra cost of building the tapped horn box would compensate for this.
According to Hornresp (which might not agree with your SPL measurements for sensitivity, but the cone travel results are similar to those in your white paper) we have the following:
-- average sensitivity 38-100Hz is 101.5dB/W for both boxes
(single box on floor, both increase by 10dB in corner or block of 4)
-- inband ripple is 3dB for tapped horn, 1dB for reflex
-- -6dB point is 34Hz for tapped horn, 32Hz for reflex
-- Average SPL at rated power is 130.5dB for tapped horn, 131.5dB for reflex
-- Xmax limited SPL inband is 126dB for tapped horn, 130.5dB for reflex
-- Xmax limited SPL at -6dB point is 119dB for tapped horn, 125dB for reflex
Conclusions
---------------
Sensitivity is similar for both boxes, dual reflex has flatter response and higher Xmax-limited SPL across the operating bandwidth.
Even allowing for a 2dB possible sensitivity discrepancy between simulations and measurements in favour of the tapped horn (which I hope Tom can explain), the Xmax-limited maximum SPL for the tapped horn is still considerably lower.
Note that I've allowed a huge 10" diameter port for *each* driver so maximum velocity even at 32Hz is only 11m/s, which should be well below the point where significant port choking or chuffing happens.
Before anyone says "it's not fair comparing 2 18" drivers to a single 15" -- remember, nobody really cares what's inside the box, only results (size, weight, cost, frequency response, maximum SPL).
On this basis it seems that the dual-18" reflex does indeed equal or exceed the performance of the single-15" tapped horn.
I await comments with interest...
Cheers
Ian
P.S. If Tom can show otherwise I'll be perfectly happy to be wrong!
Several members, including myself, have noticed that the calculated amount of excursion is nowhere near the actual excursion in a tapped horn.
For instance, mine full tilt(1400WRMS) with a 40Hz tone, is moving only 5mm( I can't accurately measure it, but its much less than the 22+mm its capable of) each way or so.
For instance, mine full tilt(1400WRMS) with a 40Hz tone, is moving only 5mm( I can't accurately measure it, but its much less than the 22+mm its capable of) each way or so.
This brings up an old question of mine and others from a few months ago. Maybe it´s worth investigating actual excursion from built horns and compare it to the simulations. Although I haven´t done measuerements of excursion on my built horns so far, my feeling, too is that cone movement is less than predicted.
judtoff said:
At frequencies close to an impedance minimum the driver won't move much even at high drive levels, just as in a reflex. near the box tuning frequency.
It's not what the driver is capable of that matters, it's how much the driver actually moves at a given frequency and drive level compared to how much the simulations say it should be moving.
If they're different this would tend to suggest that the calculations in Hornresp are in error rather than the laws of physics being wrong
David, do you have any comment on this?
Ian
Sabbelbacke said:
We've had enough trouble with feelings, impressions and speculations already in this thread -- hard facts, science or measurements are needed!
I'm sure that if simulations disagree with real life then the simulations will be the source of error, Scotty had it right here...
Ian
Exactly that´s my point. If nobody double checked excursion on an actual real TH - how can we be sure that the simulation is 100% correct? After all - simulation is one thing, checking up on reality is another. Without the latter it remains speculation. Not everything is predictable by an simulation. The often mentioned effects of port-losses in a reflex-design are one example. This has to be measured in real life.
It´s only fair to raise an eyebrow if a well known driver seems to handle much more input power in a TH than in a reflex-design. One key to explain this observation is to check up on excursion.
I know of several simulation programs which calculate excursion not as peak-peak but as "effective excursion" (don´t know if thats the correct english terms). AjHorn for example does this. So a result of 1mm excursion in the simulation has to be mulitplied by 2.83 in order to get the "real live movement" from peak to peak (or multiplied by 2.83 / 2 to get +- results comparable to the numbers in spec-sheets). I don´t know how hornresp handles this, but I am sure not everybody is aware of this when using ajHorn so many misinterpretations can occur.
It´s only fair to raise an eyebrow if a well known driver seems to handle much more input power in a TH than in a reflex-design. One key to explain this observation is to check up on excursion.
I know of several simulation programs which calculate excursion not as peak-peak but as "effective excursion" (don´t know if thats the correct english terms). AjHorn for example does this. So a result of 1mm excursion in the simulation has to be mulitplied by 2.83 in order to get the "real live movement" from peak to peak (or multiplied by 2.83 / 2 to get +- results comparable to the numbers in spec-sheets). I don´t know how hornresp handles this, but I am sure not everybody is aware of this when using ajHorn so many misinterpretations can occur.
iand said:
What you want to find out is very interesting, but maybe you should do some actual work for it instead of criticising others and proposing challenges. Constructive criticism is the key.
The real problem is, that you want other people to do the work for you. If you really want to achieve something, then DIY.
Sabbelbacke said:
I'm hoping we'll get a reply soon from David McBean (and maybe Tom) about this.
If there is any error in excursion calculations -- like x2 from confusing peak and peak-peak -- this would almost certainly apply to all simulations, since the same equations are used everywhere. So any differences in excursion between different boxes are likely to be correct -- however if the absolute numbers are wrong then this changes lots of conclusions...
In the Hornresp simulations I did for the "TH-115" (as close as I could guess) the maximum excursion was +/-0.49mm at 47Hz with 2.82V applied. In Tom's white paper (which refers to the TH-115) he says that the peak excursion with 63V applied is just over 6mm (I assume this means +/-6mm) at 46Hz, which would give 0.28mm with 2.82V applied.
This is just over half the number predicted by Hornresp, which is not good :-(
If Tom's figure is right then this also explains why my calculations showed the TH-115 to be Xmax limited at 380W; with his excursion figure this goes up to 1160W which then exceeds the thermal rating of the driver -- so my conclusion that the TH-115 was Xmax limited is then wrong.
I did clearly say that all my figures were based on Hornresp simulations and the assumption that these were correct, and that "If Tom can show otherwise I'll be perfectly happy to be wrong!" (my exact words) -- and I stand by that
Cheers
Ian