"plot5 40Hz to 320Hz is 3octaves. You must be joking?"
What am I joking about? It looks useable to 300hz. I was thinking about crossing it at 250hz to a smallish horn-loaded 10" or 12" driver.
"plot4 35Hz to 280 is 3octaves and you want to use more?"
"I would suggest that plot5 may be OK from 40Hz to 120Hz."
Why do you say only 120hz? I was thinking about a damped 1/4W stub for 150hz and maybe 225hz (like the DTS-20 uses). http://i56.photobucket.com/albums/g196/dkleitsch/dts20inside.jpg
"Plot4 will sound very different and in my opinion probably pretty damn terrible."
It was for reference only, the Lab 12 (single).
What am I joking about? It looks useable to 300hz. I was thinking about crossing it at 250hz to a smallish horn-loaded 10" or 12" driver.
"plot4 35Hz to 280 is 3octaves and you want to use more?"
"I would suggest that plot5 may be OK from 40Hz to 120Hz."
Why do you say only 120hz? I was thinking about a damped 1/4W stub for 150hz and maybe 225hz (like the DTS-20 uses). http://i56.photobucket.com/albums/g196/dkleitsch/dts20inside.jpg
"Plot4 will sound very different and in my opinion probably pretty damn terrible."
It was for reference only, the Lab 12 (single).
If you guys need a “tweener” tapped horn that won’t be asked to go much below 40Hz-50Hz, then the 18 Sound 10W400 is a good choice for a very broad bandwidth. William Cowan mentioned this woofer before, but it is buried deep within this thread. I just wanted to remind people of this very good tapped horn woofer. It is quit easy to get 2 full octaves out of this woofer in a tapped horn.
Rgs, JLH
Rgs, JLH
|18Hz Tapped Horn
Def_Const |Horn Dimensions
{
a1 = 400e-4; |Area at throat (cm^2)
a2 = 460e-4; |Area at rear of driver (cm^2)
a3 = 2200e-4; |Area at front of driver (cm^2)
a4 = 2400e-4; |Area at mouth (cm^2)
l1 = 32e-2; |Distance from throat to rear of driver (cm)
l2 = 480e-2; |Line distance from rear of driver to front of driver (cm)
l3 = 72e-2; |Distance from front of driver to mouth (cm)
}
Def_Driver 'Dr1'
| Tang Band W8Q-1071
Sd=920cm2
fs=27Hz
Qes=0.32
Qms=4.36
Vas=176.44L
Re=6.4ohm
Le=2mH
system 'S1'
Driver Def='Dr1' Node=1=0=3=4
Waveguide 'W1' Node=2=3 STh={a1} SMo={a2} Len={l1} Conical
Waveguide 'W2' Node=3=4 STh={a2} SMo={a3} Len={l2} Conical
Horn 'H1' Node=4 STh={a3} SMo={a4} Len={l3} Conical
Def_Const |Horn Dimensions
{
a1 = 400e-4; |Area at throat (cm^2)
a2 = 460e-4; |Area at rear of driver (cm^2)
a3 = 2200e-4; |Area at front of driver (cm^2)
a4 = 2400e-4; |Area at mouth (cm^2)
l1 = 32e-2; |Distance from throat to rear of driver (cm)
l2 = 480e-2; |Line distance from rear of driver to front of driver (cm)
l3 = 72e-2; |Distance from front of driver to mouth (cm)
}
Def_Driver 'Dr1'
| Tang Band W8Q-1071
Sd=920cm2
fs=27Hz
Qes=0.32
Qms=4.36
Vas=176.44L
Re=6.4ohm
Le=2mH
system 'S1'
Driver Def='Dr1' Node=1=0=3=4
Waveguide 'W1' Node=2=3 STh={a1} SMo={a2} Len={l1} Conical
Waveguide 'W2' Node=3=4 STh={a2} SMo={a3} Len={l2} Conical
Horn 'H1' Node=4 STh={a3} SMo={a4} Len={l3} Conical
FlipC said:
The problem I see is most of the common high Xmax drivers (Ciare 15.00SW and B&C 15TBX100 for example) can only do about 125dB at 40Hz in most of my simulations in 2Pi. How some one can get another 10dB out of a simular woofer is beyond me. Either we know less than what we thought, or something is fishy with those numbers. Really, another 10dB from the same woofer?
Interesting, thanks. The driver I assumed it was sims a bit better overall to my way of thinking since I can only get into the 133s from 40 - 245 Hz/~330 L net/half space/music program with the B&C, but that's all there is with Hornresp AFAIK. I'd have to learn Akabak to use more of what I know about horn design to try and squeeze a bit more out of it, but I'm using published specs and at this level it apparently doesn't take much variance to enhance/ruin it, so not going to waste any more time on it, especially since there's no use for such a sub in a HIFI/HT app.
For the latter, it looks good from below audibility to 230+ Hz/~600 L/120+ dB music program/quarter space, so laid out as a stage of sorts along the sound wall to mount speakers/whatever with a central terminus works for me if the rear wall is diffuse enough. Or if ceiling height permits it could be made into a seating riser since the close proximity of it will negate the wall loading loss.
GM
For the latter, it looks good from below audibility to 230+ Hz/~600 L/120+ dB music program/quarter space, so laid out as a stage of sorts along the sound wall to mount speakers/whatever with a central terminus works for me if the rear wall is diffuse enough. Or if ceiling height permits it could be made into a seating riser since the close proximity of it will negate the wall loading loss.
GM
FlipC said:
With a TH-115 sized TH (multiple sections, slower expansion in first part like Tom's patent) the maximum SPL at the Xmax limit I can get for one box in 2pi is as follows:
15TBX100 (Xmax=9mm): 129dB at 66V (550W/8ohms)
PA390 (Xmax=20mm) : 137dB at 145V (2600W/8ohms)
(1w efficiency is the same for both at 102dB, series inductance used to flatten response ==> L=4mH for both, maximum cone travel is at 52Hz)
So the 15TBX100 is clearly Xmax limited (Pe=1000W), the PA390 is not (Pe=2000w) -- this is pretty much the only PA-type driver I've found which isn't Xmax limited before hitting its thermal limit.
The impedance minima/peaks are in almost exactly the same places as the TH-115 data sheet so I'd expect cone travel to be similar. I can't see any way you can get 133dB (or more) over the entire bandwidth out of a TH loaded with a 15TBX100 :-(
Unless you measure with bandlimited pink noise, then only a fraction of the power is at the maximum travel frequencies and Xmax won't be exceeded until much higher power levels -- maybe this is how come the TH-115 is rated at 2000W input?
Of course if you hit it with a solid G fundamental it'll still only handle 550W...
Happy New Year!
Ian
Hi Iand, all
I am probably starting to sound like a skipping record by now but it is critical in “trusting” computer models, that the trust is formed as a result of measuring the actual item and comparing it to the models and tweaking the model where needed..
For example for small horns like discussed here, an issue relative to this, Akabak often predicts higher “Q”s than one measures on the real thing, occasionally some predicted features in the response don’t even show up at all.
Also, there is an internal loss element, which is required, which IS NOT included in the normal models, unless you have put it in.
Don’t get me wrong, Akabak is a wonderful tool, I wouldn’t have been able to make any tapped horns that worked properly without it, but it is not the last word, a good measurement is.
It is the critical tuning and specific proportions needed which prevented any of the similar looking configurations form being successful products or even existing as products today. You can’t stumble into these alignments.
So far as excursion, this is a factor here too, the peaks are usually not as high and the dips not as deep as what is generally predicted, if your going to focus on getting the max out, you must also adjust your models to fit measured reality as closely as possible.
Also, in pro sound, it IS customary to measure power etc using pink noise.
Dave Martin felt this was an important issue, a sine wave was good for mechanical testing while noise was much closer to real program material.
Both are good signals, the TEF uses a swept sine while Pat at Synaudcon uses noise for his tests.
That pink noise signal has a 6dB peak to average ratio, that is a 1000Watt rating requires an amplifier that can put out 4000Watts.
Our measurements for the TH-115 are done as stated, at 10 meters half space driven at 100Watts nominal input for that Z and measured with an HP3456a Voltmeter, Earthworks instrumentation mic and TEF machine.
What is show is EXACTLY what was measured and anyone can repeat it.
Max SPL is an imaginary figure in pro sound, NO ONE I have heard of actually measures this number or if they do, does it in a meaningful way (you might recall discussion on PSW).
Some supply a figure, which seems to have been arrived at by directing a lightning strike to the terminals. I and others have measured well respected boxes who’s actual output falls as much as 22dB short of what is claimed, this area like politics is full of nonsense.
Generally that “peak” figure is found by taking the sensitivity and power rating to calculate the max theoretical continuous output and then adding the 6dB peaks above the average level and then in some cases, add an additional 3 dB to that to account for “program power” and some add even more. This mathematical slight of hand is how a floor monitor can have a rated peak SPL of 148dB even though it has a 15 and 2 inch comp driver.
Our “max SPL” figure for our subwoofer is also imaginary but the idea of doing it the traditional way made some ill, we took the sensitivity and rated power and then added 3 dB not 6 dB. That figure is more consistent with what can measure with music and a peak hold SLM.
Reality is that power compression begins between about 1/10 and 1/8 rated power for essentially all modern drivers and this profoundly effects what happens in the top half of the systems power capacity. As a result, it normally takes much more power to reach the excursion one would predict based on a low level measurement.
For our full range boxes, we were able to use an independent lab to define th specifications, in this case, maximum output is defined the following way.
The speaker is driven with band limited pink noise, every 5 min the level is increase until the frequency response at any point has fallen –3dB from the response shape one gets at 1Watt.
Unfortunately, while this is a realistic and conservative spec for installed sound, it is not consistent with the pretend figures so popular in the live sound world.
Lastly, driver selection.
If your making something for yourself, you can use any driver you want to.
If you were in my shoes some years ago when I designed the TH-115, you would have faced some additional constraints.
First, I had to use a driver which is widely available at the time, preferably one that would be in stock around the country, I had to use a driver which already had the bugs worked out. I could not afford to experiment with anything that wasn’t proven or readily available.
The PA390 might be a cool driver but it is not clear that they are widely available in the USA, or available at all nor do I have any sense it has a “track record” or acceptance yet.
I do have some experience with “exotic drivers” and my object here has been trying to find acoustic alignments and configurations, which produce an acoustical advantage while using more conventional drivers.
So far as a “what can be done” benchmark, keep an eye out for a new box we have using two twelve inch drivers.
It is 36 by 36 by 16 inches and outperforms the TH-115.
This should be on the web site shortly.
If you wanted to go lower for a 5 string, the TH-50’s goes down lower, in the Navy Pier Imax theater in Chicago, four were flat to 10Hz in room.
They are the four tiny squares in the center on the floor, the other boxes with the lights on them are SH-96’s
Iand, you have seemed pretty skeptical about these a number of times, if your still “in the business” why not call Mike, tell him you’re an old acquaintance from the Servo days and see about getting a demo to play with.
I am probably starting to sound like a skipping record by now but it is critical in “trusting” computer models, that the trust is formed as a result of measuring the actual item and comparing it to the models and tweaking the model where needed..
For example for small horns like discussed here, an issue relative to this, Akabak often predicts higher “Q”s than one measures on the real thing, occasionally some predicted features in the response don’t even show up at all.
Also, there is an internal loss element, which is required, which IS NOT included in the normal models, unless you have put it in.
Don’t get me wrong, Akabak is a wonderful tool, I wouldn’t have been able to make any tapped horns that worked properly without it, but it is not the last word, a good measurement is.
It is the critical tuning and specific proportions needed which prevented any of the similar looking configurations form being successful products or even existing as products today. You can’t stumble into these alignments.
So far as excursion, this is a factor here too, the peaks are usually not as high and the dips not as deep as what is generally predicted, if your going to focus on getting the max out, you must also adjust your models to fit measured reality as closely as possible.
Also, in pro sound, it IS customary to measure power etc using pink noise.
Dave Martin felt this was an important issue, a sine wave was good for mechanical testing while noise was much closer to real program material.
Both are good signals, the TEF uses a swept sine while Pat at Synaudcon uses noise for his tests.
That pink noise signal has a 6dB peak to average ratio, that is a 1000Watt rating requires an amplifier that can put out 4000Watts.
Our measurements for the TH-115 are done as stated, at 10 meters half space driven at 100Watts nominal input for that Z and measured with an HP3456a Voltmeter, Earthworks instrumentation mic and TEF machine.
What is show is EXACTLY what was measured and anyone can repeat it.
Max SPL is an imaginary figure in pro sound, NO ONE I have heard of actually measures this number or if they do, does it in a meaningful way (you might recall discussion on PSW).
Some supply a figure, which seems to have been arrived at by directing a lightning strike to the terminals. I and others have measured well respected boxes who’s actual output falls as much as 22dB short of what is claimed, this area like politics is full of nonsense.
Generally that “peak” figure is found by taking the sensitivity and power rating to calculate the max theoretical continuous output and then adding the 6dB peaks above the average level and then in some cases, add an additional 3 dB to that to account for “program power” and some add even more. This mathematical slight of hand is how a floor monitor can have a rated peak SPL of 148dB even though it has a 15 and 2 inch comp driver.
Our “max SPL” figure for our subwoofer is also imaginary but the idea of doing it the traditional way made some ill, we took the sensitivity and rated power and then added 3 dB not 6 dB. That figure is more consistent with what can measure with music and a peak hold SLM.
Reality is that power compression begins between about 1/10 and 1/8 rated power for essentially all modern drivers and this profoundly effects what happens in the top half of the systems power capacity. As a result, it normally takes much more power to reach the excursion one would predict based on a low level measurement.
For our full range boxes, we were able to use an independent lab to define th specifications, in this case, maximum output is defined the following way.
The speaker is driven with band limited pink noise, every 5 min the level is increase until the frequency response at any point has fallen –3dB from the response shape one gets at 1Watt.
Unfortunately, while this is a realistic and conservative spec for installed sound, it is not consistent with the pretend figures so popular in the live sound world.
Lastly, driver selection.
If your making something for yourself, you can use any driver you want to.
If you were in my shoes some years ago when I designed the TH-115, you would have faced some additional constraints.
First, I had to use a driver which is widely available at the time, preferably one that would be in stock around the country, I had to use a driver which already had the bugs worked out. I could not afford to experiment with anything that wasn’t proven or readily available.
The PA390 might be a cool driver but it is not clear that they are widely available in the USA, or available at all nor do I have any sense it has a “track record” or acceptance yet.
I do have some experience with “exotic drivers” and my object here has been trying to find acoustic alignments and configurations, which produce an acoustical advantage while using more conventional drivers.
So far as a “what can be done” benchmark, keep an eye out for a new box we have using two twelve inch drivers.
It is 36 by 36 by 16 inches and outperforms the TH-115.
This should be on the web site shortly.
If you wanted to go lower for a 5 string, the TH-50’s goes down lower, in the Navy Pier Imax theater in Chicago, four were flat to 10Hz in room.
They are the four tiny squares in the center on the floor, the other boxes with the lights on them are SH-96’s
Iand, you have seemed pretty skeptical about these a number of times, if your still “in the business” why not call Mike, tell him you’re an old acquaintance from the Servo days and see about getting a demo to play with.
Thank you for helping drive that point home.
This inevitability escapes many who obsess about the maximum power specification of a a driver, Rather than focusing on what is done with the power, the max figure is used as a measure of performance.
A bit like determining top speed of a vehicle by virtue of the highest number on the speedometer.
Syd
Tom Danley said:
Tom, please don't get me wrong, I have absolutely the greatest of respect for your speaker design abilities, I've been following your writings (and products) for many years -- but as Scotty said, ye canna break the laws o' physics! (maybe you can bend them a bit...)
There's no doubt that you've been able to squeeze performance levels -- efficiency, HF bandwidth -- out of TH which nobody else has been able to meet, and that this is probably partly due to tweaks that nobody else has found yet ("secret sauce") and partly due to real-life losses which are not modelled.
I was taking the "no power compression" method of calculating maximum SPL, which is not realistic but is all that's possible given the lack of real power compression figures for most drivers -- it's nevertheless OK to compare one design (or driver) with another, so long as everyone understands it's not "real".
I'm not at all sceptical that the TH-115 is an excellent design, your measurements are accurate, and that simulations are just that, a prediction of performance which may or may not be very accurate -- thought it has been reported now on more than one occasion that Hornresp's predictions are close to reality, I can believe that the cone travel predictions could very easily be 10-20% in error, but this is nowhere near enough to explain the difference between 550W and 2000W power handling.
Given that the TH-115 is measured using bandlimited pink noise then the 2000W stated power handling figure is perfectly believable and accurate, just so long as it's understood that it won't handle this power as a single note anywhere in the passband without hitting the driver Xmax limit at some points -- our bass player might not be so happy with this, but it's still probably better than most (all?) other boxes on the market.
I think this is perhaps the biggest area of misunderstanding in previous discussions; I (and others) have been trying to design TH to handle such high power (for example, 2000W using the 15TBX100) without hitting the Xmax limit, and couldn't understand how this was possible given the 9mm Xmax -- and the answer is, it isn't. Now I understand that I'm much happier
I also realise the reasons behind your choice of driver, especially long-term availability -- a lot of small-volume "exotic" driver manufacturers go bust after a year or two, and I've said on many occasions that the PASUB drivers do look very attractive but are unproven and single-source, so if you design a box using them and they go bust then you're stuffed.
The TH is indeed capable of squeezing extraordinary performance out of relatively ordinary drivers in relatively small boxes, but the laws of physics still apply -- to get such high outputs *everywhere* across the bandwidth still needs more volume displacement (Sd x Xmax) than "normal" PA-type drivers like the 15TBX100 can provide (the MTX9515 in the TH-50 probably can, even with lower -3dB).
So will your new 2x12" box use "PA-style" drivers, or ones with much higher Xmax to alleviate this problem?
Happy New Year!
Ian
P.S. There's still some ambiguity in the TH-115 specs since you switched to a 4 ohm driver; the 106dB@2.82V sensitivity figure is correct (2W nominal input, so 3dB more than the 8ohm version spec of 103dB/W), but the SPL plot still says "100W at 10m" which is wrong for a 4ohm nominal speaker with 28V applied -- though the Zmin shows it's really a "6 ohm nominal" speaker even though it uses a 4ohm driver...