Its almost ironic that i came up with almost the same question sitting in english class today. My question is that if you create a curve in the corners to keep the cross sectional volume the same throughout the TL would that change the length of the line. heres a link to describe what im talking about differning lengths
the red line is the typical line people use to determine the length of a folded line, but the green one follows the center of the bend. The green line will always be .785% as long as the red one per 90 degree turn. So, which one is correct.??
the red line is the typical line people use to determine the length of a folded line, but the green one follows the center of the bend. The green line will always be .785% as long as the red one per 90 degree turn. So, which one is correct.??
OK, depending on how you treat the corners of the bend you might model a different length around the corner. But if you only have one bend, how big a difference does this really make in the total assumed length? Probably not very much. The total length of the TL is not different by 21.5% just the local bend length. Model what you have for a geometry. My guess is that the difference between the two methods of treating and modeling the bend will not make that much of a difference in the overall SPL response prediction.
I hope you were paying attention in English class and that this site does not negatively impact your final grade!
yeah, that is what I meant to say.
For this model, the cross section has a dimention of 18 3/4'' x 18 3/4'' and a line length of 63 3/4'' so the curve will take up almost half of the length of the line. The length of the line not taken up by the curve in the green line is only 27.594'' long. I do agree that in most cases, the bend and how you model it will not have much of an impact on the final result, but with this one, the curve takes up most of the length of the line, so a difference in length great enough to change the spl result is probibly present...... but I am in no way shape or form the expert on tl design.
What are you really after here? Go and find someone who's actually built some Tl's and listen to them. For me the route to audio nirvana is actually understanding what you want from your system!
I built some Olsen Ariels (200hrs!)and although they do have good reach in the bass they do not have anything like the impact of a sealed 10" woofer in the same volume. Don't expect miracles.
Good luck,
simon
the green line is about 7.7 inches shorter than the red line, so out of 63.75 thats a 12.19% difference in line length.
simon dart said:
What I am after is fidelity, and a bass that's not only loud.
The problem is that I haven't been able to find someone who has built a TL around here (please don't ask).
And, as Mr King says in one of his papers: conventional ported boxes do not account for standing waves, and I believe that this is a major drawback with ported/sealed boxes (and standard design programs).
And for Mr. King: I already looked into the sheets and the learning curve doesn't seem that "steep" at all.
Nevertheless, I've come accross some impeding difficulties for which I haven't found the answer in your papers.
One of them is this:
I design a straight TL using the MT TQWT sheet. I adjust it untill I get an acceptable SPL response.
Then I fold it, and define the discrete sections conforming to the application note of that sheet (I compute the intermediate cross sections where the dashed lines are drawn).
The SPL response that I get after that is something that's FAR from the original one.
When I looked into the "Projects" section from your website, at the Focal Two-Way Transmission Line, I saw that you defined the intermediate cross-sectional areas somewhat differently. And the difference seems to be that you consider the cross-section planes to be perpendicular to the transmission line path, and not oblique to it, like in the application note.
I promised some clarifying drawings, but I've had some troubles making them, I think I will be able to send them tomorrow.
Okay, I appreciate you can't get to hear some TLs, which is a shame. In my opinion the practical compromises that have to be dealt with to get the high quality result you're after are trickier to resolve than any other form of loading. Here's some of my thoughts on why they are such a lottery.
1. So you've done the math and decided on the area, taper and length of your line, which is not easy to hit accurately as you will not know exactly how much stuffing you need(see below) untill you've built it!
2. Line harmonics. The line, unless it is closed, infinite and critically damped is a resonant structure just the same as any other box, it just happens to deal with the backwave from the driver in a different way: by exiting a tuned pipe. Don't lump the effective line length (and its resultant loading on the driver) together with the actual resonant qualities of the line, they are not tied together except at the fundamental.
Where you position your driver in the line can make a big difference here, think about positioning at one third, fifth or one seventh of the line length to suppress the odd order harmonics.
3. You mentioned standing waves, TL's don't have standing waves? Internal reflections from the partitions, walls etc make a resonant signature just the same as any other box, only this time superimposed on the line harmonics. So it makes sense to minimise them, but how can you predict thier behavour? The internal stuctures are not obstacles to flow: How much air is moving in the line? Is it enough to even tickle the edges to into provoking significant turbulence? What's going on here is multiple orders of standing waves interacting. Those you are stuck with once the form of the line has been set.
So once the enclosure has been built that leaves attempted suppression of line harmonics and standing waves by playing with the density of the stuffing(which alters the effective line length!). To do this adequately without killing the bass which will be your greatest challenge.
The task is not hopeless but it is unpredictable, just the same as any other box. You could hit the jackpot or crash and burn. I'd sick to a proven design if you want to minimise the risk.
B&W looked at this long and hard for high performance in the TL, the result was the amazing Nautilus. No folding. Closed lines. Critically damped. Err.. isn't that a really a sealed box QTC 0.5 as far as the bass loading goes?😀
Simon
1. So you've done the math and decided on the area, taper and length of your line, which is not easy to hit accurately as you will not know exactly how much stuffing you need(see below) untill you've built it!
2. Line harmonics. The line, unless it is closed, infinite and critically damped is a resonant structure just the same as any other box, it just happens to deal with the backwave from the driver in a different way: by exiting a tuned pipe. Don't lump the effective line length (and its resultant loading on the driver) together with the actual resonant qualities of the line, they are not tied together except at the fundamental.
Where you position your driver in the line can make a big difference here, think about positioning at one third, fifth or one seventh of the line length to suppress the odd order harmonics.
3. You mentioned standing waves, TL's don't have standing waves? Internal reflections from the partitions, walls etc make a resonant signature just the same as any other box, only this time superimposed on the line harmonics. So it makes sense to minimise them, but how can you predict thier behavour? The internal stuctures are not obstacles to flow: How much air is moving in the line? Is it enough to even tickle the edges to into provoking significant turbulence? What's going on here is multiple orders of standing waves interacting. Those you are stuck with once the form of the line has been set.
So once the enclosure has been built that leaves attempted suppression of line harmonics and standing waves by playing with the density of the stuffing(which alters the effective line length!). To do this adequately without killing the bass which will be your greatest challenge.
The task is not hopeless but it is unpredictable, just the same as any other box. You could hit the jackpot or crash and burn. I'd sick to a proven design if you want to minimise the risk.
B&W looked at this long and hard for high performance in the TL, the result was the amazing Nautilus. No folding. Closed lines. Critically damped. Err.. isn't that a really a sealed box QTC 0.5 as far as the bass loading goes?😀
Simon
Fact: a properly designed/build TL has a better predicted/measured response correlation.
Can you say the same thing about closed/ported boxes?
It is a known fact that standard box design software compute the SPL response (and not only) from box volume only, completely ignoring standing waves. And this is why you get those unpredicted pits and bumps in the measured response of such an enclosure. If you know a way to cope with that, I'll go with standard ported boxes.
Can you say the same thing about closed/ported boxes?
It is a known fact that standard box design software compute the SPL response (and not only) from box volume only, completely ignoring standing waves. And this is why you get those unpredicted pits and bumps in the measured response of such an enclosure. If you know a way to cope with that, I'll go with standard ported boxes.
I'm sorry, that's way too simplistic for me. The correlation between model and measured is just one part of the story: to what degree is what you are measuring important anyway? If that was the case all hifi that measures the same would sound the same which it clearly doesn't.
Our ability to quantify sound quality by steady state amplitude anaysis is close to useless, just look at amplifier measurement, thousands of circuits measure as such within a dB or so, they all sound different !
We are all relatively insensitive to steady state variations in amplitude, which are of zero use in evolutionary terms, and massively more sensitive to change over tim, which is what natural sound is. The brain integrates and processes what we hear to a very large degree, directional cues are purely the result of such processing, we look for harmonic relationships to make sense of the nature of sounds, and we get confused if these cues are unnatural: listening fatigue.
Don't get me wrong, I feel spectral flatness is important, but so is so much else, which is often swept under the carpet, perhaps because it's a bit tricky to deal with!
Speakers are a total mess in the time domain, they are just a bunch of associated resonant systems:
Wave shapes are altered by inter-driver and crossover phase shifts. Drivers and enclosures store, smear and release energy over 10's, maybe 100's of milliseconds with not much spectral or directional consistency, adding and subtracting in space and time in a manner that is close to chaotic. Enclosures create confusing (the dimensions are often close to those of our head) patterns of reflections through edge diffraction, and so on.
None of this shows up in true relation to its sonic impact in steady state analysis. All these factors have to be dealt with to attain high performance: none of them you can successfully model as an amateur. Experience has taught me that models are useful only as a starting point. The rest is attention to every detail you have influence over. The question is then how far do you want to go?
If steady state pedictability was the overiding factor in determining sound quality, we could have all gone home long ago.
Simon
Our ability to quantify sound quality by steady state amplitude anaysis is close to useless, just look at amplifier measurement, thousands of circuits measure as such within a dB or so, they all sound different !
We are all relatively insensitive to steady state variations in amplitude, which are of zero use in evolutionary terms, and massively more sensitive to change over tim, which is what natural sound is. The brain integrates and processes what we hear to a very large degree, directional cues are purely the result of such processing, we look for harmonic relationships to make sense of the nature of sounds, and we get confused if these cues are unnatural: listening fatigue.
Don't get me wrong, I feel spectral flatness is important, but so is so much else, which is often swept under the carpet, perhaps because it's a bit tricky to deal with!
Speakers are a total mess in the time domain, they are just a bunch of associated resonant systems:
Wave shapes are altered by inter-driver and crossover phase shifts. Drivers and enclosures store, smear and release energy over 10's, maybe 100's of milliseconds with not much spectral or directional consistency, adding and subtracting in space and time in a manner that is close to chaotic. Enclosures create confusing (the dimensions are often close to those of our head) patterns of reflections through edge diffraction, and so on.
None of this shows up in true relation to its sonic impact in steady state analysis. All these factors have to be dealt with to attain high performance: none of them you can successfully model as an amateur. Experience has taught me that models are useful only as a starting point. The rest is attention to every detail you have influence over. The question is then how far do you want to go?
If steady state pedictability was the overiding factor in determining sound quality, we could have all gone home long ago.
Simon
simon dart said:
If modeled in Martin's sw you can be 90% of the way... it has been shown to be pretty accurate.
This is kinda simplistic... it may be close for an untapered line, but as soon as any taper is introduced these harmonics start shifting around -- the 3rd is the one you want to aim at killing.
Because of the damping used to kill the harmonics of the fundemental, and the shape of the line (ie these standing waves are fairly HF), they are usually mostly taken care of. Of course one has to always worry about timed smeared reflections coming back thru the cone.
Actually this appears to be a myth based on the moving fibers theory of stuffing. It is the line geometry that changes the effective length -- the heavier the taper the shorter the line can be.
Since the introduction of King's & Augspurger's models the result is WAY less unpredictable -- one thing that has been shown is that an aweful lot of TLs built on classic rules-of-thumb wer poking around in unfertile territory.
A heavily tapered half-wave transmission line.
dave
Bricolo said:
Except for the major difference that a 1/2 line is closed at both ends and a 1/4 wave is open at one end.
dave
>Actually this appears to be a myth based on the moving fibers theory of stuffing.
====
Hmm, don't know for sure how it works, but MJK's TL ws clearly predicts that stuffing does lower Fb considerably in a straight TL, hence its effective line length, and to a much lessor extent due to the mass (compression) loading at the terminus in a reverse tapered one. I'm with you though, better to tune it through line length/taper ratio/driver position and minimal stuffing than just seeing how much/what type stuffing sounds the best.
GM
====
Hmm, don't know for sure how it works, but MJK's TL ws clearly predicts that stuffing does lower Fb considerably in a straight TL, hence its effective line length, and to a much lessor extent due to the mass (compression) loading at the terminus in a reverse tapered one. I'm with you though, better to tune it through line length/taper ratio/driver position and minimal stuffing than just seeing how much/what type stuffing sounds the best.
GM
I'm not familiar with King's & Augspurger's models, so I bow to Daves better education!
I do stand by my other points though, modelling only gives you an idea of where you will be heading. Bass quality is something that is defined by many other factors than low frequency alignment (system voicing, driver characteristics, amplifier characteristics, system placement, crossover quality, room quality etc)
After all I do run some Ariels. They are TL's because the designer needed to coax as much extension out of the high F3 drivers as he could, which is a valid approach. Not because he thought TL's were any better than any other type of LF loading.
Build a TL if you're curious, but I don't accept that they have an inherently superior answer to all the compromises a designer has to try and square. The bigger picture is what gets you the higher fidelity.
Simon
I do stand by my other points though, modelling only gives you an idea of where you will be heading. Bass quality is something that is defined by many other factors than low frequency alignment (system voicing, driver characteristics, amplifier characteristics, system placement, crossover quality, room quality etc)
After all I do run some Ariels. They are TL's because the designer needed to coax as much extension out of the high F3 drivers as he could, which is a valid approach. Not because he thought TL's were any better than any other type of LF loading.
Build a TL if you're curious, but I don't accept that they have an inherently superior answer to all the compromises a designer has to try and square. The bigger picture is what gets you the higher fidelity.
Simon
planet10 said:
By 3rd harmonic you mean the first one in the pipe right? So that's the first harmonic the pipe will produce but it's the 3rd harmonic of the fundamental...
simon dart said:
Is it unpopular on this board if I say that I'm just as interested in the building as I am in the product 🙂 A TL is 2nd or 3rd on my to-do list of speakers to build just because it sounds like fun to build and they look cool. I have no idea how they sound or where I'd put it... Chances are they'll become my garage workshop speakers!!
simon dart said:
We have to live with the idea of simplification. That's part of engineering. Of course I could go with a $500 driver in a closed box, but hey, than I'd be buying Tannoys or whatever.
If I accounted for all the facts, I'd realize that electroacoustic engineering has a long way to go yet, and I'd be planning to buy audio gear in the next life. when technology will be matured.
And sticking to simplistic views, here are the promised drawings, and a few questions:
How should the discrete sections be defined? Like in the left drawing or like in the middle one? (the blue lines denote the cross-sections)
And, how should be the line length be measured? Like in the middle drawing or like in the right one? (I know this has been discussed, but since a drew it...)
An externally hosted image should be here but it was not working when we last tested it.
Sould the folded line be closed at the bottom and the divider cut, (middle), or should it (the line) be expanded downwards (right)?
An externally hosted image should be here but it was not working when we last tested it.
The server where I stored the pictures has a life of its own, Geocities is safer.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Mr Push Pull,
I would model the geometry on the left using the red line to set the length and the blue lines to define the areas. I am assuming the line exits at the top as shown by the path.
As far as the second set of plots, I don't understand the question. The left hand sketch in the first set of plots looks good to me.
I would model the geometry on the left using the red line to set the length and the blue lines to define the areas. I am assuming the line exits at the top as shown by the path.
As far as the second set of plots, I don't understand the question. The left hand sketch in the first set of plots looks good to me.
Here you can see how I did it.
I tried to keep the taper as smooth as possible through the bend.
Maybe this helps.
I tried to keep the taper as smooth as possible through the bend.
Maybe this helps.
- Status
- Not open for further replies.