Transmission line or TQWT with 2 woofers in 2 way system ?

promitheus said:
I am trying to design a 2 way speaker that has bass reproduction from a TL or TQWT till 2KHz with efficiency 95dB or more.
Is this possible? and what would be a good idea?

Certainly possible. 95 dB cuts down on the available driver units. 1st thing that springs to mind is the Fostex FE168 -- although its Vas is a little low to get the most out of a TQWT.

1st off, you want to look for a driver with Qt somewhere between 0.3 and 0.55 and then start playing with Martin King's TL modeling software.

dave
 
I wanted to be able to model a TL with 2 woofers. 1 at 1/3 and 1 at 1/5 of the line length. I have seen this a few times in different projects, as in the new Thor speaker in audioxpress.
Some people use 2xSd, Re/2, Qts same, 2xVas and so on.
Do you think calculating double cross section everywhee is enough?
 
promitheus said:
I wanted to be able to model a TL with 2 woofers. 1 at 1/3 and 1 at 1/5 of the line length. I have seen this a few times in different projects, as in the new Thor speaker in audioxpress.

The Thor was designed using the sketchy alignment tables provided by Augspurger. Augspurger's & King's SW are both in very close aggreement to one another, and whenever i look at the Thor i think "is it optimum" -- what could have been done if the actual analytical tools had been used to try out many more possibilities...

Some people use 2xSd, Re/2, Qts same, 2xVas and so on.
Do you think calculating double cross section everywhee is enough?

These are the same thing. Augspurger showed that TL volume is proportional to Vas. Use 2 drivers, you double the Vas, since line length is fixed by the resonant mode you need to double the cross-section everywhere to double the volume.

dave
 
I've been looking at the Thor and I had a couple of questions.

1) I'm thinking of switching the top or bottom woofer to face outwards - this makes the two drivers more linear. (not an Isobarik, so I don't think this would change any of the line's parameters or design.

2) If 1), should I move the cone inwards to align the center of acoustic mass?

3) If I'm phase alinging the speakers per 2), how about insetting the tweeter as well?

If I do all of this, my thought is that the top part of the enclosure could be offset by an inch or two. This would reduce the volume though, so I'd probably have to work the math.

Thoughts?
 
jgwinner said:
1) I'm thinking of switching the top or bottom woofer to face outwards - this makes the two drivers more linear. (not an Isobarik, so I don't think this would change any of the line's parameters or design.

By outwards do you mean onto the side of the enclosure? This, or taking it further and moving the other driver (& the T) to the other side and then making this the front (ie a bi-pole), will certainly require tweaking the XO.

And if one assumes the line is optimal (& i'm not convinced it is) you would need to move the drivers vertically so that they are at the average position of the drivers when in the MTM (ie preserve the driver offset from the end of the line).

dave
 
I guess I actually mean Inwards.

The idea is to flip one speaker, and wire it in opposite phase. Both cones go toward the outside, although one cone goes toward the basket and the other away.

I would have to align the cones for the best phase alignment.

It's the same net speaker design, the only difference is that (like an Isobarik) you can reduce non linear distortions considering speakers aren't symetrical.

I've read all the Thor threads on the forum, and I did see your note that the Thor may not be an optimal line length, but after reading the Audio Express article I think it's close enough for what I want.

I'd also align the tweeter in. My thought is that the top portion of the cabinet would be inset slightly, with the tweeter and the other woofer 'in' slightly.
 
jgwinner said:
The idea is to flip one speaker, and wire it in opposite phase. Both cones go toward the outside, although one cone goes toward the basket and the other away.

Now that is NOT a good idea. This driver goes up into the midrange and the extra noise from the back of the driver will deteriorate the entire speakers performance.

And i'll take a bit of a leap here, based on a comment Nick McKinney of Lambda said when i chatted with him about this wrt his drivers -- "the Lambdas are linear enuff that you won't gain anything". I expect the Excels being SEAS top-of-the-line probably fit into the same category.

I've read all the Thor threads on the forum, and I did see your note that the Thor may not be an optimal line length

That isn't what i meant. I suggest that th entire design of the TL may not be optimal. As well as line length, there is driver offset, taper (or even more complex line geometry), and terminus restriction to consider. Plus instead of guessing the stuffing one can model all sorts of schemes.

Don't get me wrong, the exposure the Thor is giving the TL is good IMO, and i'm sure it sounds really good, but i just wished Joe had taken the time to use one of the two available computer models to do more exploration -- i would think that one would want to try to take maximum advantage of such lovely (& expensive) drivers.

rant
(i also take exception to some of the hyperbole Ed Dell takes with this design -- 1st scientifically designed TL -- that honour would have to go to George A's lines or Martin King's (which was published ahead of the Thor -- Ed had a chance to publish King and rejected it) -- even Rick Shultz had a design published before the Thor -- i believe Ed had a crack at that one too)
end rant

The Thor was based on interpolating a design from a 3 example alignment table that has no hope of even scratching the surface of the possible design space. This is something i played with right after i heard Augspurger 1st present his paper. At that time i decided it was too limited (i also had advance knowledge of Martin King's model at this time).

dave
 
Now that is NOT a good idea. This driver goes up into the midrange and the extra noise from the back of the driver will deteriorate the entire speakers performance.

?? I would only be changing the orientation of the driver, not the direction the cone moves. The speakers would be EXACTLY the same as a 'stock' Thor (assumming I didn't align the cones), the only difference is that they would theoretically be more symetrical. So the issue with noise from the back of the driver is does 2x the noise go in the line, or does 1/2 go in the line and 1/2 in the room?

Does the sound from the back of the driver sound that bad? If so, wouldn't vented designs be really bad due to this noise? I have't seen any discussion of this.

I'm not say it's not a factor. In fact, it's possible that any noise from the backside is lost with the stuffing. Still, Isobarik setups with a facing orientation have nothing BUT backside sound.

Still, if they are symetrical in sound as you note then both facing out would be more cosmetic.

I don't have a full copy of Mathcad, not sure if I can redo any of the analysis ... I'd probably build the first one 'stock' anyway and play with a 2nd (or 3rd :D ) To really analyse this I assume I'd have to sum two offset line's (with one offset slightly more than another).

What did you think of phase aligning the tweeter and the two drivers?
 
jgwinner said:
Does the sound from the back of the driver sound that bad? If so, wouldn't vented designs be really bad due to this noise? I have't seen any discussion of this.

It can. I proposed this dipole concept and i got quite a "lecture" from Sig Linkwitz on the noise from the back of the driver. Vented boxes are tuned a filter out most of the sound from the box.

The isobaric subs are just that - subs. The big problems come as you move upwards in frequency.

If you time-aligned it, i assume you would probably have to redo the XO somewhat.

dave
 
Interesting. Facing the same way it is then.

I remember seeing that on your site, interesting about what Linkwitz said.

I think the 'slots' in the basket would make for some diffraction effects; I've read up about those while doing my Electrostatic investigations. However, the slots are large enough I wouldn't think it would make for that large of an effect, but I think all said the best idea is face them the same way, the gain vs the risk isn't there.

I'll double check the phasing of the crossover and see if there's anything that needs redoing if I aligned it.
 
Update:

I did some checking on prices and parts. The Odin uses the same drivers as the Thor. Zalytron also has an upgrade to the Odin but with built with Raven tweeters (Ribbon's), and can supply the same kit with the Thor components.

This is tempting but I'm not sure the Raven matches the woofers as well as the Sea's tweeter.

I have Mathcad 8 downloaded and brought up Martin's model, but I'm not sure how to model two drivers offset slightly in the same pipe.

Would using Martin's model twice, with the 2nd driver offset, and adding the response curves work? I'm not sure.
 
jgwinner said:
I have Mathcad 8 downloaded and brought up Martin's model, but I'm not sure how to model two drivers offset slightly in the same pipe.

The usual work-round is to put 2 drivers at the avaerage position.

Would using Martin's model twice, with the 2nd driver offset, and adding the response curves work? I'm not sure.

If you add them in the complex plane this would work. Don't be afraid to email Martin. He is very helpful. He has been asked about this enuff times that i am sure he has given some thot to it.

dave
 
Lately, I get asked this question at least once or twice a week. The advice you have gotten above is good and I cannot add much more. At some point, I am going to produce a model that can actually do this type of simulation for a variety of enclosure types.


Method 1 : Modeling a single driver at the average postion in the TL or TQWT, designing an optimum enclosure geometry, and then doubling the cross-section is a valid approach.

Method 2 : Model a composite driver at the average location and set the T/S parameters as follows :

fd = same as single driver

Qtd = same as single driver (Qed and Qmd)

Vad = twice that of a single driver

Sd = twice that of a single driver

Re and Le = half for a parallel connection, twice for a series connection

SPL = + 6 dB for a parallel connection, same for a series connection

The resulting cabinet design will be the actual size so the areas will not need to be doubled.


I think both of those methods will work.

It was also questioned if running two simulations, one for each driver, and then summing the result would work. I have done this but I cannot remember if it yielded the same result as the two methods described above. It might be worth trying both ways and seeing if there are significant differences.

All of these methods are approximations and until I write a two driver version of my MathCad models, probably several months away, that is about all I can recommend trying. It will be interesting to see what is the real answer. Do the standing waves interact differently with each of the two drivers due to the different positions along the length of the TL or TQWT.

Good luck,

Martin J. King
www.quarter-wave.com
 
MJK said:

A hearty welcome back to the forum Martin. I was just wondering the other if you could be enticed to join in.

DIYers, let me introduce Martin King. Between him and George Augspurger they have done the most to contribute to the development of transmission lines since Bailey's seminal article. George seems to get all the fame but Martin :king: has decidedly contributed the most to the community.

dave