@bforce As I cannot read basic literature now I asked Chat GPT and it confirmed you are right.
You won!
You won!
I need some advise on selecting the right speaker for a TL enclosure.
I was reading a report by @perrymarshall A Derivation and Analysis of the Transmission Line Speaker Enclosure and I have some questions.
Not sure whether correct, but my conclusion from it is that the right speaker for a TL enclosure should have large Mms and large Qts (how large?). In that case I get a good low frequency extension and a large SPL from the open end of TL.
On the other hand if I select a driver with a small Mms and small Qts, I get either good low frequency extension when making the TL line long (l/4) or good SPL output from the open end of TL if I shorten the TL to 2m. But never both- good SPL and good frequency extension. In the case of a driver with low Mms and low Qts the low frequency extension and SPL from the open end of TL can be traded by changing the length of the TL. If I want both, SPL and frequency extension, the only option is to use a driver with large Mms and large Qts.But the loudspeaker sensitivity will be lower. Am I right here with that conclusion?
Hornresp does not seem to take into account this trade-off, unless I am doing something wrong.
I do not see any change in the low frequency extension when varying speaker's Qts in Hornresp.
If I want a good impulse, good "punch", should I be going for a driver with large Mms and Qts, or a driver with small Mms and small Qts? Is not the case a lighter cone provides "quicker" bass, better "punch" than a heavy one?
What @perrymarshall writes in this report is that a total length of the TL can be shorter than l/4. It even must be shorter than l/4 in the case of a havy cone and large Qts otherwise I get poor low frequency extension (3rd design example shows it- when a driver with large Mms and large Qts is moved from 2m TL to 3.5M TL the low frequency extension gets worse). How do I find that shorter dimension (2m in the design example)?
I was reading a report by @perrymarshall A Derivation and Analysis of the Transmission Line Speaker Enclosure and I have some questions.
Not sure whether correct, but my conclusion from it is that the right speaker for a TL enclosure should have large Mms and large Qts (how large?). In that case I get a good low frequency extension and a large SPL from the open end of TL.
On the other hand if I select a driver with a small Mms and small Qts, I get either good low frequency extension when making the TL line long (l/4) or good SPL output from the open end of TL if I shorten the TL to 2m. But never both- good SPL and good frequency extension. In the case of a driver with low Mms and low Qts the low frequency extension and SPL from the open end of TL can be traded by changing the length of the TL. If I want both, SPL and frequency extension, the only option is to use a driver with large Mms and large Qts.But the loudspeaker sensitivity will be lower. Am I right here with that conclusion?
Hornresp does not seem to take into account this trade-off, unless I am doing something wrong.
I do not see any change in the low frequency extension when varying speaker's Qts in Hornresp.
If I want a good impulse, good "punch", should I be going for a driver with large Mms and Qts, or a driver with small Mms and small Qts? Is not the case a lighter cone provides "quicker" bass, better "punch" than a heavy one?
What @perrymarshall writes in this report is that a total length of the TL can be shorter than l/4. It even must be shorter than l/4 in the case of a havy cone and large Qts otherwise I get poor low frequency extension (3rd design example shows it- when a driver with large Mms and large Qts is moved from 2m TL to 3.5M TL the low frequency extension gets worse). How do I find that shorter dimension (2m in the design example)?
Not necessarily. That paper is dated a decade before proper modelers started appearing and is really only of acedamic interest. On. skim it contains misleading, partial, or just not so information.
The line he analyzes is likely too small.
Note that line length dependent on line taper, any mass-loading, and the design goals. Mass-loading, driver and vent offset are tools he says nothing about.
Read Martin King’s website, Augspurger’s paper, and then start studying here and asking questions, Some very skilled TL designers here.
http://quarter-wave.com/
http://p10hifi.net/TLS/downloads/AugspurgerAES107rev2.pdf
If you want to look at a bunch of TLs that pedate those models and are examples of TLs with a poor chance of being optimal (some of the virtual TLs are properly modeled..
https://t-linespeakers.org/
dave
It's an interesting enough work from a historical POV, but as Dave says, it's broadly outdated in terms of scope compared to Augspuger & King's various analysis & papers a decade later -which is no insult; as Carl Sagan pointed out, science is a self-correcting process. It's what we do in academia all the time (and occasionally have some fun debating the points or coming up with something outrageous to help stimulate new lines of investigation). Be that as it may, v. quickly as I'm heading down to London:
Hornresp simply models whatever you put into it, within the framework of what it can do (there aren't many things it can't), so it doesn't account for anything itself as such -it just shows the behaviour of whatever data entry you use.
Not really. A TL essentially 'wants' a driver suited to a vented box. It doesn't work on the same resonant basis & may not have the same objectives (depends what they are -a TL isn't a fixed alignment any more than a Helmholtz based vented is / has to be), but the requirements in that sense are broadly similar.
An echo of the above; Mms in itself is neither here nor there -it's a fundamental parameter but its practical value in this case is mostly how it works in conjunction with others. A low Qts = overdamped, so assuming a given pipe & you hold Fs, Vas static, the lower the Q, the higher the mass-corner frequency & the less the usable output you're likely to have. As far as mass is concerned, assuming everything else remains static, as it reduces, so efficiency increases, but that doesn't equate to more extension: just more output for a given alignment. In practice of course, nothing else is likely to remain static, so this is just an illustration of the principle. Ultimately, Hoffmans' Iron law applies, whatever the box type. Ignoring damping, tuning in a QW / TL variation (acoustical length) is essentially a function of physical length & taper; these determine Fb while gain is determined by Vb, and the alignment shape a function of both, plus the damping you use.
Hornresp simply models whatever you put into it, within the framework of what it can do (there aren't many things it can't), so it doesn't account for anything itself as such -it just shows the behaviour of whatever data entry you use.
@planet10 I read all the material from your links. I have known it for years. I have also build several TL loudspeakers in the past but never used Hornresp to simulate TL. I was just using theoretical equations from the sources you mentioned. I am now going to build a few more designs and the approach is to use a simulator.
Perry nicely presents design examples and graphs showing open and closed TL and frequency extension. In the past my approach was similar to what you say, select speaker that performs well in BR enclosure, with low Qts in the range 0.3 to 0.35, with low fs and design TL of a length L(fs)/4. I never run any simulations of that in the past. Now I started using Hornresp and I see that the optimum length is rarely l/4. Now, from many people I hear the opposite, I should be looking for a speaker with Qts closer to 0.5 which indicates a speaker suitable for closed enclosure. So I run a few simulations in Hornresp to confirm that. @Scottmoose And I did not come to conclusion that confirms the statement that TL prefers speakers that perform well in BR enclosure.
I posted this in another thread but I repeat it here to present my point.
I am in the process of selecting a driver suitable for a transmission line. I plan to build an enclosure based on Ekta-TL by Troels with necessary mods to accommodate the selected driver.
And I just can't figure it out what makes a driver suitable for a transmission line.
I simulated a couple of drivers and they mostly just do not perform the way I would expect. I want a good bass extension, good sensitivity, and short TL, yeah.
At first I selected SB Acoustics SB20PFCR30-8. And I got F6dB=38Hz F3dB=46Hz Sensitivity=91.4dB. Which is crap.
Then I read a report by Perry Marshall that says a good driver should have: large Qts, large Mms and large BL.
So I simulated Wavecor WF223BD02 which has it all, or so I thought, large Qts, large Mms and large BL. And I got F6dB=32Hz F3dB=39Hz Sensitivity=89.6dB. Not bad, but still way below the expectations from such a great driver.
So I simulated the driver that Troels put originally in the Ekta-TL enclosure, the ScanSpeak 18WU/4741T00, which is just the opposite of Wavecor- small Qts, small Mms, small BL. Theoretically should not work at all. Yet I got F6db=30Hz F3dB=35Hz Sensitivity=91dB. Which is excellent.
All speakers were simulated in the same TL enclosure. But I do not understand why Scanspeak performs that well.
Why does this Scanspeak perform that well?
What parameters should I be looking at when selecting a driver suitable for a TL?
Why the the optimum TL length is not l/4 ?
If I am looking for good low frequency extension, and tight, well controlled bass, good "punch", should I be looking for a driver with low Mms or large Mms? My perception so far was that a light cone is easier to move and should produce a better "punch".
The simulation results are just all over the place not allowing me to draw any conclusions.
I just can figure it out which T-S parameter makes Scanspeak's frequency response to go that low.
And what makes SB20PFCR30-8 to look so bad? It's T-S parameters are similar to Scanspeak's.
speaker | F6dB | F3dB | Sensitivity | TL length| T-S parameters
SB20PFCR30-8, 38Hz, 46Hz, 91.4dB, 190cm , small Qts, small Mms, small BL
WF223BD02, 32Hz, 39Hz, 89.6dB, 190cm , large Qts, large Mms and large BL
18WU/4741T00, 30Hz, 35Hz, 91dB, 190cm , small Qts, small Mms, small BL
Perry nicely presents design examples and graphs showing open and closed TL and frequency extension. In the past my approach was similar to what you say, select speaker that performs well in BR enclosure, with low Qts in the range 0.3 to 0.35, with low fs and design TL of a length L(fs)/4. I never run any simulations of that in the past. Now I started using Hornresp and I see that the optimum length is rarely l/4. Now, from many people I hear the opposite, I should be looking for a speaker with Qts closer to 0.5 which indicates a speaker suitable for closed enclosure. So I run a few simulations in Hornresp to confirm that. @Scottmoose And I did not come to conclusion that confirms the statement that TL prefers speakers that perform well in BR enclosure.
I posted this in another thread but I repeat it here to present my point.
I am in the process of selecting a driver suitable for a transmission line. I plan to build an enclosure based on Ekta-TL by Troels with necessary mods to accommodate the selected driver.
And I just can't figure it out what makes a driver suitable for a transmission line.
I simulated a couple of drivers and they mostly just do not perform the way I would expect. I want a good bass extension, good sensitivity, and short TL, yeah.
At first I selected SB Acoustics SB20PFCR30-8. And I got F6dB=38Hz F3dB=46Hz Sensitivity=91.4dB. Which is crap.
Then I read a report by Perry Marshall that says a good driver should have: large Qts, large Mms and large BL.
So I simulated Wavecor WF223BD02 which has it all, or so I thought, large Qts, large Mms and large BL. And I got F6dB=32Hz F3dB=39Hz Sensitivity=89.6dB. Not bad, but still way below the expectations from such a great driver.
So I simulated the driver that Troels put originally in the Ekta-TL enclosure, the ScanSpeak 18WU/4741T00, which is just the opposite of Wavecor- small Qts, small Mms, small BL. Theoretically should not work at all. Yet I got F6db=30Hz F3dB=35Hz Sensitivity=91dB. Which is excellent.
All speakers were simulated in the same TL enclosure. But I do not understand why Scanspeak performs that well.
Why does this Scanspeak perform that well?
What parameters should I be looking at when selecting a driver suitable for a TL?
Why the the optimum TL length is not l/4 ?
If I am looking for good low frequency extension, and tight, well controlled bass, good "punch", should I be looking for a driver with low Mms or large Mms? My perception so far was that a light cone is easier to move and should produce a better "punch".
The simulation results are just all over the place not allowing me to draw any conclusions.
I just can figure it out which T-S parameter makes Scanspeak's frequency response to go that low.
And what makes SB20PFCR30-8 to look so bad? It's T-S parameters are similar to Scanspeak's.
speaker | F6dB | F3dB | Sensitivity | TL length| T-S parameters
SB20PFCR30-8, 38Hz, 46Hz, 91.4dB, 190cm , small Qts, small Mms, small BL
WF223BD02, 32Hz, 39Hz, 89.6dB, 190cm , large Qts, large Mms and large BL
18WU/4741T00, 30Hz, 35Hz, 91dB, 190cm , small Qts, small Mms, small BL
Last edited:
FWIW, I've designed and built many TLs and I've gotten good to excellent results with drivers having a Qts in the range of 0.35 to 0.50. Drivers with Qts outside of that range sometimes surprise me with acceptable results. The key in my opinion is to shoot for a system tuning frequency equal to fs when Qts is equal or close to 0.40. If Qts is higher than 0.40, the optimum tuning frequency will be lower than fs. If Qts is lower than 0.40, the optimum tuning frequency will be higher than fs. Also, unless you're doing an ML-TL which has essentially a constant area throughout the line's length, I prefer a negatively tapered TL with a taper ratio of 10:1 or greater. But I've also had success with a line that's both tapered and mass-loaded (see attached).
Paul
Paul
A convoluted folding, the deflectors at the top are counter-products, the mid enclosure is a good shape thou and contributes to the deflectors. And we know the line is probably designed using classical methods ...
dave
Similar to the many lines i have drawn for Scott. A Q higher than 0.5 can often give good results, EL70 is a good example when placed in the microTower for instance.
Note: pkitt is one of those experienced TL designers i mentioned earlier.
dave
I agree. But the line is almost 2m long. You have to bend it somewhere. It is not my intention to copy this design. I want to make my own. I mention it here just to show what I have in mind for my project. I like the concept of this enclosure because it is fairly simple to build and because of the way the mid driver compartment is blended into the internal TL. It is also easy to tweak the length and cross-section area of the TL. I can also easily make it bass-reflex by adding a BR port instead of internal TL without touching the mid compartment.
I simulated this enclosure with Hornresp as a simply tapered line of 190 cm long and it all works quite well with the SS driver. (see the screenshots above).
There is plenty of drivers out there. But when it comes to select that one you need you find out it is very difficult to find anything. I can not simulate every driver out there so I want to learn to recognize the suitable candidates just by looking at the T-S parameters. I am looking for for 8" drivers.
There are opinions that drivers suitable for a TL are those that perform well in a vented enclosure. And there are opinions that drivers that are made for closed enclosure would perform better. But my simulations using Hornresp do not confirm either.
Last edited:
AFAIK Hornresp didn't model folds -- which IME were quite important at removing QW higher harmonics but also contributing many weak partial-length QW and half-wave resonances to yield a smoother overall response. Terminal stepped slot-loading can add to ML effect. Anyway I call them "TLonken", more or less like the above Troel's (without midrange and without fold-corner pieces).
IMHO, for a suitable driver and a given cabinet volume, (down-) tapered TL probably goes deepest (and sounds clearest and most dynamic). Famed MJK table 1 redux: R^1/4.75 effective line-length multiplier for R:1 taper ratio. I don't have a table of QW SPL/cab-volume data-points to work out a general near-approximation/trade-off, so I usually just start with Vas to half-Vas for the cab.
A few recently gathered posts:
IMHO, for a suitable driver and a given cabinet volume, (down-) tapered TL probably goes deepest (and sounds clearest and most dynamic). Famed MJK table 1 redux: R^1/4.75 effective line-length multiplier for R:1 taper ratio. I don't have a table of QW SPL/cab-volume data-points to work out a general near-approximation/trade-off, so I usually just start with Vas to half-Vas for the cab.
A few recently gathered posts:
TLonken for Qts low 0.3 and small Vas relative to Fs. BTW I spent a few minutes gathering TS-params on those three drivers -- although all had similar Fs, I think only the SS fit my way-oversimplified-totally-off-the-cuff rule of thumb.
MLhorn (ML-TQWT) for Qts above 0.4 and large Vas relative to Fs. Without ML, effective line-length divider see link in last msg. But with ML, Bob Brines' 3-way comp suggests (to me) they all yield the same response curve just with different cab volume and geometry (tapered TL, straight MLTL, ML-TQWT).
Fs by itself is not a good indicator.
Last edited:
Really? As far as spec. is concerned, the Scan is a perfectly ordinary 'modern' 7in midbass optimised for (acoustically small) vented box loading. So it will work fine in a resonant quarter-wave, a.k.a. a TL intended for LF support. A Qts of 0.36 (effective Qts' being higher once wire loop resistance, series R from connections, LP series inductors etc. are factored ) is not generally classed as 'small'; it's roughly in the mean range for vented enclosures. In the same vein, the SB, with a Qts of 0.37 is actually slightly higher, and I can't think of anybody who would describe that as low. The Wavecor, with a Qt of 0.435 after the suspension has been broken in, is also roughly in the mean range for vented enclosures as far as that goes, albeit more toward the upper end, and very few would describe that as 'high'.
There are no fixed categories, but loosely speaking, Q doesn't tend to be considered low (rather than small) until it drops below roughly 0.3, and probably most would give it the benefit of the doubt until you hit around 0.25 or lower. For the converse, the limit is basically Keele's 0.638 as far as vented loads go, and most would probably describe 'high Q' in more general terms as something exceeding Butterworth, i.e. 0.707.
This is in general the case for a speaker with a Qts value around 0.45 in a straight line, but there are primarily three factors that can influence the tuning frequency:
1 - The shape of the TL: A tapered TL or the presence of a coupling chamber can lower the tuning frequency (Fp) independently of the TL length.
2 - The Qts of the speaker: As previously explained, a higher Qts allows the TL to be tuned to a lower frequency (below Fs).
3 - The presence of damping material in the duct.
It a wonder anything works as predicted. 10 and 75 milliohms differences? No wonder things behave differently with different amplifiers, cables, xover inductors. Talk about a system level problem. What element dominates?
Uh, whoops. Ever feel your face turn red over something you posted? I think I'll go clean out the garage...
Only if the pipe is not tapered and has no mass loading. But don’t forget to add the end correction.
Damping has less effect than a fold does.
dave