Help me design a transmission line...
Dave Dal Farra
Much of what Bill S. writes is in contradiction to both my practical
and theoretical experiences with loudspeakers.
Rather than assume either of us is full of BS (but I have my
suspicions [smiley]), I'll post my experiences on a point by
point manner and wait for the flames/comments.
William Spencer, bi...@verdix.com writes:
>he...@ego.psych.mcgill.ca (Henry Pasternack) writes:
>
>> ...I desire tight,
>>clean bass that is free from excessive warmth or wooliness. The
>>drivers I have are ideally suited for closed box enclosures (high
>>Q, low fs). I am wondering if a TL would work even better.
>
>In my experience, to get this sound you need a speaker that goes deep
>with no peaks and possibly a very gradual rolloff. This objective is
>best realized using a ported design.
Non-assisted ported designs roll off at 24 dB/octave. This is not a
gradual roll-off.
If a gradual *initial* roll is used (they are all still asymptotic to
24 dB/octave eventually), my experiences show that you're losing the
real benefits of a ported in the first place, which is a lower -3 dB
point than a sealed. The gradual taper gives an earlier 3 dB
down point.
>The typical sealed box rolloff goes down low
>enough to exite room resonances but not below them, effectively
>exaggerating the resonances.
I have a real problem with this one. This is such a sweeping
generalization that it's just not true, not even "typical". The
variables considered: room size, speaker placement, room construction,
speaker 3 dB down point and roll-off all combine to set this
relationship.
Your above statement is true, not "typically", but only of a sealed box
with a 3 dB down point around the lowest room fundamental given poor
speaker and listener placement.
Similarly, one could replace the words "sealed" with "ported" in both
your statement and mine. Or does a "typical" ported
go below "typical" fundamental room modes (sarcastic smiley).
>The low slope of sealed box rolloff doesn't do much
>for you because a 12 dB down signal can't compete with a +10 dB room
>peak.
Again, this "10 dB" peak comment makes some pretty wild-ass
assumptions about "typical" room size, speaker and listener location,
room construction etc. Did you ever consider the scenario of a large
room with a slowly rolling sealed box alignement, where placement was
used to augment and smooth the low end roll-off? I think you missed
the major thrust of Allison's work.
>The ported box has a bad name because they are often used by those
>wanting a lot of bass.
They get a bad name because:
a) there are a lot of poorly tuned mid-fi ported boomers out there;
b) ported designs are more sensitive to manufacturing tolerances than
sealed boxes, resulting in a higher possibility of a slightly
mis-tuned ported system in the field over a slightly mis-tuned
sealed.
>Also, a nonstandard alignment must be used to get a gradual taper.
>A low Q sealed box with a very low resonant frequency
>would also do this but will be larger and have less dynamics and more
>distortion.
It would be larger, but "less dynamics and more distortion"?
First, I assume you're relating that cone motion for a given
size driver and f3 is less for a ported than a sealed as the port takes
over some (most) of the radiation duties.
This is very different than your statement.
The theoretical transient response of a high pass filter
(i.e. loudspeaker) is determined by the ringing time, the rise time and
the overshoot of the particular alignment/order. Higher order systems
(i.e. ported) have a longer ringing time and typically more overshoot
than a similar alignment-type of lower order (i.e. sealed). In my books,
it looks like the sealed has better "dynamic" characteristics.
(Sorry I left out rise time, but I can't remember the realtionship).
Maximum SPL abilities at resonance are another matter. The ported should
fair better due to the driver not reaching xmax (due to the port's
contribution) as early as in a sealed.
Since a ported speaker goes through less motion at F3 than a sealed,
the cone distortion is lower here. But what of port distortion?
I'm not familiar with the magnitude of these effects in relation,
but I'd be surprised if the ported faired significantly better
than the sealed.
>> So. We build a line with a length of 6-8 feet. We make the line
>>taper from 2.5 times Sd at the driver to Sd at the end.
>
>I think you should throw out this rule of thumb. The "throat" ... of the
>line is the area of the driver, not the area right behind it. If it
>suddenly expands to 2.5 times this area a resonant chamber is formed.
>.....The area of the line in
>this one expanded *smoothly* from the cone area to 4 times that
>area at the end. This flare was used just because it was an easy
>way to build a tubular line, it seems to neither improve or hinder
>performance.
Every piece of literature I've read on TL's say that it's best to start
with a line cross-sectional area at least twice Sd, than to taper the line
to sd at the open end. This supposedly results in a sufficient duct
area to provide adequate back absorption material, and is also supposed
to provide a better impedance match with free space at the port end.
>> Questions:
>> 1) Do TL's really work?
>
>It seemed to avoid having the backwave from the driver from bouncing
>back. If the smaller cross-section lines can be made to work,
>they may beat the size to performance ratio as good as any other designs.
>They are good at producing a desirable gradual rolloff characteristic.
Eliminating the "back wave bouncing back" is construction method dependant
with TL's. I do agree that mid range resonances caused by internal
reflections will be less with a properly designed TL (read, sloped back
feeding the duct opening behind the driver) than with a lightly to medium
density-stuffed sealed or ported.
Transmission lines have a gradual roll-off? I haven't seen any
Theil/Small type modelling done on TL's, but I'd expect them to
be closer to 4-th order high passes, like a ported.
Comments, character slams and spell checking are ecouraged.
Dave D.F.
"It's true they say that money talks. When mine spoke it said
'Buy me a Drink!'."
Richard D Pierce
>
>Much of what Bill S. writes is in contradiction to both my practical
>and theoretical experiences with loudspeakers.
>
Dave goes on to post what seems to be some conclusions about how
transmission lines REALLY work based on maybe empirical data and some
theoretical logic. For example, he asserts that a line may, in fact be
nothing more than a somewhat strangely tuned 4th order (i.e. vented system).
Unfortunately for transmission line proponents, his conclusions parallel
mine rather closely. I spent a good 3-4 years between 1973 and 1978
studying lines, and initially fell into the camp that many current
afficianados seem stuck in. All of the theoretical models based on the
concept of 1/4 wave operation, reduced propogation velocity, and such,
failed to come even close to predicting the actual measured response
characteristics of what we call transmission line speakers.
In the long run, the ONLY model which came even remotely close to
accurately predicting the response (and it comes REAL close) is simply a
mis-tuned reflex. The "mis" tuning comes several ways. First, you have a
much larger ratio of "port mass" (the acoustic inertance of the bulk of
the line) to the enclosure compliance (the first part of the line really
acts more like a compliance). Second, your vent losses are fairly high,
orders of magnitude higher than for a normal reflex. Building a model
based on this assumption and taking what you would expect for the effect
complaince and mass of a lumped element model comes remarkably close to
really showing what a line does.
Further, look at the bloody electrical impedance curve. It has all the
hallmarks of a reflexm with its two bumps (actually it's really one large
bump, due to the motional impedance of the driver, with a hole in the
middle caused by the acoustical impedance of the enclosure resonance).
And what about the so-called reduction in propogation velocity. Much of
this comes from measurement of the relative phase of the woofer and line
output, which seems to be in phase at some impossibly low frequency
(compared to what the normal propogation delay would suggest). Well, of
course the line output would be in phase at some very low frequency
(compared to the line length) if the entire enclosure is resonant at that
frequency, JUST LIKE A REFLEX!
One assertion can be made that stands pretty well on its own. at these
kinds of low frequencies, where the wavelength is poximal (or larger) than
the length of the line, IT IS NOT A TRANSMISSION LINE.
So, one may ask, does a transmission line have any advantages? In my
opinion, yes, it does, and they are both substantial:
1: At higher frequencies, especially the midbass and lower midrange,
the enclosure IS acting as a well-terminated (or very lossy)
transmission line, and, as a result, succeeds in effectively
loosing much, maybe most, of the rear radiation of the cone, a
source, in my opinion, of much of the deleterious cabinet colorations
taht, to me at any rate, are disturbing
2: as a SECONDARY effect, the construction methodology of typical
transmission lines results in a cabinet which is extremely well
braced and very heavy, further reducing cabinet coloration
substantially.
The best transmission line I have heard is one that, in fact, was so
heavily damped that the line output was effectively nil. It behaved, at
low frequencies as a second order system until very low (below about 15
Hz) where the effective absorbtion of the material in the line had dropped
enought where it then behaved as 4th order.
Many lines that try to use the output of the line for reinforcement do so
at the cost of a significant hole in the system respones in the lower
midbass and multiples thereof. This seems to occur at frequencies that
correspond to the wavelength of the line, PRESUMING NEAR-NORMAL
PROPOGATION VELOCITIES. If the velocity through the line was significantly
changed by the damping, why does it not affect these frequencies, but others.
Adding or subtracting damping didn't seem to change the center frequency
much, but it does change the Q (there is a slight change in center
frequency, but not to the extent that many claim, like a reduction in
velocity to 70% of the original. More like 3-5%).
After spending a lot of time building prototypes, measuring, calculation,
modeling, and so on, I reached a couple of conclusions:
1: A properly designed line can sound marvelously clean and effortless
in a region of the audio spectrum that, for some reason, I seem
to be sensitive too: the region between 150 and 1000 Hz. In
this region, transmission lines probably have a significant advantage
over normal geometries simply because they reduce coloration in a
reasonably predictable fashion.
2: At low frequencies: if it measures like a reflex, has the transient
response of a reflex, has the impedance curve of a reflex, models as
a reflex, it's probably a reflex, albeit a strangely tuned and very
lossy one. The reflex model is the best fit for the measured reality
so far.
I have fond memories of my experiments, including one pair using a KEF
B-110 5" woofer in a 3 foot long line that was -3 dB at 40 Hz. The cabinet
was something like 7" wide, 12" deep and 28" high, weighing 50 lbs. I was
able to achieve the SAME cutoff with flatter passband response and far
better power handling in a more conventional reflex half the size. But the
midbass and lower midrange one the line was obviously better.
Accelereometer measurements of the enclosure walls confirmed amplitudes
that were at least 1-2 orders of magnitude less in the line, again because
of the inherent bracing.
--
| Dick Pierce |
| Loudspeaker and Software Consulting |
| 17 Sartelle Street Pepperell, MA 01463 |
| (508) 433-9183 (Voice and FAX) |
C.M. Hicks
<< LOTS DELETED IN AN EFFORT TO PRESERVE BANDWIDTH >>
>So, one may ask, does a transmission line have any advantages? In my
>opinion, yes, it does, and they are both substantial:
> 1: At higher frequencies, especially the midbass and lower midrange,
> the enclosure IS acting as a well-terminated (or very lossy)
> transmission line, and, as a result, succeeds in effectively
> loosing much, maybe most, of the rear radiation of the cone, a
> source, in my opinion, of much of the deleterious cabinet colorations
> taht, to me at any rate, are disturbing
> 2: as a SECONDARY effect, the construction methodology of typical
> transmission lines results in a cabinet which is extremely well
> braced and very heavy, further reducing cabinet coloration
> substantially.
>The best transmission line I have heard is one that, in fact, was so
>heavily damped that the line output was effectively nil. It behaved, at
>low frequencies as a second order system until very low (below about 15
>Hz) where the effective absorbtion of the material in the line had dropped
>enought where it then behaved as 4th order.
I think this is probably how the pair that I built are behaving. I have
measured the port output (albeit only in the near field, since this is all
I could do...) and its output peaks at around 80Hz, where it is some 10dB
below the output of the driver at the same frequency. I am sure there are
inaccuracies in my rough measurements, but I reckon they are in the right
ball park.
>Adding or subtracting damping didn't seem to change the center frequency
>much, but it does change the Q (there is a slight change in center
>frequency, but not to the extent that many claim, like a reduction in
>velocity to 70% of the original. More like 3-5%).
I found this too - polyester wool just does not affect gamma significantly.
It does, however, seem to be very effective at damping higher-frequency
standing waves in the line.
>After spending a lot of time building prototypes, measuring, calculation,
>modeling, and so on, I reached a couple of conclusions:
> 1: A properly designed line can sound marvelously clean and effortless
> in a region of the audio spectrum that, for some reason, I seem
> to be sensitive too: the region between 150 and 1000 Hz. In
> this region, transmission lines probably have a significant advantage
> over normal geometries simply because they reduce coloration in a
> reasonably predictable fashion.
This is interesting. I find this the least satisfactory range of mine,
possibly because I have a significant chamber behind the driver. The
region I like best is 30 to 100Hz where the notes float out of the
cabinet, no matter how low they go. The bass isn't particularly strong,
but it does go very low.
> 2: At low frequencies: if it measures like a reflex, has the transient
> response of a reflex, has the impedance curve of a reflex, models as
> a reflex, it's probably a reflex, albeit a strangely tuned and very
> lossy one. The reflex model is the best fit for the measured reality
> so far.
Haven't tried applying a reflex model, as my measurements are not accurate
enough at sub-50Hz frequencies. It does, however, seem entirely reasonable.
When I have time, I will measure the impedance curve properly, and will
report back to the net.
>I have fond memories of my experiments, including one pair using a KEF
>B-110 5" woofer in a 3 foot long line that was -3 dB at 40 Hz. The cabinet
Hmm... Last Summer my mother was none too impressed with 20 feet of 8
inch diameter yellow plastic gas pipe in the back garden...
>was something like 7" wide, 12" deep and 28" high, weighing 50 lbs. I was
>able to achieve the SAME cutoff with flatter passband response and far
>better power handling in a more conventional reflex half the size. But the
>midbass and lower midrange one the line was obviously better.
If you have them to hand, it would be interesting to see the -3, -10 and
-20dB points of each of these two designs.
>Accelereometer measurements of the enclosure walls confirmed amplitudes
>that were at least 1-2 orders of magnitude less in the line, again because
>of the inherent bracing.
Bathroom scales measurements of the cabinet weights confirmed my suspicions
that they are bloody heavy :-)
Christopher
--
==============================================================================
Christopher Hicks | Paradise is a Linear Gaussian World
c...@uk.ac.cam.eng | (also reported to taste hot and sweaty)
==============================================================================
Lionel Tun
>Hmm... Last Summer my mother was none too impressed with 20 feet of 8
>inch diameter yellow plastic gas pipe in the back garden...
Tell me more! I having been chewing over this very idea ever since
British Gas did that work all over the country with those yellow
pipes. Did you put an 8" driver into the 8" pipe? Did it work?
It seems to me a very easy way of getting a long transmission
line, if say you can just lay the pipe along a corridor or
something. Is it better than a wooden cabinet?
--
________ Lionel Tun, lio...@cs.city.ac.uk ________
/ /_ __/\ Computer Vision Group /\ \__ _\
/___/_/_/\/ City University, London EC1V 0HB \ \___\_\_\
\___\_\_\/ 071-477 8000 ext 3889 \/___/_/_/
JOHN I LIPP
>>
>>Much of what Bill S. writes is in contradiction to both my practical
>>and theoretical experiences with loudspeakers.
>>
>
[D.Pierce writes:]
>Dave goes on to post what seems to be some conclusions about how
>transmission lines REALLY work based on maybe empirical data and some
>theoretical logic. For example, he asserts that a line may, in fact be
>nothing more than a somewhat strangely tuned 4th order (i.e. vented system).
>
>Unfortunately for transmission line proponents, his conclusions parallel
>mine rather closely. I spent a good 3-4 years between 1973 and 1978
>studying lines, and initially fell into the camp that many current
>afficianados seem stuck in. All of the theoretical models based on the
>concept of 1/4 wave operation, reduced propogation velocity, and such,
>failed to come even close to predicting the actual measured response
>characteristics of what we call transmission line speakers.
>
>In the long run, the ONLY model which came even remotely close to
>accurately predicting the response (and it comes REAL close) is simply a
>mis-tuned reflex. The "mis" tuning comes several ways. First, you have a
>much larger ratio of "port mass" (the acoustic inertance of the bulk of
>the line) to the enclosure compliance (the first part of the line really
>acts more like a compliance). Second, your vent losses are fairly high,
>orders of magnitude higher than for a normal reflex. Building a model
>based on this assumption and taking what you would expect for the effect
>complaince and mass of a lumped element model comes remarkably close to
>really showing what a line does.
I'll vote for this. This is the kind of model that I have come up with based
on my Thiel/Small and acoustic modeling knowledge. Never got to try it out...
but it seems it has already been done...
My question is, what is the difference in sound between a transmission line
with and without the port opening sealed? That is, how much (bass extension,
clarity,...) is gained by using the port?
/***************************************************************************\
* * Loudspeaker design pseudo-expert and software developer *
* John Lipp * (Working on amplifier and surround-sound psuedo-expertise) *
* li...@mtu.edu * *
* * 3rd Year Doctorate Candidate in Electrical Engineering *
\***************************************************************************/
William Spencer
>>
>>In my experience, to get this sound you need a speaker that goes deep
>>with no peaks and possibly a very gradual rolloff. This objective is
>>best realized using a ported design.
>Non-assisted ported designs roll off at 24 dB/octave. This is not a
>gradual roll-off.
I'm talking about above Fb.
>If a gradual *initial* roll is used (they are all still asymptotic to
>24 dB/octave eventually), my experiences show that you're losing the
>real benefits of a ported in the first place, which is a lower -3 dB
>point than a sealed. The gradual taper gives an earlier 3 dB
>down point.
And possibly loosing the disadvantages as well. If the resulting response has
the desired shape, it should be used. At any rate, any valid ported box will
have some increase in output at some low frequency.
>>The typical sealed box rolloff goes down low
>>enough to exite room resonances but not below them, effectively
>>exaggerating the resonances.
>I have a real problem with this one. This is such a sweeping
>generalization that it's just not true, not even "typical". The
>variables considered: room size, speaker placement, room construction,
>speaker 3 dB down point and roll-off all combine to set this
>relationship.
The desired roll-off relates to the other things. But yes it matters,
that was my point. Yes rooms matter, perhaps speakers should be designed
specifically for different rooms. But every room with the usual 8'
ceiling I've experienced has peaks around 80 Hz. Where you are in the room
only varies the severity of the effect.
>Your above statement is true, not "typically", but only of a sealed box
>with a 3 dB down point around the lowest room fundamental given poor
>speaker and listener placement.
At any rate, I found that my first attempt at non-boomy response, a low-Q
sealed box, only made matters worse, much to my surprise, while going to
the allegedly boomy ported box made matters better. So much for that.
>Again, this "10 dB" peak comment makes some pretty wild-ass
>assumptions about "typical" room size, speaker and listener location,
>room construction etc. Did you ever consider the scenario of a large
>room with a slowly rolling sealed box alignement, where placement was
>used to augment and smooth the low end roll-off? I think you missed
>the major thrust of Allison's work.
I looked for a house in this area with a large room. Sorry, I ain't rich.
At any rate, both a design that would work well in this case would also be
better in my room than the typical (sorry, that word again) 10", very much
non-sub, sealed box woofer which doesn't go down low enough to work below
room resonances or get much low end augmentation.
>>The ported box has a bad name because they are often used by those
>>wanting a lot of bass.
>They get a bad name because:
>a) there are a lot of poorly tuned mid-fi ported boomers out there;
>b) ported designs are more sensitive to manufacturing tolerances than
Also a factor but I hope you're not saying my statement was wrong.
>>A low Q sealed box with a very low resonant frequency
>>would also do this but will be larger and have less dynamics and more
>>distortion.
>It would be larger, but "less dynamics and more distortion"?
>[...] Higher order systems
>(i.e. ported) have a longer ringing time and typically more overshoot
>than a similar alignment-type of lower order (i.e. sealed). In my books,
>it looks like the sealed has better "dynamic" characteristics.[...]
>Maximum SPL abilities at resonance are another matter. The ported should
>fair better due to the driver not reaching xmax (due to the port's
>contribution) as early as in a sealed.
That's what I meant.
>>> So. We build a line with a length of 6-8 feet. We make the line
>>>taper from 2.5 times Sd at the driver to Sd at the end.
>>
>>I think you should throw out this rule of thumb. The "throat" ... of the
>>line is the area of the driver, not the area right behind it. If it
>>suddenly expands to 2.5 times this area a resonant chamber is formed.
>>[...]
>Every piece of literature I've read on TL's say that it's best to start
>with a line cross-sectional area at least twice Sd, than to taper the line
>to sd at the open end.
I've seen articles where the auther had good results with the area starting
about 25% over the area of the driver. My own experiments show it is a good
thing to try.
> This supposedly results in a sufficient duct
>area to provide adequate back absorption material, and is also supposed
>to provide a better impedance match with free space at the port end.
I really didn't want to get into it, but actually I should think the best
impedance match would flare the line like a horn at the end.
>Eliminating the "back wave bouncing back" is construction method dependant
>with TL's.
Yes, I forgot to mention my line was 2' straight, which also had a lot to do
with it not absorbing as much garbage as I would have liked. The corner tiles
are important for folded lines, at least for the first couple bends.
>Transmission lines have a gradual roll-off? I haven't seen any
>Theil/Small type modelling done on TL's, but I'd expect them to
>be closer to 4-th order high passes, like a ported.
Yes, but very wierd ported combining something a like a real low Q
sealed box with a blip of port reinforcement way down low, dieing below that.
(And that's the good designs ;-) ).
>Comments, character slams and spell checking are ecouraged.
Sorry, no character slams.
Bill s
William Spencer
>Dave goes on to post what seems to be some conclusions about how
>transmission lines REALLY work based on maybe empirical data and some
>theoretical logic. For example, he asserts that a line may, in fact be
>nothing more than a somewhat strangely tuned 4th order (i.e. vented system).
He claimed no emperical data regarding transmission lines. I don't
have that much experience either except for one experiment, but it suggests
that any TL using the rule of large line area behind the driver may lead to
false conclusions. It adds a resonant box chamber and therefore causes the
response to be more like that of some box-TL hybrid. I really wouldn't be
unhappy if you're right, just one question reaining open.
BTW, to build a TL with a constant line area equal to driver will almost
surely require some type of fairing cone to be attached to the back of the
driver magnet.
>Further, look at the bloody electrical impedance curve. It has all the
>hallmarks of a reflexm with its two bumps (actually it's really one large
>bump, due to the motional impedance of the driver, with a hole in the
>middle caused by the acoustical impedance of the enclosure resonance).
As I said, I saw one hump, but whether it was for the reason I thought
or just the same as with a mis-tuned reflex I can't conclude.
>Many lines that try to use the output of the line for reinforcement do so
>at the cost of a significant hole in the system respones in the lower
>midbass and multiples thereof. This seems to occur at frequencies that
>correspond to the wavelength of the line, PRESUMING NEAR-NORMAL
>PROPOGATION VELOCITIES.
Interesting....
Although the TL may not be like a dipole in operation,
as I suggested earlier, it certainly is like a dipole with its production
of peaks and dips. These can either improve or degrade in-room response,
depending on tuning.
>I have fond memories of my experiments, including one pair using a KEF
>B-110 5" woofer in a 3 foot long line that was -3 dB at 40 Hz. The cabinet
>was something like 7" wide, 12" deep and 28" high, weighing 50 lbs. I was
>able to achieve the SAME cutoff with flatter passband response and far
>better power handling in a more conventional reflex half the size.
It appears you also used what I am calling oversize line areas with this
one. With the area I suggest it could be smaller than the ported box for
that line length.
I have also compared TLs to dipoles regarding size. There is no relation
between dipoles and boxes because dipoles don't work on enclosed volume.
Yet dipoles just happen to work out taking up a similar amount of space
(but different shaped). This could also be true for TLs, unless you're
right about the similarity to ported boxes.
Bill S