Scottmoose said:
The stuffing certainly makes the line act longer at the l/4 frequency. This has been documented and measured many times. I've made the measurement myself. You seem to be ignoring the adiabatic to isothermal shift of the air as a result of the stuffing. TL's are a distributed mass/compliance element, and the stuffing increases the compliance per unit length. Some have claimed that a heavily stuffed line completely absorbs the rear wave, this is not true, however the l/4 resonance is certainly reduced in amplitude in a heavily stuffed line. One should consider the velocity/presssure profile down the line when stuffing a TL.
Pete B.
How do you get even multiples of the fundamental 1/4 wave in a TL like the Thor? Or any other closed at one end and open at the other TL.
My measurements indicate that the shift from adiabatic towards isothermal occurs but is not that significant. I believe that this is less than a 10% shift in the speed of sound. How much longer do you think the TL acts? If the stuffing changes the compliance of the TL, are you saying that motion of the fibers is a contributor to a TL design as proposed by Bradbury?
Just to add my tuppence worth, no-one is denying that QWRs produce harmonic resonances higher than the 1/4 wavelength fundamental you wished to generate in the first place (with the caveat Martin mentions in that you can't actually have even multiples of the fundamental in a pipe which is closed at one end...;-). But the objective is to use the fundamental 1/4 wave resonance to prop up the low frequency output. Hence my preference for the QWR name, which, er, sums it up rather nicely. I suppose we could call them quarter-wave-with-additional-higher-harmonic-resonators, but it doesn't exactly trip off the tongue does it? The link to transmission lines on the other hand I find much more tenuous, given that it is primarily an electrical term. I would only apply it to those lines where the object is to produce the flattest impedence curve possible, usually through massive over-stuffing, which absorbs the fundamental. Especially effective at the line terminus, as Martin in particular has shown.
As regards stuffing making the line 'longer' Martin has covered that in the above post. Stuffing as far as I'm concerned is only there to damp out those harmonics you can't flatten by engineering them out of the enclosure in the first place, and should be used very sparingly. It's not a substitue for a proper enclosure design.
Regards
Scott
As regards stuffing making the line 'longer' Martin has covered that in the above post. Stuffing as far as I'm concerned is only there to damp out those harmonics you can't flatten by engineering them out of the enclosure in the first place, and should be used very sparingly. It's not a substitue for a proper enclosure design.
Regards
Scott
MJK said:
Hello Martin,
I should probably point out that my disagreement is with Scott's dismissal of the phrase transmission line, when in fact you use it generously on your site. It is a minor, and not worth arguing about. However, Scott suggests an error in the work of the past and is dismissive of it, when in fact we all built and learned from past work, even that of lesser quality. I enjoyed reading Olney's original work from the mid 1930s, fascinating, largely ignored, and better than much of the work that was done in the 60s and 70s.
I read your site some time ago and you've done a fine job, I do have a few comments that we can discuss offline if you'd like. I just noticed that you used the Quarter Wave name, which I have no problem with as long as we do not dismiss the transmission line analogy. It is routine, as I'm sure you know to model loudspeakers, indeed most acoustical systems with electrical analogies to acoustical networks.
Concerning your point, the input impedance of an lossless line is essentially the characteristic impedance of the line multiplying a tan function, it is always complex or reactive. Now, how we model this depends on the choice of a mobility or impedance analogy, one case being the reciprocal of the other. In any case, rather than get into all the details, the behavior is that the line is a short circuit from a voltage to volume velocity analogy at l/4 and odd multiples, and an open circuit at even multiples. The nature of a TL is that it exhibits both series, and parallel resonance in it's input impedance. Both are resonance conditions. It is also important to consider what I call the ATL volume velocity transfer function, but perhaps this is for another time.
First, I should point out, for clarity, that the l/4 frequency is a minimum, between the LF peaks, in the electrical input impedance of a TL system. It is analogous to Fb in a vented system. I'll take 10%, but I have seen 15 to 20% even more in heavily stuffed lines. Considering the low end and how difficult it is to get every last Hertz, a reduction from 40 Hz to 36 as a 10% example is nothing to sneeze at, certainly is not insignificant as was suggested in the original post.
I believe that the behavior of damped lines is not well understood, and it seems that the l/4 frequency sees a greater reduction than the ideal 1.4 change in compliance due to isothermal conditions. I believe that it depends on the material and that damped lines are a good area for more research and experimentation. A bonded damping material where the fibers are bonded together will have less motion and might offer a good test case, however one must also take into account other changes due to the bonding. I've wondered about Bradbury's paper in the past, but now having looked at more data wonder if he might be correct at least for some cases. I've not done any detailed experiments to test the fiber movement theory.
Pete B.
First let me say thanks for the sims that you offered here, interesting material.
Pete B.
Do you have some analysis to support your claim or are you just going to ride on the claims of Martin?....;-)Scottmoose said:
Your statements suggest here "flattest impedence curve possible" that you do not recognize resonant TLs which are a building block in microwave design, stub and double stub tuning. Martin repeatedly uses the phrase transmission line on his own site, how can you dismiss it? I suggest we agree to disagree as this is unproductive. BTW TL rolls off the tongue just fine.Scottmoose said:
I have often commented on how important it is to dampen higher resonant modes with damping material, and here we agree. However, the reduction in l/4 in lines such as the Thor cannot be ignored, it is significant. Look at the reduction in the lower impedance peak below l/4 in the Thor even with the minimal stuffing case. The input impedance is an indicator of the mechano-acoustical load as it is reflected into the primary of the driver. There is a drastic change in behavior, around and even below the l/4 frequency as indicated in the input impedance plot. To dismiss this point clouds the understanding of their behavior.Scottmoose said:
Pete B.
Hi Pete,
Unfortunately, I have not read the paper ny Olney so I cannot comment. I have read the work of Bailey and Bradbury from the 60's and 70's and in my opinion Bradbury's paper sent transmission line understanding down the wrong path for over 30 years.
I am not an electrical engineer, I am a mechanical engineer dabbling just enough in electrical circuit theory to model speaker systems. I have read the electrical engineering theory version of transmission lines and wave guides and almost understood part of what was being presented. The problem I have with classic closed form solutions is that they are only accurate for a very limited number of ideal geometries. I have seen efforts to use electrical transmission line and wave guide theory to model quarter wave speaker systems and it gets very messy quickly, the work approaches Phd level documentation and exceeds my limited knowledge in that area. Personally, I need to stick with something closer to a mechanical analogy for me to understand. I think that mechanical modal analysis theory is the next step in understanding what goes on inside a tuned acoustic pipe and this is the direction in which I am headed.
I need a simpler explanation to understand. If I plot the velocity or pressure along the length of a classic TL geometry at resonance, I see 1/4 or odd multiples of 1/4 cosine or sine waves respectively. I don't see any even multiples, which would be half waves, in my simple mind's visualization.
The minimum that occurs between the impedance peaks of an unstuffed TL or a BR, the curves will look very similar, is not a system resonant condition. It is the midpoint between two system resonances which produce the humps in the electrical impedance curve above and below this frequency.
In my MathCad worksheets I do account for some slowing of the speed of sound as a function of stuffing density, I think approximately 10% is the maximum reduction I apply. For a given length of transmission line, the best way to reduce the tuning frequency is with geometry and not with the adiabatic to isothermal transition created by adding stuffing. For small signal linear operation of TL's, I have seen no evidence of fiber motion and the claimed significant reduction in the speed of sound. I have seen claims of fiber motion under extreme operation, 1812 cannon shots for example, but do not believe that for most music played even at even very loud volumes that fiber motion occurs and impacts the system design and performance. If somebody is counting on fiber motion dramatically increasing the line's effective length then I think thay have missed other more significant design variables.
I cannot agree with what you have written above. People send me measurements of the speakers they have designed using the MathCad worksheets several times a year and the correlation in usually extremely good. When the correlation is not so good there is typically something that was not simulated correctly, garbage in-garbage out like manufacturer's inconsistent T/S parameters, or a measurement technique problem. Making good measurements is harder then doing the engineering and construction. This tells me that the current tools are good enough engineering approximations to the way standing waves behave in an enclosure that designing with confidence is appropriate, but recognize with any engineering approximation there is always room for improvement. I worked for a manager once that used to preach better is the enemy of good enough.
Unfortunately, I have not read the paper ny Olney so I cannot comment. I have read the work of Bailey and Bradbury from the 60's and 70's and in my opinion Bradbury's paper sent transmission line understanding down the wrong path for over 30 years.
I am not an electrical engineer, I am a mechanical engineer dabbling just enough in electrical circuit theory to model speaker systems. I have read the electrical engineering theory version of transmission lines and wave guides and almost understood part of what was being presented. The problem I have with classic closed form solutions is that they are only accurate for a very limited number of ideal geometries. I have seen efforts to use electrical transmission line and wave guide theory to model quarter wave speaker systems and it gets very messy quickly, the work approaches Phd level documentation and exceeds my limited knowledge in that area. Personally, I need to stick with something closer to a mechanical analogy for me to understand. I think that mechanical modal analysis theory is the next step in understanding what goes on inside a tuned acoustic pipe and this is the direction in which I am headed.
I need a simpler explanation to understand. If I plot the velocity or pressure along the length of a classic TL geometry at resonance, I see 1/4 or odd multiples of 1/4 cosine or sine waves respectively. I don't see any even multiples, which would be half waves, in my simple mind's visualization.
The minimum that occurs between the impedance peaks of an unstuffed TL or a BR, the curves will look very similar, is not a system resonant condition. It is the midpoint between two system resonances which produce the humps in the electrical impedance curve above and below this frequency.
In my MathCad worksheets I do account for some slowing of the speed of sound as a function of stuffing density, I think approximately 10% is the maximum reduction I apply. For a given length of transmission line, the best way to reduce the tuning frequency is with geometry and not with the adiabatic to isothermal transition created by adding stuffing. For small signal linear operation of TL's, I have seen no evidence of fiber motion and the claimed significant reduction in the speed of sound. I have seen claims of fiber motion under extreme operation, 1812 cannon shots for example, but do not believe that for most music played even at even very loud volumes that fiber motion occurs and impacts the system design and performance. If somebody is counting on fiber motion dramatically increasing the line's effective length then I think thay have missed other more significant design variables.
I cannot agree with what you have written above. People send me measurements of the speakers they have designed using the MathCad worksheets several times a year and the correlation in usually extremely good. When the correlation is not so good there is typically something that was not simulated correctly, garbage in-garbage out like manufacturer's inconsistent T/S parameters, or a measurement technique problem. Making good measurements is harder then doing the engineering and construction. This tells me that the current tools are good enough engineering approximations to the way standing waves behave in an enclosure that designing with confidence is appropriate, but recognize with any engineering approximation there is always room for improvement. I worked for a manager once that used to preach better is the enemy of good enough.
I don't think I've seen it online, and I don't have it scanned at the moment.MJK said:
I agree that the theory is very difficult, however it seems to me that the electrical engineering side is the most developed offering many methods that simplify common problems. Smith charts for example can help in the understanding of TLs. Yes I agree that the simplified models often do not offer enough flexibility for modelling real systems, however they're fundamental to understanding the basics. I cannot emphasize how important this is. SPICE could be used with coupled transmission lines for complex geometries but I've not tried this approach. I wondered about the basis of your analysis and it seems that we're each more comfortable with analogies in our own areas of specialty, as would be expected. The partial differential equations are the same for analogous systems so no matter how its done we should end up with the same answer.MJK said:
MJK said:
It seems that you only recognize peaks as resonance, however we should think about the definition. Resonance is when the negative reactive component is equal to the postive reactive component so that they cancel, with only the resistive losses remaining. The phase is zero because the reactive components cancel. Impedance peaks in a response occur with parallel resonance, or what we call a tank circuit in electronics. The common use is tuning the antenna in a radio circuit, the impedance is high at resonance, infinite for the lossless case. However, series resonance must also be considered as is seen in speaker notch filters where the impedance goes to a low value, zero for the lossless case.
What is interesting in the vented and TL cases is that we have coupled resonance systems. The driver is a parallel resonance system where we mainly see the piston mass, suspension compliance, and the loss factors in parallel. However, a vented system places a series resonance circuit, the box compliance and port mass, across the parallel resonance circuit of the driver. The analysis of this coupled system is complex, a very interesting feature is that for the high Q case, the series circuit, box and vent completely short out the parallel resonance circuit and the system response is only dependent on the box, vent, and motor characteristics, not the cone, loss, or suspension. This is where we see the minimum at Fb and it is indeed resonance independent of the upper and lower peak. The TL input impedance is analogous at and around l/4 and is in fact the minimum between the lowest frequency peaks in Zin. The peaks above and below Fb or l/4 are a form of parallel resonance but are actually a complex interaction between the driver and load. You can see in your own experimental data that the Zin phase crosses zero at the upper and lower peaks, but also at the minimum in between, each one is a resonance condition.
I agree that we should not rely on motion of the fibers to build a better system as it would be non-linear but where I believe it comes into play is in heavily stuffed situations where the lower peak seems to vanish, or probably move very low in frequency. I agree that heavy stuffing is not the way to go, but what do we do when we want to analyze the situation? I believe that Thor is over stuffed and we might note that the simulations offered here do not agree very well with measurements by the author, or SEAS. This is where I believe more work is needed even just to prove that heavily stuffed is not the way to go.MJK said:
What I also find interesting is that the more recent suggestions for tapered lines and driver offset down the line, is moving the behavior closer to that of a traditional vented system. Your woek is the best I've seen so far and useful for cases where reasonable amounts of stuffing are used.
Pete B.
PB2 said:
Here is a reference on TLs for those who are interested in the math behind them:
http://hibp.ecse.rpi.edu/~crowley/java/Transline/transinfo.htm
We can take the reciprocal of the delay time to get prop delay:
propagation delay = 1/sqrt(LC) m/sec
L and C are per unit length parameters of the line
I just wanted to clarify that if the compliance/unit length is increased by a factor of 1.4 the prop delay is only reduced by a factor of 1/sqrt(1.4) or .85. I offer this to point out that we do not need to see a full factor of 1.4 reduction in support of the isothermal/adiabatic view.
Pete B.
Re: Thor Design Comments
I see some negative comments about the Thor design and I want to point out that my comments concern minor issues.
I reread the original article and now remember what the issue was with the crossover. The author provided measured input impedance for each driver, and electrical transfer functions for the crossover. I entered the driver impedances and crossover design, including inductor losses, in a simulator and the tweeter filter response which normally has excellent agreement had a different shape at and around the crossover region. The Q was different and from memory the response shape was different by about 1 or 2 dB. There were no acoustic measurements involved in this particular comparison ruling it out as a source of error. I was about to dismiss it when I saw another reliable source mention similar findings on the Madisound board. I was curious to determine the reason for the difference, which we never did. It is a minor point.
Pete B.
PB2 said:
I see some negative comments about the Thor design and I want to point out that my comments concern minor issues.
I reread the original article and now remember what the issue was with the crossover. The author provided measured input impedance for each driver, and electrical transfer functions for the crossover. I entered the driver impedances and crossover design, including inductor losses, in a simulator and the tweeter filter response which normally has excellent agreement had a different shape at and around the crossover region. The Q was different and from memory the response shape was different by about 1 or 2 dB. There were no acoustic measurements involved in this particular comparison ruling it out as a source of error. I was about to dismiss it when I saw another reliable source mention similar findings on the Madisound board. I was curious to determine the reason for the difference, which we never did. It is a minor point.
Pete B.
GUNGNIR (Odin's Spear)
I am thinking about the front half of my Home Theater, and I had the idea of a 'stretched Thor' by simply unfolding the pipe.
These would be called GUNGNIR (Odin's Spear)
My idea was that they could be thinner (about 14-15" deep) and could then be integrated into a built in book case. An acoustically transparent retractable screen would cover the center speaker.
The terminus would point straight up, which in my case might not be too bad as that side of the living room is 12' high.
Would this sound pretty similar to the Fat Thor's?
The reason I'm asking is that I'm using two Thor's on the side of the screen with an Odin in the center, and I think a pair of Fat Thors and a Gungnir in the center, or 3 Gungnir's would sound better.
What do y'all think?
I am thinking about the front half of my Home Theater, and I had the idea of a 'stretched Thor' by simply unfolding the pipe.
These would be called GUNGNIR (Odin's Spear)
My idea was that they could be thinner (about 14-15" deep) and could then be integrated into a built in book case. An acoustically transparent retractable screen would cover the center speaker.
The terminus would point straight up, which in my case might not be too bad as that side of the living room is 12' high.
Would this sound pretty similar to the Fat Thor's?
The reason I'm asking is that I'm using two Thor's on the side of the screen with an Odin in the center, and I think a pair of Fat Thors and a Gungnir in the center, or 3 Gungnir's would sound better.
What do y'all think?
Has the 3/2004 Hobby HiFi article, written in German, covering Thor been discussed here? I believe that they offered a slightly different crossover design the reduced the system output around 3 kHz.
As I recall the system input impedance showed a larger peak in the lowest resonance indicating that something was different in the stuffing. Distribution or amount used.
Pete B.
As I recall the system input impedance showed a larger peak in the lowest resonance indicating that something was different in the stuffing. Distribution or amount used.
Pete B.
A different resonant system
I want to offer this as a thought exercise, it is a lumped system but it might help with understanding the different resonant modes. Let's consider the case where we take a speaker and mount it with the voice coil in the horizontal plane, this speaker has no cone or suspension and the voice coil is self centering and massless. We attach a massless spring between a mass on a frictionless surface and the voice coil. Now we drive it with a current source, we should add some damping to the mass but it can be very small in considering the high Q case:
Consider:
the shape of the input impedance
the shape of the voice coil velocity versus frequency
the velocity transfer function from the voice coil to the mass - roughly
the impedance at resonance
This is obviously just a thought exercise.
Pete B.
I want to offer this as a thought exercise, it is a lumped system but it might help with understanding the different resonant modes. Let's consider the case where we take a speaker and mount it with the voice coil in the horizontal plane, this speaker has no cone or suspension and the voice coil is self centering and massless. We attach a massless spring between a mass on a frictionless surface and the voice coil. Now we drive it with a current source, we should add some damping to the mass but it can be very small in considering the high Q case:
Consider:
the shape of the input impedance
the shape of the voice coil velocity versus frequency
the velocity transfer function from the voice coil to the mass - roughly
the impedance at resonance
This is obviously just a thought exercise.
Pete B.
![image_091_[1].jpg](/community/data/attachments/53/53903-9a825146e65d76c40b3d29fc9dc61a7d.jpg)
