Scott
I was interested in the results you were getting of in-room simulation, and was a bit surprised by the magnitude of the bass boost, although I knew this was significant.
So I went back to some of the reviews of speakers done in the early nineties, when they actually measured speakers, and was able to compare anechoic and "averaged" listening room results.
They compare well with the results you are getting.
I was interested in the results you were getting of in-room simulation, and was a bit surprised by the magnitude of the bass boost, although I knew this was significant.
So I went back to some of the reviews of speakers done in the early nineties, when they actually measured speakers, and was able to compare anechoic and "averaged" listening room results.
They compare well with the results you are getting.
That correlates with my own experiences too.
Dave -good question. I can't do it myself properly at present, as I don't have the facilities to make my own data of much use in a properly controlled scientific study. However, whenever I find an enclosure I am interested in, that has been properly measured in an anechoic chamber, I try to model its dimensions in the worksheets and then compare the two graphs. I can't really separate the sheet from the stuffing, so to speak, as they tend to go hand in hand. However, this one might be of interest to you.
Below, I present the anechoic model of the Seas Thor kit (I can't seem to escape this one, but it's a really good example) Lovely drivers and crossover, insofar as I can see (I haven't heard them), but the enclosure... well, let's say it has issues. Keep with me here -it is relevant to your particular interest, as I shall make clear in a minute. For these, by the way, I'm going to stick with anechoic simulations and measurements for the moment.
Dave -good question. I can't do it myself properly at present, as I don't have the facilities to make my own data of much use in a properly controlled scientific study. However, whenever I find an enclosure I am interested in, that has been properly measured in an anechoic chamber, I try to model its dimensions in the worksheets and then compare the two graphs. I can't really separate the sheet from the stuffing, so to speak, as they tend to go hand in hand. However, this one might be of interest to you.
Below, I present the anechoic model of the Seas Thor kit (I can't seem to escape this one, but it's a really good example) Lovely drivers and crossover, insofar as I can see (I haven't heard them), but the enclosure... well, let's say it has issues. Keep with me here -it is relevant to your particular interest, as I shall make clear in a minute. For these, by the way, I'm going to stick with anechoic simulations and measurements for the moment.
Attachments
...and here's the measurements of the enclosure taken by Bjorn Idland at the Seas anechoic chamber in Norway. The overall response curves match extremely well as you can see. Please ignore the higher gain the sim implies -my apologies, and entriely my fault: I mucked up entering a parameter when paralleling the drivers. Change it back and it snaps into line, and here, I'm interested only in the shape of the response curve.
Pay particular interest to what happens below 200Hz.
Pay particular interest to what happens below 200Hz.
Attachments
So, what am I driving at?
Well, the twin drivers used in Thor have an Fs of 31Hz. I mention this not thorugh any misguided belief that a QWR should be tuned to the driver Fs, but to illustrate that these magnesium woofers are capable drivers, that can cheerfully go very low indeed.
Does that look like a good anechoic response from a TL? No, it doesn't to me either. It's rapidly falling away below 200Hz. I'm all in favour of compensating for the boost room-gain will provide, but this is daft. What's going on? Well, my own initial reaction was to assume the cabinet was physically too small for the drivers.
Thor was nominally designed by Dr. J. D'Apollito using Mr George Ausperger's TL alignment tables (though why Dr J.D didn't just ask him for the software is a mystery to me.) However, int he article, Dr. J.D. quite openly states that he arbitarily threw away 1/3 of the cabinet volume. Which I leaped upon, baying like a bloodhound. I should have kept my mouth shut. This is what happens when you double the internal volume of Thor:
Well, the twin drivers used in Thor have an Fs of 31Hz. I mention this not thorugh any misguided belief that a QWR should be tuned to the driver Fs, but to illustrate that these magnesium woofers are capable drivers, that can cheerfully go very low indeed.
Does that look like a good anechoic response from a TL? No, it doesn't to me either. It's rapidly falling away below 200Hz. I'm all in favour of compensating for the boost room-gain will provide, but this is daft. What's going on? Well, my own initial reaction was to assume the cabinet was physically too small for the drivers.
Thor was nominally designed by Dr. J. D'Apollito using Mr George Ausperger's TL alignment tables (though why Dr J.D didn't just ask him for the software is a mystery to me.) However, int he article, Dr. J.D. quite openly states that he arbitarily threw away 1/3 of the cabinet volume. Which I leaped upon, baying like a bloodhound. I should have kept my mouth shut. This is what happens when you double the internal volume of Thor:
Attachments
So, it's not that. What else?
Stuffing. (told you there was a point to it!). Ausperger apparantly recommended 0.78lbs ft^3 of stuffing throughout the enclosure. Now in my own experience, both with models, and with building enclosures, that is excessive in the extreme -I would never go pbeyond 0.5lbs ft^3, and even then, I wouln't be too happy. So, I reduced the stuffing, in the simulation, to 0.1lbs ft^3 to see what happened.
Stuffing. (told you there was a point to it!). Ausperger apparantly recommended 0.78lbs ft^3 of stuffing throughout the enclosure. Now in my own experience, both with models, and with building enclosures, that is excessive in the extreme -I would never go pbeyond 0.5lbs ft^3, and even then, I wouln't be too happy. So, I reduced the stuffing, in the simulation, to 0.1lbs ft^3 to see what happened.
Attachments
I posted a long thread on Thor models and designs over on the other, multi-way page, which is where I lifted these graphs from for speed.
As I said over there, instant conclusion, examining the bass performance: Ausperger's stuffing advice is way off, at least in this case. The object of a QWR is to create a fundamential resonance where you wish, to support the bass, correct? But we know that over-stuffing can damage this fundamental, because instead of being used in moderation simply to damp higher resonances, it's starting to damp out the very thing we wanted to generate. Hence my suggestion over on the other thread to Thor owners to rip the stuffing out of the second half of the line for starters. As stuffing appears to work best in high-velocity areas, if they can cope with a touch more ripple, this could be a good way forward for them, so the rear-wave isn't being forced to run though a completely, heavily stuffed line. At present, my summary is that this enclosure, like Bailey's original 'Transmission Lines', and also arguably the Alpha TLs, is actually behaving closer an aperiodic enclosure, not a QWR.
This is a more detailed measurement in some ways of the Thor enclosure in the Seas anechoic chamber on a different occasion. It's pretty similar. However, note the ripple between 100Hz-200Hz, which I think is the third harmonic. It's flattened in the 0.78lbs ft^3 MathCad sim, but you can see evidence of it in the reduced stuffing model (the higher, not the lower frequency -the lower will be flattened by any increase in stuffing at the response plummets). So, again, my own conjecture is that this is due to, perhaps, a slight aberration in the quantity of stuffing used in the enclosure -a slight reduction down to, perhaps, around 0.7lbs, or a movement away from the highest-velocity areas perhaps. I'll see if I can run a sim later to see if that is in fact the case.
I'll try to dig out some more examples when I can -nothing very profound or original here at all I'm afraid, but it seems to go a little way toward illustrating that stuffing strategies need to be carefully worked out and that Martin's MathCad modelling appears to predict the effects with a high degree of accuracy.
For myself, I haven't in practical terms found any difference yet in the materials I have used for this purpose -Dacron, firbreglass, long-hair wool etc., but I try not to use too high a density anyway, so I imagine any aberrations would be lessened. Whilst I think stuffing certainly has a place and a value in damping higher resonances, it can be over-used, as you know, to correct problems best supressed by careful engineering of the physical enclosure dimensions. It looks to be a highly interesting study however, and as and when I find more data or measurements that might be of use to you, I will, of course let you know.
As I said over there, instant conclusion, examining the bass performance: Ausperger's stuffing advice is way off, at least in this case. The object of a QWR is to create a fundamential resonance where you wish, to support the bass, correct? But we know that over-stuffing can damage this fundamental, because instead of being used in moderation simply to damp higher resonances, it's starting to damp out the very thing we wanted to generate. Hence my suggestion over on the other thread to Thor owners to rip the stuffing out of the second half of the line for starters. As stuffing appears to work best in high-velocity areas, if they can cope with a touch more ripple, this could be a good way forward for them, so the rear-wave isn't being forced to run though a completely, heavily stuffed line. At present, my summary is that this enclosure, like Bailey's original 'Transmission Lines', and also arguably the Alpha TLs, is actually behaving closer an aperiodic enclosure, not a QWR.
This is a more detailed measurement in some ways of the Thor enclosure in the Seas anechoic chamber on a different occasion. It's pretty similar. However, note the ripple between 100Hz-200Hz, which I think is the third harmonic. It's flattened in the 0.78lbs ft^3 MathCad sim, but you can see evidence of it in the reduced stuffing model (the higher, not the lower frequency -the lower will be flattened by any increase in stuffing at the response plummets). So, again, my own conjecture is that this is due to, perhaps, a slight aberration in the quantity of stuffing used in the enclosure -a slight reduction down to, perhaps, around 0.7lbs, or a movement away from the highest-velocity areas perhaps. I'll see if I can run a sim later to see if that is in fact the case.
I'll try to dig out some more examples when I can -nothing very profound or original here at all I'm afraid, but it seems to go a little way toward illustrating that stuffing strategies need to be carefully worked out and that Martin's MathCad modelling appears to predict the effects with a high degree of accuracy.
For myself, I haven't in practical terms found any difference yet in the materials I have used for this purpose -Dacron, firbreglass, long-hair wool etc., but I try not to use too high a density anyway, so I imagine any aberrations would be lessened. Whilst I think stuffing certainly has a place and a value in damping higher resonances, it can be over-used, as you know, to correct problems best supressed by careful engineering of the physical enclosure dimensions. It looks to be a highly interesting study however, and as and when I find more data or measurements that might be of use to you, I will, of course let you know.
Apricot and cranberry, with sausage meat. Roast in a medium oven, a few potatos, chipollatas, sprouts etc and a nice gravy with red-wine added. Excellent stuff.
Hey dave,
Maybe you missed it way back because of the activity on this thread, but I tried to ask what the problem was with your OB setup.
Is it a simple problem of space, or is it some other reason? I'm asking because I have two 6x2-foot baffles just about to get holes sawed into them. I've got the space for two of them laid end to end, in a space unoccupied by you-know-who.
I love my basement office/den/AV room!
Dave
Maybe you missed it way back because of the activity on this thread, but I tried to ask what the problem was with your OB setup.
Is it a simple problem of space, or is it some other reason? I'm asking because I have two 6x2-foot baffles just about to get holes sawed into them. I've got the space for two of them laid end to end, in a space unoccupied by you-know-who.
I love my basement office/den/AV room!
Dave
Hi Scott,
Above and beyond the call with the data, thanks very much. Its a great match. Now, did you play with the density assumptions to get the match? Both material type and density as you know affect the response, it looks like Bjorn didn't publish his assumptions here? Any light you could shed?
0.8 lb/ft^3 is over the top if extending the low end via the port is a goal; A port 20 dB down justs steepens the ultra low end roll off. Getting much above 1lb/ft^3 also starts to eat volume IME.
Getting back to TLs. Back in '93 Bob Bullock's students were looking into fiber in TLs. Bullock was harmonizing Putland's fiber models with Letts TL work. The model didn't quite strike a good relationship with measured results. Todd Jenkins (Bob's student at the time) investigated various materials (shared the results with me) and found Bradbury's models didn't quite sit right. Speed of sound didn't slow near what Bradbury assumed, tangles didn't really move as expected. Still, he found velocity reduction greater than Martin did, especially below 30 Hz, where 180 m/s was common (measured, not simmed) at 20 Hz. Ramon Cancel (Bullock's student in 95) shared further results with me at the time, corroborating Todd's earlier work. That the velocity reduction increased below 30 Hz indicate the fibers may be moving, which implies a divergence from Martins assumptions, below 30 Hz. For ultra low subs, it may be prudent to assume c reduces faster than the model credits.
So this left us with Bradbury's work not quite right, substantiated later by Bullock's students and by George A, then Martin, and still some question marks remain regarding material type effects.
I'm looking at Todds old attenuation measures and he's showing wool and Madisound poly with the highest attenuation, & Acoustuff and Dacron about 60% less. All show about half the attenuation at 20 Hz (1.5 to 4 dB per m) vs 100 Hz. I've been carrying this data around with me for over 10 years waiting for some bright individual with the time to corroborate of augment it!
I should also mention some experimentation I've made with Rigid Fiberboard. Densities up to 6 lb/ft^3, great looking absortion coeffs on paper in small sizes. I bought a 4x2 foot bail, 8 pcs 2" thick/ea. Thought about using it to line vented and TL enclosures, but ran some tests as a floor absorber when taking in room measures. It showed significant skimming reflection, up to 2 kHz! To get the absorption, you need to hit it normal. Still, I think this could be great for lining a back wall. I'd never use it elsewhere in the enclosure.
Dave
PS My experience with long fiber wool, fiberglass and audio poly gave me the following behaviours in several sealed cabinets: wool dries out the bass; very attenuative down to very low frequencies, good to correct high Q cabinets. Garden-variety audio Poly subjectively is not attenuative enough in the mid bass for my liking, results in less definition. R13 fiberglass is just-right for a properly designed enclosure, as far as I'm concerned. Just be wary of getting it sucked in the voice coil gap, as you know.
Above and beyond the call with the data, thanks very much. Its a great match. Now, did you play with the density assumptions to get the match? Both material type and density as you know affect the response, it looks like Bjorn didn't publish his assumptions here? Any light you could shed?
0.8 lb/ft^3 is over the top if extending the low end via the port is a goal; A port 20 dB down justs steepens the ultra low end roll off. Getting much above 1lb/ft^3 also starts to eat volume IME.
Getting back to TLs. Back in '93 Bob Bullock's students were looking into fiber in TLs. Bullock was harmonizing Putland's fiber models with Letts TL work. The model didn't quite strike a good relationship with measured results. Todd Jenkins (Bob's student at the time) investigated various materials (shared the results with me) and found Bradbury's models didn't quite sit right. Speed of sound didn't slow near what Bradbury assumed, tangles didn't really move as expected. Still, he found velocity reduction greater than Martin did, especially below 30 Hz, where 180 m/s was common (measured, not simmed) at 20 Hz. Ramon Cancel (Bullock's student in 95) shared further results with me at the time, corroborating Todd's earlier work. That the velocity reduction increased below 30 Hz indicate the fibers may be moving, which implies a divergence from Martins assumptions, below 30 Hz. For ultra low subs, it may be prudent to assume c reduces faster than the model credits.
So this left us with Bradbury's work not quite right, substantiated later by Bullock's students and by George A, then Martin, and still some question marks remain regarding material type effects.
I'm looking at Todds old attenuation measures and he's showing wool and Madisound poly with the highest attenuation, & Acoustuff and Dacron about 60% less. All show about half the attenuation at 20 Hz (1.5 to 4 dB per m) vs 100 Hz. I've been carrying this data around with me for over 10 years waiting for some bright individual with the time to corroborate of augment it!
I should also mention some experimentation I've made with Rigid Fiberboard. Densities up to 6 lb/ft^3, great looking absortion coeffs on paper in small sizes. I bought a 4x2 foot bail, 8 pcs 2" thick/ea. Thought about using it to line vented and TL enclosures, but ran some tests as a floor absorber when taking in room measures. It showed significant skimming reflection, up to 2 kHz! To get the absorption, you need to hit it normal. Still, I think this could be great for lining a back wall. I'd never use it elsewhere in the enclosure.
Dave
PS My experience with long fiber wool, fiberglass and audio poly gave me the following behaviours in several sealed cabinets: wool dries out the bass; very attenuative down to very low frequencies, good to correct high Q cabinets. Garden-variety audio Poly subjectively is not attenuative enough in the mid bass for my liking, results in less definition. R13 fiberglass is just-right for a properly designed enclosure, as far as I'm concerned. Just be wary of getting it sucked in the voice coil gap, as you know.
Interesting stuff re the materials Dave. Thanks for sharing that with us. Whilst we're on this matter, I don't know if you were aware that Bob Brines doesn't use any stufing at all in his own MLTL enclosures I understand. He used to, but now lines the enclosure's internals with thin concrete panels, I assume to deaden any MDF resonances (more than people automatically assume), and then with stiff accoustic yellow fibreglass. However, I don't know how he models it, as I know he uses Martin's software: he's taken a licence to use it (good man!). Knowing Bob, I very much doubt if he just rams it in, then takes measurements, adjusts, more measurements, etc., and does it that way. I would guess that he probably came up with a pair of test mules, one stuffed, one lined, and compared various quantities / strategies of both until he found he could model it accurately by setting a specific stuffing density into the worksheets. That's a guess of course; I'd have to ask him, and as he'd moved closer to commercial, I don't know if he'd be willing to share his experiences. Oddly, I've been meaning to fire him an email on this very question for some time, and the Ariel MLTL (bit rough and ready, but I'm convinced that I / we can beat the existing enclosure) I came up with a couple of pages ago and your own contrbutions have rather set me off again. I'll ask him: can but try!
Sorry Dave, no, I haven't yet re-run the sim -had too much to do last night and I clean forgot. My apologies: I'll try to get that done sometime today & get back to you on the other matters too. However, regarding your latter point, I believe that is the case, going by memory of the Thor's components. On the up side, every time I've run a sim and compared it to the anechoic response of an enclosure, it matches (assuming I've entered up the dimensions correctly!) very well. Not 100% perfect of course; that would be next to impossible, but nothing significant to worry about.
Sorry Dave, no, I haven't yet re-run the sim -had too much to do last night and I clean forgot. My apologies: I'll try to get that done sometime today & get back to you on the other matters too. However, regarding your latter point, I believe that is the case, going by memory of the Thor's components. On the up side, every time I've run a sim and compared it to the anechoic response of an enclosure, it matches (assuming I've entered up the dimensions correctly!) very well. Not 100% perfect of course; that would be next to impossible, but nothing significant to worry about.
Hi Dave
I've re-run the sim; this is what happens with 0.6lbs ft^3 of stuffing. As you can see the ripple at F3 (I believe!) increases. Not a vast amount, but looking at the response curve I think the fractional peak at between 150-200Hz is the same as the slight increase the worksheet is predicting at 200Hz.
This is conjecture on my part, but I suspect what happened is that the stuffing in the enclosure had moved slightly before it was measured. D'Apollito notes in the original article that the stuffing in the second half of the line, with it being wider at the bottom than the top had a habit of settling, albeit after some months -I imagine movement into and out of the chamber used to take the measurements would have similar results. I can't model that effect, as Martin's TL Offset Driver sheet, which is what I used for modelling Thor, assumes a uniform stuffing density throughout. However, I think this is the most likely explanation. It's very minor anyway, within a db or so, and as Martin points out, he's not chacing every sight difference. For what it's worth, I'm satisfied the prediction is accurate. I checked the other points you raised too just to make sure and I think the model and the measurement are pretty much as close as we're going to get, at least for the moment. I've emailed Bob asking if he can share any experience regarding lining and its equivelenace to stuffing -haven't heard back from him yet, but it is the weekend!
All the best
Scott
I've re-run the sim; this is what happens with 0.6lbs ft^3 of stuffing. As you can see the ripple at F3 (I believe!) increases. Not a vast amount, but looking at the response curve I think the fractional peak at between 150-200Hz is the same as the slight increase the worksheet is predicting at 200Hz.
This is conjecture on my part, but I suspect what happened is that the stuffing in the enclosure had moved slightly before it was measured. D'Apollito notes in the original article that the stuffing in the second half of the line, with it being wider at the bottom than the top had a habit of settling, albeit after some months -I imagine movement into and out of the chamber used to take the measurements would have similar results. I can't model that effect, as Martin's TL Offset Driver sheet, which is what I used for modelling Thor, assumes a uniform stuffing density throughout. However, I think this is the most likely explanation. It's very minor anyway, within a db or so, and as Martin points out, he's not chacing every sight difference. For what it's worth, I'm satisfied the prediction is accurate. I checked the other points you raised too just to make sure and I think the model and the measurement are pretty much as close as we're going to get, at least for the moment. I've emailed Bob asking if he can share any experience regarding lining and its equivelenace to stuffing -haven't heard back from him yet, but it is the weekend!
All the best
Scott
Dave,
Do your measurement results include with and without stuffing plots of impedance and SPL?
Do you have a detailed drawing of what was tested as far as enclosure geometry?
Do you have the raw data? Can you share what you have?
Without all the information about these tests it is very difficult to form an opinion on the accuracy or applicability of the results. If the test set-up was the same as was used in Bullocks' AES TL paper, then I think the geometry of the TL masked the results, he was seeing the results of fiber and geometry mixed together.
I think one mistake TL poeple have made for years is to not accounting for taper when looking at the predicted and mesurured fundamental resonant frequency of a TL.
When I measured my polyester filled test line, I also measured long fiber wool. The results were virtually identical. The other thing I have observed over the years is that measuring SPL below about 40 Hz can be very problematic. It is very difficult to get good low frequency SPL results.
Do your measurement results include with and without stuffing plots of impedance and SPL?
Do you have a detailed drawing of what was tested as far as enclosure geometry?
Do you have the raw data? Can you share what you have?
Without all the information about these tests it is very difficult to form an opinion on the accuracy or applicability of the results. If the test set-up was the same as was used in Bullocks' AES TL paper, then I think the geometry of the TL masked the results, he was seeing the results of fiber and geometry mixed together.
I think one mistake TL poeple have made for years is to not accounting for taper when looking at the predicted and mesurured fundamental resonant frequency of a TL.
When I measured my polyester filled test line, I also measured long fiber wool. The results were virtually identical. The other thing I have observed over the years is that measuring SPL below about 40 Hz can be very problematic. It is very difficult to get good low frequency SPL results.
Martin
-that matches my own experiences; as I mentioned to Dave in one of my previous posts, I haven't found any difference (at least not one that I can hear or measure in SPL terms at any rate) between stuffing types. I think it's easy to get hung up on that subject: it's only there to damp out higher harmonics as far as I'm concerned. That said, it would be an interesting study perhaps, in academic terms, proviing it's done properly.
I assume it's difficult to measure bass accurately due to the fact that it's omni-directional, and that it tends to be the room that dominates below 40Hz, rather than the speaker? By the way -do you have any thoughts or views regarding lining, rather than stuffing the cabinet? (oh boy, I've started at it now!)
All the best
Scott
-that matches my own experiences; as I mentioned to Dave in one of my previous posts, I haven't found any difference (at least not one that I can hear or measure in SPL terms at any rate) between stuffing types. I think it's easy to get hung up on that subject: it's only there to damp out higher harmonics as far as I'm concerned. That said, it would be an interesting study perhaps, in academic terms, proviing it's done properly.
I assume it's difficult to measure bass accurately due to the fact that it's omni-directional, and that it tends to be the room that dominates below 40Hz, rather than the speaker? By the way -do you have any thoughts or views regarding lining, rather than stuffing the cabinet? (oh boy, I've started at it now!)
All the best
Scott
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.