This is especially for Mr. MJK and Mr. GM. I have simulated this enclosure using Mr. King's sheets and the response seems pretty flat with good bass extension and little stuffing. My question is: is the setup OK, because from what I gather, with TLs the length of the enclosure must be much bigger that its width, otherwise the simulations aren't accurate.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I have not rerun the simulation so I am making an educated (maybe not so educated. I am actually winging it) guess about your enclosure.
The length is 30 inches. Assuming a square cross-section, at the top it is approximalety 9 inches wide and deep. At the bottom it is 9 inches wide and just over 2 inches deep. All of these are internal dimensions. So the length is over 3 times longer then any other dimension, the first couple of resonances should be along the modeled direction.
But this is a rather short enclosure and it may actually be acting more like a BR then a ML TL. I say this because of the very small amount of fiber density required to damp the waves and the lack of significant peaks and nulls in the summed response. If you modeled the enclosure as rectangular of the same volume, does the response change much?
I don't see anything wrong with what you have modeled, the MathCad TL worksheets will also model a BR accurately. Looks OK to me.
Hope that helps,
The length is 30 inches. Assuming a square cross-section, at the top it is approximalety 9 inches wide and deep. At the bottom it is 9 inches wide and just over 2 inches deep. All of these are internal dimensions. So the length is over 3 times longer then any other dimension, the first couple of resonances should be along the modeled direction.
But this is a rather short enclosure and it may actually be acting more like a BR then a ML TL. I say this because of the very small amount of fiber density required to damp the waves and the lack of significant peaks and nulls in the summed response. If you modeled the enclosure as rectangular of the same volume, does the response change much?
I don't see anything wrong with what you have modeled, the MathCad TL worksheets will also model a BR accurately. Looks OK to me.
Hope that helps,
I believe the used parameters are wrong. Certainly Sd is. It is listed as 221.67cm^2 in my focal catalog. Fs is 27.4, Vas 106l, Qts .30, Qms 2.9, Qes .33, Bl 9.4, Re 7.8, Le .9.
It's possible that the factory specs are wrong of course, but I've had good correlation between measured and listed specs in the past with all focals I've used.
I have a focal with very similar specs to those listed, and it works very well in a tower BR/ML-TL. My first sim attempt would be a straight pipe about 39 inches tall, with the driver at 1/3, and a port at the bottom. The Q is a little too low for a 'traditional' 2m TL IMO.
GB
It's possible that the factory specs are wrong of course, but I've had good correlation between measured and listed specs in the past with all focals I've used.
I have a focal with very similar specs to those listed, and it works very well in a tower BR/ML-TL. My first sim attempt would be a straight pipe about 39 inches tall, with the driver at 1/3, and a port at the bottom. The Q is a little too low for a 'traditional' 2m TL IMO.
GB
Greg,
The Focal label is a leftover from the sample problem run in the downloaded worksheet. People don't seem to change the title, only the T/S parameters, when they run a simulation for their drivers. By not updating the title it stops others from double checking the driver's specs they are using, somebody might actually have measured T/S results for the particular driver being sim'ed.
The Focal label is a leftover from the sample problem run in the downloaded worksheet. People don't seem to change the title, only the T/S parameters, when they run a simulation for their drivers. By not updating the title it stops others from double checking the driver's specs they are using, somebody might actually have measured T/S results for the particular driver being sim'ed.
>My question is: is the setup OK, because from what I gather, with TLs the length of the enclosure must be much bigger that its width, otherwise the simulations aren't accurate.
====
Looks fine to me also and should work well in a small room or if placed near a wall. If it is a larger room or they will be well away from the wall, then I recommend a straight pipe with a lower Fb. FWIW, here is my version of a T/S max flat:
L = 39.5"
zd = 14.63"
zp = 37"
So/SL = 63"^2
density = 0.2
rp = 1.25"
Lp = 4.125"
The vent looks a little small though and may 'chuff' if the driver requires much power around/at Fb. I recommend at least a 2.5" diameter vent.
Also, the BL value is off quite a bit. When I do a T/S max flat alignment it produces a dip in the response, so change it to 4.1757 to see how it affects your sim.
BTW, which driver is this, so I can label my sim?
====
>But this is a rather short enclosure and it may actually be acting more like a BR then a ML TL. I say this because of the very small amount of fiber density required to damp the waves and the lack of significant peaks and nulls in the summed response.
====
It's long enough, the 4:1 reverse taper combined with a vent really loads both the driver and vent.
====
>neatness and accuracy are everything and pay attention to the details!
====
You bet! When I was learning how to build race engines, my mentor had two golden rules: Cleanliness is next to Godliness, and God is in the details.
GM
====
Looks fine to me also and should work well in a small room or if placed near a wall. If it is a larger room or they will be well away from the wall, then I recommend a straight pipe with a lower Fb. FWIW, here is my version of a T/S max flat:
L = 39.5"
zd = 14.63"
zp = 37"
So/SL = 63"^2
density = 0.2
rp = 1.25"
Lp = 4.125"
The vent looks a little small though and may 'chuff' if the driver requires much power around/at Fb. I recommend at least a 2.5" diameter vent.
Also, the BL value is off quite a bit. When I do a T/S max flat alignment it produces a dip in the response, so change it to 4.1757 to see how it affects your sim.
BTW, which driver is this, so I can label my sim?
====
>But this is a rather short enclosure and it may actually be acting more like a BR then a ML TL. I say this because of the very small amount of fiber density required to damp the waves and the lack of significant peaks and nulls in the summed response.
====
It's long enough, the 4:1 reverse taper combined with a vent really loads both the driver and vent.
====
>neatness and accuracy are everything and pay attention to the details!
====
You bet! When I was learning how to build race engines, my mentor had two golden rules: Cleanliness is next to Godliness, and God is in the details.
GM
Greg B said:
This is all my fault, I discovered after returning home (I post from work), that I forgot to change the name of the driver. It is a Visaton W 170 S / 4 Ohm. Sorry for leading the thread in that direction, I am the idiot. And, those are the factory specs, not the meausred ones (I don't owm that driver now), so this is more of an estimation.
Thanks for taking the time to answer my question. I will try your suggestions, they will probably lead to better results.
I already answered this...
But, aren't the principles used for the both sheets the same? I asked my question about the length because someone told me that the sheets assume the length of the enclosure is much bigger than its width, because otherwise the transversal waves become comparable in strength with the longitudinal ones, and the sheets only account for the longitudinal ones. Actually, the guy who told me that learned it from a failure with a square box, and Mr. King told him that the source of the problem is the shape of the box. I have to admit I'm pretty confused on this one...
OK, I deserve that. To be frank, I thought that my mistake may lead to confusion, and I appologize again. That's the effect of saturation, I have spent to many hours looking at MathCAD
The worksheets were written assuming that one dimension was so long that it dominated the low frequency resposne enough to ignore the other two dimensions. Between 10 Hz and 100 Hz this is easily accomplished and the T/S modeling is accurate.
But then I plotted the responses out to 1000 Hz so that the harmonics of the fundamental could be seen and placement of the driver to suppress the harmonics could be studied. At some point above 100 Hz the model startes to become less accurate for a number of reasons one being standing waves in the remaining two dimensions of the enclosure. Hopefully stuffing will attenuate these out of plane waves and the plotted results will be reasonable.
As an estimate of the limits over which the worksheet is accurate you might consider the following calculations.
1) The worksheet will determine the tuning frequency in the direction modelled and it is easily seen in the acoustic impedance plot when no stuffing is present as the first large peak.
2) The first driver to back wall standing wave's frequency can be estimated by the following calculation
f = c / (4 x d) where d is the depth of the air in the cabinet
3) The first side to side standing wave'e frequency can be estimated by the following calculation
f = c / (2 x w) where w is the width of the air in the cabinet
These are rough estimates of the out of plane resonant frequencies. The calculations assume a rectangular enclosure. If the difference between the transmission line tuning frequency and the out of plane frequencies is large, then the model is valid for the frequency range below the lower out of plane frequency. For a more exact estimate of the out of plane frequencies for any geometry I would recommend a FE solution, I personally use ANSYS to verify the applicability of my MathCad worksheets to a particular geometry. If the frequencies are all reasonably close, like in a square box, I would be careful using the results and would expect some surprises in the finished design.
How far apart do the frequencies have to be? I have no idea of the exact number, but I would probably look for a factor of at least four.
I am not sure I completely understand the question. The bottom line for any design is the plot you reference. The other plots help you understand why you get the final SPL trace. Can you only look at one plot? Sure, but the understanding might not be as complete. I think the goal is a flat response so I am not sure what you are asking in the second question. But remember, all of the plots assume that the sound is produced by the speaker in an infinite baffle and in the real world a baffle step response should be superimposed on the plots. I recommend after sketching the final enclosure design, that you consult "The Edge" to get a more complete representatin of the real SPL response.
Hope that helps,
>But, aren't the principles used for the both sheets the same? I asked my question about the length because someone told me that the sheets assume the length of the enclosure is much bigger than its width, because otherwise the transversal waves become comparable in strength with the longitudinal ones, and the sheets only account for the longitudinal ones.
>Actually, the guy who told me that learned it from a failure with a square box, and Mr. King told him that the source of the problem is the shape of the box. I have to admit I'm pretty confused on this one...
====
Golden or acoustic ratios average out any standing waves in the cab and below the lowest mode there is a state of ~uniform particle density (pressure). All standard box programs assumes this is the case. Once one dimension gets long enough to upset this balance, it transitions to a resonant pipe of increasing amplitude with increasing length.
Normally my goal WRT ML-TL designs is to find the right length/Vb that maximizes extension without causing peaking at Fb, a driver position that minimizes the need for excessive stuffing, and use other means to get the driver(s) at whatever height is required for the app.
It appears though that most folks pick a height/driver position, then use Vb/stuffing to 'force' an acceptable FR. From a purely technical POV the performance theoretically suffers, but folks rave about how good they sound. Not having done any comparisons between mine and others, I do not know if there is any audible difference, especially since the room dominates down low, so as always, YMMV.
Still, per "God is in the details", I will continue to suggest what I consider to be optimum based on the available info because it is all I have to work with if a height limitation and/or driver position is not specified.
====
>What if one completely ignores all the graphs except the SPL response? More exactly, what is the worst case when obtaining a perfectly flat and extended SPL response and ignoring all the others? And how audible is the difference?
====
A max flat response can have too much gain down low when combined with the room's gain, requiring either room treatment, a different speaker placement, rolling off the speaker's response, or a combination of them all to get a ~flat in-room response. IOW it can make it sound 'boomy', 'bloated', and/or 'slow' depending on how excessive it is.
A max flat does not have the best transient response (TR) either, so those of us who are more sensitive to LF 'attack/decay' require a more damped response (rolled off). Unless you know for sure what alignment sounds best in your room then it is better IMO to have too much acoustic gain and reduce it as required than to not have it and have to use some form of EQ to compensate, which increases distortion.
GM
>Actually, the guy who told me that learned it from a failure with a square box, and Mr. King told him that the source of the problem is the shape of the box. I have to admit I'm pretty confused on this one...
====
Golden or acoustic ratios average out any standing waves in the cab and below the lowest mode there is a state of ~uniform particle density (pressure). All standard box programs assumes this is the case. Once one dimension gets long enough to upset this balance, it transitions to a resonant pipe of increasing amplitude with increasing length.
Normally my goal WRT ML-TL designs is to find the right length/Vb that maximizes extension without causing peaking at Fb, a driver position that minimizes the need for excessive stuffing, and use other means to get the driver(s) at whatever height is required for the app.
It appears though that most folks pick a height/driver position, then use Vb/stuffing to 'force' an acceptable FR. From a purely technical POV the performance theoretically suffers, but folks rave about how good they sound. Not having done any comparisons between mine and others, I do not know if there is any audible difference, especially since the room dominates down low, so as always, YMMV.
Still, per "God is in the details", I will continue to suggest what I consider to be optimum based on the available info because it is all I have to work with if a height limitation and/or driver position is not specified.
====
>What if one completely ignores all the graphs except the SPL response? More exactly, what is the worst case when obtaining a perfectly flat and extended SPL response and ignoring all the others? And how audible is the difference?
====
A max flat response can have too much gain down low when combined with the room's gain, requiring either room treatment, a different speaker placement, rolling off the speaker's response, or a combination of them all to get a ~flat in-room response. IOW it can make it sound 'boomy', 'bloated', and/or 'slow' depending on how excessive it is.
A max flat does not have the best transient response (TR) either, so those of us who are more sensitive to LF 'attack/decay' require a more damped response (rolled off). Unless you know for sure what alignment sounds best in your room then it is better IMO to have too much acoustic gain and reduce it as required than to not have it and have to use some form of EQ to compensate, which increases distortion.
GM
I have to admit I'm an ignorant, because it's just now that I find the importance of the impulse response, which I have completely ignored until now. And now I find that MathCAD never plots it when I run the simulations! Not even with the unmodified examples provided by Mr. King! Do you have any idea why is that? I haven't changed any configuration in MathCAD or anything.
I also found that the impulse response was not displayed when I ran simulations. But it appeared when I printed out a result, also they are shown if you look at the page when you go to " print preview " ?? Another problem I sometimes get, is the " Far Field " TL/IB SPL responses occasionally do not change when I alter any of the line parameters, but if I save and reopen the file I find it has worked. Again ??? I have Mathcad 2001 and Mr King told me that this edition has caused some problems which were solved when he went back to a previous or later edition. Perhaps you should contact him
Roy
Roy
I have no idea why this is happening, I have not had this problem with MathCad. I have been using MathCad since version 4 and have not seen this with plots. My only suggestion would be to scroll up or down until the plot is no longer shown and then come back to it. One other trick that I have used is to press "control-R" to refresh the screen. Other then those two suggestions I have no idea how to resolve the problem.
Hope that helps,
Hope that helps,
mr_push_pull,
The worksheets that are downloaded from my site have not changed too much in the past couple of years. The changes that were made are minor corrections and adjustments thet do not relate to the IFFT section that calculates the impulse response. Roy indicated that he could see the plot in print preview or by printing the page, is this the case for you also? Many people are using these same worksheets and I have not heard of this problem before, does it happen in the original files that you downloaded?
At work I have some very expensive FE software and everytime I find a problem the vendor blames me, my PC, anything but the software. I have taken my PC completely apart and reloaded the software to "fix the problem" in my equipment only to find that it really was a bug in their software. So I am not trying to make light of your problem, but I am having trouble concluding it is the worksheet itself. Have you tried installing and running the worksheets on a different PC?
Hope that helps,
The worksheets that are downloaded from my site have not changed too much in the past couple of years. The changes that were made are minor corrections and adjustments thet do not relate to the IFFT section that calculates the impulse response. Roy indicated that he could see the plot in print preview or by printing the page, is this the case for you also? Many people are using these same worksheets and I have not heard of this problem before, does it happen in the original files that you downloaded?
At work I have some very expensive FE software and everytime I find a problem the vendor blames me, my PC, anything but the software. I have taken my PC completely apart and reloaded the software to "fix the problem" in my equipment only to find that it really was a bug in their software. So I am not trying to make light of your problem, but I am having trouble concluding it is the worksheet itself. Have you tried installing and running the worksheets on a different PC?
Hope that helps,
To Mr. King:
I haven't tried running the simulations on other computer, neither did I reinstall MathCAD, but I will try to do that today when I'll get home. Probably it is a MathCAD bug. But to change the subject....
To everyone (including Mr. King):
I am familiar for some time now with Mr King's work, and I'm thinking about building a TL as my first DIY project (well not 100% done by myself, my woodworking skills and knowledge are close to zero, and I don't have the equipment either). So, looking for a TL design with cheap drivers to see if I can benefit from someone else's experiments (I can afford to experiment on my budget only to a point) I came accross the CT 142 enclosure with Visaton drivers (designed by Klang&Ton in '93 I think) http://www.visaton.de/bilder/forum/ct142-1.jpg, http://www.visaton.de/bilder/forum/ct142-2.jpg.
I thought it's a nice and cheap TL, and the reviews seemed good. I also thought about simulating it usign Mr. King's sheets, and correlating the results with those from the Baffle Diffraction Simulator. This is when I truly realized how accurate are those two tools. The response I obtained by simulation has two humps, one at approximately 50Hz and one around 100 Hz, the second one being attenuated by the baffle diffaction. This is exacly what the measured response from Klang&Ton looks like! Now.... I friend of mine wants to build these speakers too, but I told him that the design could use some improvements (the BSC circuit is a must by me, and a much flatter response could be obtained using GM's suggestions from above for the enclosure). But the guy won't trust my judgement. Well, would you guru's of the DIY world be the judges and share your more educated opinions? I really think that the enclosure suggested by GM together with a BSC could be way better than the CT 142, those two humps don't look too good to me, no matter what the reviews may say.
I haven't tried running the simulations on other computer, neither did I reinstall MathCAD, but I will try to do that today when I'll get home. Probably it is a MathCAD bug. But to change the subject....
To everyone (including Mr. King):
I am familiar for some time now with Mr King's work, and I'm thinking about building a TL as my first DIY project (well not 100% done by myself, my woodworking skills and knowledge are close to zero, and I don't have the equipment either). So, looking for a TL design with cheap drivers to see if I can benefit from someone else's experiments (I can afford to experiment on my budget only to a point) I came accross the CT 142 enclosure with Visaton drivers (designed by Klang&Ton in '93 I think) http://www.visaton.de/bilder/forum/ct142-1.jpg, http://www.visaton.de/bilder/forum/ct142-2.jpg.
I thought it's a nice and cheap TL, and the reviews seemed good. I also thought about simulating it usign Mr. King's sheets, and correlating the results with those from the Baffle Diffraction Simulator. This is when I truly realized how accurate are those two tools. The response I obtained by simulation has two humps, one at approximately 50Hz and one around 100 Hz, the second one being attenuated by the baffle diffaction. This is exacly what the measured response from Klang&Ton looks like! Now.... I friend of mine wants to build these speakers too, but I told him that the design could use some improvements (the BSC circuit is a must by me, and a much flatter response could be obtained using GM's suggestions from above for the enclosure). But the guy won't trust my judgement. Well, would you guru's of the DIY world be the judges and share your more educated opinions? I really think that the enclosure suggested by GM together with a BSC could be way better than the CT 142, those two humps don't look too good to me, no matter what the reviews may say.
Roy Lewis said:
Thanks Roy, that solves the problem for me too, now I can look at the impulse response by printing the sheets as PDFs
It seems that the graph isn't displayed for a certain zoom only, even at print preview (I think that the problem occurs at 100% zoom). I think it is a MathCAD bug.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.