Re: Re: No side panels ?
If it were that simple, would SL have used a Linkwitz Transform (LT) circuit and not bothered the 6 dB/oct. shelving lowpass?
Actually, I seem to remember the first version of the W-baffle woofer had a driver with Q around 0.5, but the driver went out of production. He mentions somewhere that when a lower Q driver is used, the premature rolloff must be accounted for through a LT circuit.. I believe in the XLS version of the W-woofer he does this by having the shelving filter extend lower rather than bothering with a LT circuit.
Last not least, the 6 dB/oct rolloff of both H and W-baffle woofers started resonably close to the theoretical value.
I don't get it. What are the isseues we should be concerned with then?
Regards,
Eric
Kuei Yang Wang said:
If it were that simple, would SL have used a Linkwitz Transform (LT) circuit and not bothered the 6 dB/oct. shelving lowpass?
Actually, I seem to remember the first version of the W-baffle woofer had a driver with Q around 0.5, but the driver went out of production. He mentions somewhere that when a lower Q driver is used, the premature rolloff must be accounted for through a LT circuit.. I believe in the XLS version of the W-woofer he does this by having the shelving filter extend lower rather than bothering with a LT circuit.
Last not least, the 6 dB/oct rolloff of both H and W-baffle woofers started resonably close to the theoretical value.
Kuei Yang Wang said:
I don't get it. What are the isseues we should be concerned with then?
Regards,
Eric
Konnichiwa,
Hmmm. WELL. This would imply first of all a woofer with an Fs of << 28Hz and a Qt of >>0.7. Pretty rare animal. The problem is that you cannot really divirce baffle and driver.
Unless your driver still has full output at a given frequency, the dipole will be down in output even with a truely infinite baffle.
Modern Drivers are invariably made and designed for use in what I'd call "acoustically assisted" enclosurers (Horn, TL, Reflex, Sealed) where the resonance of the enclosure pushes up the LF output.
A dipole is what I'd call a "acoustically disassisted" system, so below a given point (determined by the size in a more complex relationship than Olsons formulas suggest) the system Q is lower than the Qt of the driver and hence LF output is reduced.
Now if you take (for arguments sake) a Driver with a Qt of 0.4 (very commonly found in Reflex box drivers) and 28Hz Fs on a Baffle as described you will likely have a System Q of < 0.2 at the 28Hz resonance.
The output of the driver itself will be 7db down at Fs compared to midband output, the baffle will be 6 or 9db down at the same frequency (need to check my reference to work out if 6 or 9db).
So, compared to the midband the output at 28Hz will be down 13 - 16db, HOWEVER the contribution of the baffle to this rolloff will only amount to 6..9db. The rest is the driver. This kind of rolloff in a normal (pressure transducer) would easily be offset by roomgain, not for the dipole.
Now lets take a theoretical driver having Fs = 28Hz and Qt = 1.5. Here the driver in a truely infinite baffle will have a lift almost 3db at 28Hz. The baffle loss will be still 6-9db @ 28Hz but the system will show -3 to -6db @ 28Hz, equivalent BTW to 0.707 or 0.5 System Q.
So, I hope this makes the relations a little clearer.
To come back to our baffle and driver from the original example, the driver would be around Qt = 0.7 and Fs = 55Hz. The driver in an infinite baffle would be around -4db @50Hz and -1db @ 100Hz. According to both Backman and the referenced German study the Baffle should show around -3db @ 50Hz and +2db @ 100Hz.
For net this would suggest 50Hz @ -7db and a slight lift at 100Hz. Observed was closer to -3db @ 50Hz in room, suggesting some LF room gain was present or the drivers Qt was higher than stated (probably the latter).
So, mathematically, theoretically (wave theory) and empirically we get similar data. I would not quite get as far as claiming that this is full proof, but the indications are strong.
I have just taken delivery of a pair of Supravox 215 Signatue Bicone to be build into a "Glass Open Baffle" (actually Acrylic for the Baffle) like below and will make measurements then.
The Driver is specified as Fs 50Hz Qt 0.8 (Vas etc can be ignored for dipole use) which should for the driver on its own give around -2db @ 50Hz. Quite nicely build BTW the driver.
The Baffle should contribute around -3db so I will be expecting 50Hz/-5db and 100Hz +2.5db with a slight improvement in the Lf through the baffles coupled air load and slight room gain. Let's see how close the theory matches the practice some time next week or two, when the Metalworker delivers my "stand" and RS delivers my Acrylic sheets.
Sayonara
(PS, in case you don't note, based on theory and empirical research I'm happy to drop a significant shekel on the project simply to have new, stylish looking speakers before I get married, so I can keep them...
)
sfdoddsy said:
Hmmm. WELL. This would imply first of all a woofer with an Fs of << 28Hz and a Qt of >>0.7. Pretty rare animal. The problem is that you cannot really divirce baffle and driver.
Unless your driver still has full output at a given frequency, the dipole will be down in output even with a truely infinite baffle.
Modern Drivers are invariably made and designed for use in what I'd call "acoustically assisted" enclosurers (Horn, TL, Reflex, Sealed) where the resonance of the enclosure pushes up the LF output.
A dipole is what I'd call a "acoustically disassisted" system, so below a given point (determined by the size in a more complex relationship than Olsons formulas suggest) the system Q is lower than the Qt of the driver and hence LF output is reduced.
Now if you take (for arguments sake) a Driver with a Qt of 0.4 (very commonly found in Reflex box drivers) and 28Hz Fs on a Baffle as described you will likely have a System Q of < 0.2 at the 28Hz resonance.
The output of the driver itself will be 7db down at Fs compared to midband output, the baffle will be 6 or 9db down at the same frequency (need to check my reference to work out if 6 or 9db).
So, compared to the midband the output at 28Hz will be down 13 - 16db, HOWEVER the contribution of the baffle to this rolloff will only amount to 6..9db. The rest is the driver. This kind of rolloff in a normal (pressure transducer) would easily be offset by roomgain, not for the dipole.
Now lets take a theoretical driver having Fs = 28Hz and Qt = 1.5. Here the driver in a truely infinite baffle will have a lift almost 3db at 28Hz. The baffle loss will be still 6-9db @ 28Hz but the system will show -3 to -6db @ 28Hz, equivalent BTW to 0.707 or 0.5 System Q.
So, I hope this makes the relations a little clearer.
To come back to our baffle and driver from the original example, the driver would be around Qt = 0.7 and Fs = 55Hz. The driver in an infinite baffle would be around -4db @50Hz and -1db @ 100Hz. According to both Backman and the referenced German study the Baffle should show around -3db @ 50Hz and +2db @ 100Hz.
For net this would suggest 50Hz @ -7db and a slight lift at 100Hz. Observed was closer to -3db @ 50Hz in room, suggesting some LF room gain was present or the drivers Qt was higher than stated (probably the latter).
So, mathematically, theoretically (wave theory) and empirically we get similar data. I would not quite get as far as claiming that this is full proof, but the indications are strong.
I have just taken delivery of a pair of Supravox 215 Signatue Bicone to be build into a "Glass Open Baffle" (actually Acrylic for the Baffle) like below and will make measurements then.
An externally hosted image should be here but it was not working when we last tested it.
The Driver is specified as Fs 50Hz Qt 0.8 (Vas etc can be ignored for dipole use) which should for the driver on its own give around -2db @ 50Hz. Quite nicely build BTW the driver.
The Baffle should contribute around -3db so I will be expecting 50Hz/-5db and 100Hz +2.5db with a slight improvement in the Lf through the baffles coupled air load and slight room gain. Let's see how close the theory matches the practice some time next week or two, when the Metalworker delivers my "stand" and RS delivers my Acrylic sheets.
Sayonara
(PS, in case you don't note, based on theory and empirical research I'm happy to drop a significant shekel on the project simply to have new, stylish looking speakers before I get married, so I can keep them...
)
Re: german study
@Kuei Yang Wang
For the -6db point Linkwitz gives the equation
f=0.17*V/D
where V=345m/S and D=W/2.
Your equation looks suspiciously similar, except for 0.07 <> 0.17 (which makes for a huge difference). Could you lead us to the recent german study, you were talking of? I would like to test its credibility.
Sayonara
@Kuei Yang Wang
For the -6db point Linkwitz gives the equation
f=0.17*V/D
where V=345m/S and D=W/2.
Your equation looks suspiciously similar, except for 0.07 <> 0.17 (which makes for a huge difference). Could you lead us to the recent german study, you were talking of? I would like to test its credibility.
Sayonara
When talking about the implications of different values of Q, I always find this diagram very helpful. It shows graphically, what Kuei Yang Wang talked about in his last reply. I never saw it published on the net – so here we go.
It shows SPL against w/w1, where w/w1 is frequency divided by resonance frequency fs. x=1,00 means fs, x=2,00 is double fs and so on ... The author says, the diagram is valid for infinite baffle as well as for a closed box.
At fs (x=1) a loudspeaker with Q=0,7 may be -4 dB down compared to its midrange SPL. A loudspeaker with Q=0,2 will be -14 dB down at its fs.
If we take fs=50 Hz and look at the SPL at 200 Hz (x=4), the loudspeaker with Q=0,2 will still be down -4 dB compared to -1 dB of the other LS (Q=0,7).
In my view this diagram is a very strong statement in favor of high Q drivers for dipole implementations.
It shows SPL against w/w1, where w/w1 is frequency divided by resonance frequency fs. x=1,00 means fs, x=2,00 is double fs and so on ... The author says, the diagram is valid for infinite baffle as well as for a closed box.
At fs (x=1) a loudspeaker with Q=0,7 may be -4 dB down compared to its midrange SPL. A loudspeaker with Q=0,2 will be -14 dB down at its fs.
If we take fs=50 Hz and look at the SPL at 200 Hz (x=4), the loudspeaker with Q=0,2 will still be down -4 dB compared to -1 dB of the other LS (Q=0,7).
In my view this diagram is a very strong statement in favor of high Q drivers for dipole implementations.
Attachments
Re: Re: german study
Konnichiwa,
It came out of thge NXT "soundfield" and evaluated primarily the NXT type "diffuse field dipole" against conventional cone dipoles, but it had loads of data on cone dipoles for comparison. IIRC it was an AES preprint, or if not just some (well researched) marketing document.
Backmans stuff is AES preprint anyway and comes to more or less the same figures.
And yes, Siegfrieds numbers and reality leave some gap.
Sayonara
Konnichiwa,
Rudolf said:
It came out of thge NXT "soundfield" and evaluated primarily the NXT type "diffuse field dipole" against conventional cone dipoles, but it had loads of data on cone dipoles for comparison. IIRC it was an AES preprint, or if not just some (well researched) marketing document.
Backmans stuff is AES preprint anyway and comes to more or less the same figures.
And yes, Siegfrieds numbers and reality leave some gap.
Sayonara
Rudolf,
you may find some of the papers at www.wvier.de, but not all of these are making sense to me.
Regards,
Eric
you may find some of the papers at www.wvier.de, but not all of these are making sense to me.
Regards,
Eric
Re: Re: No side panels ?
I disagree. I have read on several sites that the size of the "wings" do make a difference because they too become part of the baffle. The further the bass has to travel around the other side of the baffle to cancle out the back wave, the lower the bass is going to be before that happenes.
My dipoles for instance: 24"W x 24"D x 48"H. If you were to take the "wings which extend 12" in each direction, the baffle would actually be 48"W x 96"H.
That's why I use vertually no equalization and have a flat responce down to 20Hz, if not lower.
Now if I made the baffle and "wings" smaller, I would be lossing bass probable around 40-50Hz and needing a lot of equalization to overcome the small baffle size.
And my dipoles definantly do not need any help in the low bass department!
Just go through my thread "My first Dipole Subwoofer!!!". That is where I learned everything I know about dipoles. Before that, I knew vertually nothing about them. And my first attempt was and still is a BIG HIT!!!
Kuei Yang Wang said:
I disagree. I have read on several sites that the size of the "wings" do make a difference because they too become part of the baffle. The further the bass has to travel around the other side of the baffle to cancle out the back wave, the lower the bass is going to be before that happenes.
My dipoles for instance: 24"W x 24"D x 48"H. If you were to take the "wings which extend 12" in each direction, the baffle would actually be 48"W x 96"H.
That's why I use vertually no equalization and have a flat responce down to 20Hz, if not lower.
Now if I made the baffle and "wings" smaller, I would be lossing bass probable around 40-50Hz and needing a lot of equalization to overcome the small baffle size.
And my dipoles definantly do not need any help in the low bass department!
Just go through my thread "My first Dipole Subwoofer!!!". That is where I learned everything I know about dipoles. Before that, I knew vertually nothing about them. And my first attempt was and still is a BIG HIT!!!
Re: Re: Re: No side panels ?
Konnichiwa,
So, did you actually make a comparison yourself? Just curious....
Taking your use of gramar it would seem you have not actually made the comparison?
Sayonara
PS, I am not saying wings make no difference, but adding wings which summed are equal to the baffle width will not extend the cutoff frequency by the factor 2, try it.
Konnichiwa,
chops said:
So, did you actually make a comparison yourself? Just curious....
chops said:
Taking your use of gramar it would seem you have not actually made the comparison?
Sayonara
PS, I am not saying wings make no difference, but adding wings which summed are equal to the baffle width will not extend the cutoff frequency by the factor 2, try it.
My "use of gramar" ????!!!! I did not know we were in English class here. 
I do not need to make a comparison because I already have the proof in my listening room, not to mention that Mr. Linkwitz agreed with my dimensions through e-mail.
If flattened out, my baffle area is 4 feet by 8 feet for two 15" drivers per side, with a driver Fs of 26Hz!
I went with what I was told by Mr. Linkwitz, and I will take his word way before I would take yours!
My first Dipole Subwoofer!!!
I do not need to make a comparison because I already have the proof in my listening room, not to mention that Mr. Linkwitz agreed with my dimensions through e-mail.
If flattened out, my baffle area is 4 feet by 8 feet for two 15" drivers per side, with a driver Fs of 26Hz!
I went with what I was told by Mr. Linkwitz, and I will take his word way before I would take yours!
My first Dipole Subwoofer!!!
Konnichiwa,
I did find the article I was refering to, at wvier.de:
Theoretical and Practical Aspects for the design of Dipole Speakers
The Article is in German BTW, but I'll abstract the key issues.... First however a little graph, namely the attached one.
What this graph shows is the relation between baffle width compared to the wavelength NORMALISED with curves for baffles with a "width" equal to 2.2, 4.4 and 8.8 times the width of the active radiating area. Note, these are PLAN baffles.
Now the Wavelegth (Lambda) is (IIRC) 345/F. Or expressed differently, 345m = Wavelength * Frequency.
If we go back to our example Golden Ratio 65cm X 105cm Baffle and a radiating diameter of around 18cm we find we have average baffle width (which is how this MUST be calculate) of (65 + 105)/2 cm or 0.85m.
From this we know that Lambda & average baffle width coincide at 405Hz, for simplicity we use 400Hz. So, on the graph we can label "1" as 400Hz and 0.1 as 40Hz. Our ratio between bafflewidth and radiating surface with is around 4.8, so we will use the D=4.4d graph.
From this we first find that for an absolutely flat response to DC driver our -9db point is around 40Hz, that output is around 0db at around 100Hz and that around +6db are reached at around 240Hz. At 400Hz a small dip is observable with a small peak at around 600Hz above which the response flattens out.
In reality the golden ratio spacing of the driver on the baffle will reduce the ripple effect and the peak at 240Hz will be reduced, as well as the degree of rolloff at 40Hz (slightly so).
We must remember now two things.
1) if the baffle is placed directly on the floor it's size becomes virtually doubled, at low frequencies, reducing the in room rolloff at LF further.
2) This rolloff curve must be overlaid and added to the frequency response exhibited at LF by the driver alone with "free air" loading, meaning a near infinite size sealed box.
Using a suitable driver we can get pretty good results, I'll post a graph comparing a few Drivers in the next post.
Using the Supravox 215 Signature Bicone the resulting system ANECHONICALLY suggests a frequency response pattern that will be 10db down (Driver & Baffle combined) at around 50Hz and a peak in the 300Hz region.
If we now drive the Supravox from a Zero Feedback Triode Amplifier (SE or PP) with around 3 Ohm output impedance the frequency response becomes flatter reaching "reference" level (95db/2.83V) at 70Hz and -10db @ 50Hz.
Taking into account the room effects (especially floor standing placement) we should have an LF rolloff reduced by around 4 - 6db (guestimate).
Certainly the above laid out notes are concurrent with observations made on plan baffles. In comparison, folded baffles failed to materially lower the -3db & -10db points, for whatever specifical reasons that may be so.
Sayonara
sfdoddsy said:
I did find the article I was refering to, at wvier.de:
Theoretical and Practical Aspects for the design of Dipole Speakers
The Article is in German BTW, but I'll abstract the key issues.... First however a little graph, namely the attached one.
What this graph shows is the relation between baffle width compared to the wavelength NORMALISED with curves for baffles with a "width" equal to 2.2, 4.4 and 8.8 times the width of the active radiating area. Note, these are PLAN baffles.
Now the Wavelegth (Lambda) is (IIRC) 345/F. Or expressed differently, 345m = Wavelength * Frequency.
If we go back to our example Golden Ratio 65cm X 105cm Baffle and a radiating diameter of around 18cm we find we have average baffle width (which is how this MUST be calculate) of (65 + 105)/2 cm or 0.85m.
From this we know that Lambda & average baffle width coincide at 405Hz, for simplicity we use 400Hz. So, on the graph we can label "1" as 400Hz and 0.1 as 40Hz. Our ratio between bafflewidth and radiating surface with is around 4.8, so we will use the D=4.4d graph.
From this we first find that for an absolutely flat response to DC driver our -9db point is around 40Hz, that output is around 0db at around 100Hz and that around +6db are reached at around 240Hz. At 400Hz a small dip is observable with a small peak at around 600Hz above which the response flattens out.
In reality the golden ratio spacing of the driver on the baffle will reduce the ripple effect and the peak at 240Hz will be reduced, as well as the degree of rolloff at 40Hz (slightly so).
We must remember now two things.
1) if the baffle is placed directly on the floor it's size becomes virtually doubled, at low frequencies, reducing the in room rolloff at LF further.
2) This rolloff curve must be overlaid and added to the frequency response exhibited at LF by the driver alone with "free air" loading, meaning a near infinite size sealed box.
Using a suitable driver we can get pretty good results, I'll post a graph comparing a few Drivers in the next post.
Using the Supravox 215 Signature Bicone the resulting system ANECHONICALLY suggests a frequency response pattern that will be 10db down (Driver & Baffle combined) at around 50Hz and a peak in the 300Hz region.
If we now drive the Supravox from a Zero Feedback Triode Amplifier (SE or PP) with around 3 Ohm output impedance the frequency response becomes flatter reaching "reference" level (95db/2.83V) at 70Hz and -10db @ 50Hz.
Taking into account the room effects (especially floor standing placement) we should have an LF rolloff reduced by around 4 - 6db (guestimate).
Certainly the above laid out notes are concurrent with observations made on plan baffles. In comparison, folded baffles failed to materially lower the -3db & -10db points, for whatever specifical reasons that may be so.
Sayonara
Attachments
I think I'd still take SLs data and advice, especially as they matched my own measurements when I was deciding on baffle width and shape for my speakers.
Then again, I have no problems using active EQ since I don't believe in using flea powered amps.
BTW, whether good, bad or indifferent, 'grammar' usually has two 'm's.
Steve
Then again, I have no problems using active EQ since I don't believe in using flea powered amps.
BTW, whether good, bad or indifferent, 'grammar' usually has two 'm's.
Steve
corner placement
The corner placement will radiate the same power into half the space both from the front and the back but path length would be the same ( one side is just chopped off !).So I suspect that it might not improve the LF extension but will probably improve the SPL.
I was wondering about the extended wings. I couldn't see how it will not work . But I do see that having it rounded or stepped will not increase the path length as the wavelengths are comparatively so long. The path length will probably be the direct line from the driver to the end point of the wing. So , for example, a 15 inch wide wing (at 90 degrees to the front baffle) on either side may not produce much improvement in 'effective path length' as compared to a flat front baffle 30 inches wider. But I would expect 'some' improvement! The effect of the wing keeps getting reduced as the front panel gets wider. We are talking of reasonable lengths of course ( in several inches not feet).
So the statement ' wings do not help' probably means that the improvement is negligible in most cases. It's probably better to make them just wider -- but then what about the 'comb' effect?
Back to compromises again I guess!
In addition -- what was the golden rule for placement of the driver on the baffle ? My memory is getting rusty.
Cheers.
The corner placement will radiate the same power into half the space both from the front and the back but path length would be the same ( one side is just chopped off !).So I suspect that it might not improve the LF extension but will probably improve the SPL.
I was wondering about the extended wings. I couldn't see how it will not work . But I do see that having it rounded or stepped will not increase the path length as the wavelengths are comparatively so long. The path length will probably be the direct line from the driver to the end point of the wing. So , for example, a 15 inch wide wing (at 90 degrees to the front baffle) on either side may not produce much improvement in 'effective path length' as compared to a flat front baffle 30 inches wider. But I would expect 'some' improvement! The effect of the wing keeps getting reduced as the front panel gets wider. We are talking of reasonable lengths of course ( in several inches not feet).
So the statement ' wings do not help' probably means that the improvement is negligible in most cases. It's probably better to make them just wider -- but then what about the 'comb' effect?
Back to compromises again I guess!
In addition -- what was the golden rule for placement of the driver on the baffle ? My memory is getting rusty.
Cheers.
wow...you guys are talking great big baffles. to reduce baffle width and size of dipole i was considering using one of these 2 ideas...
either making a dipole using 2 drivers configured like shown in chops's post but rearranging the drivers to face down or up.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=14179&highlight=dipole
or making a dipole like shown in Phear's post but having the drivers facing sideways.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=16867&highlight=dipole
in this case the depth of Phear's box is my baffle width.
either making a dipole using 2 drivers configured like shown in chops's post but rearranging the drivers to face down or up.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=14179&highlight=dipole
or making a dipole like shown in Phear's post but having the drivers facing sideways.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=16867&highlight=dipole
in this case the depth of Phear's box is my baffle width.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.