Ok, let's see if I remember...
The volume of the enclosure - that's a bit tricky. Do you know how to calculate the area of a D shape cross section? It is a 12" cylinder (11.25" internally) but with the 'front' flattened for a baffle approximately 8.5" wide. It's 62" long, but then subtract 3/4" for the top cap insert. So, if we were to assume it was just an 11" cylinder, that would come out to about 3.36 cubic feet or 95 liters. I figure subtract at least 0.3 cubic feet for the drivers, crossover, port, and extra wood behind the baffle, so let's call it 3 cubic feet or 85 liters. I suppose it could be a bit less, so if anyone knows how to calculate area of a D shape, we could get more exact.
In any case, this is the volume (85 liters) I used to tune the enclosure. I initially started with a 3" x 17" port (flared), which WinISD tells me was about 18Hz (I was still thinking the 850122 had an Fs of 37Hz). That sounded a bit too thin, so I cut the center section in half, giving me 11" of port length. This in theory is somewhere around 22Hz. I could probably go down to the minimum length of 5", and expect the sound might not change all that much because the Fb would still be way below Fs. F3 would go down a little and output above 30 Hz would be greater. But, that would push the group delay peak farther up into the audible range.
The drivers were played pretty hard for a couple weeks before I measured Fs, so they should have been broken in. The fact that they measured between 49-52Hz seems to point in that direction, as most notes I have seen put the new batch at around 55Hz Fs. I also noted that the Fs was a few Hz lower at higher signal levels (like 4-6V+).
Of course now, I don't use them full range, because I have my subwoofer, so all that bass is lost. Truth be told though, my new room doesn't have nearly as much gain as my old apartment did, which is bad for my ET's full range performance, but good for my giant subwoofer.
The volume of the enclosure - that's a bit tricky. Do you know how to calculate the area of a D shape cross section? It is a 12" cylinder (11.25" internally) but with the 'front' flattened for a baffle approximately 8.5" wide. It's 62" long, but then subtract 3/4" for the top cap insert. So, if we were to assume it was just an 11" cylinder, that would come out to about 3.36 cubic feet or 95 liters. I figure subtract at least 0.3 cubic feet for the drivers, crossover, port, and extra wood behind the baffle, so let's call it 3 cubic feet or 85 liters. I suppose it could be a bit less, so if anyone knows how to calculate area of a D shape, we could get more exact.
In any case, this is the volume (85 liters) I used to tune the enclosure. I initially started with a 3" x 17" port (flared), which WinISD tells me was about 18Hz (I was still thinking the 850122 had an Fs of 37Hz). That sounded a bit too thin, so I cut the center section in half, giving me 11" of port length. This in theory is somewhere around 22Hz. I could probably go down to the minimum length of 5", and expect the sound might not change all that much because the Fb would still be way below Fs. F3 would go down a little and output above 30 Hz would be greater. But, that would push the group delay peak farther up into the audible range.
The drivers were played pretty hard for a couple weeks before I measured Fs, so they should have been broken in. The fact that they measured between 49-52Hz seems to point in that direction, as most notes I have seen put the new batch at around 55Hz Fs. I also noted that the Fs was a few Hz lower at higher signal levels (like 4-6V+).
Of course now, I don't use them full range, because I have my subwoofer, so all that bass is lost. Truth be told though, my new room doesn't have nearly as much gain as my old apartment did, which is bad for my ET's full range performance, but good for my giant subwoofer.
Hi,
Don’t understand the relation Vb needs to be somewhere below 2x Vas. With most speakers I end up with a much smaller BR box than Vas, to over 4 times smaller and rarely over 1x time Vas. Note that Qt also plays a big role here.
Be careful with Vas. If you want to measure Vas yourself then be sure the speaker is run-in fully. Vas reflects the compliance of the speaker and this can vary a lot after running in.
Don’t understand the relation Vb needs to be somewhere below 2x Vas. With most speakers I end up with a much smaller BR box than Vas, to over 4 times smaller and rarely over 1x time Vas. Note that Qt also plays a big role here.
Be careful with Vas. If you want to measure Vas yourself then be sure the speaker is run-in fully. Vas reflects the compliance of the speaker and this can vary a lot after running in.
Hi MJK,
The constructed enclosure was just a test enclosure. I will build a new enclosure with a very large dimension comparing the other two dimensions, so the enclosure will act as a TL. In you reply you mention the word “Classic TL”. All your discussion is referring only to classic TL (TL offset), or also to all your sheets (Ported Box and ML TQWT)? Do you think that in a classic ported box(modeled with WinISD for example) the pressure is equal everywhere inside the box and in a ML TQWT the pressure is not equal?
The constructed enclosure was just a test enclosure. I will build a new enclosure with a very large dimension comparing the other two dimensions, so the enclosure will act as a TL. In you reply you mention the word “Classic TL”. All your discussion is referring only to classic TL (TL offset), or also to all your sheets (Ported Box and ML TQWT)? Do you think that in a classic ported box(modeled with WinISD for example) the pressure is equal everywhere inside the box and in a ML TQWT the pressure is not equal?
Pjotr said:Hi,
Don’t understand the relation Vb needs to be somewhere below 2x Vas. With most speakers I end up with a much smaller BR box than Vas, to over 4 times smaller and rarely over 1x time Vas. Note that Qt also plays a big role here.
Vb does not need to be 2 x Vas at all. Its that there's not much
point going beyond 2 x Vas in trying to extend bass because
the speaker compliance begins to dominate.
At Vb = Vas the compliances are equal and I'd call this a big reflex
box, as you say Qt is usually arranged such that Vb is typically
1/2 Vas but this does depend on the Qt of the driver.
Vb >/= ~ 1.5xVas generally needs very low port tuning to
avoid bass peaking if it is possible at all with the driver.
I'd also say if Vb > Vas then its a rule of thumb an alternative
driver would give better performance with the box volume,
especially in bass SPL capability, as the implication is a bigger
driver could have been used.
sreten.
Ok, thanks Sreten,
That makes it clearer. For BR boxes Vas is not very useful as a selection criterion IMHO. Qt is more suitable. To high a Qt and you end up with a large box indeed and sometimes it is impossible to tune to a decent alignment at all. It is not for nothing Qt of units intended for BR’s is somewhere between 0.2 and 0.3.
For a TL, a Qt of around 0.5 is a more suitable choice.
Cheers
That makes it clearer. For BR boxes Vas is not very useful as a selection criterion IMHO. Qt is more suitable. To high a Qt and you end up with a large box indeed and sometimes it is impossible to tune to a decent alignment at all. It is not for nothing Qt of units intended for BR’s is somewhere between 0.2 and 0.3.
For a TL, a Qt of around 0.5 is a more suitable choice.
Cheers
Pjotr said:Ok, thanks Sreten,
That makes it clearer. For BR boxes Vas is not very useful as a selection criterion IMHO. Qt is more suitable. To high a Qt and you end up with a large box indeed and sometimes it is impossible to tune to a decent alignment at all. It is not for nothing Qt of units intended for BR’s is somewhere between 0.2 and 0.3.
For a TL, a Qt of around 0.5 is a more suitable choice.
Cheers
IMO you have to consider Qt, Vas and Fs because its
simple to make a low Qt and Fs driver by having a high Vas.
What you need is low Fs combined with moderate Vas
and a Qt around 0.3 to 0.35 for good extension.
For a 88dB/W 6.5" driver something like :
30Hz Fs, 25 litre Vas and Qt = 0.35.
I regard an innappropriate Vas as a rejection criteria.
sreten.
Pjotr said:
I am a little surprised to hear that. A speaker has to have a Qt well over .4 and be tuned below Fs for a box much larger than Vas to make sense.
Thomas W took a HE 15 and put it into an enclosure about twice it's Vas. It was the only way he could get it to have output below the Fs of 25 Hz. He used an equalizer to take care of the dip above the tuning frequency of 18 Hz. That was not the ideal situation, but it was what he was forced to rely on to use a driver with a 1.5" one way excursion to get low bass output out of it.
In most cases, there is a wide range of choices for drivers with Qts of .4 or below. Not that many take drivers with Qts over .5 to use for reflex encloures. As Small has shown here, the transient response gets worse the greater the Vb is compared to the Vas.
http://www.diyaudio.com/forums/showthread.php?postid=303547#post303547
If anything, on a few threads here I found myself the only person arguing that a driver with a Qts of .5 can be put into an enclosure larger than it's Vas and be tuned to below it's Fs and produce a perfectly usable box, as long as Vb is not that much larger than Vas. Making a reflex box with the Vb four times the Vas is cetrtainly beyond normal practice-very few would be inclined to do so, considering that there is no Thiele-Small alignment which calls for it.
Re: Ok, let's see if I remember...
I did some figuring. Assuming similar Vas/Fs ratios, if your drivers Fs was 50 Hz, the Vas would shrink to about 17 liters. Which would make your enclosure about six times the Vas.
You tuned the enclosure to 22 Hz. Just as a quick experiment, have you considered running a tone of say, 25 Hz or so through your speakers, and see if there is any difference when you cover the port and when you leave it clear? There are freeware online tone generators, such as this one provided by the Taylor family on their home page:
http://www.satsignal.net "Audio Tools"
aaroncgi said:... so let's call it 3 cubic feet or 85 liters. I suppose it could be a bit less, so if anyone knows how to calculate area of a D shape, we could get more exact.
In any case, this is the volume (85 liters) I used to tune the enclosure. This in theory is somewhere around 22Hz.
....The drivers were played pretty hard for a couple weeks before I measured Fs, so they should have been broken in. The fact that they measured between 49-52Hz seems to point in that direction, as most notes I have seen put the new batch at around 55Hz Fs.
I did some figuring. Assuming similar Vas/Fs ratios, if your drivers Fs was 50 Hz, the Vas would shrink to about 17 liters. Which would make your enclosure about six times the Vas.
You tuned the enclosure to 22 Hz. Just as a quick experiment, have you considered running a tone of say, 25 Hz or so through your speakers, and see if there is any difference when you cover the port and when you leave it clear? There are freeware online tone generators, such as this one provided by the Taylor family on their home page:
http://www.satsignal.net "Audio Tools"
Pjotr said:Hi Sreten,
Low Fs and a moderate/small Vas are contrary to each other in some way. If a speaker has both its efficiency is unavoidable low then.
Cheers
Indeed they are - and as you say efficiency is implied to be low.
Once you've decided on your approximate box size and the
amount of bass extension you want you select the driver on
Fs, Qt and Vas parameters - but as you say your effectively
choosing the eficiency of the driver to suit your requirements.
The thing is you need to know the other two parameters to
judge the third, you cannot use any of them in isolation for
driver selection.
sreten.Hi tda,
My discussion was simplified to a classic TL, not mass loaded by a port, so that the wavelengths of sound are directly related to the lengths. Once you mass load, add a port, then the long dimension becomes much shorter and closer to the width and depth dimensions. I wanted a simple explanation without this additional complication.
When you use a lumped parameter model, like WinISD, you are assuming that the pressure in the box is uniform. That is how the equations were formulated. In reality at some frequency this assumption breaks down and the simulation is very inaccurate. The frequency at which this occurs is related to the dimensions of the enclosure. There are always standing waves in enclosures! The program will not tell you when this happens. The user needs to be aware of the limitations of any computer model and double check that they are not asking for something outside the scope of the assumptions used to derive the equations. Garbage in, garbage out, and you don't get any warning.
My MathCad worksheets extend the lumped parameter model to include one long dimension and the associated standing waves. If your enclosure is dimensioned to be consistent with this type of geometry, the worksheets are accurate. If your dimensions are nearly equal then you have stepped outside the intent of the worksheets. Garbage in and garbage out again. This is true of all my MathCad worksheets.
But there is one situation that will still work, if your enclosure is small enough that you do not get standing waves in any of the three dimensions in the bass frequency range then the MathCad worksheets will simulate a constant pressure in the box and a lumped parameter solution.
Bottom line, whenever you use any computer model you need to have a real good understanding of the assumptions and limitations that are associated with the derivation of the equations used in the simulation. Just because you can enter data and produce a SPL response curve, there is no double check to make sure that the results will be accurate. These programs are only as smart as the user.
My discussion was simplified to a classic TL, not mass loaded by a port, so that the wavelengths of sound are directly related to the lengths. Once you mass load, add a port, then the long dimension becomes much shorter and closer to the width and depth dimensions. I wanted a simple explanation without this additional complication.
When you use a lumped parameter model, like WinISD, you are assuming that the pressure in the box is uniform. That is how the equations were formulated. In reality at some frequency this assumption breaks down and the simulation is very inaccurate. The frequency at which this occurs is related to the dimensions of the enclosure. There are always standing waves in enclosures! The program will not tell you when this happens. The user needs to be aware of the limitations of any computer model and double check that they are not asking for something outside the scope of the assumptions used to derive the equations. Garbage in, garbage out, and you don't get any warning.
My MathCad worksheets extend the lumped parameter model to include one long dimension and the associated standing waves. If your enclosure is dimensioned to be consistent with this type of geometry, the worksheets are accurate. If your dimensions are nearly equal then you have stepped outside the intent of the worksheets. Garbage in and garbage out again. This is true of all my MathCad worksheets.
But there is one situation that will still work, if your enclosure is small enough that you do not get standing waves in any of the three dimensions in the bass frequency range then the MathCad worksheets will simulate a constant pressure in the box and a lumped parameter solution.
Bottom line, whenever you use any computer model you need to have a real good understanding of the assumptions and limitations that are associated with the derivation of the equations used in the simulation. Just because you can enter data and produce a SPL response curve, there is no double check to make sure that the results will be accurate. These programs are only as smart as the user.
A) What are the dimensions of the original box? Inside or outside will do, as long as you tell us which, and the thickness of your building material.
B) Tda is using a 4 inch woofer, which naturally has a Vas which is a small fraction of the 27 liter, (almost 1 Ft³), box. What if tda replaced that with a 6½" or even 8" woofer with a Vas nearly 27 liter or even larger? The standing wave situation should not change with the change of woofer size or Vas size, will it?
If tda can dig up a 6½" (160 mm), woofer or 8" (200mm), woofer, stick it in the same or similar sized box and finds that the output is more nearly that of the predicted response, then we have a pretty good indicator that the irregularity is due to the Vas being too small relative to box size rather than any standing wave situation.
On the other hand, if the larger woofer exhibits the same situation as the original woofer and does not seem to follow the predicted output, then we know that it is due to the standing wave situation inside the box.
Remember, this is for a bass reflex box only. Transmission Lines are a whole different story, even though Thiele-Small parameters are used in the calculations.
Might be fun to try.
B) Tda is using a 4 inch woofer, which naturally has a Vas which is a small fraction of the 27 liter, (almost 1 Ft³), box. What if tda replaced that with a 6½" or even 8" woofer with a Vas nearly 27 liter or even larger? The standing wave situation should not change with the change of woofer size or Vas size, will it?
If tda can dig up a 6½" (160 mm), woofer or 8" (200mm), woofer, stick it in the same or similar sized box and finds that the output is more nearly that of the predicted response, then we have a pretty good indicator that the irregularity is due to the Vas being too small relative to box size rather than any standing wave situation.
On the other hand, if the larger woofer exhibits the same situation as the original woofer and does not seem to follow the predicted output, then we know that it is due to the standing wave situation inside the box.
Remember, this is for a bass reflex box only. Transmission Lines are a whole different story, even though Thiele-Small parameters are used in the calculations.
Might be fun to try.
kelticwizard,
tda designed and optimized a box based on the properties of his small driver. Granted the size of the dirver and box made the system a very extreme alignment. If he starts switching drivers then he is not working to understand his own design. He is just randomly changing his system hoping for some improvement. Whatever results will not be an optimum combination. Nothing wrong with that but I would be more interested in undersatnding what went wrong in the first place.
I still recommend doing an impedance measurement for the unstuffed cabinet and seeing what standing waves can be seen in this plot. Then I would add the cardboard internal baffles to push the speaker design towards the assumptions in the simulation, only allow standing waves in the "long" direction and see if things improve.
tda designed and optimized a box based on the properties of his small driver. Granted the size of the dirver and box made the system a very extreme alignment. If he starts switching drivers then he is not working to understand his own design. He is just randomly changing his system hoping for some improvement. Whatever results will not be an optimum combination. Nothing wrong with that but I would be more interested in undersatnding what went wrong in the first place.
I still recommend doing an impedance measurement for the unstuffed cabinet and seeing what standing waves can be seen in this plot. Then I would add the cardboard internal baffles to push the speaker design towards the assumptions in the simulation, only allow standing waves in the "long" direction and see if things improve.
Hmm,
You can better have a look at the phase change of the impedance, it is more sensitive. But with proper damping of a BR box, standing waves shouldn’t be a problem.
From Keele’s pocket calculator method to find the optimum box size for a BR:
Vb = 15 * Qt^2.87 * Vas
This will give an alignment without hump or dip.
Tda, if you have a chance get also this book by David Weems:
http://topics.practical.org/browse/Loudspeakers
It is much more structured than Dickason and has a good introduction.
Cheers
You can better have a look at the phase change of the impedance, it is more sensitive. But with proper damping of a BR box, standing waves shouldn’t be a problem.
From Keele’s pocket calculator method to find the optimum box size for a BR:
Vb = 15 * Qt^2.87 * Vas
This will give an alignment without hump or dip.
Tda, if you have a chance get also this book by David Weems:
http://topics.practical.org/browse/Loudspeakers
It is much more structured than Dickason and has a good introduction.
Cheers
Thanks for the great software link!
Kelticwizard,
That is a really awesome signal generator, Audio SweepGen! I have a standalone signal generator at work, but it's accuracy is in question, so I may switch over to this computer based one.
Anyhow, I did your little experiment, blocked off the port, and came up with some interesting and somewhat unexpected results:
- Both versions have strong output down to 30Hz. I find this amazing in a sealed box using a woofer with an Fs of 50Hz!
- The sealed box starts to noticeably lose output below about 40Hz, but very smoothly and gradually all the way down to 20Hz, where it's audible but barely.
- The vented box seems to have a dip from 27-30Hz, a broad peak from 21-26Hz, and still has significant output at 18Hz. It has much more output below 30Hz than the sealed box.
- Woofer excursion in the vented box appears lowest right at 20Hz, so that seems pretty close to expected.
- There is something rattling in my right speaker (suspect a loose crossover nut).
- My subwoofer tune is right where it's supposed to be, at 11-12Hz.
- I won't need an air conditioner in the summer; just set the frequency to 10Hz, crank the volume way up, and attach an elbow to direct air from my subwoofer ports.
Of course, putting either speaker in a corner seriously helps the low end, and above 30 Hz can either make up or totally skew the difference between them. This would explain why I had so darned much bass in my apartment with these speakers, as both L/R were in a corner. Now, the L speaker is about 10' from either side wall, and the R speaker about 2.5' from the side.
Next week I'll grab my SPL meter from work and bring it home, so I can put some actual numbers to each frequency.
Based on this, it would certainly seem there really is not a huge benefit to be gained from my port except below 30Hz. The sealed response seems a bit smoother, and extension down to 30 Hz should be just fine for most kinds of music. Since I use my subwoofer 99.9% of the time, low end extension is not relevant anyway. I can certainly experiment and make the port shorter, but I suspect what would happen is the peak at the bottom end would grow higher in amplitude and frequency.
I'm still amazed just how much low end there is without the port. Is this a sort of infinite baffle reponse, with a super shallow rolloff below F3 (which may be quite high)?
Cheers,
Aaron Gilbert
Kelticwizard,
That is a really awesome signal generator, Audio SweepGen! I have a standalone signal generator at work, but it's accuracy is in question, so I may switch over to this computer based one.
Anyhow, I did your little experiment, blocked off the port, and came up with some interesting and somewhat unexpected results:
- Both versions have strong output down to 30Hz. I find this amazing in a sealed box using a woofer with an Fs of 50Hz!
- The sealed box starts to noticeably lose output below about 40Hz, but very smoothly and gradually all the way down to 20Hz, where it's audible but barely.
- The vented box seems to have a dip from 27-30Hz, a broad peak from 21-26Hz, and still has significant output at 18Hz. It has much more output below 30Hz than the sealed box.
- Woofer excursion in the vented box appears lowest right at 20Hz, so that seems pretty close to expected.
- There is something rattling in my right speaker (suspect a loose crossover nut).
- My subwoofer tune is right where it's supposed to be, at 11-12Hz.
- I won't need an air conditioner in the summer; just set the frequency to 10Hz, crank the volume way up, and attach an elbow to direct air from my subwoofer ports.
Of course, putting either speaker in a corner seriously helps the low end, and above 30 Hz can either make up or totally skew the difference between them. This would explain why I had so darned much bass in my apartment with these speakers, as both L/R were in a corner. Now, the L speaker is about 10' from either side wall, and the R speaker about 2.5' from the side.
Next week I'll grab my SPL meter from work and bring it home, so I can put some actual numbers to each frequency.
Based on this, it would certainly seem there really is not a huge benefit to be gained from my port except below 30Hz. The sealed response seems a bit smoother, and extension down to 30 Hz should be just fine for most kinds of music. Since I use my subwoofer 99.9% of the time, low end extension is not relevant anyway. I can certainly experiment and make the port shorter, but I suspect what would happen is the peak at the bottom end would grow higher in amplitude and frequency.
I'm still amazed just how much low end there is without the port. Is this a sort of infinite baffle reponse, with a super shallow rolloff below F3 (which may be quite high)?
Cheers,
Aaron Gilbert
Aaron,
While the bass response of the closed and ported versions sounds like an interesting trade-off, take a look at the driver excursion when you sweep this frequency range. I think that you may find that for the ported box the driver's motion is significantly attenuated at the tuning frequency. This may give you some advantage when you really push the speakers to higher volume in terms of distortion and running out of displacement. On the other hand the ported box will unload the woofer below the tuning frequency but it sounds like this is low enough in frequency to not be a concern, unless you are battling the old turntable's rumble.
While the bass response of the closed and ported versions sounds like an interesting trade-off, take a look at the driver excursion when you sweep this frequency range. I think that you may find that for the ported box the driver's motion is significantly attenuated at the tuning frequency. This may give you some advantage when you really push the speakers to higher volume in terms of distortion and running out of displacement. On the other hand the ported box will unload the woofer below the tuning frequency but it sounds like this is low enough in frequency to not be a concern, unless you are battling the old turntable's rumble.
MJK,
Quite right you are. I did notice the much increased cone excursion with the port sealed. These 7" have a pretty healthy Xmax of 5.5mm though, so with two per speaker, that gives me slightly more cone area than a single 12", albeit still with a low 5.5mm Xmax. At the tuning frequency, the ported version's woofers aren't moving much at all, as you would expect, whereas with sealed, they are going crazy and not giving much output to boot.
For normal music it's true, the tuning is low enough that unloading is not a concern. I have some CDs and DVDs that have significant below 20Hz information though, and for that material, the subwoofer gets the nod, as it doesn't start to unload until 10Hz and below.
All that said, I deal with commercial speakers every day at my job, and four Peerless 850122's vented kick the crap out of a LOT of subwoofers. But it's not very often that I see a commercial subwoofer use six cubic feet of enclosure. In fact, I don't know that I have ever seen one of those - yet.
Aaron Gilbert
Quite right you are. I did notice the much increased cone excursion with the port sealed. These 7" have a pretty healthy Xmax of 5.5mm though, so with two per speaker, that gives me slightly more cone area than a single 12", albeit still with a low 5.5mm Xmax.
At the tuning frequency, the ported version's woofers aren't moving much at all, as you would expect, whereas with sealed, they are going crazy and not giving much output to boot.For normal music it's true, the tuning is low enough that unloading is not a concern. I have some CDs and DVDs that have significant below 20Hz information though, and for that material, the subwoofer gets the nod, as it doesn't start to unload until 10Hz and below.
All that said, I deal with commercial speakers every day at my job, and four Peerless 850122's vented kick the crap out of a LOT of subwoofers.
But it's not very often that I see a commercial subwoofer use six cubic feet of enclosure. In fact, I don't know that I have ever seen one of those - yet.Aaron Gilbert
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.