Member
Joined 2009
Paid Member
Looking again at Cal's special box design (see post 7 of this thread) - with an open area specifically behind the mid driver. This is a box speaker for sure - there's lots of 'box' everywhere. But Cal doesn't say he finds it 'boxy' and it sounds better with the back section behind the mid to be open. So it's about allowing the mids to 'breath' like a dipole.
So how about this one - the well-regarded InConcert Miles speaker. Look at the attached image. The bass reflex ports are huge, positioned right behind the drivers. These drivers are XO'd high enough to say that there must be a prodigious amount of 'mid' leakage through the ports out of the back of the speaker. The thread is here: AVforum.no - Bygging av, Tangen AVdesign inConcert Miles.
Doesn't this remind you of Cal's open back-mid speaker. Perhaps this is the secret-sauce ?
Here's a quote from somebody who built a version of it:
"Because the ports are so large, the cabinet also seems to breathe easily, which (in my mind at least) accounts for the clean and open sound. Normally a bass reflex cabinet has a certain character to it, which I find interestingly lacking with these - which is a good thing BTW!"
My latest and largest HE speaker build
So how about this one - the well-regarded InConcert Miles speaker. Look at the attached image. The bass reflex ports are huge, positioned right behind the drivers. These drivers are XO'd high enough to say that there must be a prodigious amount of 'mid' leakage through the ports out of the back of the speaker. The thread is here: AVforum.no - Bygging av, Tangen AVdesign inConcert Miles.
Doesn't this remind you of Cal's open back-mid speaker. Perhaps this is the secret-sauce ?
Here's a quote from somebody who built a version of it:
"Because the ports are so large, the cabinet also seems to breathe easily, which (in my mind at least) accounts for the clean and open sound. Normally a bass reflex cabinet has a certain character to it, which I find interestingly lacking with these - which is a good thing BTW!"
My latest and largest HE speaker build
Attachments
Well, there really isn't any other place to put those giant ports than the rear side of the cabinet, so they ended up right behind the drivers. Have no idea of there is significant midrange leakage through the ports.
I had to use a lot of damping material in this speaker to attenuate the "boxy" sound of internal standing waves, but not enough to kill it entirely....
I had to use a lot of damping material in this speaker to attenuate the "boxy" sound of internal standing waves, but not enough to kill it entirely....
Just the way my mind works, but I can't help noticing that musical valve amps, tweeter attenuators and open back (or infinite baffle in the original correct sense) speakers all have one thing in common.
They reduce the Q of the drivers. Or reduce the resonant peaks and consequent phase bend and time delay if you like.
Is there a consensus on the Qt of drivers that work best on open baffles. Presumably low?
They reduce the Q of the drivers. Or reduce the resonant peaks and consequent phase bend and time delay if you like.
Is there a consensus on the Qt of drivers that work best on open baffles. Presumably low?
My experience is that it is the middle range that suffers most from box colourations, so you should be careful to avoid that when building. But it might be that those problems are only discovered when you mount the drivers and start to play, so you should listen then if you got any disturbing colourations and try to get rid of them.
How to brace the cabinet and so on is the art to master.
How to brace the cabinet and so on is the art to master.
Just the way my mind works, but I can't help noticing that musical valve amps, tweeter attenuators and open back (or infinite baffle in the original correct sense) speakers all have one thing in common.
They reduce the Q of the drivers. Or reduce the resonant peaks and consequent phase bend and time delay if you like.
Is there a consensus on the Qt of drivers that work best on open baffles. Presumably low?
No, Q is actually increased with additional series resistance like you get with tube amps and attenuators.
There are two schools when it comes to OB and Q: High Q and low/medium Q. Both works, and both requires EQ. Low Q requires less EQ though, at the cost of lower overall sensitivity.
Is there a consensus on the Qt of drivers that work best on open baffles. Presumably low?
Let's presume you want a maximum flat system bass FR (Q=0.77) or no overshoot in system impulse response (Q=0.5?). Then you want to start with a driver Qts above those values to minimise EQ.
If you believe in the need of low driver Qts to "control" cone movement, you will have an ever harder time to supply the huge EQ amount needed for a flat bass response curve.
For drivers starting five octaves above their resonance frequency - take any Q you want.
Rudolf
Thanks guys. I didn't express myself well on the Q business. You'd compensate the box and crossover to allow for a valve amp of course.
Can I ask you it another way. How do you control cone excursion with open baffle. Is it just the car backshelf approach of a bass driver with a Qts of between 0.7 and 1? Would you consider a RLC series compensation of the driver Fs resonance?
Don't know much about this. But interested.
Can I ask you it another way. How do you control cone excursion with open baffle. Is it just the car backshelf approach of a bass driver with a Qts of between 0.7 and 1? Would you consider a RLC series compensation of the driver Fs resonance?
Don't know much about this. But interested.
How do you control cone excursion with open baffle.
Don't know much about this. But interested.
Linkwitz explains the dipole
Martin J. King explains how to EQ OBs.
In theory it shouldn't be but in practice damping material isn't always used right.. what, where and how much for instance.
But as in this example of an 8" woofer, the backwave rounds the cabinet in a wavelength at around 600Hz, and 1" of damping can give 20dB of attenuation to around 200Hz (possibly better when the positioning is favourable). The 16" cabinet height would be a half wavelength at 400Hz.
@Steve, I was thinking of using this for wall mounting around a TV, is it the same as using a roundover/back with a suitable radius for the distances involved?
Attachments
Member
Joined 2009
Paid Member
Hi - I certainly wasn't intending to criticise the design. To the contrary - I'm suggesting that there's hidden genius in allowing for more breathing space behind the drivers. It's actually a desirable thing.
How should we design these ports to maximize the benefits of mid-range dipoleness ?
Those Linkwitzlab links never work for me, Rudolf. No idea why. Just can't load them in the UK. But M.J. King on open baffle was fine. Thanks.
It does seem to be received wisdom that you aim for Butterworth maximally flat bass response, yet the more overdamped low Q bell-shaped gaussian or bessel shapes have something going for them too. Bell-shaped is certainly what works best with midrange driver response.
Sorry, Allen. That gaussian shape is not really a box design as such, and your guess is as good as mine. More the front baffle that best corrects the diffraction fault of a driver not being totally flush with the wall or baffle and the consequent droop in response. I doubt if a 5x8x13 box is going to be too bad anyway. BTW, I like your reflection approach to standing waves.
It does seem to be received wisdom that you aim for Butterworth maximally flat bass response, yet the more overdamped low Q bell-shaped gaussian or bessel shapes have something going for them too. Bell-shaped is certainly what works best with midrange driver response.
Sorry, Allen. That gaussian shape is not really a box design as such, and your guess is as good as mine. More the front baffle that best corrects the diffraction fault of a driver not being totally flush with the wall or baffle and the consequent droop in response. I doubt if a 5x8x13 box is going to be too bad anyway. BTW, I like your reflection approach to standing waves.
Graphs explained.
The in-room graph is consistent with what we should get if we are not correcting for the room. The variations you are seeing at 52Hz (-6db valley) and 104Hz (+3dB peak) are from the longitudinal modes in the room. They are actually pretty benign when compared to what regular box speakers produce in the same position.
(Note: Ignore the response below 40Hz, it's LF noise. I can assure you this speaker is NOT putting out much at 20Hz)
Notice the relative absence of other modes!
OB will reduce the contributions of lateral and vertical modes. OB will still excite longitudinal modes, so will still get some peaks and dips related to the room length. I would say this graph seems to support the theory.
The in-room response graph is a single point measurement, taken at the main listening position. In this same position regular speakers create huge response aberrations on the order of +/- 15dB, with significant peaks and dips all the way to 300-400Hz.
Also, the graph has a very expanded scale, 1dB/div. which tends to visually emphasize the differences in linearity.
Perhaps I should throw in a response curve from an unequalized conventional box speaker in the same position... the peaks and valleys go to about +/- 15dB, and there are many more of them. But it's too ugly. I can't bear to look at it!
Markus your graphs look really great. That's a nicely tuned system. How would those same speakers look if non-optimally placed in a different room that has really bad resonant modes?
The point here is not to say look how good I can get these speakers when I optimize them in the room, it's to illustrate that a good OB design can give pretty good results even when used in a very BAD room. Without special treatments or room eq.
Besides which, the area I'm really most concerned with (musically) is not the low bass below 100Hz. OB with the addition of a good subwoofer array is the answer there.
The area where I notice the most gain in clarity with OB is 100Hz to 400Hz, and this is the core of where most music occurs. It appears to me that the lion's share of the colorations in that area come form lateral and vertical room modes, and with the OB exciting them much less, the net gain in clarity and articulation is phenomenal.
Yes, treating the room to solve these issues is the best solution, even when using an OB (StigErik's system is a nice example of that). But there are many, many, many people out there who do not have the resources or desire or WAF to address the room in these ways. My new OB system is designed with that in mind, and gives excellent results in a very wide variety of rooms. How many conventional box speakers can do the same? Not many at all.
Now, for those that DO have the resources and a really big dedicated music room, yes we can do custom solutions based around my omni-directional reference system, but that's a different kettle of fish, and gold-plated ones at that. An expensive solution indeed.
What's really cool is that with this simple little OB speaker we can get results that approach the quality of the reference system/special room/finely tuned scenario... in a regular, untreated living room and with very high WAF.
THAT's the beauty of good OB!
Depends on what you are comparing to!
The in-room graph is consistent with what we should get if we are not correcting for the room. The variations you are seeing at 52Hz (-6db valley) and 104Hz (+3dB peak) are from the longitudinal modes in the room. They are actually pretty benign when compared to what regular box speakers produce in the same position.
(Note: Ignore the response below 40Hz, it's LF noise. I can assure you this speaker is NOT putting out much at 20Hz)
Notice the relative absence of other modes!
OB will reduce the contributions of lateral and vertical modes. OB will still excite longitudinal modes, so will still get some peaks and dips related to the room length. I would say this graph seems to support the theory.
The in-room response graph is a single point measurement, taken at the main listening position. In this same position regular speakers create huge response aberrations on the order of +/- 15dB, with significant peaks and dips all the way to 300-400Hz.
Also, the graph has a very expanded scale, 1dB/div. which tends to visually emphasize the differences in linearity.
Perhaps I should throw in a response curve from an unequalized conventional box speaker in the same position... the peaks and valleys go to about +/- 15dB, and there are many more of them. But it's too ugly. I can't bear to look at it!
Markus your graphs look really great. That's a nicely tuned system. How would those same speakers look if non-optimally placed in a different room that has really bad resonant modes?
The point here is not to say look how good I can get these speakers when I optimize them in the room, it's to illustrate that a good OB design can give pretty good results even when used in a very BAD room. Without special treatments or room eq.
Besides which, the area I'm really most concerned with (musically) is not the low bass below 100Hz. OB with the addition of a good subwoofer array is the answer there.
The area where I notice the most gain in clarity with OB is 100Hz to 400Hz, and this is the core of where most music occurs. It appears to me that the lion's share of the colorations in that area come form lateral and vertical room modes, and with the OB exciting them much less, the net gain in clarity and articulation is phenomenal.
Yes, treating the room to solve these issues is the best solution, even when using an OB (StigErik's system is a nice example of that). But there are many, many, many people out there who do not have the resources or desire or WAF to address the room in these ways. My new OB system is designed with that in mind, and gives excellent results in a very wide variety of rooms. How many conventional box speakers can do the same? Not many at all.
Now, for those that DO have the resources and a really big dedicated music room, yes we can do custom solutions based around my omni-directional reference system, but that's a different kettle of fish, and gold-plated ones at that. An expensive solution indeed.
What's really cool is that with this simple little OB speaker we can get results that approach the quality of the reference system/special room/finely tuned scenario... in a regular, untreated living room and with very high WAF.
THAT's the beauty of good OB!
Attachments
Last edited:
Well, if this is what you wanted to show then the data for the closed box speaker is missing. And, as I said before, 1/3 smoothing is way too coarse to show anything meaningful at low frequencies.
Same data, different smoothing:
An externally hosted image should be here but it was not working when we last tested it.
P.S. If the 20Hz peak in your graph is the noise floor then you're not measuring loud enough and your graph is probably not really valid.
Not so much the roll-off
About the reduction in room boom, it's not so much the roll-off, with the eq being applied via DSP, my OB speaker is anechoically flat down to 39Hz. See the attached "anechoic" and in-room graphs.
Most of the bothersome room-boom is in the 60 to 200Hz range, so if the speaker extension is flat well below this, other factors must be at play for it to sound so much cleaner.
Here's the reason why:
OB speakers have a figure 8 directivity pattern, with almost complete cancellation when measured 90 degrees off axis. Because of this, when facing forward, they dramatically reduce the excitation of vertical and lateral modes in the room. (And yes, Toeing them in or angling them back would start adding more lateral and vertical excitations back in.)
This OB arrangement also dramatically reduces the bass output as frequency decreases, and as a result many OB designs sport woofers with very high Qt (up to 1.5 or even 2), to get back a little of the lost bass through resonant peaking. And even so, OB usually requires LOTS of eq to make up for that missing 10dB or 15dB or even 20dB of bass at 40Hz.
When I first looked at this I was concerned that the bass would be muddy and slow. But it's not. As Frenchy would say, "Au contraire, mon ami, eet ees queeck, fast and eempactful."
So, although it might seem counter-intuitive, using a very high Q woofer in an OB can yield very good sound, IMO much better even than the majority of well-designed conventional box speakers using low-Q drivers.
Where OB has its biggest limitation is in achieving high spl at low frequencies, due to excursion limitations. If we want to get down to 20Hz at 110 dbSPL, we're gonna need a whole bank of 15" drivers to get it. In practice, it may be better to simply blend in a really good servo-sub at about 40-50Hz.
Hi Steve, thanks for the kind words.
About the reduction in room boom, it's not so much the roll-off, with the eq being applied via DSP, my OB speaker is anechoically flat down to 39Hz. See the attached "anechoic" and in-room graphs.
Most of the bothersome room-boom is in the 60 to 200Hz range, so if the speaker extension is flat well below this, other factors must be at play for it to sound so much cleaner.
Here's the reason why:
OB speakers have a figure 8 directivity pattern, with almost complete cancellation when measured 90 degrees off axis. Because of this, when facing forward, they dramatically reduce the excitation of vertical and lateral modes in the room. (And yes, Toeing them in or angling them back would start adding more lateral and vertical excitations back in.)
This OB arrangement also dramatically reduces the bass output as frequency decreases, and as a result many OB designs sport woofers with very high Qt (up to 1.5 or even 2), to get back a little of the lost bass through resonant peaking. And even so, OB usually requires LOTS of eq to make up for that missing 10dB or 15dB or even 20dB of bass at 40Hz.
When I first looked at this I was concerned that the bass would be muddy and slow. But it's not. As Frenchy would say, "Au contraire, mon ami, eet ees queeck, fast and eempactful."
So, although it might seem counter-intuitive, using a very high Q woofer in an OB can yield very good sound, IMO much better even than the majority of well-designed conventional box speakers using low-Q drivers.
Where OB has its biggest limitation is in achieving high spl at low frequencies, due to excursion limitations. If we want to get down to 20Hz at 110 dbSPL, we're gonna need a whole bank of 15" drivers to get it. In practice, it may be better to simply blend in a really good servo-sub at about 40-50Hz.
Attachments
Oh, it's valid enough!
It's the noise from the Air Conditioning system. I had another measurement without it, and managed to erase it by accident, but the main response was identical above 40Hz. LF noise doesn't change the data at higher frequencies if it's not overloading the mic.
As for the contention that the 1/3 octave resolution not being sufficient, well... Your graphs do illustrate the point you're making, and the amplitude swings ARE dramatically different. But do we actually HEAR it that way? Less than it might seem. 1/3 octave is accepted by most in the pro-audio sector as being fairly close to how our hearing integrates sound.
That being said, I agree, smoother with a high resolution is better. But your contention that 1/3 octave invalidates the measurements is not something that most industry veterans would automatically agree with. I certainly don't.
Also, your examples are mostly concerned with the modes in the sub 100Hz range. IMO, it is more important to get the 100 to 400 Hz range working as well as possible, and this is where the OB works best. Again, I think OB's greatest contribution is in the reduction of lateral and vertical excitation, and due to "average" room dimensions, usually in the 80Hz to 400 Hz range
In practice, to address the low bass issues, I would add a distributed servo sub-woofer array with two at the far front, two in the far back, adjust delay so the arrival from the front subs is in exact alignment with the signal from the main speakers, and run the rear subs out of phase (again with the appropriate delay) so as to cancel the pressure at the back wall, and voila, those main longitudinal modes disappear almost completely, and not just at the sweet spot. It's been done before. A brilliant idea, almost completely eliminates the need for any eq. in the low bass.
But again, if I had to rank the relative importance of getting the 100-400Hz range right, I would give it a 9/10 .
As for getting the sub 100 Hz stuff just right? Well IMO that's about a 2/10... because most music is happening in the 100 to 400Hz range.
Please note, I'm not underestimating the importance of the low bass: my speaker is adjusted to be anechoically flat to 39Hz. As a result it's pretty good for just about any musical genre, and only shows limitations on a very small number of recordings that feature extreme infra-bass. For that, I would invoke the rule of the holy sub-woofer: use one.
Last edited:
Did you measure the noise floor?
Pardon? When reading through the current scientific literature there is very broad agreement that 1/3 resolution below the Schröder frequency was a very bad mistake. 1/3 wasn't chosen because of its perceptual validity but because of technical limitations.
Important is a frequency response that is complete. Around the Schröder frequency is as important as above or below it. So while a dipole might indeed have benefits when compared to a boxed speaker, you're missing to show any data that would prove your claim.
You're also just showing a single point in space. How does the frequency response look like at the left and the right ear??
Last edited:
Hi Markus, I too am of of the opinion that 1/3 doesn't describe everything there is to know about sound. Our hearing is a LOT more discriminating both in time and frequency than most studies reflect, especially our subliminal hearing... which almost no one has studied because it requires a lot of brain-wave measurements.
However, I'm also not in agreement that 1/3 oct. data is worthless and that the use of 1/3 oct makes a measurement invalid. Especially when said 1/3 octave data doesn't look all that different from the 1/12th data excepting it's prettier to look at.
As for missing any data that would prove my claim, I clearly disagree with your stance.
Although the data I have chosen to show here is limited, it is a subset which IS representative of other data that I have on hand. I could present reams of data, but this is not the place for that! I'll be happy to present more information, but please allow me to do that in the Vendor's section rather than doing it here. (I wish to respect the forum rules).
As for averaged response across many points, and deviance from the average, that's one of the criteria I study the most. It's hugely important. Again, I'll be happy to post the data for this in a more appropriate place. But I will say the OB I've referenced here does pretty well in that regard.
So, to bring this back on-topic, my contention is that while cabinet resonances and panel resonances are indeed audible, and desrving of solid construction, damping and bracing, that the room resonances are even more important, and that OB reduces the room issues significantly.
Perhaps the "absolute method" would be to measure at 1/24th octave with good conventional enclosures and my new OB design and compare extensively in 10 different rooms, and then publish the results in an 80 page white paper. But when it comes to giving helpful hints to the OP, please forgive me for not publishing "complete" data according to the criteria you state is necessary. What I've given actually IS a pretty good indicator, if only you have eyes to see.