AVS Forum banner

1 - 20 of 54 Posts

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #1 ·
Over at SpeakerDesignWorks.com, there is an article on sealed vs. vented enclosure designs. Among other things, this article shows the response of a driver in three enclosures.




Here is the description of the graphic from the site :

In a burst plot, a sine wave is suddenly started at the origin (0 volts) and stops just as fast after a few cycles. The light blue trace is the input signal, and of course the ideal driver would track this response perfectly. Unfortunately, trivial details such as mass and inertia get in the way, so this doesnt happen. The cone takes a cycle or so to get up to speed, and likewise when the signal stops. How quickly it responds to the changes is an indication of its transient response. The magenta trace is the RS225 in the sealed enclosure Qtc .707. The white trace is the vented enclosure described previously. The Green trace is the RS225 in a traditional QB3 alignment. Note that while the low tuned alignment is not quite as good as the sealed alignment, it is far superior to the QB3 response, -which, although not shown, oscillates for yet another half cycle before settling down.


While the focus is on the differences between the three curves, what is perhaps more interesting is the behavior of the curves relative to the input signal.


The author suggests that the delay in amplitude response is attributable to mass and inertia. Extrapolating from his conclusion, lighter cones that have less mass and in turn less inertia will more quickly get their amplitude to match the input signal while a heavier cone with more mass and in turn more inertia will take longer to match the input signal in amplitude.


What is tricky about this amplitude distortion is that it doesn't show up in the traditional tests for distortion. It is not harmonic distortion and since after a couple of cycles it matches the input signal in amplitude, it won't show up in the typical multi-cycle burst test.


This may help explain why drivers with lighter moving masses and stronger motors tend to have a little bit more zip in the upper bass, such as HSU employs in his mid-bass module.


...


The authors observation that the sealed enclosure, the magenta trace, rises more quickly and stops more quickly is consistent with the observation that sealed enclosures have "faster" bass, if by that we mean how quickly the driver gets up to full amplitude and how quickly it stops when the signal stops.


This is a test of one driver in several different enclosures and not a comparison between drivers that are optimized for various enclosures, so it can't be said that sealed has better transient response universally, simply that for a particular driver, the sealed enclosure will have better transient response.


Thoughts?
 

·
Registered
Joined
·
7,989 Posts
I've read this before, and it seemed to agree with my intuition. I've often looked at mass, compliance, and BL when considering drivers. I have wondered if it also applies to higher frequencies. You hear the word detail thrown around. Perhaps a heavy stiff suspension doesn't reveal the details like a light compliant one would. Because it doesn't get moving with such small signal inputs. Not quite what the article is about, but something it makes me think about. I've heard people say they have to turn up certain systems quite loud to get detail. Possibly to get those signals high enough for the heavy driver to react well enough.
 

·
Registered
Joined
·
6,980 Posts

Quote:
Originally Posted by LTD02 /forum/post/21472834

In a burst plot, a sine wave is suddenly started at the origin (0 volts) and stops just as fast after a few cycles. The light blue trace is the input signal, and of course the ideal driver would track this response perfectly. Unfortunately, trivial details such as mass and inertia get in the way, so this doesnt happen. The cone takes a cycle or so to get up to speed, and likewise when the signal stops. How quickly it responds to the changes is an indication of its transient response. The magenta trace is the RS225 in the sealed enclosure Qtc .707. The white trace is the vented enclosure described previously. The Green trace is the RS225 in a traditional QB3 alignment. Note that while the low tuned alignment is not quite as good as the sealed alignment, it is far superior to the QB3 response, -which, although not shown, oscillates for yet another half cycle before settling down.


While the focus is on the differences between the three curves, what is perhaps more interesting is the behavior of the curves relative to the input signal.


The author suggests that the delay in amplitude response is attributable to mass and inertia. Extrapolating from his conclusion, lighter cones that have less mass and in turn less inertia will more quickly get their amplitude to match the input signal while a heavier cone with more mass and in turn more inertia will take longer to match the input signal in amplitude.


This may help explain why drivers with lighter moving masses and stronger motors tend to have a little bit more zip in the upper bass, such as HSU employs in his mid-bass module.

...


The authors observation that the sealed enclosure, the magenta trace, rises more quickly and stops more quickly is consistent with the observation that sealed enclosures have "faster" bass, if by that we mean how quickly the driver gets up to full amplitude and how quickly it stops when the signal stops.


This is a test of one driver in several different enclosures and not a comparison between drivers that are optimized for various enclosures, so it can't be said that sealed has better transient response universally, simply that for a particular driver, the sealed enclosure will have better transient response.


Thoughts?

Aaaack! Is even the DIY community still confused on this one?



While the signals do have different rise and settling behavior, the above author completely missed the WHY. The common error made here in comparing to the electrical input signal greatly confuses the issue. While the comparison does make a case for differing bandwidth and order designs, there is zero info that should infer mass/inertia or motor strength are to blame. This is simply bandwidth and roll off qualities seen in any Fourier decomposition.


The blue reference signal is not what we should compare to. We should look at the input signal with at least a comparable low pass filter that mimics the high frequency response of the woofer. Even better is to electronically low pass the driver under test within its pass band as it might be used in practice so the roll off and associated behavior is known. The woofer is not a full range speaker, and is not supposed to perfectly match the displayed waveform unless it extends past 10kHz and is used that high.


FACT: You cannot create the sharp start/stop shape at each end of the blue waveform without extended high frequency content. Right off the top, part of the softening of each end is the lack of high frequency content.


This is why the tone bursts used by Keele, Linkwitz and the CEA measurements ramp up and down over 5-6 cycles to keep the required bandwidth of the signal within the range of interest (ie ~1/3rd octave wide).


Quite to the contrary of what is asserted, you could add some mass to the RS225 woofer, use a larger sealed box resulting in the same Qb, and you would BETTER follow the input signal.
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #5 ·
"Perhaps a heavy stiff suspension doesn't reveal the details like a light compliant one would."


what you say is consistent with many subjective reports of many of the best drivers. some of the best, like the old altec 416-8c had a compliance of 637 um/n or the jbl 2245 at 439 um/n, which are off the charts by most of today's standards, subjectively were incredible at detail. like you say, a slightly different but related topic.
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #6 ·
"Aaaack! Is even the DIY community still confused on this one?"


the corner behavior isn't the focus. the focus is how long it takes for the driver to reach the amplitude of the input signal.


here is a 40hz sine wave (top) and the same 40hz sine wave low passed 2nd order at ~80hz.


you can see that all the high frequencies required to make the corner are gone, but the amplitude of the wave rises to its full peak at the first crest. it does not require several cycles to "build up" amplitude.


 

·
Registered
Joined
·
6,980 Posts

Quote:
Originally Posted by LTD02 /forum/post/21473103


"Aaaack! Is even the DIY community still confused on this one?"


the corner behavior isn't the focus. the focus is how long it takes for the driver to reach the amplitude of the input signal.

All one in the same. The ramping up of the signal and tailing down happen with different shapes due to the bandwidth of the playback device and the relative phase shift through the spectrum. Any driver with the exact same frequency response and phase response will have the same behavior, even if it's 10x heavier or 10x lighter.
 

·
Registered
Joined
·
2,405 Posts
It is a nice graph. But, these are differences that most people will never really get to evaluate. There are more variables than simply moving mass at play. Motor Strength, inductance, suspension compliance, air spring (or not), etc.


Comparing a light moving mass, higher Fs, high sensitivity driver to a heavier moving mass, lower Fs, lower sensitivity driver is just not a good apples/apples. They are designed for different things. One is designed for more output at higher frequencies for a given input signal. The other is designed to be able to generate more output for a given inout signal at lower frequencies. Even this comparison is waayy too simplified.


For more info on the subject, you need to pick through some of this:

http://www.baudline.com/erik/bass/thomas_danley.html


Or read what Mark said above.


I tell folks the following:


If you do not have near identical clean (read: low distortion, however you want to measure it, but a baseline must be set, whatever it is) output and extension with flat frequency response, you cannot really compare two alignments with regards to group delay, transient response, etc. without significant bias. This means that in a small room, the 2 compared speakers must be in the exact same location when being compared, as differing locations could lead to differing freq responses. Especially in small rooms. Outdoors, not so much.


I don't know of too many folks who have compared 2 LF (or fullrange, or whatever) units that have fulfilled the above criteria.


Also, there's what I call the 'paraxial criteria' of audio. The less the driver has to move, the more accurate the reproduction of the signal, just like in optics, the closer you are to the optical axis, the lower the aberrations. Very large Xmax drivers may be great for spl, but not necessarily for 'starting and stopping on a dime' like a 2226H.... Then again, not apples/apples, and different freq ranges with differing excursion needs, and......you can go round and round on this topic. Just too many variables. In general, spread the load (be it with a horn/waveguide or multiples) and don't make a driver work too hard...



JSS
 

·
Registered
Joined
·
6,980 Posts

Quote:
Originally Posted by LTD02 /forum/post/21473103


"Aaaack! Is even the DIY community still confused on this one?"


the corner behavior isn't the focus. the focus is how long it takes for the driver to reach the amplitude of the input signal.


here is a 40hz sine wave (top) and the same 40hz sine wave low passed 2nd order at ~80hz.


you can see that all the high frequencies required to make the corner are gone, but the amplitude of the wave rises to its full peak at the first crest. it does not require several cycles to "build up" amplitude.


Thanks for the graphs, that's 1/4 of the way there. Now include the stop of the waveform (sine wave x step function of t=3 periods) and we're 1/2 way there.


Next add a 30Hz 2nd order high pass with Q=0.7 and then overlay that with a 30Hz 4th order Butterworth high pass. Now shift the Fc of the high pass to 15Hz and compare...


If you keep playing and compare different Q values at the same frequency, you can create various behaviors, and they all follow the response shape of the driver and box tuning. Any driver with a Q of 0.7 and Fb of 30Hz and the same high frequency extension will give the same waveform for this test, no matter how heavy/light, weak/strong the driver is. What changes with those variables is box size to achieve the alignment and the Voltage required to produce the same level.
 

·
Registered
Joined
·
904 Posts

Quote:
Originally Posted by Mark Seaton /forum/post/21473076


While the signals do have different rise and settling behavior, the above author completely missed the WHY. The common error made here in comparing to the electrical input signal greatly confuses the issue. While the comparison does make a case for differing bandwidth and order designs, there is zero info that should infer mass/inertia or motor strength are to blame. This is simply bandwidth and roll off qualities seen in any Fourier decomposition.

The blue reference signal is not what we should compare to. We should look at the input signal with at least a comparable low pass filter that mimics the high frequency response of the woofer. Even better is to electronically low pass the driver under test within its pass band as it might be used in practice so the roll off and associated behavior is known. The woofer is not a full range speaker, and is not supposed to perfectly match the displayed waveform unless it extends past 10kHz and is used that high.


FACT: You cannot create the sharp start/stop shape at each end of the blue waveform without extended high frequency content. Right off the top, part of the softening of each end is the lack of high frequency content.


Quite to the contrary of what is asserted, you could add some mass to the RS225 woofer, use a larger sealed box resulting in the same Qb, and you would BETTER follow the input signal.
Quote:
Any driver with a Q of 0.7 and Fb of 30Hz and the same high frequency extension will give the same waveform for this test, no matter how heavy/light, weak/strong the driver is. What changes with those variables is box size to achieve the alignment and the Voltage required to produce the same level.

I agree wholeheartedly. IF the objective was the attempt at tracing the input signal you would need a single loudspeaker capable of a huge bandwidth (super low, maybe even negative, inductance) in order to do so. Well put Mark.
 

·
Banned
Joined
·
15,491 Posts
The main problem with quantifying 'fast' versus 'slow' bass is that the cues that our ear/brain use to make that determination don't come from the frequencies below 80 Hz, they come from the frequencies from roughly 80 to 500Hz. So yes, transient response is very important, but not the transient response of the subs, but rather the transient response of the midbasses. Earl Geddes would have a field day over this one.
 

·
Premium Member
Joined
·
24,846 Posts
Somewhere in the archives is a thread that examined this verycomprehensively; I wondered how a vented box good work very well, as it's a spring/mass system that takes a number of cycles to pump up to full amplitude.


Illka provided some very interesting analyses and plots that showed the acoustic outputs for the first few cycles of sealed and vented systems.
 

·
Registered
Joined
·
21,345 Posts
Discussion Starter · #14 ·
i'm using audacity for this analysis and it isn't quite as flexible as we would like, but this is the behavior with two high passes (essentially a second order high pass) added to the low passed response from the previous graphic. top graphic no eq. bottom graphic high and low passed.


mark identified correctly that the high pass leads to the amplitude build up that is shown in the op. since these are theoretical/electrical signals and they pretty much match the behavior of the drivers in the op, it seems reasonably safe to conclude that this is the dominating effect.


i don't have any intuition for fourier transforms, so this behavior seems quite alien to me. it is what it is though.


thanks mark for weighing in on this one.


 

·
Registered
Joined
·
1,536 Posts
Fascinating stuff, i learned something new today
Once again what seems logical is not necessarily the truth.
 

·
Registered
Joined
·
919 Posts
One error to avoid in stuff like this. "A sine wave that starts at..." is most definitely NOT a sine wave. Sine waves don't start or end. Don't be fooled into thinking that using a sinusoid shape in a part of the waveshape limits you to effects near the sinusoid's frequency or period. This is maybe a gated sinewave, which contains energy over a much wider bandwidth than a sinewave (which has ideally, none), so you can't make conclusions about characteristics of just the highpass natures of the types of boxes.
 

·
Registered
Joined
·
5,545 Posts
Great posts Mark.





Quote:
Originally Posted by noah katz /forum/post/21474780


Somewhere in the archives is a thread that examined this verycomprehensively; I wondered how a vented box good work very well, as it's a spring/mass system that takes a number of cycles to pump up to full amplitude.


Illka provided some very interesting analyses and plots that showed the acoustic outputs for the first few cycles of sealed and vented systems.

I seem to remember that one. No idea what it was called or when though. It's been a few years at least.
 

·
Registered
Joined
·
405 Posts
What Mark said.


Hate to be a repetitive pedant, but people really should read up on linear systems, particularly the fundamental frequency-time duality, if they really care to understand this stuff. Largely, if you know the FR, you can infer the time-domain response, and vice versa. Raw intuition can be misleading, sometimes grossly so. And what's commonly "known" can be wrong.


And one other thing. Practically, all of this is likely swamped out by what the system does in the room at the listening position, which is what really matters. As long as a given woofer can be cleanly driven to excursion and in-room response properly eq'd, alignment doesn't really play. Many ways to skin the cat: cubic inches, cubic dollars, cubic sweat...
 
1 - 20 of 54 Posts
Top