Thanks for the reply! A bit of an unexpected answer (to me), but certainly not the first time someone's used Excel for fairly heavy-duty engineering calculations.
My personal life is Microsoft-free (and also Apple-free), and has been that way since 2001, so if I ever find the time and interest to pursue my old MFB speaker research, I guess I'll have to find another software approach. There is probably a suitable open-source computing language option out there somewhere (one possibility is Gnu Octave, a Matlab-alike.)
-Gnobuddy
It depends a bit on what you see as a problem - a soft suspension means the equivalent volume of air (Vas) will be very large. This is even more true if the piston area (cone area) is large - a large piston compresses more air for the same piston excursion, so to keep the air-spring equally soft, you need an even bigger enclosure.
So that means the "light cone, soft suspension, large cone area, low resonance frequency" driver is likely to need an unacceptably large enclosure volume by todays standards.
In this era of gobs of cheap, clean, audio power, it's a much more viable option to make a heavy (and yes, less efficient) speaker, because it will allow a smaller-volume enclosure to be used with it.
A couple of the 15" woofers I looked at on the Parts Express website had sensitivities around 91 dB per watt at 1 metre, so efficiency is not too shabby, heavy cones and all.
One would think the smaller Vas is less important for dance-club installations, but I think even there, there is now less tolerance for having to deal with 50-kilo speaker enclosures the size of small refrigerators.
I remember a Carver subwoofer from the late 1990s that had a 10" woofer, and a 10" passive radiator, mounted on opposite sides of a cubical enclosure that was only 11"x11"x11" - less than one cubic foot on the outside, not to mention it contained a power amplifier inside taking up even more air volume.
Both the woofer and passive radiator were almost ridiculously heavy, resulting in a pretty low resonance frequency - claimed to be 18 Hz - as installed in the tiny box.
Carver's trademark too-much-power-is-just-enough approach completed the design, with 2700 watts (!!) of built-in audio power, if you believe the manufacturer's claims.
This "irresistible force meets immovable object" design supposedly resulted in deep, loud, bass from an almost absurdly small speaker system.
I found a little more information on this bizarre beast online, here: Welcome Secrets of Home Theater and High Fidelity
-Gnobuddy
No quibbling about your nicely presented analysis.
Ordinarily I too would emphasize the cheapness of power today (along with buying 30 yr old amps at the Salvation Army store as the perfect sound quality match for the best speakers today). But with very low freq EQ, the demands of power escalate to surprisingly high values by the magic of db arithmetic.
Yes, some of those modern drivers have heavy cones, high Bl motors, and are efficient. Isn't that because they have oodles of wire in the magnetic gap?
Nobody says the heavy-weight Corvette is a shabby performer, but I'm still lusting for another light and nimble Lotus (in addition to the aesthetics of the engineering concept, wastes a lot less gas).
Ben
Hi,
mass as such is not negative in any way.
In fact, without sufficient mass there´d be not enough low-bass at all.
Now one can argue about how low in frequency a woofer needs to play, but if You want real deep bass mass is inevitable.
Unfortunately drivers after the car-boom-box school typically feature too small and too stiff (and as such too progressive) spiders and very lossy surrounds, which imho results in that terrible bass quality and horrible inefficiency that requires 100s of watts before the driver gets his *** up and going.
Drivers like the upper TCSound range for example feature larger spiders which are much softer and responsive at low levels.
Those respond to small music power levels, resulting in fine resolution, high-precision low-bass in very small cabinets.
If required they can take some wattage abuse and can deliver very high levels of clean low- an lowest bass.
Before that experience I´d fully agreed with Ben about making bass simply big
jauu
Calvin
mass as such is not negative in any way.
In fact, without sufficient mass there´d be not enough low-bass at all.
Now one can argue about how low in frequency a woofer needs to play, but if You want real deep bass mass is inevitable.
Unfortunately drivers after the car-boom-box school typically feature too small and too stiff (and as such too progressive) spiders and very lossy surrounds, which imho results in that terrible bass quality and horrible inefficiency that requires 100s of watts before the driver gets his *** up and going.
Drivers like the upper TCSound range for example feature larger spiders which are much softer and responsive at low levels.
Those respond to small music power levels, resulting in fine resolution, high-precision low-bass in very small cabinets.
If required they can take some wattage abuse and can deliver very high levels of clean low- an lowest bass.
Before that experience I´d fully agreed with Ben about making bass simply big
jauu
Calvin
I beg to differ. Sound output is determined by displacement volume as the only parameter (for sealed box). Voltage based sensitivity is not important at all, only power efficiency is, and that is best with the strongest motor, softest supension and lightest cone (in order of importance). Amp need lots of voltage headroom when massive pre-EQ is required but with class-D this is not an efficiency penalty. Average current demand will be low and total system efficiency is good. It will become stellar when you place the system resonance in the middle of a (rather wide) passband, so of course too light a cone (or too stiff a suspension) might not be fully optimum, efficiency-wise, when your passband is the very lowest ranges only and the system resonance is outside (too high).
Last edited:
Sorry to quote in full.
Yes, odd for Calvin to say something obviously wrong about cone weight without explaining how he is looking at the issue. Bet it has something to do with sim's, just like the way the dubious "ideal" Q is derived.
But KSTR brings in another issue. And it is related to figuring out how to assess MF systems for overshoot and resonance correction (other than by posting nice 'scope pictures: how much power is needed to correct overshoot and how to conceptualize the process.
Lots of talk in other threads of the power needed when you EQ very low tones; when you do the basic db math, always looks like you need 10,000 watts. But as KSTR posts, what level of power is needed for MF correction and how do you know when you have enough moxie to kill a satisfying amount of overshoot and undershoot and ringing?
Ben
Last edited:
For non-MFB speakers there is sort of an optimum theoretical Q, which is when it yields close to aperiodic damping of cone excursion, so about Q=0.6. The reason is that this is the Q at work for any "external" exitation of the cone, including the driver's own distortion and we sure don't want to have ringing induced by any error signal. Vastly overdamped isn't ideal either because then any re-centering after a dynamic DC-offset incident takes longer than necessary.
Thanks for the spreadsheet - Excel is incredibly powerful & the more one learns about it the more one realizes how almost limitless it is.
I dont know if this link below (yes, it is long - just copy and paste) is going to work but I have tried to recreate at least a portion of the Sony MFB circuit (no where near as detailed and elegant as Steph-tsf's simulation but condensed down to something I think I can sort of understand). The link below should be the simulation, shown in this pic is what is should look like in the browser.
I am not sure if the ground I have (lower left next to 100 Hz input) that goes to Pin 3 (A) of main board is ground or something else. If someone wants to check and make corrections and repost new link that would be great.
The lower left 100 Hz input signal is what I think would be coming from the 0.22 Ohm MFB resistor based on my previous scope measurements (ball park) - I was thinking still it may be possible to pipe in the piezo output here with the goal of getting deeper and tighter bass.
The upper right 100 Hz signal is what I think would be coming in from the preamp, before being combined with the MFB signal.
Right clicking on the components allows one to change values, phase, etc. The virtual scope on the bottom can be removed and different parts of the circuit can be view on the scope just by right clicking on the circuit pathway wherever you would like to observe.
Thoughts, corrections & critiques invited.
NOTE: WORKS BEST IN CHROME, IE tends to crash.
SONY MFB SUBWOOFER SUBCIRCUIT
http://tinyurl.com/hmu7ev2
Last edited:
I do that too - most recent find was a Yamaha RX-360 receiver for $15 or $20, I forget which. It would have cost me twice as much to buy a pig-pink silicone cellphone-sheath with the Holy Emblem Of Steve Job imprinted on it.
And for having mentioned thrift-store receivers on diyAudio, you and I may both end up tarred, feathered, and run out of town on a rail!
They probably also have more magnetic flux in the gap - deeper pole-pieces and/or a stronger magnet.
I never had a Lotus, but day-dreamed about owning a Lotus Seven for years.
The analogy isn't exactly right, though. Both the Corvette and the Lotus have tires made of the same material (rubber). It turns out that rubber tires produce less lateral acceleration when you put more weight on them, all else being the same. (The grip goes up with weight, but it goes up by a smaller percentage than the weight increase, so the actual lateral G's decrease as the tire is loaded more heavily.) So the Lotus really does have an advantage in handling.
The speaker mass/damping situation is a bit different. Imagine a super-light cone, damped to achieve a Qts of 0.4. It's the equivalent of a butterfly's wing, damped only by the thin air around it.
Now imagine a big heavy speaker cone, also damped to achieve the same Qts of 0.4. This time the equivalent is a mammoth caught in a tar-pit; both the mass and damping went up enormously, in such a way that the big mass is controlled to exactly the same degree as the light one was. (But we had to use very viscous tar to achieve that same degree of control.)
The point being, Newton's equations of motion tell you that the big heavy cone and the light feathery one will produce exactly the same transient response, if both have the same Qts and resonant frequency.
But - it takes a lot more power to drive those mammoth mammoth limbs through the tar. The mammoth gets very poor fuel mileage compared to the butterfly.
Light cars with decent power are wonderful things!
I owned a '73 Datsun 240Z for a while. It was old and had been badly mistreated by previous owners. I fixed all the mechanical stuff, and upgraded the steering, brakes, suspension, wheels, and tires. It was still cosmetically rough, and loud and noisy inside - but also the most fun car I ever owned. I hated having to give it up.
-Gnobuddy
I was surprised when I read this article a while back, as their test results were counterintuitve to my thinking. Here they demonstrate that adding mass to a woofer does NOT reduce its speed in an impulse test. Efficiency goes down but not the 'speed' of the woofer.
http://www.adireaudio.com/Files/WooferSpeed.pdf
http://www.adireaudio.com/Files/WooferSpeed.pdf
Hi,
Oh well Ben .. regarding Your own special 'trueth' its rather funny to see You accuse others of beeing 'obviously wrong'.
Well my remarks hold true for non-equalized speakers ... I didn't point that out explicitely.
Equalization and motional feedback can certainly stretch the limits ... but not until ultimo.
As KSTR pointed out correctly the equalization requires more amp power.
This in turn asks for larger dimensioned, heavier coils that can take the power.
This in turn asks for larger magnetic gaps which then require bigger stronger magnets ... and so on and so on.
This simply means that in praxis certain limitations are quickly reached that can't be pushed any further, or things become uneconomical.
In other words, if You want deep bass in realistically compact casings high moving mass is helpful.
As KSTR also said, it'd be good -efficiency wise- to put the base resonance in the middle of the pass band.
With too lightweight cones that resonance could end up at the top end or even above the passband (known over here as URPS ... Under Resonance Principle Speaker).
The equalizer must then counter the -12dB/oct drop below the resonce, eventually over the full passband.
Power requirements shoot up, demanding beefy and heavy coils.
Also the rather tiny cabinet volumes ask for very stiff cones that come at the cost of more mass.
So moving mass is not good or bad in itself for bass quality, but just another parameter.
If large sized cabinets aren't feasable, then adding mass to the moving parts is rather a must than an valuable option, regardless of the use of additional equalization or mfb.
jauu
Calvin
Oh well Ben .. regarding Your own special 'trueth' its rather funny to see You accuse others of beeing 'obviously wrong'.
Well my remarks hold true for non-equalized speakers ... I didn't point that out explicitely.
Equalization and motional feedback can certainly stretch the limits ... but not until ultimo.
As KSTR pointed out correctly the equalization requires more amp power.
This in turn asks for larger dimensioned, heavier coils that can take the power.
This in turn asks for larger magnetic gaps which then require bigger stronger magnets ... and so on and so on.
This simply means that in praxis certain limitations are quickly reached that can't be pushed any further, or things become uneconomical.
In other words, if You want deep bass in realistically compact casings high moving mass is helpful.
As KSTR also said, it'd be good -efficiency wise- to put the base resonance in the middle of the pass band.
With too lightweight cones that resonance could end up at the top end or even above the passband (known over here as URPS ... Under Resonance Principle Speaker).
The equalizer must then counter the -12dB/oct drop below the resonce, eventually over the full passband.
Power requirements shoot up, demanding beefy and heavy coils.
Also the rather tiny cabinet volumes ask for very stiff cones that come at the cost of more mass.
So moving mass is not good or bad in itself for bass quality, but just another parameter.
If large sized cabinets aren't feasable, then adding mass to the moving parts is rather a must than an valuable option, regardless of the use of additional equalization or mfb.
jauu
Calvin
Last edited:
I smell a very large rat...
I have no time to do more than a quick skim of the article right now, but the equation for acceleration is: a = (BL/m) * i
Which means that, yes, the acceleration is proportional to "i" (the voice coil current), as stated in the article. But the acceleration is also proportional to (BL/m)!
So, as you can see from the equation, if you double the moving mass, you halve the acceleration you get for the same voice coil current. Triple the mass, get one third the acceleration. And so on.
Now, this result applies when (a) the speaker is not mounted in an enclosure, and (b) the frequency being applied is well above the fundamental resonance. (Below resonance, the mechanical and electromagnetic damping also slows down the cone, not just it's mass.)
I don't yet know how that paper came to that conclusion, and exactly what their measured data is showing. But, like you, I am skeptical of the correctness of the conclusion.
If I can find some free time one of these days, I'll read through that paper and see if I can figure out what other problems it has (besides leaving out the BL/m proportionality constant in the equation for acceleration).
-Gnobuddy
Hi,
as long as the ´required´ acceleration -dictated by the input signal- is lower than the possible acceleration the driver is capable to, there´s nothing to worry about.
Which means that the signal frequency only must stay below the upper bandwidth limit of the driver.
If You´d ask Usain Bolt to pass a distance of 100m in exactly one minute he´d be able to do so regardles of him going naked or transporting a 50L keg of beer.
He´d probabely be happier though with the beer afterwards ;-)
jauu
Calvin
as long as the ´required´ acceleration -dictated by the input signal- is lower than the possible acceleration the driver is capable to, there´s nothing to worry about.
Which means that the signal frequency only must stay below the upper bandwidth limit of the driver.
If You´d ask Usain Bolt to pass a distance of 100m in exactly one minute he´d be able to do so regardles of him going naked or transporting a 50L keg of beer.
He´d probabely be happier though with the beer afterwards ;-)
jauu
Calvin
I think that is *not* what I said. I said more voltage headroom is required with an overdamped driver but average current is lower (peak currents can by high, though). In-box impedance curve is telling. Best for efficiency (with normal music material) is a wide impedance peak centered in the pass band. The wider the better.
Agreed, but how did our anonymous friend at Adire Audio set up his/her experiment in order to produce the same acceleration from two woofers of different Mms?
If you applied the same impulse testing signal - say, a 1-volt step function - to both drivers, then Newton's laws of motion tell you that the heavier driver will most definitely be slower than the lighter one.
As you said, Usain Bolt would require extra incentive to get to the end of the 100m strip in 1 minute, with 50 kg strapped to his back. (I'll assume that beer has the same density as water, and that the keg itself is weightless.
)Same thing with the subwoofer - if I double Mms, I would need to also double the current in the voice coil to create the same acceleration. So I would have to go out of my way to apply a 1-volt step function to the light driver, and a twice-as-big 2-volt step function to the heavy driver, and then use the resulting measurements to "prove" that cone mass has zero effect on cone dynamics.
I would think that a person would either have to be quite confused (ignorant) - or quite deceitful - to apply different-sized step voltages to two drivers that he was trying to prove were identical. So, it seems that either the experiment was rigged to prove the (wrong) point, or the measurement was mis-understood, and was performed incorrectly.
Perhaps Adire Audio's point is that if you have to travel 100m hanging upside down from a glass-plate ceiling, then, an ordinary garden snail would be faster than Usain Bolt?
(Incidentally, just like Usain Bolt, I'm told that snails and slugs like beer too. Gardeners used to trap slugs by setting out a shallow bowl of beer for them. They would crawl in, get drunk, and drown.
)-Gnobuddy
Added mass, as long as it is attached rigidly to the point of force injection (voice coil), does not change the high frequency range in terms of bandwith / shape of response. It changes the LF behaviour, obviously, and it changes efficiency, also obviously. Glue a ring of lead to the joint of former and cone with epoxy and you will see absolutely nothing changes at frequencies above the mass corner except an overall level decrease. Current induced distortion in a lesser motor may go up, though, as more current is needed to get the same SPL.
So far, we've had a battle of the sims and an examination of frequency response in relation to weight. And I think it is fair to say, some of the dispute is just a matter of points of view (starting with Calvin's), not contradictory views of Newton's laws.
But, as with other peculiar discussions on this forum, FR is not the whole story. For example, there is the euphemism, "group delay" which normal people call boom.
I think we can agree, that the Lotus and numerous muscle cars can go 175 miles and hour. But as Lotus owners will rush to tell you, speed going in a straight line is a small and maybe cheesy aspect of driving pleasure.
So, can the posters offer thoughts on other aspects of sound quality for heavy versus light cone assemblies as it applies to MF?
Ben
Last edited:
Hi,
well, I should have pointed out more clearly that I assumed a properly low-pass filtered system, lets say for convinience a 100Hz upper bandwidth limit.
Your points are valid for non-LP-filtered conditions ... that´s like asking Usain to race the 100m as fast as he can.
In that case mass counts as it reduces the acceleration rate.
Under the LP-filtered condition it´s like telling him to be at the finish line in exactly 1min.
He could easily do that carrying considerable added mass.
Regarding drivers, there´s no difference in acceleration as the heavier driver´s acceleration capability is still vastly higher than required to accurately follow the signal.
A lowpass filtered step-signal is hardly a step anymore.
There´s mostly a difference in amplitude response.
Even though of course efficiency will drop with increasing mass, may the amplitude response show higher SPLs towards the low end, due to sinking Fs and rising Q.
Typically then the equalization costs on amplifier power.
jauu
Calvin
well, I should have pointed out more clearly that I assumed a properly low-pass filtered system, lets say for convinience a 100Hz upper bandwidth limit.
Your points are valid for non-LP-filtered conditions ... that´s like asking Usain to race the 100m as fast as he can.
In that case mass counts as it reduces the acceleration rate.
Under the LP-filtered condition it´s like telling him to be at the finish line in exactly 1min.
He could easily do that carrying considerable added mass.
Regarding drivers, there´s no difference in acceleration as the heavier driver´s acceleration capability is still vastly higher than required to accurately follow the signal.
A lowpass filtered step-signal is hardly a step anymore.
There´s mostly a difference in amplitude response.
Even though of course efficiency will drop with increasing mass, may the amplitude response show higher SPLs towards the low end, due to sinking Fs and rising Q.
Typically then the equalization costs on amplifier power.
jauu
Calvin
Agreed. If you take the same frequency response curve, and lower it a few decibels, that means that every signal amplitude (y-axis number) on the curve is a smaller number than before, i.e. the entire frequency response has been multiplied by a constant, and the constant is a number smaller than unity. If the first speaker has the frequency response f(x), the second has the frequency response a* f(x), where "a" is a positive number less than one.
Now take the inverse Fourier transform of both frequency responses, and you get the corresponding impulse responses. Since the inverse Fourier transform is a linear process, the inverse transform of the second one (mass loaded woofer) will have the same shape as the first, but will be multiplied by the same constant, smaller than unity.
Which means that every number representing the impulse curve is smaller (compared to the lighter woofer). In other words, the second, more massive woofer will indeed be "slower", as it has lower acceleration at every point on the frequency response curve. And this is an inevitable consequence of the lowered sensitivity.
If the BL product was increased to bring the sensitivity of the heavy woofer back up to be identical with the light one, then we do indeed get the same acceleration, same impulse response, same frequency response from both. Alternatively, if the heavier woofer was driven with more current, to raise the cone excursion until it was identical with the light woofer, we'd have the same result (identical impulse responses), but at the penalty of more input power for the heavy woofer.
So I still don't understand the Adire Audio white-paper's claims. In this case, both magnets are identical, as are both BL products. Only the moving mass is changed, and therefore, the sensitivity too.
So, unless the test involved driving the heavy woofer with a bigger signal voltage, it should be impossible to get the same impulse response from both - you can only get one of the same shape, but scaled down, from the heavier woofer!
-Gnobuddy