George (on Topic Only)

I'm upset......................................................Butt.


And so my post doesn't get erased:

I have found that Vas can get you a general idea of how large the enclosure a subwoofer will need. But my experiences are somewhat lacking in comparison to many others here, so I thought I would see if they would agree. Do you guys use Qes instead, or EBP? Have you found Vas being way to inacurate to be able to predict enclosure size? I know I have seen a couple examples where it was inacurate.

Now I know that you can trow it into WinISD and get a much more accurate idea. But I'm just talking about getting a very general idea just by a quick glance at the numbers.

 

Within discussions about distortion levels of woofers while comparing how stable the motor strength stays, I have seen lately Dan Wiggins mention recently that it isn't the actual THD number that is important, but how it corrisponds to IMD IIRC. So my guess is that two drivers can have the same THD numbers at a certain frequency, but one may have more audible distortion? Can someone elaborate on that?

 

How about a general description of the T/S parameters. You know, what they mean, how they are derived, and how to use them properly. I myself know jack Sh*t about them really and I never really look at measurements anyways. Perhaps if I knew more, I would look at them, but for ME, ultimately, it just comes down to my ear. I don't care if it measures up like a VW, as long as it sounds like a Bentley.

Just a suggestion for those of us that know alot, but not enough.

 

Vb is proportional to Vas (double Vas, double Vb). Vb is proportional to Qes squared (halve Qes, quarter Vb). This is for the same Qtc/alignment... And since Qes is inversely proportional to BL squared, if you double BL you cut Qes by a factor of 4, meaning the box volume for a given alignment is cut by a factor of 8.

Now, that is all well and good, but consider what Fb does in relation to Qts/Qes... Fb is inversely proportional to Qts, thus as you raise the BL, and lower the Qts, you raise the Fb accordingly. Double BL, cut Qts by a factor of 4, meaning Fb just jumped up 2 octaves!

As far as subjective quality of THD/IMD, it's more the spectrum that matters. If the THD generated was all second order, it's not too objectionable (audible, yes - objectionable, no). It's when you get higher order THD and IMD components that things get nasty. In fact, you can measure a THD of 10% and have most people prefer it to a THD of 1%, based upon the spectrum of the distortion products. Dr. Earl Geddes has been working on subjective evaluation of THD/IMD spectrum - I look forward to talking at length with him at the ALMA meeting prior to CES...

Dan Wiggins
Adire Audio

 

Tirefyr,

There are only 6 fundamental physical parameters of a driver:

Re - the DC resistance of the voice coil (often called DCR). This includes the voice coil and tinsel leads. Measured in Ohms.

Le - the inductance of the voice coil. Usually measured in millihenries, sometimes microhenries.

Sd - surface area of the cone, usually measured in square centimeters.

Cms - compliance/stiffness of the suspension, usually measured in mm/N. That is, the suspension will sag X millimeters per Newton of force (there are ~4.5 Newtons per pound).

Mms - moving mass of the driver. Typically measured in grams.

BL - motor force factor of the driver. Measured in Tm (Tesla-meters) or N/A (Newtons per Ampere). I prefer the latter, because it is much more explicit about the force factor - X Newtons of force for every Amp of current you put through the driver (note that 1 Tm = 1 N/A).

Xmax isn't really a physical parameter of the driver - it's a "side effect" of a given driver design. Note that you can have the same 6 physical parameters, but Xmax values all over the place. And Xmax does not affect any of the derived T/S parameters.

T/S parameters are typically:

Fs: resonant frequency, set by the moving mass and compliance.
Qms: loss of the compliance, set by the Rms (mechanical impedance) of the suspension.
Qes: loss of the motor. Set by Mms, BL, Re, and Cms.
Qts: Q of the driver - combinatiom of Qms and Qes.
Vas: the stiffness of the driver, scaled by the cone size (Cms and Sd).

Dan Wiggins
Adire Audio

 

Interesting. I remember you saying once that even order distortion isn't as bad as odd order distortion. IIRC wasn't it because even order distortion shows up only at certain frequencies? I would be very curious to see Dr. Geddes's work as well. Is there a link to it that you could provide? Or is it available in book form?

Also, how does all this relate to your parameter stability with excursion white paper? In other words, which parameters create what type of distortion and how much? I know Fs changes effect frequency response, so would that be an example of even order distortion?

One thing I also noted in the white paper is that most parameters are effected by the Kms linearity. In fact only Spl seems to be effected by BL linearity. Yet Kipples research shows that 70% of distortion comes from BL nonlinearities, and only about 30% comes from loss of Cms. Now I know that any parameter change will effect the other parameters. So loss of BL will effect not only SPL, but Fb as you described above, as well as Qts, etc.

One possibility I can think of is that BL changes on the curve are much more magnified in what we hear than Kms changes. In other words a slight variance in BL will be much more audible than the same variance in Kms. Is this true?

I hope that made sense. I'm just thinking out loud, hoping some people will be able to correct my wrong thinking or fill in the blanks.

 

Stephen,

BL nonlinearities tend to be worse because they scale with power (BL is measured in Newtons per Ampere - the more power, the more current, thus the BL nonlinearity increases). Additionally, BL nonlinearities tend to introduce lots of odd and even distortion products, as well as IMD products; Cms nonlinearities tend to introduce even order products only.

As far as parameter stability goes, you can have perfectly stable parameters and still have high distortion products. In other words, if the Cms and BL curves were complementary such that the T/S parameter changes were eliminated, you'd still have increasing distortion from the curvature of each parameter.

Seth,

You are correct about the material really not affecting the inductance. Inductance is much more about the physical construction of the voice coil and motor rather than the voice coil metal.

Your point about Mms and Cms affecting Fs are exactly correct; however, note that as Mms is added, driver efficiency goes down (efficiency is, at its root, just BL and Mms). Thus you want to be as light as possible, but within reason. If you're too light, Fs will be too high, unless you loosen the suspension. Of course, Vas then skyrockets and so does box size for a given alignment...

As in all things, moderation is the key! You can have too little or too much moving mass, likewise with BL, Cms, and the other physical parameters.

Dan Wiggins
Adire Audio

 

I am just waiting to see what direction we are heading into..
I've got some thoughts on the history of these specs...

That is, we used to be dependent on T/S specs.
The flaw with them is that they are small-signal analysis. That means, tiny wattage. So little, the speaker doesn't move... essentially measuring the speaker's specs at 0mm of excursion, aka it's "at rest" position.

Things change dramaticallly, however, with the speaker moving... because it's oscillating between the limits of it's excursion, across it's "at rest" position quickly... many times per second.
Hz is "cycles per second".. makes it easy to referenc how many times per second.

But, since a voice coil is by nature a coil of wire suspended in a magnetic gap, it makes sense that as the cone moves in and out, the voice coil is moving up and down in that magnetic field.. if it moves far enough, the coils will start to leave that magnetic field.
Also, the suspension is getting stretched if you will.
With more power in the voice coil, it heats up.
All this leads to parameters shifting from their small-wattage, at-rest-position specs.

Today, we have new measurement techniques that are starting to analyze speakers not just "at rest", but also with respect to excursion, measurement systems like DUMAX and Klippel.

That's important not just for what they are....
But also because it's promoting a greater awareness of how excursion factors in, how limited traditional loudspeakers have been, and what's been revealed by these excursion-considering measurement systems of some old classic loudspeaker designs.
Recently, driver design has been largely focused on increasing excursion - and specifically increasing excursion without shifting the specs as these higher excursions are reached.

 

I think we are still dependant on those parameters. It's just that now we have an almost 3 dimentional view of them. What's really nice is that with new technology within the motors (XBL2, W7's motor), they are much more accurate beyond the 1 watt measurement.

 

Stephen... I agree 100%. The T/S concepts are all good, it's just that they fell short of "real world performance" because they were simply 'at rest' measurements. Now, measuring them across the excursion range is certainly a step towards the goal of predicting "real world performance". B)

Seth...
We can discuss motors up and down... in basic form, just repeating the old trusty "Loudspeaker Design Cookbook" (by Vance Dickason... highly recommended), as regards overhung and underhung designs. XBL^2 is a departure from that... particularly in multi-gap form (more than 2). The "inverse XBL^2" design (such as TC is doing, but is apparently also an XBL^2 patent item) is a similar departure, but there are more, way-out-there motor designs that are really the sort of thing that get me giddy... things like servo-motors, electrostats/ribbons/planars, and plasma drivers (sound producing hi-fi flamethrowers... how odd is that?

 

Dan... I'd also love an [even simplified] description of the Klippel system, how it works, what it measures... and how it compares/contrasts to the DUMAX results. B)

I did have the honor and privilege of visiting DLC labs, where we got to see and play with some [at that point in time] pre-production prototype units, meet David Clark in person, see some of his inventions, including the DUMAX machine, where we actually did personally DUMAX a few drivers... including one of my 15" XBL^2-enabled prototypes, and an Adire Koda 6.5" driver.

What's interesting about the DUMAX measurement system is how simple it really is...
Yet how high-tech, and state-of-the art at the same time.

I can describe the machine very simply.
It's a vacuum chamber. B)

OK, the high tech part is this:
It's got a measurement laser mounted inside the chamber, which simply monitors the position of the motor.
It's got a measurement laser mounted on a bridge, that's suspended over the cone, after the driver is mounted up, which tracks the cone position.
And it's got a wire to attach to the voice coils, to take measurements by, and to stimulate the driver into resonance. However, the power input into the driver is at all times very minimal... small-signal analysis, still.
The wire feeds into a computer (which also monitors the lasers) - a simple PC - which has some custom software written for the testing, to interface these devices.

You mount the driver into the vacuum chamber, and the chamber is either pressurized to manipulate the cone outwards, or the chamber can be depressurized to manipulate the cone inwards.

During the test, the subwoofer is driven to it's resonance naturally - and almost mysteriously (even David Clark couldn't explain to us exactly how it did it.. it involved both a small electrical signal and vacuum manipulations), to obtain Fs.
The subwoofer is then run through an impedance sweep, Linear-X "LMS" style, to obtain the traditional T/S parameters.

The BL and Kms curves are the most interesting measurements produced though.
To obtain these, the vacuum chamber manipulates the cone into position, forward and backward, the front laser tracking the position of the cone, and the rear laser correcting for any movement of the entire driver - isolating out truly just cone position.
As I recall, the Kms measurement is taken as the amount of pressure required to move the cone to a certain position.
And the BL measurement is taken by manipulating the cone to a certain position, applying a small DC voltage to the voice coil, and monitoring the resultant movement, with respect to (ie. factoring out) the Kms measured previously.

But hey, why listen to my description of DUMAX, when there's a great description by inventor David Clark himself. B)
And here's a good link for other DUMAX information and pictures of the machine in operation.

 

I thought they used microphones to get their measurements within the DUMAX machine.


One question though, can the machine give differing readings sometimes even while it measures the same driver? If so, what causes this? I heard it was due to external noises in the room interfering with the microphone.

 

No microphones at all...
But...
Consider this:
What's the difference between an electrical motor, and a generator?
Nothing, really, right? Well then...
What's the difference between a loudspeaker, and a microphone?
Ahhhhhh...

Some corruption can definitely be had from noises in the room... but it doesn't involve any microphones in the measurement (there aren't any)...rather, the driver itself picking up the noises, and the resulting signal being picked up by the computer... that's one possibility, but that's also one rather quiet facility. Off the beaten path, and with acoustical considerations made, internally.

I've heard of several times where subsequent DUMAX tests have yielded inconsistent results. Stephen Ponte commented to us that there is in fact a bit of operator skill involved in setting up the machine and driver, and even in the running of the partially-automated tests, to yield good results.
There are factors that can lead to the inconsistencies...
But to blame them on "noises in the room" doesn't seem impossible... but maybe less likely than other potential factors.

 

A long time ago I mentioned to Stephen Ponte that it would be nice to have a modeling program that we could input the DUMAX report, and get accurate, high level modeling of the driver. That would give you a great deal of information from distortion levels at different outputs, highly accurate output levels with a given input since BL compression can be included in the estimate, and anechoic frequency response of the system at different output levels. He said it would never happen since many companies don't release their results, which is too bad. But I still think it would be nice.

Now do you think it would work for a company to utilize Adire's measurement system, and write their own program as mentioned above? They could start buying speakers from different companies (or have people send them in) and test them; then input the results into the program. I can see them being able to sell that program for a great deal of money, and I think people would buy it due to it's superior accuracy.

What do you think? Would there be any legal issues with this?

 

Wow, very good point. I am slapping my head as we speak for not thinking of that.


So differences between drivers with reguards to suspension stiffness, cone area, and Mmd can actually be effected more or less by this, couldn't they? :detective:

 

I HAVE A TECH QUESTION! WOOHOO.

how is it that you can have increasing excusion that doesnt change (or barely changes) the suspension stiffness? isnt that like trying to divide by zero? i guess the thing that baffles me most about this stuff is actually the spider and the cms/kms curves.

any in depth analysis on this stuff?

eric

 

one more thing to add...addressed to dan specifically, but anyone who knows can answer. how does the buckle system work? i know they cut the dust cap off, but beyond that, i understand nothing.

eric