diyAudio logo
home  .  forums  .  OldWiki

Advanced driver design features

[Paul Spencer]

Sources of driver distortion

There are two basic types of distortion - linear distortion (abberations in the frequency, phase and time response) and nonlinear distortion where extra frequency components are added. Most commonly discussed are THD and IMD, and these are nonlinear distortions.

This article will focus on nonlinear distortion.

Sources of distortion include:


inductance linearity
Typically the value of inductance varies with frequency and excursion. According to Adire, inductance is the one parameter which gives the best indication of transient response. It influences the time taken for a speaker to begin to reproduce a transient impulse. Its value can be reduced with a shorting ring, as well as how much its value varies with stroke. B&W indicate on their whitepaper on the development of their 700 series that they have managed to get inductance almost perfectly flat with a combination of a shorting ring and a phase plug.

Subwoofer drivers may have a high inductance. Inductance increases with longer and larger diameter voice coils. Longer coils are required for high xmax, and larger diameter coils are required for high thermal power handling.

suspension linearity
The stiffness of the suspension varies with excursion. This is less of a problem with drivers like the Peerless XLS series in which there is a fairly large difference between xsus (suspension excursion limit) and xmax. However, most drivers with a similar suspension system to the XLS have a larger xmax, hence distortion based on suspension linearity is likely to be more of a problem. This linearity also applies to passive radiators. The ideal passive radiator or driver would have a perfectly flat suspension vs excursion curve.

BL linearity
As the VC moves out of the gap, the effective force of the motor system on the VC decreases. Higher excursions mean less motor strength to restore the driver to its rest position. An ideal driver has a completely flat BL vs excursion curve, with no loss of force over the excursion range used. Adire have created a motor system called XBL^2 which primarily address this source of distortion in bass drivers. It has two gaps which results in a shorter VC with lower inductance, and the motor system is far more linear in BL force with respect to excursion.

Flux modulation
Flux modulation is caused by the interaction between the permanent field generated by magnet and alternative field generated by current in the VC. It causes an increase in distortion based on power input. It shows up as strong odd order harmonic distortion, primarily 3rd and 5th order.

Where high efficiency can be achieved for a subwoofer, flux modulation will be reduced for a given output. However, high efficiency is rarely achieved below 40 Hz since this requires very large vented boxes, bass horns or bandpass enclosures. A large 18" PA driver will not normally be any more efficient than an 18" high excursion hifi driver at 20 Hz.

power compression
At high power, the VC heats up and the resistance changes and this changes the frequency response. Efficiency is reduced. Power compression can be reduced by driver design that includes better heat transfer and larger diameter VCs, but this will increase inductance. Typical high efficiency drivers have larger diameters VCs, lower power compression and higher power handling than hifi drivers. Horn loading decreases power compression by increasing efficiency while at the same time reducing current going in to the coil since horn loading increases electrical impedance.

Shorting rings


Shorting rings are also comonly referred to as Faraday rings. They reduce flux modluation and inductance. Distortion related to increased power input is reduced. You might also infer that transient response is improved by lowering inductance, both in its value and variance with respect to excursion.

Dan Wiggins was asked why he didn't use Faraday rings on XBL^2 drivers:

We do... The Extremis 6.8, the FR125S, and many other drivers use Faraday rings. If the driver design overall needs it, we'll add it. But for many cases, the improvements are subtle so we don't use them. It's all about what the specific driver design requires.

The flux modulation of the Tumults is pretty low; the BL�/Re is pretty high, up around 90, so the field is dominated by the gap flux, not the voice coil flux. Additionally, the rest of the motor is balanced to reduce flux modulation (there are techniques you can use to reduce flux modulation WITHOUT resorting to copper or aluminum shorting rings, and no I won't say what they are...)

this is a work in progress, please come again later!


[Dan Wiggins posts] (to be added)

Regarding distortion in general, Faraday rings reduce distortion by reducing flux modulation.
And reducing inductance modulation.

BL-based distortion is displacement based. Flux modulation is power-based. Take a driver in a ported box, operating at resonance. Very little stroke, so BL-based distortion is low. But the flux compression would be VERY high, because of all the power you can drop in at resonance.

"If so, why not use Faraday rings in the XBL drivers?"
We do... The Extremis 6.8, the FR125S, and many other drivers use Faraday rings. If the driver design overall needs it, we'll add it. But for many cases, the improvements are subtle so we don't use them. It's all about what the specific driver design requires.

The flux modulation of the Tumults is pretty low; the BL�/Re is pretty high, up around 90, so the field is dominated by the gap flux, not the voice coil flux. Additionally, the rest of the motor is balanced to reduce flux modulation (there are techniques you can use to reduce flux modulation WITHOUT resorting to copper or aluminum shorting rings, and no I won't say what they are...).

"What kind of distortion product would flux modulation show up as on measurements?
Can you comment on how it varies among your different speaker lines?"

It shows up as a strong odd-order harmonic (meaning very little even order distortion, pretty much just odd order, primarily 3rd and 5th) that increases/decreases with power, not stroke. So it's not necessarily linear with SPL; for example, below Fb of a sealed box, with EQ you can put a lot more power in, not fall as fast, but greatly increase the flux modulation distortion.

That's really why you don't see plots - or much talk - of flux modulation. It's a power effect, which means it may or may not occur for a given excursion level; it's in the hands of the operator completely. So design for minimal flux modulation at, say, 50-70% of rated power and call it "good enough"...

"So it seems that the distortion products as a % of output should be *less* for ported systems."
Kind of... It's all about how the distortion is generated. You can have really high Le-based distortion, and still have less THD than a driver with ZERO Le-based distortion, and moderate amounts of BL or Cms distortion...

Now, with flux modulation it comes from eddy currents in the pole (usually) cancelling some of the static flux from the magnet system. The eddy currents are a function of power into the voice coil AS WELL AS the coupling of the voice coil's magnetic field to the steel! So it does have a positional effect.

Imagine that you have 1T of flux in the pole, right below the gap (total flux in the system is 1T). Say your voice coil generates 0.5T of flux. Now, if your voice coil is covering the entire gap, then you have (1-0.5) 0.5T in the pole at minimum, and (1+0.5) 1.5T in the pole at maximum (minimum and maximum being the relative phase of the voice coil's magnetic field with respect to the static field, based upon how the voice coil is driven from the AC signal).

Now, move the voice coil up, so that it is covering only one quarter of the gap. The gap has 0.75T "unmolested", and 0.25T that will be modulated (and assuming an equal percentage of the voice coil is out of the gap region) by approximately 0.125T. So now we go not from 0.5T to 1.5T, but from 0.875T to 1.125T. A lot lower modulation.

So position will affect the flux modulation in the system. If we hold the voice coil at rest, we maximize the flux modulation. If we move the voice coil away from rest, we start to decrease flux modulation.

What does a ported box do? Among other things, it lowers excursion over an octave or two centered on the tuning frequency. And it also lowers IMPEDANCE at the tuning frequency as well, meaning more power delivered at resonance (as compared to a sealed box). It actually enhances the ability of the driver to flux modulate, based upon its operation, as compared to a sealed box!

HOWEVER, there is more to distortion than just flux modulation! There is BL, Cms, and Le nonlinearities with excursion, and often - for a given driver - it is better to take higher flux modulation and lower the other sources, by reducing excursion.

"But the ported driver will also have a lower xmax, therefore making it more susceptable to flux modulation."
BINGO.

For those familiar with flux modulation effects, you've no doubt seen that it is MAGNIFIED when the driver is in - voice coil towards the back. More steel/conductive material around the voice coil, better coupling for flux modulation. So the net effect is increased as you move backward, and reduced as you go forward, for the very reasons I gave.

Also note that impedance over the subwoofer frequencies tends to average higher for sealed boxes than ported boxes, because of the typically higher AND broader peak at resonance. Meaning that for a given applied voltage, less power is delivered to the voice coil, meaning lower flux modulation.

Ported systems can really reduce lots of distortion issues related to nonlinearities; I love them for that! However, they are more susceptible to other distortion mechanisms, like flux modulation, cabinet resonances, and cone reflections (higher pressures in the box being the issue). So it's a set of tradeoffs to make.

In general, I prefer to accept the downfalls of ported boxes, and overcome the issues with sealed boxes, but they both have definite driver-related issues that arise from the operation of the driver.

In general, sealed boxes have higher THD components from BL, Cms, and Le nonlinearities.

In general, ported boxes have higher THD components from flux modulation and greater thermal compression.

Choose your box, choose your poison...

...

For flux modulation, less power is just as important as it is for power compression. Also, you want the B field to stay as high as possible over as much of the stroke as possible. A BL domiated by static flux - not from the voice coil - is the key. So you want the flux to be as peaked as possible, over a wide a range as possible. And you want to use as few turns as possible, since that means for a given BL you're more B based, rather than L based.

Now, an overhung motor with a short gap is a great way to go; the flux in the gap is typically the highest possible of all topologies (given the same size magnet available). HOWEVER, it also has high flux modulation, because the voice coil must have lots of turns - length - to get decent stroke.

Underhung, or long-gap topologies aren't quite as good; flux density is lower in the gap, which means you are getting more of your BL from windings, not from static flux.

XBL� actually turns out quite well here. The flux in the gaps isn't quite as high as in a short-gap overhung, but the voice coil tends to have fewer turns, and more of the voice coil - percentage-wise - in one or both of the gaps. This means you actually have more of the BL dominated by the static flux, not the voice coil windings.

There are some other types, where split coils are used (a la Babb in 1973 and others claiming it's a "new" concept), or variable winding depths are used. However, both of those end up with even more turns, meaning greater susceptibility to flux modulation.
...
But remember, it's POWER based - current into the voice coil will create greater flux modulation. And look at the impedance at Fb for a ported box, compared to Fb for a sealed box.

In general, the average impedance below 80 Hz tends to be HIGHER for a sealed box than for a ported box, and that will affect the results.

You do get some advantage for ported boxes; it's hard to turn away from 4-6 dB more output! But you don't get all of that in terms of flux modulation; rather than a 70-100% reduction in flux modulation (which you might surmise from the extra output), you get a 20-30% reduction, and predominantly above and below Fb, at the impedance peaks.

"rather than a 70-100% reduction in flux modulation (which you might surmise from the extra output), you get a 20-30% reduction, and predominantly above and below Fb, at the impedance peaks."
OK, so it *is* less with ported, thanks.

At some frequencies, and if you average over the two impedance peaks, in general. But that is not always valid. You're looking for an "it's always better" type statement, and that is precisely what I am NOT giving.

For example, consider at Fb. Impedance is at a minimum! And, in fact, driver excursion is at a minimum as well, meaning the voice coil is pretty much fully containing the pole. You get your HIGHEST flux modulation at this point, 3-5 times HIGHER than a sealed box at the same frequency.

Of course, the total MEASURED distortion is lower, but that is because you're eliminating contributions from BL, Cms, and Le nonlinearities with respect to excursion.

So, in the case of running at tuning, you are actually maximizing your flux modulation distortion, while minimizing all the other distortions. The net total is lower, but specifically NOT because of lower flux modulation.

...
Is it to different things, though? Look at the average impedance from 20 Hz to 80 Hz of the following most-popular alignments (at least for our drivers):

- Brahma 12 in 1 cubic foot sealed
- Brahma 12 in 1.5 cubic feet tuned to 33 Hz

The average impedance of the sealed box is actually a bit higher when you look at the range between the impedance peaks of the ported box. Meaning for a given applied voltage, you have LESS current flowing, and therefore less flux modulation.

Of course, the ported box has more total output, and around Fb will have lower distortion because of the reduced excursion. But if we're talking about flux modulation - which is what I thought we were discussing - the flux modulation of the ported box, on average, will be higher.

If you're talking about distortion output versus SPL, then yes, a ported box - properly designed - will win. High Q isn't the way to go, though, you want lower Q with the port contributing over as wide a bandwidth as possible.

Lower the average impedance and you increase flux modulation for a given voltage. If the sensitivity of the system is high, then you will need less applied voltage for a given SPL level, so you may in fact end up with lower flux modulation. That is more of a side-effect of the alignment, though, than a direct cause.

Ported boxes reduce excursion AND reduce impedance around Fb, which both INCREASE flux modulation. It is only because you have higher sensitivity around Fb that you see a net decrease in the actual THD from flux modulation for a given SPL.

"Dan, realistically at levels where SQ is an issue (ie low power levels), power based flux modulation shouldn't be a big deal, correct? At higher levels where flux modulation would become a problem, you're already at a point where BL based distortion is audible, right?
I'm looking for the cleanest bass I can get, but the big question is, at what point does it no longer matter?"

Completely depends upon the situation. In a ported box at Fb, the driver's contribution to distortion is nearly completely flux modulation based. BL, Cms, and Le nonlinearities are irrelevant, since the driver is barely moving. That means the driver's contribution to THD is primarily from non-excursion related nonlinearities, which would mainly be flux modulation.

Does it matter? Depends upon how loud you like it... For most people, it's not that big of an issue; in fact, flux modulation in general is a pretty low priority. I use shorting rings for inductance linearizing, not flux modulation, as I believe that Le nonlinearities are a lot worse - audibly - than flux modulation.

...
The comparison was a ported system at resonance, and a sealed system at resonance - I think that's comparable. It wasn't the raw driver at resonance.

And I like to look at the average over the range, since music consists of multiple frequencies. Just looking at or very close to Fb isn't too instructive, IHMO, because it typically is just half an octave of a system that typically covers 2-3 octaves.

As far as the Q, if you lower Q you either lower the center frequency OR widen the bandwidth (Q = Fc / BW). Lowering the Q of a ported box does result in a bit less peak output, but you get MORE port output over a wider range, meaning the port is assisting you over a wider range. Rather than have the port do 100% of the job over 1/3rd of an octave, I'd rather have it do 50% of the job over a full octave.

Just my preference, though...

...
Take a Brahma 12mkII, and put it in an SBB4 box - 55.4 liters, tuned to 25 Hz. Put 200W into the driver, and you'll see the above-Fb excursion peak at 20mm at 35 Hz, and the below-Fb excursion breaks 20mm at 22 Hz.

Now, put it in what I would call a detuned box. Put the driver in 35 liters, tuned to 21.5 Hz. Put the same 200W into the driver. The above-Fb peak in excursion is 20mm at 35 Hz, just like before. But we don't cross 20mm below Fb until 18 Hz, and the excursion at Fb is a few mm less, too.

Basically, we're making the port share the load over a wider frequency range. Yes, we have lost output between 20 and 45 Hz, but we've also gained output below 20 Hz.

And, when factored in with a typical room or car gain curve, the gentler slope of the detuned box will, IMHO, blend better overall. I think you'll get smoother in-room/in-car response (the response matches a 2nd order roll-off down to ~16 Hz).

So, with a detuned, lower Q alignment, you lose peak output, but you affect a wider bandwidth - the port is contributing over a wider range, and reduces excursion of the driver over a wider range as well. And I hold that you end up with an alignment that is also inherently friendlier to rooms/cars, in the way it rolls off on the bottom end.

Does that shed some light on my position?

[Post by Thorsten Loesch]:

Flux modulation is caused by the interactions of two magnet fields. It can only be resolved in the magnet design. If you saturate the Magnet structure and ideally use a fieldcoil magnet with sufficient negative source resistance to overcome the fieldcoil DCR flux modulation will diappear too.

[Post by kelticwizard]:
As you run the tone generator down through the bass range, you will finally approach and then hit the box tuning frequency. At that point, unless your speaker has a shorting ring or other special aspect, you will suddenly notice the speaker will actually get sucked into the enclosure, so that instead of the speaker going back and forth from it's centerpoint, it goes to the inward end of it's excursion and moves only in the forward direction from there. In other words, half the waveform becomes clipped off.

You would think such a situation would sound terrible, but in fact that waveform has very high second harmonics, (overtones). Incredibly, it sounds quite musical. The only problem is, of course, is that you are missing your deepest bass notes and are having the notes from the octave above subsitituted for them.

The reason this happens is that in most speakers the magnetic structure is located mostly BEHIND the voice coil. Near ported resonance, this unbalance asserts itself, and pulls the speaker back toward the inside of the box.

There are different ways of handling this. Some manufacturers change the suspension of the speaker to compensate. Some extend the pole piece-the round bar the voice coil surrounds-forward so it juts into the cone area, (a larger dustcap hides it). That method eliminates the unbalance. And some manufacturers, like Peerless and others, put on a shorting ring. Precisely HOW a shorting ring works, I do not know, but it somehow eliminates the unbalance between the amount of magnetic material in the front of the voice coil and the back. This eliminates the waveform clipping that results in such high second harmonic distortion, because the speaker does not travel into the speaker near resonance-it stays right at it's customary centerpoint and moves back and forth from there.


[kelticwizard]:

An article in Audio magazine by the well-known D.B. Keele dealt with the "suck-in" issue. To prevent it, one of 3 things is necessary.

A) A shorting ring, whether aluminum or copper;

B) An extended pole piece, so that there is an equal amount of magnetic material on both sides of the voice coil. There is a variation on this I will explain later;

C) A special progressive suspension designed to counterract the "suck-in" phenomeneon. Although Keele mentioned this a few times, I get the impression there might be only one or two hi-fi companies that use this approach. I have not seen it mentioned in any drivers available to the public.

The "suck-out" phenomenon occurs when the voice coil "sees" less magnetic material in one direction than in the other. So it gets drawn more to one side than the other. The extended core, (pole piece), simply puts more magnetic material on the outside end of the pole piece-the side that is usually truncated-to restore the balance. I have never totally understood the Faraday ring's action, but I gather from several explanations that it cuts short the magnetic field when in operation-hence "short circuits" the magnetic filed past a certain point-and restores symmetricality that way.

The variation on the "extended pole piece"? Apparently the well-respected McCauley? and JBL lines of PA speakers cut a notch in the pole piece on the part that is near the magnet. Instead of extending the pole piece farther out to restore symmetricality, it shaves off a little of the pole piece that is on the other side. This restores balance.

I also gather that the Faraday ring reduces inductance, and therefore has benefits for the speaker at higher frequencies. Perhaps that was the original reason for putting them on, before the benefits of the "suck-in" phenomenon were observed.

Granted that any feature must be implemented correctly to work, I would think that any company that goes to the trouble of putting these features on will do so competently, considering that many companies don't even bother with them at all. Therefore, if you want to prevent "suck-in", go for a design that uses a shorting ring or extended pole piece, (or it's variation, the "notch" in the pole piece).