Barkhausen noise in drivers and inductors

planet 10 wrote:

Heinrich Barkhausen, a German physicist, discovered in 1919 that a slow,
smooth increase of a magnetic field applied to a piece of ferromagnetic
material, such as iron, causes it to become magnetized, not continuously but
in minute steps._The sudden, discontinuous jumps in magnetization may be
detected by a coil of wire wound on the ferromagnetic material; the sudden
transitions in the magnetic field of the material produce pulses of current in
the coil that, when amplified, produce a series of clicks in a
loudspeaker._These jumps are interpreted as discrete changes in the size or
rotation of ferromagnetic domains._Some microscopic clusters of similarly
oriented magnetic atoms aligned with the external magnetizing field increase
in size by a sudden aggregation of neighboring atomic magnets; and, especially
as the magnetizing field becomes relatively strong, other whole domains
suddenly turn into the direction of the external field." -_

Too late to go digging to-nite, but there is a JoeList thread on this...

ok, some questions spring to my mind immediately:
- would noise really sacrifice resolution as long as it is noise?
- why are Neodynium and AlNiCo (whatever that is) free from this effect?
- considering that apart from tweeters, there is only one midrange by Excel and the Manger MSW to reproduce lower frequencies with a Neodynium magnet, it would be interesting to know if there are differences in different ferrites that can be used in drivers? After all, the nonlinearities in inductors are very different depending on what kind of ferrite was used.
- wouldn't the Barkhausen effect also be a problem in crossover inductors?

barkhausen etc

Would noise limit the resolution?

Well, the resolution (or dynamic range) of a system is limited by on the one hand the max level, on the other hand the noise floor. For 16 bit resolution you need (top of my head) 96dB. So if the barkhausen noise is higher than 96dB below the max level, it does limit resolution to below 16 bit.

Cheers, Jan Didden
Most audio electromagnetic systems (speakers, cartridges, dynamic microphones) are plagued by the Lenz effect as well as Barkhausen noise. In fact, the work that I have done in phono cartridges suggests that although the Lenz effect gets less mention than Barkhausen noise, it actually has a greater impact on the sound.

And with loudspeaker drivers that have significant peak-to-peak stroke (woofers, full-range units etc.), the inductance of the voice-coil will change depending on where it is in the gap.

BTW, the strength of neodymium plummets like a stone at higher temperatures (like 70 degree centigrade), so anyone attempting to use it in a loudspeaker driver should definitely design the physical structures with a good amount of thermal dissipation in mind.

regards, jonathan carr
Wiegand effect

capslock said:
planet 10 wrote:a slow, smooth increase of a magnetic field applied to a piece of ferromagnetic material, such as iron, causes it to become magnetized, not continuously but in minute steps._The sudden, discontinuous jumps in magnetization may be detected by a coil of wire wound on the ferromagnetic material

Sounds like really bad news for transformers then, doesn't it?

The Wiegand effect is related.

capslock said:
- considering that apart from tweeters, there is only one midrange by Excel and the Manger MSW to reproduce lower frequencies with a Neodynium magnet

You are probably missing more than a few. I think Morel makes some Neo stuff, and if Aura ever gets their act together.

There is a lot of alnico about, particularily if you include mining old speaker cabinets & consoles. And once the US decides that cobalt doesn't need to be a strategic metal (or the Russians start churning it out) the price of new alnico should come down.

FE103As are destined to be my next FR/midranges, i have a set of alnico 15mm dome supertweeters, and if i can ever afford to keep them, i'll gather up a set of 4 8 or 10" Isophon to fill in the bottom.

Yeah, I have missed a few. Ciare makes a 15" woofer, and there are quite a few midranges around.

I wonder where the resolution problem is.

Noise probably is the least harmful of the gremlins we deal with. In the form of dither, it is even used to increase the resulution of a digital recording, trading time for amplitude resolution, which is always possible except for the highest highs where it isn't needed anyway because we don't hear to well there.

As far a distortion reported in tests, Nedynium systems are ususally average to excellent, but not any better than the best ferrite systems.

I could be wrong, but in principle, why would the Barkhausen effect apply to speakers, since my understanding is that you are not magnetizing or demagnetizing anything (at least in the ideal case). I could be wrong, but I have always assumed that the pole pieces are saturated or nearly so.

I know this is an idealized description, and that nothing is perfect. If I'm wrong, I'd like to find out.

Eric (capslock):

>Isn't the Lenz effect just another way to describe inductance?<

I didn't mean to mix the Lenz effect and inductance, which is one reason why I mentioned each term in its own paragraph.

To clarify, when I mentioned the Lenz effect, I specifically had eddy currents in mind. When I later mentioned inductance (in a different paragraph), I was thinking of voice coil inductance, and how the value affects the characteristics of a passive speaker crossover network.

>Why would it depend on the position of the voice as long as the coil is still in the linear region?<

The inductance value of a coil will change significantly depending on whether it is in free air or it has iron around it. Take an air-core coil and measure the value. Now put a permeable core into the same coil, and measure the value again. Note what happened.

Now, most speaker units with significant peak-peak excursion are built as asymmetrical magnetic structures with magnetic gaps that are considerably shorter than the voice coil. As a result of this type of structure, the amount of iron that the voice coil "sees" is greater on the pull part of its stroke - when it is heading back towards the magnet, and smaller on the push side of the stroke - when the coil is moving out away from the magnet. And if the amount of iron that the voice coil sees is affected by its position, the inductance value will be affected as well.

To some extent, this effect can be countered by clever design of the magnetic system (for instance using an extra-long center polepiece, and/or strategically located copper or silver rings). A gap that is equal in length to the peak-peak throw would likely be an even better solution, but at considerably higher cost.

regards, jonathan carr

thank you for your explanations. I am not sure I understand them. Let's assume a gap that is shorter than the coil and a field that is confined to the gap (idealization). As long as one end of the coil does not actually enter into the gap, there will always be a constant length of coil inside the gap and hence inductance will remain constant, as well as the force on the coil. This is by the way the definition of the linear stroke (we all know it is idealized).


copper ring to screen eddy currents

I have seen a note in Berndt Stark's "Lautsprecher-Handbuch" that JBL and Isophon have a patent on using a copper ring around the pole piece to keep eddy currents from entering the magnet. The author says that he has listened to otherwise similar drivers with and without this devices, and the difference is supposedly very strong.

Also, Excel/Seas advertises that their solid copper phase plugs extend to the center pole piece and that the whole magnet is anodized with copper.

The copper would act as a short, hence increasing mechanical losses but screening to some degree the field that produces eddy current in the ferrite.

Anybody tried to estimate how big this effect is? or heard an otherwise identical drivers for comparison?


Did you see

The idea of applying high-conductivity coatings and platings to the polepieces has been around for a very long time. I know that some companies (for example, SEAS) have utilized advanced variants of this technique to minimize the change in inductance along the stroke.

I have done a number of experiments in raising the surface conductivity of polepieces (mainly for phono cartridges, but with the odd speaker unit here and there), and there can be a fairly large audible and measured effect (like 15~20dB reduction in distortion).

However, another very audible source of distortion is the polepiece iron itself. Various grades of iron may not show up on the measurements as having so greatly different distortion characteristics, but they sure sound different.

regards, jonathan carr