At zero crossing it might just be one domain which flips direction. Or perhaps two flip direction one way, while another flips back the other way. You do understand that 'zero' does not mean no domains, but merely random domains? As far as I understand it, nothing much happens at zero which does not happen at other fluxes.6A3sUMMER said:
I think what he's suggesting, is that in an inductor where the current never reverses polarity, the core is only working in Q1 - Q2 (or Q3 - Q4) of the BH hysteresis loop. There is more or less domain strain / ordering as the current waxes and wanes, but the domains don't ever have to flip. I believe that much of the hysteresis loss in the core is due to the work done in toggling domains - and if they're not toggling, then that loss and its associated artifact is eliminated.
Which may or may not be correct - or even what he was trying to get at.
If zero crossing were a problem then EI cores would be a catastrophe. In EI cores for at least 1/3 of the path the field is not aligned in the easy magnetization direction. That's a significant loss but they work just fine!
That's not correct. You can make domain flipping audible with simple equipment and prove it to yourself.
This might be a good reference for demonstrating the Barkhausen effect: https://skullsinthestars.com/2012/10/01/making-magnets-speak-the-barkhausen-effect/
Hey B-Mike,
Did we misunderstand one another? I never said that domain flipping / toggling in a ferromagnetic core +didn't+ produce an artifact. Nothing like that at all. In fact, I'm fairly sure that it +is+ an established source of artifact.
What I said, or tried to say, is that I was interpreting 6A3sUMMER's comments to mean that if the current through an inductor doesn't ever reverse - simply rises & falls within a single polarity - then the domains never need to toggle to the opposite magnetic polarity.
Again, not only may the theory be incorrect, but I may well have misinterpretated 6A3sUMMER's comments. Anyway, at this point it's so far out in the weeds that I'm retiring from the convo - it's classic 'net confusion.
Yes, it would appear to be on the scale of pollution induced, random pop-corn style leakage current noise across insulation surfaces in my valve amps output transformer that drives me nuts when I'm really just wanting to listen to the silent passages between songs.
My understanding is that when the flux points in one direction, but below saturation, you don't have all the domains pointing weakly in that direction. Instead, because of the strong coupling within a domain, all the domains point strongly but in different directions so the average points weakly in the given direction. Domains walls move about due to random thermal fluctuation etc. As the flux increases, so the proportion of domains pointing on that direction increase. In order to achieve this some domains may flip, which can be audible. Nothing changes at zero flux. If low flux was a problem, how would a microphone or MC transformer work?
It's not about low flux per se, it's related to the Br (point of remanence) of the core material. When you push / pull the core all the way up the curve, it tends to retain a magnetic flux for the same reason that you can make a 'permanent' magnet by using strong current to magnetize a piece of iron, etc. When you need to reverse the current to the same level, you end up walking around a different path on the BH curve to arrive at the new opposite H point - which again, has a point of remanence of the opposite polarity.
An externally hosted image should be here but it was not working when we last tested it.
This diagram shows the respective +/- points of remanence at 'b' and 'e'. The root of the dashed line is where the process would start with a neutral (unmagnetized) core. Once you pull it up to +saturation, the core then follows the solid line anti-clockwise (widdershins..) around the loop from then on. This is the concept of magnetic hysteresis.. which is an unavoidable loss mechanism.
Last edited:
I think we're all happy for a core material to have a BH curve characteristic that incorporates a loss when signal voltage is applied to a winding.
However trying to attribute a noise component due to the core flux excursions in a typical amplifier context (ie. an output transformer with loads on windings) is not expected to generate a level that can be measured - well not for present day measurement equipment that only go down to circa -120 to -140dB noise floor levels.
However trying to attribute a noise component due to the core flux excursions in a typical amplifier context (ie. an output transformer with loads on windings) is not expected to generate a level that can be measured - well not for present day measurement equipment that only go down to circa -120 to -140dB noise floor levels.
Hysteresis is a loss mechanism, and potentially a distortion mechanism. It has nothing to do with zero crossing. Just because the usual hysteresis curve you see in books is drawn around zero does not mean that hysteresis only occurs around zero. It happens everywhere, so any AC flux will encounter it.
Zero crossing is the problem, even when you have all magnetic domains aligned with the easy axis, at zero crossing they must flip 180º (π if you want) and the work made by the magnetic field is bigger than if you do not have zero crossing, not to mention Barkhausen noise.
Totally agree, but to avoid that people do not trust in books, let me clarify some cases, supposing that fields are changing harmonically in time.
i) If current flows always in the same direction, the magnetic field vector, B, always conserve the same direction also, this subtle fact has enormous importance because it needs less energy to orientate magnetic domains inside transformer core, i.e. less losses/more linearity and minimize Barkhausen noise due to lower movability of magnetic domains.
The only condition to avoid zero crossing is that current always flows in the same direction, i.e. SE operation.
ii) If you have some DC current component, flowing always in the same direction, and an AC current component arbitrarily flowing, the magnetic hysteresis curve is not centered around zero (B=H=0) and you still can have zero crossing depending on the field values.
iii) If you have an AC current arbitrarily flowing, if the magnetic field is harmonic in time, the magnetic hysteresis curve is centered in zero (B=H=0)
People who are still confused by the misinformation in this thread should see https://en.wikipedia.org/wiki/Magnetic_domain. This says what I have been saying, that at zero magnetisation the domains are arranged randomly. As the flux increases the domains aligned with the field grow, and those opposed to the field shrink. There is no mass flipping of domains, either around zero or any other flux. There is occasional domain flipping at any flux change below saturation. There is no zero crossing problem.
Zero crossing or flipping is just a non-existent problem. The amount of info that is potentially lost is peanuts. The induction is high only at low frequency and it drops more and more as frequency increases greatly reducing the importance of the core. On top of this such "big problem" happens (fortunately) where human sensitivity is very poor and there is little or no music. It doesn't affect audio performance in any way. Barkhausen is of the same entity.
Instead the higher loss of EI cores is reflected in the (clear) better efficiency achievable with C cores. Even so they still work fine.
Last edited:
Barkhausen noise
https://www.youtube.com/watch?v=BLXVLDysroY
https://www.youtube.com/watch?v=Q3c0mICFKLE
Ugly, isn't it? 😛😀
https://www.youtube.com/watch?v=BLXVLDysroY
https://www.youtube.com/watch?v=Q3c0mICFKLE
Ugly, isn't it? 😛😀
Great! There are awesome animations; some definitions are correct some are not.
This is wrong, zero magnetization and zero magnetic field are different things, you should know that, by definition
B = H + 4π M
At zero external magnetic field, magnetic domains are not arranged randomly (As Weiss originally supposed) they are known now to be magnetized along certain directions.
Seems that you forgot the fact that ferromagnetic materials are anisotropic, so please do not ask me again for another demonstration.
As the external magnetic field increases, the magnetic domains whose magnetization is aligned with the external field grow, because those whose magnetization is opposed to the external field shrink, due to most individual magnetic moments of the atoms are flipped, you can call this process "flipping of magnetic domains", if you want you can call it "flipping of magnetization of magnetic domains", and if you want to be more formal you can call it "flipping of most magnetic moments of the atoms of magnetic domains"
If this is more of your British humor, it is not funny. And stop confusing field with flux.
Magnetic field = B
Flux of magnetic field = ∫B.dS
Under certain circumstances, e.g. B is constant over S, you can write
Flux of magnetic field = ∫B.dS = B S
Magnetic field B is a vector, flux of B is a scalar, so you can call the modulus of B as a flux density, and anyway it was proved that field and flux are different things.
Flipping of magnetic domains can be produced at any external magnetic field below saturation (not flux change), if the external magnetic field is reversed, i.e. B→-B, aka zero crossing, magnetic domains must flip again, now at an angle π.
This means more work for the external magnetic field, which in turns means more losses, which in turns produce a wider magnetic hysteresis curve, which in turns it traduces into less linearity. In addition, Barkhausen noise reaches a maximum, which is perfectly logic; note that Barkhausen effect was the proof of the existence of magnetic domains.
Only in Fantasyland there is no zero crossing problem.
Last edited:
Zero magnetization only equates to zero field, in the case of a core with zero remanence. Otherwise, there is always remanent magnetism of given orientation(s), and it takes +work+ to reverse the material polarity to the opposite point.
It has been elsewhere suggested** that in a loose sense, you can think of the BH curve as a +power curve+ and that the area underneath / inside of the curve represents energy, aka work performed over time.
(** LoL)
Last edited:
- Status
- Not open for further replies.