You need to read the JBL JAES paper. I thought it was Mark Gander but I'm wrong.
Alnico is indeed far poorer for H especially recoil perm. Design, construction & manufacturing must take this non-trivial point into account. There used to be a couple of Engineering Notes from Swift Levick on this. Dunno if they are still in business.
However, the final (properly designed) magnet assembly is far less susceptible to changes due to evil voice coil current etc. All explained in the JBL paper which was written about the time they realised they had to go to ferrites cos the increased cost of Cobalt.
Samarium Cobalt is a rare earth magnet and is nearly as good (about 1/2) as Alnico.
But ferrites, and even worse, Neo are terrible.[*] They also have HUGE temperature coefficients.
Speakers come under stuff that is certainly NOT Linear or Time Invariant. Its just often convenient for us to assume that they are. If we couldn't assume this lie, much of the DSP advances in speaker EQ, (and I argue this has been the biggest advance in speaker tech. in the last 2 decades) would be worthless.
[*] I'm told the new Neos with some even rarer earth sh*t are less temperature dependent though still not as good as Feroba. But I've not seen detailed BH curves for any magnet material this millenium. They used to be easily available but they seem to keep these secret from the unwashed masses and beach bums. There's a lot of Mickey Mouse stuff on the net.
(The magnet gurus will realise I'm cheating here cos some esoteric design issues but the fact remains that a properly designed Alnico magnet greatly helps LTInvariancy in speakers.)
There is no chance at any time soon of Class-D amplifiers approaching the linearity of the best linear amplifiers. No chance whatsoever.
There is no chance at any time soon of Class-D amplifiers approaching the linearity of the best linear amplifiers. No chance whatsoever.
Obviously you haven't heard of Hypex NCore.
You're going to have to explain to me then in what way 0.0007% THD across the audio bandwidth at 200 watts into 4 ohms and <3 mOhm output impedance across the audio band is not "approaching the linearity of the best linear amplifiers".
Bearing in mind that this is a class-D amplifier that can handle reactive loads far better than a linear amplifier, this amplifier can maintain its ultra-low distortion (do you deny that 7ppm is ultra-low distortion?) over a wider range of loading conditions than any sensible linear amplifier. And all of that with <5 W quiescent power consumption.
Oh I see! You think that Hypex in their datasheets and Bruno in his published papers are just lying? You concede that if their amplifiers perform as stated that these are indeed very high fidelity amplifiers?
It's got a fifth-order control loop so the numbers are not dubious at all.
Someone must have verified them by now surely? They are certainly a target for it.
Harry, but what would be the Slew Rate and damping factor for this amplifier Class-D ? This has always been a problem for class D.
Output impedance is < 3 mOhm across the audio bandwidth
Distortion performance at 20 kHz signal frequency proves that Slew Rate is not an issue.
I've tried ferrite in the bases of the drivers and they work very well. However, like you I also ended up going for an RC damper network. Maybe it's time to ignore those claims they they affect the sound and just use them because it makes perfect sense.
I also experimented putting ferrites on the input- so threading the input signal wire through one, and also could not detect anything bad.
Can you provide a reference?
I have often seen the claim that Alnico has a more linear B-H curve for speakers than ferrite but the evidence appears to be the opposite.
This myth is in a book by the late John E*, a JBL employee who was reportedly a fine chap, but who must have been off form when he wrote this.
The charts in his own book contradict it.
Alnico has low flux modulation because it automatically provides a shorted turn, not due to its BH properties, but this seems to have been misunderstood.
Temperature coefficients are definitely worse.
Evidence for the other claim?
Best wishes
David
Du.uuh! As I can't remember the author even and beach bums dun get to read JAES, this might be difficult. I would have posted a link directly to the paper if I could.
Dave, I see you are a magnet guru. Your point about the shorted turn (and other stuff) is what I was referring to when I said I cheated. Are you in the business?
My point being that when you design an Alnico magnet properly, it is far better for many of the faults that Klippel et al describe. Klippel came into prominence while I was in da bush so I don't really know much of their pontificating except friends who are still in the business rate them.
Of course if you drop an Alnico magnet, you have problems which ferrite/rare sh*t magnets shrug off.
BTW, my supa dupa magnet would use Alnico as the main material but probably Neo for the pole pieces. They are excellent for this.
Did Maths/Science back when university was practically free in Australia, so did some bonus physics and mathematical techniques of theoretical physics.
Much misinformation about magnetics in the audio world, even by respected authors like John E* and a few others. Hard to correct, not much informed discussion. The "Alnico is more linear" myth is particularly persistent.
Alnico speakers can sure be done well however. The JBL 1500Al series are the best woofers I know of, pity they won't sell them. I have an idea for an improved version that I dream to make one day.
Best wishes
David
Re - Hypex NCore
I looked @ a few of their PDF's last night, & it "appears" they seem to know their stuff.
*
Yeah the magnet strength weakens, but what about the other types, are they affected the same ?
I knew that temperature affects Neos worse than Ferrite, & i guess the driver manufactures know this too. Low power drivers won't suffer as much as higher powered ones, so it's peculiar that for eg, 1kW + drivers use Neo ! As i'm sure you know too, the magnet strength to weight ratio is significantly better with Neo. So i presume that even though they cost a LOT more, this must be the ONLY reason they choose it ?
I looked @ a few of their PDF's last night, & it "appears" they seem to know their stuff.
*
Yeah the magnet strength weakens, but what about the other types, are they affected the same ?
I knew that temperature affects Neos worse than Ferrite, & i guess the driver manufactures know this too. Low power drivers won't suffer as much as higher powered ones, so it's peculiar that for eg, 1kW + drivers use Neo ! As i'm sure you know too, the magnet strength to weight ratio is significantly better with Neo. So i presume that even though they cost a LOT more, this must be the ONLY reason they choose it ?
They are brittle materials and will probably crack so it is a moot point.
Temperature affects two different parameters, coercivity and remanence and each behaves differently. Your implication that Neodymium is an inferior material is not correct.
Best wishes
David
Hi keantoken,
Yes, saturation in ferrite beads is a problem. In my day job I design very high-speed digital boards for undersea optical communications, with some boards having data throughput of 100Gb/s. They have numerous very large BGA ICs on them (some 1500 pins) that have multiple power supply voltages at very high currents. In some cases, numerous supplies to the chip of the same voltage need to be individually filtered, as for those powering the 10gb/s CMOS I/O transceivers. The boards also have a very large number of components, on the order of 2500, so space is at a premium. In many places I have used ferrite beads for filtering, but I have had to be very careful of the current flowing through them, and have had to consult spec sheets that show the degradation of bead impedance at higher currents. In the cases where the currents are high, I usually have to resort to surface mount inductors.
Cheers,
Bob
5 poles. That's unusual, AFAIK ! BUt if it works, it works
I don't pretend to understand a lot of if, but enough to be interesting
Yes, most designers steer clear of such high-order control loops because they are conditionally stable. You have to be very confident that you've designed it robustly so that it can detect when the loop is going to go unstable (e.g. onset of clipping) and reduce the order of the loop to keep it stable.
Check out the patent.
Also check out this MSc thesis from a student that was supervised by Bruno. Makes use of a lot of the same theory but for a fixed-frequency comparator-based modulator instead of self-oscillating like the Hypex amps.