This is where I would like some input from the experts. Is constant current amplification feasible with waveguided CD with their usual non-smooth impedance? For midrange it may be good idea if the Fb resonance is taken care of.
I tried to fix some problems (flux modulation -> distorted current) in a subwoofer by current-drive, but in the end it didn't work well enough. And only sealed alignment is somewhat workable. But at least any thermal compression was eradicated!
John simple is fine if you dont oversimplify and miss the point. You oversimplified. Its the time constants that you example misses. With displacement or velocity nonlinearity the effect is always in synch or at the same time constant as the signal. With thermal its not. The thermal can happen at much slower time constants, but still fast enough to be audible. You should certainly be aware of the different between time invariaent coefficiencts and nonlinear terms in a Diff EQ - they are certainly NOT the same things.
Borat - I don't think you have a grasp on the situation. IMD is NOT time varient.
To both of you. I don't deny that thermal aspects are a nonlinearity, but they are certainly not the same form as displacement nonlinearity and there is no reason to believe that they should be of comparable audibility. That thermal problems are minimized - but not elliminated - by current sources is well know, but current sources are not readily available and not without their problems either. That these problems can be minimized with DSP is also well know, JBL was doing this long before EAW. The speakers at a concert most likely do have this kind of processing applied.
But all of these solution are impracticle in home use. Simply understanding the problem and designing for it makes it go away without all of the ridiculous costs involved with your "solutions". Its just called "good engineering".
Borat - since copper and alluminum have very similar thermal coefficients they will act the same thermally.
And don't try and tell me that these problems don't exist - they do. We can argue about the level of audibility, but not about their existance and that they are different from normal nonlinearities. I am uncertain about the level of audibility, I'll admit that, I am completely certain that the effect is real, that it is substantially different from one loudspeaker to another, that it is completely different than component nonlinearity, and that there is an easy fix once you understand the issue.
or maybe it's called denial ? 😉
as far as i can tell your thermal distortion is nothing but intermodulation distortion where a tone is modulated by the integral of its own envelope.
per unit of volume maybe. but not per unit of mass or conductance.
i thought your idea was interesting because i can easily design a motor which will have 99.9% linear BL curve. at which point the thermal effects might become more significant than effects of BL nonlinearity. but to say that BL linearity is practically irrelevant is just a very Geddes thing to do 🙂
Last edited:
I wonder. I'd say that most horns have enough speaker-level manipulations applied that the actual impedance swings at the driver are largely swamped. The CD eq applied to any of them will tend to give them a much higher impedance over the entire passband, and between that, the highpass, and any other filter components, you're unlikely to see much ripple. I used a couple notches to get my own JBL 2426H down to 4 ohms +/- .55 in the passband, and after padding and CD highpass we're in the "tens of ohms +/- 2%" range- certainly any level artifacts will be largely suppressed.
Wow "integral" that's a sophisticated word to be using. You are partially correct in that it similar to "intermodulation distortion where a tone is modulated by the integral of its own envelope", but we are talking about music and this gross oversimplification does not get at the issue. Further more this integral that you suggest would go to infinity for large time (the envelope is always positive), so its not that simple.
I have studied nonlinear systems theory extensively so you are not going to catch me making a math error on this stuff.
Not "maybe" - per unit volume is all that matters - DA! All thermal coefficients are per unit volume.
Yes, I don't accept the standard dogma about driver design or loudspeaker design. But if you think for one instant that I do this because I am nieve or don't understand the issues then you are truely more arrogant than even I thought.
I would not want and don't use, drivers with a flat BL curve. Its a waste of time and money. It doesn't hurt anything, but it doesn't help either. So why bother. A gradual BL curve has very low orders of nonlinearity, low orders are not very audible, so you are making a linear BL curve for no reason. In fact for higher output where the coil comes close to the inevitable BL drop, the sharper it is the worse it will sound, so the gradual drop will sound better at these excursions and they will both sound the same at lower excursions.
I agree with your point in principle, and yes the higher series impedance to a compression driver does help, but my tests still show a change in the driver with thermal modulation. As I said, it is much much smaller than a direct radiating tweeter.
And there is also the woofer, which in my case goes pretty high in frequency, so this is a concern as well.
It's okay with me, I've enough measurements and listening variation to solve some issues.
I have listened to Wolfgang Klippels demonstrations, and compared it with some drivers that have BL and Km curves, and induction variation curves as well, and I clearly hear the effects. I'm sorry that others don't, and probably it's a blessing as well. People that listened with me noticed something funny, but could not pinpoint the problem, but they did reflect their impression quite clearly to me.
I'm am sharing my own experience rather than quoting the work of some other, the conditions of which are questionaable. Show me the research you referr to any time, and I'm 99% sure I will find some critical conditions not mentioned.
you're just lucky that nobody on this forum except me understands anything about this subject so you can get away with such a statement.
i suggest you talk to Kyle Lee of TC Sounds on AVSForum about this since i don't think you would believe me.
i wouldn't say that. i just think you either are looking from the wrong perspective or aren't expressing your ideas effectively enough.
it's part of my charm 😛
Earl, it's all there. I tried to explain it to you before in terms of the difference between the the rate of heat generation and heat rejection. Temperature is like velocity. Rate of change in temperature is like acceleration.
There is certainly a meandering of the VC temp over time which will alter the system sensitivity, but what is the effect of a short term dynamic input? It yields a high rate of change in temperature, but over a very short period of time which may contributes little to the heating of the VC compared to the more or less constant heat generation by the sustained level. Percussive impacts, for example, may not be likely to suffer significant compression of their own making. They are over too quickly. On the other hand, machine gun fire might exhibit compression as the chain of shots continues on because even if the change in temperature associated with a single shot is small, if the VC does not return to the initial temperature before the next shot the temperature will ultimately rise exponentially until it has reached a new equilibrium where, due to the increase the heat generated can be rejected to the surroundings at a sufficiently high rate. Thus the longs the chain of shots lasts, the less dynamic they may become.
Multiple impulse testing allows you to investigate this because the duration of the impulse can be altered, as can the recovery time between impulses.
John, This is incorrect because heat is a first order DifEQ, not 2nd, there is no effect comparable to acceleration. One can use resistors and capacitors in thermal modeling, but not inductors. There is no thermal inertia. The caps represent thermal mass.
You keep try to "explain things" to me but you keep getting it wrong.
As to your other points, everything depends on the thermal time constants involved because without those you cannot determine the effects that you correctly describe. Your impulse test will not give those directly, mine will. There are short term time constants - most notably the voice coil which can be very short (the coil heats instantly, but changes temperature more gradually, but still very rapid), medium term like the former and center pole and then longer term like the magnet and the whole speaker system. Without knowing these time constants you can't say much of anything except the long term thermal compression that everyone talks about. Its the short term stuff that interest me.
Finally, no matter how hot the voice coil gets, its temperature change with heating is the same - we are way up on the Kelvin scale and the changes are linear so the delta is the same no matter what the absolute value.
Last edited:
i think now you're missing the point too.
the long term effects are irrelevant because they merely affect loudness and frequency response but neither generate new frequencies nor shift existing frequencies around.
a short but rapid change in temperature ( due to a transient ) on the other hand has the potential to do just that. however the effect would be the same as a very low frequency component modulating the transient by pushing the VC out of the gap at the same time as the transient is being played.
in other words it doesn't matter whether the VC temp is rising at a rate X or the VC is exiting the gap at rate X - the net result is still the same change in BL^2/Re which is what matters.
Geddes arbitrarily labels one of the two phenomena as time-invariant. I fail to see how what you call something matters if the result is the same.
Last edited:
Now it is audible.😀 I agree it's not quite audible. I would not have noticed it in one design if I didn't listened to it in a relatively dead room for recording and mixing. Wolfgang has somepapers with some possible solution. The concept seems sound, just how good the real time implementation will be and how flexible for adaption to various configurations need to be investigated. I still have an analog design that sort of has a similar function, and the piano dynamics is still the best I have ever heard. So it would be interesting to see if Mr. Klippel can pull it off.
It seems we need to both consider the cause of change in dynamics as defined technically, and as perceived. Both may not have the same source of problem.
😱
and there i was thinking that sitting alone in my room 24/7 was going to build my people skills.
- Status
- Not open for further replies.