No, I do not consider your simulation invalid. You had been basically saying that my concept would not work, I did not agree with that. Because you were alluding to that, I assumed your model was giving you erroneous output.
Now it seems, your model has been providing you exactly what I've been predicting. The fact that it produced for you the exact same result as current drive is significant. It isn't exact of course, but to see that will require hardware.
While measurement of acoustic distortion is the final metric, my interest lies in looking at the system on the fly, microsecond by microsecond. I want to see the actual error signals, I want to see the vc (x) modulations on the drive signal.
I've used both FFT and waveform cancellation techniques to troubleshoot systems. FFT is pretty good, but it doesn't directly give you non linear diagnostics...For example, the last waveform analysis, I subtracted the fundamental, then third, then fourth. The residue remaining was a product of fundamental and fourth. Within the system being tested, it was very easy to then find the problem, as the force ripple was pushing hard on a nonlinearity that generated fourth.
If we drive current and look at difference, we should see errors that are not flux coupled directly.
That said, how are you modulating Le(x) and BI(x) in your model? Also, how do you model eddy dragging as a function of position and velocity?
jn
Last edited:
Another way of looking at current drive is that you place the driver in the feedback loop. In other words, driver behaviour that translates into distortions of the electrical signal gets error corrected. (At least, the distorted electrical signal is, whether the driver follows is a different question.)
Non linearities in the magnetic domain will be reflected into the electrical signal, which will be error corrected in a current drive setup.
In that sense, current drive uses a single VC both as motor and as pick up element.
Does this fundamentally differ from what Jn proposes, although his method to accomplish a similar effect is a bit more complicated? I am just catching up with this discussion, so consider it shooting from the hip.
Non linearities in the magnetic domain will be reflected into the electrical signal, which will be error corrected in a current drive setup.
In that sense, current drive uses a single VC both as motor and as pick up element.
Does this fundamentally differ from what Jn proposes, although his method to accomplish a similar effect is a bit more complicated? I am just catching up with this discussion, so consider it shooting from the hip.
I agree that this concept has promise, but does it really give us much advantage? It seems that current drive really does 'something' that is very useful. It would appear, that on the surface, the JN drive will give similar, but perhaps somewhat better results, depending on a number of factors.
Now this concept of noting the 'back EMF' from a coil modulated by magnetic material is not entirely new. It could, in theory, be used to sense the magnetic tape distortion coming from an analog recorder recorder tape head, and has been incorporated in some limited situations. (read while writing) However, a SIMPLE solution proved impractical 49 years ago when we first discussed it at the Ampex Research Dept., but who knows? There may be hope today.
Now this concept of noting the 'back EMF' from a coil modulated by magnetic material is not entirely new. It could, in theory, be used to sense the magnetic tape distortion coming from an analog recorder recorder tape head, and has been incorporated in some limited situations. (read while writing) However, a SIMPLE solution proved impractical 49 years ago when we first discussed it at the Ampex Research Dept., but who knows? There may be hope today.
Last edited:
Please extend on how JNs concept may reduce magnetic tape distortion. How would one deploy it in that context?
//
//
It is the idea that reading the change in flux and using the difference signal to 'correct' the drive is what I was addressing. In the case of analog magnetic recording, it was reading the reproduce signal that is being generated while recording AND ITS NONLINEARITIES due to tape saturation and simply correcting it with feedback to reduce its overall amplitude. (sound familiar?) JN's approach is different than mine was 49 years ago, that was just as speculative, back then.
If he reads voltage errors, the equivalent is he reads force errors. So, he cannot correct for velocity errors. That is what some of us are saying from the beginning and that is why it is just a more complicated current drive. Only more troubles.
I noticed the slight time adjustment, if no one else did 😉, you are funny, I appreciate your posts 🙂
Force on the voice coil from reactively stored energy is not the same as force on the voice coil due do dissipative release.
The latter cannot be seen by a current drive. As a result, the cone motion will not be exactly as predicted by the current. However, if the vc is braking or flux dragging, that will show up in the voltage..
What "some of you have been saying from the beginning" shows me that you do not fully understand all the E/M.
You have set your bar too low.
I repeat my question: How are you modulating Le(x) and BI(x) in your model? Also, how do you model eddy dragging as a function of position and velocity?
jn
Last edited:
John, “eddy dragging as a function of position and velocity” I understand it.
What is “flux dragging”?
George
Oh, you mean 49 years ago, instead of 50, Scott? Well, 49 years ago, I moved to the Research Department at Ampex from the Audio Department where I made my original discoveries in 1968, or 50 years ago.
Gotcha. Yes, though it might have been your little joke because of what JN said about you always talking about 50 years ago 😉
Trying to move a current carrying conductor alongside a conductive surface. You are in essence, attempting to drag the magnetic field caused by that wire.
I assume it is a lower order effect.
I didn't know what else to call it..
jn
No, it does not answer my question as to how PMA is modelling those non linearities.
jn
Thanks.
It’s the same as trying to move a permanent magnet alongside a conductive surface.
Eddy current braking.
Eddy currents work as 1st, 2nd, 3rd… order effects. We experience the resultant final effect from the combination of all the ‘ordered’ effects.
George
I'm still not entirely sure what this equation is for. There are no terms for defining eddy current parameters, so my only other guess is that it has to do with Le shedding stored energy as it increases and decreases. But if this is the case, then modeling Le modulation should provide the reluctance without the need for a separate model. (in the figure 6 model Le(x) does not appear to include turns modulation, only inductance modulation, so maybe I'm on the right track here).
The arbitrary flux inductor in LTspice behaves very badly which leads me to suspect it is not correctly hardcoded in the simulator. I don't trust it so I'm trying to work out my own model.
In the case of a voicecoil the inductor EMF is responding to changing amp-turns. As the coil moves into the magnet, instead of changing current we are increasing the number of turns. As flux equals amp-turns and inductance is the square of turns, then
V(Le) = ddt(I(Le)*sqrt(Le(x))*sqrt(Le))
Where sqrt(Le) is just a static scaling factor to get the right inductance value. This behaves exactly like the flux inductor model in LTspice, but without the matrix solver glitches. I will see what the LTspice developer has to say.
You are on the mark.. that comes from V = LdI/dt + I dL/dt.
The units Klippel used were weird, as it's more a power term.
But that term does indeed cover the energy that is being taken into the coil's field as it's inductance increases moving to the back, and shedding as it moves forward.
You are on the right track there..
jn
- Status
- Not open for further replies.