Confused about DVC and Q

[Ron E]

Originally posted by john k...
Nothing good ever comes easy. :-)

Cute. Putting yourself on a pedestal? A ford tempo with a $1000 paint job don't go any faster, but it might make you proud to own it

Your ODE model might be useful in a nonlinear simulation, but concepts like Qms would then have no meaning... Besides, it's incomplete, there is no acoustical stuff in there....

Put it in a lumped (or T/S) model and what happens to "Qds"? It gets lumped...

Good is a value judgement, and just yours at that.

[john k...]

Originally posted by Ron E


A ford tempo with a $1000 paint job don't go any faster, but it might make you proud to own it

Ron, you know I drive a Ferrari!

Your ODE model might be useful in a nonlinear simulation, but concepts like Qms would then have no meaning... Besides, it's incomplete, there is no acoustical stuff in there....

Put it in a lumped (or T/S) model and what happens to "Qds"? It gets lumped...

Oh poppy ****. In a lumped prameter model everthing gets lumped. Qtc is all you need to deterimine the system response. But the more you lump together the less you know about the actualy system. Anyway, where do you think all those lumped parameters came from in the first place? The problem with extracting the T/S parameters from the impedance curve is that it is incomplete for the dual VC case. Besided you can't actually extract Qms from the impedance anyway. What you get Qes associated with the amplifier loop and a parameter, typically referred to as Qms, that represents all additional damping in the system. That is, Qms, as extracted from the impedance curve does not allow you to then calcualte Rms, given Fs and Mms unless the only additional damping comes from Rms. What is extracted is, as you put it, a lumped Q, call it Q1, but not Qms if we accept the common definition of Qms = Fs x Sqrt(Mms/Rms). Now if you take two impedance measurements, one with the the second VC opened and then one with the second VC loaded by a resistor the second measurement will allow extraction of another lumped Q, Q2. And from the two lumped Q's it would be possible to extract what I called Qds, and another Q which would be Qms if there were no other damping mechanisms present.

But why bother with any of this since all that really matters is the final lumped prarmater, Qts (or Qtc).

Lastly, my model isn't any good for nonlinear distortion, nor is Small's, since they are linear moddels. But they can be extended to include nonlinear Bl, nonlinear compliance, etc.

This is great, I'm actually learning how to use smiles and stuff.

[Ron E]

Originally posted by john k...
Ron, you know I drive a Ferrari!

And a miata! I'm not a fan of 308's. Maybe a 246?

In a lumped prameter model everthing gets lumped.

Right, so un-lumping it is adding unnecessary complexity - which may make you feel as if you "discovered" something, but since anyone capable of doing that analysis knows where the damping comes from anyway, it is a moot point. Why analyse a bunch of parallel resistors when you can work on the equivalent...

Lastly, my model isn't any good for nonlinear distortion, nor is Small's, since they are linear moddels. But they can be extended to include nonlinear Bl, nonlinear compliance, etc.

Right and IMO that is the only good reason to use an ODE model, but you would have to extend yours to use nonlinear fundamental parameters rather than T/S to make it useful. This is why I likened it to a painted tempo. Then you would have to measure those nonlinear parameters to characterize the driver, another challenge.

[john k...]

Originally posted by Ron E



Right, so un-lumping it is adding unnecessary complexity - which may make you feel as if you "discovered" something, but since anyone capable of doing that analysis knows where the damping comes from anyway, it is a moot point. Why analyze a bunch of parallel resistors when you can work on the equivalent...

I think we are looking at this from different points of view. If all that is cared about is the final result then we can just write:

(24) Qts = Qms Qes / (Qes + c Qms)


where


(25) c = 1 + Re /(Re + Rl)

and it would look to all the world that the result is just a modification to Qms, which is one way to express the result. But I am, and I believe there are others who might be, interested in how that result is obtained. Even if they are capable of doing the analysis themselves that doesn’t mean they have nor have the desire to do so. It’s just information for those who are interested, to be ignored by those who aren’t.