My example was that of a single transistor emitter follower, chosen only to show that while it handles a resistive load well enough, it cannot handle reactance cleanly.
Your right with your circuit, but it is a model evolved beyond my simplistic example. I intended only to show that there is a distinction caused by the load, and that within the full amp topology it is important to assure that all elements remain within their proper realm of operation. Using a DC supply to pull an amp off origin while it's outputting a DC value and a sound card/amp/resistor to measure damping seems a bit easier to me than designing an amp properly to include 4 quad operation. Others here certainly can design to that extent, I just test things.
John
Last edited:
By that I mean that modelling is a very useful and very powerful tool at our disposal, especially if it accurately portray's the entity being modelled. Your understanding of the variables in your own models is unknown to me.
I've been down that road with simple coils of wire as well (room temp ones, not cryo). And yet, nobody seems to have published an accurate model for proximity dissipation on an instantaneous basis. Yes, frequency, current, flux distibution/crowding, but providing a clean analytical model of current crowding on a slew rate basis, sigh.. Most of the work has been on simple dissipation.
When I wind an inductor with a specific copper conductor, I have to actually measure it's frequency response.. ewww. it just makes me feel so...dirty..
John
1: DF is defined with a theoretical, constant, resistive, nominal load in the denominator. This is the only way it can be used to describe an amplifier alone in general, not a particular amplifier+speaker combination.
2: The term "impedance" is related to linear networks. It can not describe a nonlinear element properly with large signals. (Approximations are possible upon special considerations.)
3: The term "impedance" is defined for harmonic signals (sine, cosine). With other signals in a linear system it can be used as a frequency dependant complex variable quantity, with the extended operators, as usual in circuit theory. Excitation signal can be generated by fourier transformation of the signal in time domain. This is simple to deal with a simulator.
3.a: Momentary impedance is generally not a meaningful quantity unless it is constant over frequency. Frequency dependancy is meaningless at a fixed time, since frequency needs a certain time duration to have meaning. And even if you defined a series of nonzero time durations, you also have to define how you transform signal in time to spectra. Different methods and different parameters gives different result. This is a very problematic approach, not really good for basic understanding. Mixing different domains are always problematic. It's analogous to uncertainity relations in phisics.
4. If you need a model accurately represents any kind of nonlinearity in frequency domain: you sucked. Those models are too difficult and/or limited in usability. If you want to understand what happends, you have to assume linearity, or think only in time domain and forget the term impedance. It's easier to make nonlinear physical modells (in time domain), set every parameters and then analising numerically.
5: what is your goal? Understanding DF definition, or understanding why is it important? Unfortunately these require different modell to think in. Definition is based on ideal load, understanding is best on ideal (resistive) amplifier. Combining every nonideality is a perfect way to confusion.
I was flicking through an almost 20 year old WhatHifI amplifier group test and one of their criteria was output impedance. One amplifier was rubbished for 0.2R, another ticked off for 0.1R and some praised for <0.05R.
Once you get below 0.2R, further reduction has negligible effect on speaker frequency response and is giving warning of very high levels of negative feedback
Once you get below 0.2R, further reduction has negligible effect on speaker frequency response and is giving warning of very high levels of negative feedback
An amplifier can be designed for a specific V/I output space.
The damping factor of the amp should be consistent throughout that space. It should remain constant at all frequencies despite the quadrant of operation being used at any instant.
A speaker can easily be measured to determine the V/I space it will require for proper operation. Or, one can simply derate using the nominal DC resistance of the VC's.
As I've detailed, the proper V/I space is NOT what would be defined by the steady state sweep response typically referred to. It must reflect the transient load response as music would produce. Different speaker topologies will produce different transient behaviour. In addition, 2 and 3 way speakers will sum the currents of each driver, so the V/I space of an 8 ohm speaker is NOT what an 8 ohm resistive load will do. There will be locations on the LF ellipse where mid and high frequency V/I patterns will add to and subtract from the voltage and current being requested. Anybody who has had experience with a spirograph knows the type of V/I plots I refer to.
The primary culprit I would look at is the protection circuitry if I suspected an issue. Protection of output devices in quadrants 2 and 4 are quite difficult due to operation close to the opposing rail voltage.
My secondary concern would be the linearity of all components range of operation in the circuit during excursion in Q2 and Q4. Assuming that condition has been met may be an incorrect path of thought.
John
The damping factor of the amp should be consistent throughout that space. It should remain constant at all frequencies despite the quadrant of operation being used at any instant.
A speaker can easily be measured to determine the V/I space it will require for proper operation. Or, one can simply derate using the nominal DC resistance of the VC's.
As I've detailed, the proper V/I space is NOT what would be defined by the steady state sweep response typically referred to. It must reflect the transient load response as music would produce. Different speaker topologies will produce different transient behaviour. In addition, 2 and 3 way speakers will sum the currents of each driver, so the V/I space of an 8 ohm speaker is NOT what an 8 ohm resistive load will do. There will be locations on the LF ellipse where mid and high frequency V/I patterns will add to and subtract from the voltage and current being requested. Anybody who has had experience with a spirograph knows the type of V/I plots I refer to.
The primary culprit I would look at is the protection circuitry if I suspected an issue. Protection of output devices in quadrants 2 and 4 are quite difficult due to operation close to the opposing rail voltage.
My secondary concern would be the linearity of all components range of operation in the circuit during excursion in Q2 and Q4. Assuming that condition has been met may be an incorrect path of thought.
John
Last edited:
Thanks for replying.
-----
I read few things about magnetic field. Found some interesting research to bring back magnetic field to zero in induced magnetic materials. If voice coils are made with such design, we can reduce back emf. Researchers change coercivity of material by patterning surface. I Don't understand it fully though. It says with such patterning of nickel ferrite one can reduce the losses and increase effeciency. Which may mean less impedance in our case. Right ?
Regards.
Agreed. If I want to accurately represent a kick drum, a speaker which derives it's efficiency via storage mechanisms would be the last on my list. Closed baffle Q=.7071, or horn would be my first choice. Bass reflex ain't bad, but it does depend on what you're looking for.
My use of the kick drum was part of an explanation of the distinction between steady state response ala curve as presented, and what is actually going on under the V/I hood during music transients.
John
I posted those pics just to show the principle of magnetic braking in action (probably not an exact analogy to viscous damping). I was hoping someone closer to this problem could show how to fit measurements like this to the bog standard electro-mechanical model of a speaker/ideal amplifier. A better analogy might be the super-magnet on a spring inside and outside of a copper pipe. I was interested in the simple stuff first which I don't think the current vs voltage drive folks all get.
When it comes to magnetic storage, I always tend to see things that are missed. Comes from 3 decades of attending meetings where my intellectual understanding gets beaten to a pulp.
I just noticed that DVC thread, and lo and behold, remember the paper I sent you on coupled coils for that 30 foot tall solenoid? Turns out they adopted my design. Cool.
(I received no credit of course, but hey.
)I just gonna post a slightly modified jpg in that thread (you'll recognize it without the references to supers), that's a conceptual thingy I like..
edit: I cringe at the though of trying to critically damp a system which is underdamped, while only having access to the drive terminals. Negative resistance gives me the hives...
John
Last edited:
Thanks for replying.
-----
I read few things about magnetic field. Found some interesting research to bring back magnetic field to zero in induced magnetic materials. If voice coils are made with such design, we can reduce back emf. Researchers change coercivity of material by patterning surface. I Don't understand it fully though. It says with such patterning of nickel ferrite one can reduce the losses and increase effeciency. Which may mean less impedance in our case. Right ?
Regards.
No. In every way. No, because they increase the impedance. No, because you confuse magnetic material with conductor. No, because reducing back EMF is a veeeery bad idea. Back EMF is not an enemy, it is your best friend, without it you didn't have any sound from speaker. (Or energy conservation theory didnt work.) Back EMF is because of the same structure as what moves the cone: B×l. As long as energy conservation applies, you can not separate B×l in F=I×B×l from B×l in V=v×B×l (=EMF). So if you want movement, you need EMF. And actually it helps, reduces current, reduces heat generation. Why would you want to eliminate it?
Yes, speaker manufacturer try to do this. Unfortunately the best conductors (copper, silver...) are very heavy (or need supercooling), and high mass decrease sensitivity. The material has the best conductivity/mass ratio is aluminium. This is quite hard to work with, and since not as good conductor as copper, needs wider air gap, bigger or better magnet, so the speakers are more expensive. With really big magnet, much higher pole height than voice coil one can build a really efficient speaker, but this is not an optimal solution in terms of cost, because power handling capacity is lower. A bigger amplifier with traditional speaker costs less. That high sensivity speakers shows stronger back-EMF, resonance is wider, damping is more effective, bass quality is better, but they cant produce more bass SPL.
One maybe could make an amplifier entirely using switched capacitor method and relatives.
And the result of all this is that many HF drivers use an aluminium voice coil.
Weight is so important to maintaining acceptable efficiency in an HF driver.
KT/C noise can be derived by an integral as switch resistance goes to 0 in the limit. In the end the work done on/by the charge is lost to the environment.
This is true. I omitted this aspect only because HF drivers have tipically almost negligible back EMF in the used band.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.