Unless you were trying to measure the output impedance of the amplifier by back-driving it
The error can be significant in some cases.
Consider a simple 'blameless' topology that is fully DC coupled (simpler to show the concept in DC terms) amp with BJT input. Whatever you manage to get to the input diff. pair feedback side, will make the amplifier produce the same base current on the input side to satisfy the balance condition imposed by GNFB. Forgetting to put a realistic impedance on the input often means replacing something on the order of a few hundred ohms with the usual 47k input resistor - easily a factor of 100 increase. This means that any current injected into the diff pair on the feedback side will now require 100x the voltage drop on the input side to be balanced inside the diff pair, compared to a realistic situation. What does this do to the output voltage?
Newbies world translation: so the less voice coil moves, the less EMF ?
Better to have the lowest Xmax, the shortest voice coils length and so bigger Sd ? Bad for Open Speakers with Linkwittz transforms , but good for High effcienty & large speaker with low cone mvt (horn, BR load, bass array to minimize cones mvt) ? What trade offs, we speak for a while of the couple amp/speaker in the Olson thread ?!
.... Or just buy amps with low DF and pray ?
.... me have a nervous breakdown now to read this thread and want to buy a Staxx headphones !
Better to have the lowest Xmax, the shortest voice coils length and so bigger Sd ? Bad for Open Speakers with Linkwittz transforms , but good for High effcienty & large speaker with low cone mvt (horn, BR load, bass array to minimize cones mvt) ? What trade offs, we speak for a while of the couple amp/speaker in the Olson thread ?!
.... Or just buy amps with low DF and pray ?
.... me have a nervous breakdown now to read this thread and want to buy a Staxx headphones !
Last edited:
This is truely complex. I tend to look at it very simplistic. When the forces in the speaker are positive and the release of those aids cone motion, then impedance rises and thus it draws less current and when the speaker needs an extra push the impedance goes down a bit as the inductance driven impedance is lowered. This is off course a continuous process with ever changing loading of the amplifier thru a cycle. The good thing is that with voltage drive (low output impedance) the system is to some degree self balancing.
Clever driver and cabinet designs aims to reduce the problems that lies inherent in dealing with spring loaded mass systems. Lowering VC inductance is a primary concern, an other is controlling cone motion by air flow rather than by air springs (like you do in a Thile Small optimised cabinet)
Clever driver and cabinet designs aims to reduce the problems that lies inherent in dealing with spring loaded mass systems. Lowering VC inductance is a primary concern, an other is controlling cone motion by air flow rather than by air springs (like you do in a Thile Small optimised cabinet)
For the same frequency, yes.
The voltage produced in the coil by motion is proportional to the magfield strength and the rate of change of the field through the coil. Coil velocity determines the actual voltage generated. I've included a link to the georgia state hyperphysics site, I hope the link goes where I think... It explains motional EMF caused by a loop of wire going through a magnetic field..
http://hyperphysics.phy-astr.gsu.edu/hbase/electric/elevol.html#c2
It is Faraday's law that describes the effect.
John
Likely a Completely Dumb newbie question
Wouldn't a Voice coil generating electricity against a magnet produce DC?
Depends on the motion of the voice coil. When the voice coil reverses direction, it produces a voltage of the opposite polarity. Normally voice coils vibrate which means reversing direction frequently. That implies that the generated voltage is reversing polarity frequently. We generally call that Alternating Current (AC).
The magnitude of voltage that the voice coil produces is related to the speed at which the voice coil moves through the magnetic field.
The amplitude is related to frequency and speed or vice-versa by the usual Newtonian laws of motion.
John how do I get an idea for what the generated EMF would be as a fraction (I guess) of the amp output current? Is here an easy, simple ratio?
{I know: for every complex and difficult question there is a simple, easy to understand, wrong answer}
The back EMF is always less than the applied voltage if the speaker is passive and reasonably linear.
Introduce a really strong nonlinearity like a motor-driven switch and the back voltage can go all over the place and even sometimes create a back current.
In more common and realistic situations, sufficient nonlinearity can be the result of a speaker voice coil being driven way out out the magnetic field. This requires a speaker with far more Xsus than Xmax. Of course, with all this nonlinearity, we're not talking high fidelity audio any more. But this sort of thing can happen with MI speakers and rock PA.
Intellectually, I know the back-EMF there from the relevant principles of EE, and from an experimental viewpoint, as a DIY-er I've measured it.
That all said, I find it far more helpful to dwell more on the consequences of the back voltage which is to increase the apparent impedance of the speaker.
John how do I get an idea for what the generated EMF would be as a fraction (I guess) of the amp output current? Is here an easy, simple ratio?
{I know: for every complex and difficult question there is a simple, easy to understand, wrong answer}
If the load is lossless, approaching resonance can cause the generated emf to be larger than the drive. I've never calculated out nor tested this to compare low vs high efficiency speakers, nor to compare a sealed baffle/horn enclosure against a 4th order bandpass.
Personally, I'd put together a simple test set to look at the VI space. You could put together a reflection bridge to look at the send and return waveforms independently, as Cyril Bateman presented complete design and assembly guides to do just that.
To me, the most important aspect of viewing the VI characteristics is to see where the VI point goes with respect to rail voltage and current limit when driving a two or three way, and especially how close the VI gets to a foldback limiter on the output stage. Also, you can include the hyperbola of constant output device dissipation to see if you're getting close to secondary breakdown areas. As one approaches that boundary, the gain will tend to go crazy and the pre drivers may not be able to cleanly turn the device off.
As a plus, you can see what happens when the VI crosses the zero current axis while keeping output at the same polarity. Most discussion centers about the resistive characteristics of the output crossing distortion, where V and I both approach zero.
Sorry if I didn't answer your question well.
John
Hmmm… sort of. More like the force on the speaker 'motor' coil is proportionate to the current passing through it (in the fixed magnetic field). Magnetics is all about current. Voltage is the consequence of changing current thru the inductor AKA motor.
dI/dt = V(t)/L … plus other terms
So, while one could have a speaker cone traveling forward (truly), a negative applied voltage will cause its current-flow to head towards reversing… decelerating the coil and finally causing it to reverse direction.
Just saying…
Darn integrals of time aren't quite so well behaved.
GoatGuy
Good points thanks, I must look up the I/V stuff.
Intuitively I thought that as the impedance of the driver rises at resonance, the drive current would go down for the same motion/speed. So the EMF would stay up (because the motion continues in my simplistic view) with the drive going down.
And I can see now that it translates into a phase shift between drive voltage and current. I wonder whether it would go all the way to π radians.
Jan
Not so much. If you figure out how to falsify or make Faraday's Law irrelevant...
Where? I've never found it published. Never looked much of course, but hey...maybe it should be published in LA?
I was pretty loose with the terms there, your intuition is correct, the drive current will go down. When the load is close to resonance, the amp current will climb if the amp needs to control it, such as the tail end of a kick drum feeding a 4th order bandpass (although I'm not sure if it would be identified as a kick drum in one of those). I want those beasts to hit my chest hard, then instantly go away... kick drums are not bells...
As opposed to Q radians?So the EMF would stay up (because the motion continues in my simplistic view) with the drive going down.
And I can see now that it translates into a phase shift between drive voltage and current. I wonder whether it would go all the way to π radians.
Jan
John
edit: I see Ben Duncan does discuss VI of output stage, but he only describes one quadrant. I discuss all four, and the result of reactive load.
SWTPC mentioned (about 1977) reactive load, but show only a simplistic bent curve superimposed on the load line graph, again one quadrant. I've taken it a tad farther than that in the intervening years, but never researched to determine if any of the biggies published enhancements on the understandings.
Edit 2: This is a logical extension of phasor notation. My first foray into this was back in 1974, the text was Energy Conversion, Edward M Walsh, Ronald Press Company, 1967
Last edited:
Of course (which is why I only mentioned the injected current into the feedback side). What actually happens is that you have an additional high resistance in series with the source. Now look at it from a standpoint of open loop gain but seen from the actual low impedance source - keep in mind that the differential input impedance of a typical input pair is not that high!
At speaker resonance, the driver's impedance is maximized (sealed box, aside from HF VC voice coil effects)) (for the vented box impedance is minimized). In either case the voltage and current are in phase or if you will the phase angle of the impedance is zero.
(Please note that these are real world measurements and may deviate slightly from the ideal.)
For example:
Sealed box:
Revel Performa3 F208 loudspeaker Measurements | Stereophile.com
An externally hosted image should be here but it was not working when we last tested it.
Vented box:
Dynaudio Excite X14 loudspeaker Measurements | Stereophile.com
An externally hosted image should be here but it was not working when we last tested it.
newbie alert !
Can we simulate and exaggerate this back EMF artificially, so we can measure and audition its effect.
Regards.
Sure - just put another power amp in series with the one is already there, and of course deal with the grounding issues (e.g. drive it though a high quality audio transformer) so you don't end up frying anything.
In the case of a loudspeaker, the primary and largest magnetic field is generally that of the permanent or electro magnet that has to be there.
You said:
"More like the force on the speaker 'motor' coil is proportionate to the current passing through it (in the fixed magnetic field)"
Since the fixed field in a modern speaker comes from a permanent magnet, the fixed magnetic field is not due to any electrical current.
The speaker doesn't work if there is no magnetic field to interact with the one coming from the voice coil.
The magnetic field due to the fixed magnetic field is usually the far stronger magnetic field that is involved.
Where? I've never found it published. Never looked much of course, but hey...maybe it should be published in LA?
I was pretty loose with the terms there, your intuition is correct, the drive current will go down. When the load is close to resonance, the amp current will climb if the amp needs to control it, such as the tail end of a kick drum feeding a 4th order bandpass (although I'm not sure if it would be identified as a kick drum in one of those). I want those beasts to hit my chest hard, then instantly go away... kick drums are not bells...
As opposed to Q radians?
John
edit: I see Ben Duncan does discuss VI of output stage, but he only describes one quadrant. I discuss all four, and the result of reactive load.
SWTPC mentioned (about 1977) reactive load, but show only a simplistic bent curve superimposed on the load line graph, again one quadrant. I've taken it a tad farther than that in the intervening years, but never researched to determine if any of the biggies published enhancements on the understandings.
Edit 2: This is a logical extension of phasor notation. My first foray into this was back in 1974, the text was Energy Conversion, Edward M Walsh, Ronald Press Company, 1967
Actually many years ago I build Peter Baxandal's clever V/I monitor in a Tek TM500 custom case.
Must see if I can find it back.
I remember that the display was not very clear and I needed lots of creative interpretation.
Jan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.