In the most general possible terms, multi-driver loudspeakers are impedance-compensated by parallel networks across each driver, then by using all-pass crossover networks. This is a gross generalization, but a place to start thinking about the issues.
A conventional speaker driver has a fundamental resonance (meaning mass x compliance, like all resonantly oscillating things - a rock on a spring, etc.) and a significant series inductance, from the motor's coil. Both contribute to the driver's terminal impedance, but may or may not matter much to the crossover's frequency response (with real drivers) depending on where they happen and how far away from the crossover region.
Fundamental resonances are impedance-compensated with a damped series resonant shunt (series LCR) and coil inductance is compensated with a shunt Zobel (series RC).
There is no performance advantage to the driving amplifier to these added shunt loads, ever, at all, period. But they do give the crossover filter a well defined and easily calculable termination. As computer modelling improves, this is becoming less important.
There are two separate related issues of amplifier load termination and speaker wire transfer termination. Zobels are the answer to both issues.
All good fortune,
Chris
A conventional speaker driver has a fundamental resonance (meaning mass x compliance, like all resonantly oscillating things - a rock on a spring, etc.) and a significant series inductance, from the motor's coil. Both contribute to the driver's terminal impedance, but may or may not matter much to the crossover's frequency response (with real drivers) depending on where they happen and how far away from the crossover region.
Fundamental resonances are impedance-compensated with a damped series resonant shunt (series LCR) and coil inductance is compensated with a shunt Zobel (series RC).
There is no performance advantage to the driving amplifier to these added shunt loads, ever, at all, period. But they do give the crossover filter a well defined and easily calculable termination. As computer modelling improves, this is becoming less important.
There are two separate related issues of amplifier load termination and speaker wire transfer termination. Zobels are the answer to both issues.
All good fortune,
Chris
Not sure you're saying as much, but if ever there is a case where the amplifier is sensitive to its load, the effect might be seen regardless of whether the crossover is active or not.
This is a link for a demo version ( in italian at the moment) of the most advanced sw tool to play with loudspeaker and filters and compensation in various forms
LDS/ALAB descrizione e versioni demo - AudioReview
Walter
LDS/ALAB descrizione e versioni demo - AudioReview
Walter
Yes, that is true also. But a variable speaker load is One Thing, adding some power sucking bits to make it level( at some particular value), is IMO a second-place solution.
cheers,
Douglas
Chris, is this the case for tube amps as well? It was my assumption that the big deal with SETs would be unusually low impedance in the bass and midrange regions but I was not sure under what conditions a rising impedance might cause problems with the OPT reactance. My intuition says that a rising impedance just flattens the load line.
My instinct is to keep the crossover as simple as possible but don't mind complexity where needed.
How to flatten impedance curves? According to these articles from Audiopax, it has to be done in a system with amplifier and speaker together. (The formula equations in the articles are not very well formatted).
Last edited:
If the article says it "has to" be done together, I'd suggest the words are getting in the way. While it can be I would start by pointing out that making the impedance resistive works the same for different output impedances, different amps.
The article does mention output transformer self-resonance damping, and LF system damping, both of which are important issues in their own right.
The article does mention output transformer self-resonance damping, and LF system damping, both of which are important issues in their own right.
That is certainly one use of RC & RLC networks.
But it in no way describes why one would use either in a FR loudspeaker with no XO.
That one example shows that the generalization made in the first senstence is decidely wrong. In these case the network is for the amplifier.
dave
2 generalizations:
Single Ended amplifiers tend to have an Un-symmetrical damping factor (positive alternation versus negative alternation of a sine wave; the same goes for the instantaneous polarity of the music signal).
Push pull amplifiers tend to have a Symmetrical damping factor (positive alternation versus negative alternation of a sine wave; the same goes for the instantaneous polarity of the music signal).
It is much more complicated than this, but it is often true.
Loudspeaker drivers tend to have Un-symmetrical excursions (positive alternation versus negative alternation of a sine wave; the same goes for the instantaneous polarity of the music signal). The loudspeakers do have 2nd Harmonic distortion, Right?
All Generalizations Have Exceptions
Just my Opinions
Single Ended amplifiers tend to have an Un-symmetrical damping factor (positive alternation versus negative alternation of a sine wave; the same goes for the instantaneous polarity of the music signal).
Push pull amplifiers tend to have a Symmetrical damping factor (positive alternation versus negative alternation of a sine wave; the same goes for the instantaneous polarity of the music signal).
It is much more complicated than this, but it is often true.
Loudspeaker drivers tend to have Un-symmetrical excursions (positive alternation versus negative alternation of a sine wave; the same goes for the instantaneous polarity of the music signal). The loudspeakers do have 2nd Harmonic distortion, Right?
All Generalizations Have Exceptions
Just my Opinions
Last edited:
And you can go further on the PP description; assume for a moment an equal a-a>speaker load. The AB amp is going to have a different damping factor from an A amp.
cheers,
Douglas
cheers,
Douglas
Bandersnatch,
Agreed.
But . . .
Class AB push pull is either in Class A during low and medium signal level voltages, then there is a transition from moderate positive and moderate negative voltages, to Class B for the larger positive peaks, and the larger negative peaks.
During each large signal, the operation goes through all these states (as in a sine wave, but music with large signal will do the same thing too).
During the Class A time, both tubes are on, the damping factor is highest, and equal in both the positive and negative directions.
During the Class B time (large positive voltage, and large negative voltage), only one tube is on at a time). During that time the damping factor is about 1/2 of what it is during the Class A time when both tubes are on.
The Class A time damping factor is symmetrical to itself.
The Class B time damping factor is symmetrical to itself.
To simplify, I will make the stipulation that this is true, if there is no negative feedback from the output stage to an earlier stage.
But now, lets add negative feedback . . .
Using negative feedback from output stage to an earlier stage does not completely change this effect.
But the effect of very high negative feedback from the output stage to an earlier stage will make the damping factor so high, that the open-loop damping factor of the class B time is almost completely "covered up" by that high open loop versus closed loop gain ratio.
I did Not say that the damping factor was a Constant, as the signal went from Class A, to transition, to Class B.
I said the damping factor was Symmetrical, not Constant.
I hope that better explains what I meant to communicate.
Perhaps there should be a new term, instantaneous damping factor, or instantaneous output impedance.
Constantly changing in SE and PP that do not have negative feedback.
Agreed.
But . . .
Class AB push pull is either in Class A during low and medium signal level voltages, then there is a transition from moderate positive and moderate negative voltages, to Class B for the larger positive peaks, and the larger negative peaks.
During each large signal, the operation goes through all these states (as in a sine wave, but music with large signal will do the same thing too).
During the Class A time, both tubes are on, the damping factor is highest, and equal in both the positive and negative directions.
During the Class B time (large positive voltage, and large negative voltage), only one tube is on at a time). During that time the damping factor is about 1/2 of what it is during the Class A time when both tubes are on.
The Class A time damping factor is symmetrical to itself.
The Class B time damping factor is symmetrical to itself.
To simplify, I will make the stipulation that this is true, if there is no negative feedback from the output stage to an earlier stage.
But now, lets add negative feedback . . .
Using negative feedback from output stage to an earlier stage does not completely change this effect.
But the effect of very high negative feedback from the output stage to an earlier stage will make the damping factor so high, that the open-loop damping factor of the class B time is almost completely "covered up" by that high open loop versus closed loop gain ratio.
I did Not say that the damping factor was a Constant, as the signal went from Class A, to transition, to Class B.
I said the damping factor was Symmetrical, not Constant.
I hope that better explains what I meant to communicate.
Perhaps there should be a new term, instantaneous damping factor, or instantaneous output impedance.
Constantly changing in SE and PP that do not have negative feedback.
Last edited:
Well 6A3, aside from the idea that an AB amp has *ANY A bit of operation, you have expanded the description nicely. There is no 'A' operation in an 'AB' amp. Class A is not defined so simply as 'a condition where both are conducting'. If it is only at low power, then it is by definition AB. This 'x' Watts Class A, and 'XX Watts at full power', is purely Marketing Cattle Exhaust, Male, Solid Type.
cheers,
Douglas
cheers,
Douglas
For an alternative interpretation see:
Part 4: Of Loadlines, Power Output and Distortion
All good fortune,
Chris
Part 4: Of Loadlines, Power Output and Distortion
All good fortune,
Chris
Bandersnatch,
One more attempt to describe one effect / characteristic of a push pull amplifier.
For this concept, let's talk about the total absence of negative feedback.
Whatever you want to call the amplifier, a push pull Class AB amplifier does not have equal instantaneous output impedance, all the way from the most negative signal peak, to zero signal, to the most positive signal peak.
Suppose a small 20kHz sine wave signal from a moderate resistance, is applied to the amplifier output at the same time the amplifier input is driven with a 200Hz signal, that results in a large signal at its output. Then, measure the 20kHz signal level superimposed on the 200Hz signal, you would find that the amplitude of the 10kHz signal varies depending on the instantaneous value of the 200Hz signal.
Damping factor is normally taken with a sine wave, at one particular signal level.
Then, by using various test methods, the answer is obtained.
But change that signal level, and there will be a different damping factor number.
I believe that "Symmetrical" centered about zero signal voltage, is a very well established fact of properly operating Class AB push pull amplifiers, and Class A push pull amplifiers, and Class B push pull amplifiers.
Then compare that to a non-negative feedback class A single ended amplifier, the output impedance of the SE amp is Not Symmetrical about Zero Volts.
So, rather than considering the classical damping factor measurement, lets consider the instantaneous output impedance of the amplifier.
The loudspeaker does not see a constant output impedance versus over the complete plus, zero, and minus signal voltage.
I hope this 3rd try illustrates the concept.
Yes?
No?
One more attempt to describe one effect / characteristic of a push pull amplifier.
For this concept, let's talk about the total absence of negative feedback.
Whatever you want to call the amplifier, a push pull Class AB amplifier does not have equal instantaneous output impedance, all the way from the most negative signal peak, to zero signal, to the most positive signal peak.
Suppose a small 20kHz sine wave signal from a moderate resistance, is applied to the amplifier output at the same time the amplifier input is driven with a 200Hz signal, that results in a large signal at its output. Then, measure the 20kHz signal level superimposed on the 200Hz signal, you would find that the amplitude of the 10kHz signal varies depending on the instantaneous value of the 200Hz signal.
Damping factor is normally taken with a sine wave, at one particular signal level.
Then, by using various test methods, the answer is obtained.
But change that signal level, and there will be a different damping factor number.
I believe that "Symmetrical" centered about zero signal voltage, is a very well established fact of properly operating Class AB push pull amplifiers, and Class A push pull amplifiers, and Class B push pull amplifiers.
Then compare that to a non-negative feedback class A single ended amplifier, the output impedance of the SE amp is Not Symmetrical about Zero Volts.
So, rather than considering the classical damping factor measurement, lets consider the instantaneous output impedance of the amplifier.
The loudspeaker does not see a constant output impedance versus over the complete plus, zero, and minus signal voltage.
I hope this 3rd try illustrates the concept.
Yes?
No?
Last edited:
Bandersnatch,
OK.
I get your point.
I missed the point.
1. Either all loudspeakers need to have a perfectly flat impedance, over the full frequency range.
2. Or all power amplifiers need to have infinite damping factors.
Then whenever either number 1 is not met, Or number 2 is not met, we will still have perfectly flat frequency response for proper music reproduction.
Even though my 'missing the point' different discussion brought up another factor of SE and PP amplifiers, and that factor can interact with loudspeakers,
is that factor true?
OK.
I get your point.
I missed the point.
1. Either all loudspeakers need to have a perfectly flat impedance, over the full frequency range.
2. Or all power amplifiers need to have infinite damping factors.
Then whenever either number 1 is not met, Or number 2 is not met, we will still have perfectly flat frequency response for proper music reproduction.
Even though my 'missing the point' different discussion brought up another factor of SE and PP amplifiers, and that factor can interact with loudspeakers,
is that factor true?
Last edited:
Just to be clear, my disagreement was on your mis-use of the term Class A as you applied it to an AB amp.
cheers,
Douglas
cheers,
Douglas
I trust some marketing people as far as I can throw a 500 pound amplifier.
I think that Class A push pull amplifier is often just a marketing term.
Example:
Send the marketeer's push pull Class A amplifier to me, and I will drive the input with increasingly larger signals. Then, at some point, one of the tubes will be cut off, and then the other tube will be cut off, during the peaks of either the sine wave or the real world music signal.
Therefore, some "Class A" 'rated' amplifiers can enter "Class AB" (that is one of my points).
The Non-marketing term of such an amplifier should be 'Class AB'.
But that will lower the perceived value, and sales $$$ value of the amplifier (scaring marketeers).
Now there are some push pull amplifiers that will not ever cut off either one tube or the other tube.
That is a real Class A amplifier.
I think that Class A push pull amplifier is often just a marketing term.
Example:
Send the marketeer's push pull Class A amplifier to me, and I will drive the input with increasingly larger signals. Then, at some point, one of the tubes will be cut off, and then the other tube will be cut off, during the peaks of either the sine wave or the real world music signal.
Therefore, some "Class A" 'rated' amplifiers can enter "Class AB" (that is one of my points).
The Non-marketing term of such an amplifier should be 'Class AB'.
But that will lower the perceived value, and sales $$$ value of the amplifier (scaring marketeers).
Now there are some push pull amplifiers that will not ever cut off either one tube or the other tube.
That is a real Class A amplifier.
I'd use a slightly more restrictive term than 'some'. In audio, effectively none; they lie like they's selling jalopies.
cheers,
Douglas
cheers,
Douglas