Replace resistor by low-inductance resistor - Is it audible?

[Francis Vaughan]

Hello All,
There would be no damping factor if the the driver impedance and series crossover resistors were not in the amplifier feedback loop.

This makes no sense. The load is not part of the feedback loop in a voltage source amplifier. Damping factor is the ratio of the output impedance of the amplifier to the load impedance. For an amplfier using negative feedback output impedance is the impedance of the output devices divided by the feedback factor. Feedback serves to reduce the apparent output impedance of the output devices, and thus define the overall output impedance of the amplifier. There is no load in the feedback loop.

Damping factor is a pretty useless metric at best. Obviously it changes with the impedance of the load, but that is simply how it is defined. So most amplifiers have the damping factor calculated into an 8ohm load. But that is just for marketing. You can obviously calculate the actual output impedance of the amplifier by dividing 8 ohms by the damping factor. That gets a more useful metric - one that is independent of the load.

A conventional loudspeaker is designed to be driven by a voltage source. So how close to a voltage source the amplifer is can matter. Back in the days of steam driven Hi-Fi output impedances were much higher: tubes inherently have a much higher plate impedance than transistors, there is a transformer with its copper and iron losses, and the reduced open loop gain and bandwidth meant low feedback factors. Output impedances of the order of ohms not uncommon. So in such cases, speakers could indeed be sensitive to the amplifier. But even here, damping factor is a just silly figure of merit.

 

[DualTriode]

This makes no sense.

This makes perfect sense.

Feedback is what reduces the amplifier output impedance.

The driver will typically have a high resonance peak or high "Q" where the driver resonates and generates a voltage.
From the point of view of the amplifier this is a Back EMF that goes back through the amplifier negative feedback loop and damps the drivers resonate frequency.

Not so silly after all.

Thanks DT

 

[Francis Vaughan]

This makes perfect sense.

Feedback is what reduces the amplifier output impedance.

The driver will typically have a high resonance peak or high "Q" where the driver resonates and generates a voltage.
From the point of view of the amplifier this is a Back EMF that goes back through the amplifier negative feedback loop and damps the drivers resonate frequency.

Not so silly after all.

Thanks DT

None of what you wrote puts the speaker in the feedback loop. This is a common misconception. The amplifier looks like a voltage source. The feedback loop makes it look like a better voltage source. The Thevenian equivalent is a perfect voltage source (zero impedance) and a series resistance. You cannot tell externally how this is implemented.
It is no different to the thermostat on a heater. Opening the window in a room causes the thermostat to turn up the heater. The window is not part of the feedback loop.

Back EMF is one part of looking at how a conventional loudspeaker works, and is a core part of the basic analysis by Neville Theile. The source impedance forms part of that analysis, and a very low source impedance makes speaker design easier. There is no intrinsic reason why you need a very low source impedance, and in the extreme, current source loudspeakers have a certain niche following. They have the advantage of being insensitive to dynamic changes in speaker impedance, both as a result of voice coil heating and changes in voice coil inductance with excursion. You need to compensate for the rising frequency response, but they work fine.

 

[DualTriode]

Wonder how amplifiers deal with voice coil inductances? Or speaker wire?

Yes, minuscule. 2/3 of a microhenry. Most tweeters have two orders of magnitude greater inductance. Step up a couple more orders of magnitude for midranges and woofers.

2/3 of a microhenry is not a very good example of a inductive resistor. I think.

Chances are that there a lot of resistors used in crossovers that have a lot more than 2/3 of a microhenry inductance.

How much change in resistor impedance and phase shift are you willing to accept as being inaudible?

 

[Bruce Morgen]

2/3 of a microhenry is not a very good example of a inductive resistor. I think.

Chances are that there a lot of resistors used in crossovers that have a lot more than 2/3 of a microhenry inductance.

How much change in resistor impedance and phase shift are you willing to accept as being inaudible?

Phrases like "I think" and 'Chances are" rob any reply of the slightest semblance of persuasive effect.

 

[SIY]

2/3 of a microhenry is not a very good example of a inductive resistor. I think.

Yet that's what the OP measured. And it's pretty typical of the values I've measured as well. That is absolutely negligible at 20kHz, and even more negligible compared to the other inductances in the circuit. This is absolutely a non-issue.

Cheap Chinese 3R9 25W sandcast:

Dummy load of 2 100W cheap Chinese 4 ohm sandcast at the end of 2 meters of wire:

2" midrange/tweeter:

 

[DualTriode]

None of what you wrote puts the speaker in the feedback loop. This is a common misconception. The amplifier looks like a voltage source. The feedback loop makes it look like a better voltage source. The Thevenian equivalent is a perfect voltage source (zero impedance) and a series resistance. You cannot tell externally how this is implemented.
It is no different to the thermostat on a heater. Opening the window in a room causes the thermostat to turn up the heater. The window is not part of the feedback loop.

Back EMF is one part of looking at how a conventional loudspeaker works, and is a core part of the basic analysis by Neville Theile. The source impedance forms part of that analysis, and a very low source impedance makes speaker design easier. There is no intrinsic reason why you need a very low source impedance, and in the extreme, current source loudspeakers have a certain niche following. They have the advantage of being insensitive to dynamic changes in speaker impedance, both as a result of voice coil heating and changes in voice coil inductance with excursion. You need to compensate for the rising frequency response, but they work fine.

Amplifier negative feedback is taken from the amplifier output and routed through a voltage divider to an earlier inverting stage in the amplifier.

The driver attaches to the same amplifier output.

The driver is a motor that at resonance becomes a generator that pushes back EMF at the amplifier output connection. The back EMF at the amplifier output loops back through the amplifier feedback loop. This is the definition of negative feedback.

The open window is a poor analogy of the feedback loop. Closer is warm air from the grille blows directly on the heat sensor, the sensor warms and the call for heat is throttled. This is closer in concept to an anticipator heater. It reduces the wide swing in temperature before the next call for heat. This is known in the HVAC world as negative feedback.

 

[SIY]

The driver is a motor that at resonance becomes a generator that pushes back EMF at the amplifier output connection. The back EMF at the amplifier output loops back through the amplifier feedback loop. This is the definition of negative feedback.

So you're claiming that the couple of microhenries (generously!) of a crossover resistor is going to shoot uncontrollable back EMF into an amplifier? But the 40-100 uH of a tweeter won't bother the amp?

That is... novel.

 

[Francis Vaughan]

The driver is a motor that at resonance becomes a generator that pushes back EMF at the amplifier output connection. The back EMF at the amplifier output loops back through the amplifier feedback loop. This is the definition of negative feedback.

The speaker is still not part of the loop. The loop is the section of the circuit that leaves the output and returns to the inverting input, maybe via a potential divider. There is a reason why it is called a loop. Heck, look at a unity gain amplifier, or an inverting amplifier with gain. It should be obvious that the speaker is not in the feedback loop. Maybe you are confused by the common way of drawing an non-inverting amp.

 

[DualTriode]

@SIY,

I will not disturb your fantasy of uncontrolled back EMF shooting around in your head.

I am assuming that the amplifier is nearly perfect with sub ppm distortion into an ideal resistive load. I am also thinking that parts per hundred of frequency dependent, variable impedance may be important.

How much impedance change and phase shift does it take to be audible?



@Francis Vaughan,

It should be obvious, it is not that simple. “It is still not part of the loop.”

There may not be a single loop. Wrap your head around it. There may be global feedback. There may be local feedback. There may be summing nodes. There need not be only a single feedback loop. For the amplifier we are speaking of we do not need to know the details of the feedback. All we need to know is that the amplifier and the driver interact. That is all that is necessary to say that there is feedback. If it helps you, call the amplifier output a summing node.

 

[SIY]

I am also thinking that parts per hundred of frequency dependent, variable impedance may be important.

You can think anything you like. However, lacking even a shred of evidence of "importance," we're back to fairies in the garden. Fifteen seconds with the voltage divider equations might prove enlightening to you, and I strongly suggest that. Much better than hand-waving.

It should be obvious, it is not that simple. “It is still not part of the loop.”

There may not be a single loop. Wrap your head around it. There may be global feedback. There may be local feedback. There may be summing nodes. There need not be only a single feedback loop. For the amplifier we are speaking of we do not need to know the details of the feedback. All we need to know is that the amplifier and the driver interact. That is all that is necessary to say that there is feedback. If it helps you, call the amplifier output a summing node.

I think what Francis is trying to tell you is that there's two input and two output terminals to the amplifier. ANY amp feedback loop exists somewhere between them. The load is external. "Interact" is NOT the same thing as "the load is part of the amplifier's feedback network."

 

[Francis Vaughan]

There may not be a single loop. Wrap your head around it. There may be global feedback. There may be local feedback. There may be summing nodes. There need not be only a single feedback loop. For the amplifier we are speaking of we do not need to know the details of the feedback. All we need to know is that the amplifier and the driver interact. That is all that is necessary to say that there is feedback. If it helps you, call the amplifier output a summing node.

You know the words. Seriously, how about using them in the accepted manner? You insist the speaker is within the feedback loop. You know the definitions of the terms, so how about using them correctly? The speaker is not part of the feedback loop. That is by accepted definition of terms. It is, as you say connected to the output node. Nobody disputes that. Nor does anyone dispute that a real life feedback loop can be disturbed by reactive loads. But if you insist on using technical terms in a manner that is different to the accepted definition, you can hardly expect people to be mind readers and divine your intended meaning.

 

[RayDunzl]

Easy Answer:

The speaker is not part of the feedback loop if you can remove it and the loop is still there ready for its next victim.

 

[DualTriode]

Yes I do know the words and I am using them in the normal sense. It is just not in the way you used to thinking about it.

Think of a large sub-woofer as a large moving machine. Once it gets moving it wants to stay moving until it rolls to a stop. Now think of the large woofer as a large motor. Connect an oscilloscope across the driver red and black terminals, tap the driver cone with your finger, and observe the oscilloscope screen you will see a decaying signal at the resonate frequency of the driver. Now add a resistor equal to the driver impedance across the speaker terminals. Now tap the driver cone again, you will see the signal on the oscilloscope screen stop much sooner this time.

Next we connect our amplifier output to the speaker, our frequency generator to the amplifier input and our oscilloscope probe to the amplifier inverting input location where the feedback loop lands.

Now adjust everything to a moderate amplifier output at the driver resonate frequency. While watching the oscilloscope screen turn off the frequency generator. Now what you see on the oscilloscope screen is the voltage generated by the resonating driver. It will not last long because the voltage generated by the voice coil is amplified 180 degrees out of phase and is the input to the driver. What we have is the amplifier actively damping the motion of the diver motor. This is feedback, loop network whatever you want to call it. Servo woofers are designed on this concept.

 

[RayDunzl]

To me, that's like saying the hill becomes part of the car as the Cruise Control feeds a little more fuel and air to the engine (or ups the voltage delivered to the drive motor in an e-car) to maintain speed.

It's an external "something" the feedback loop attempts to negate the effect of.

I don't understand your argument.

Carry on!

 

[solderdude]

A feedback loop does respond to signals injected in a feedback loop but that injected signal is not part of the feedback loop.

 

[Francis Vaughan]

. It will not last long because the voltage generated by the voice coil is amplified 180 degrees out of phase and is the input to the driver. What we have is the amplifier actively damping the motion of the diver motor. This is feedback, loop network whatever you want to call it. Servo woofers are designed on this concept.

Sorry, again no. There is no active damping. Again, we need to be clear, we are talking about conventional voltage source amplifiers for audio reproduction. The anti-phase output from the amplifier will act to exactly hold the output of the amplifier at the value dictated by the input within the margins allowed by output device impedance and feedback factor. It will not change further than this. It will not act to counter the cone movement with a reverse drive. The output node of the amplifier is indistinguishable from a resistor of the same value as the effective output impedance of the amplifier. If you perform your experiment, the voltages seen and movement of the woofer will be identical whether you use a voltage source amplifier or a passive resistor.
Now, it is possible to build systems that sense the back-emf and use it to predict movement of the cone in a manner that allows for active damping. (Another application is speed control for simple universal AC motors.) But these are not what we are talking about. Servo subwoofers could use it, most prefer to use external sensing of cone movement. Such systems add another control loop (feedback or often a feedforward or other control network) that is separate to the negative feedback loop around the amplifier. These systems require significant additional analysis, they contain many sources of potential instability. Currently they only exist for bass drivers.
You can't start moving the goalposts and say that "well there are other control systems that could react to back emf." Conventional passive loudspeakers, ones with passive crossover components, which is where we started, do not and cannot act as you describe. I'm staying with where we started - conventional voltage source amplifiers with negative feedback driving a full range multi-way speaker with a passive crossover. The assertion was made that inductance in resistors within the crossover somehow formed part of the feedback loop of the amplifier. This was, and still is, simply incorrect.

 

[Dimitri]

I don't understand your argument.

If you understood the misundrestanding, I would be worried

 

[pma]

For those who have a clue what the loopgain stability is, there is a comparison of 0R22 resistor non-inductive and with 1uH inductance, as a Re resistor in output stage of the class AB amplifier with global negative feedback. I am sorry I would be tired to argue with anyone that resistor inductance does not matter.

 

[JohnYang1997]

This particular test the OP proposed is a specific place where resistor inductance does not matter. Of course there are many other places the inductance surely will matter.