A Zobel is an example of a non-resonating damper. A simple shunt resistor is too.
As for the frequency of the horn/driver's impedance peak, it is dependent on the acoustic load provided by the horn. In my experience, an exponential radial horn shifts this peak about 20% lower than a comparably sized conical or catenary horn with the same horizontal coverage angle.
Of course you would want a damper to provide impedance compensation if the impedance peak is in the passband. I'm not particularly concerned with what kind of damper is used for impedance compensation. Just as long as the impedance peak is brought in line, it's fine.
What I am concerned about, are horns that create response ripple even when connected to a voltage source, e.g. output impedance very low. In this case, we're not talking about impedance peaks potentially interacting with the crossover, we're talking about a resonant horn.
I should probably make a distinction between impedance peaks and response peaks. This is important when talking about electrical filters, whether they are used for impedance compensation or for response shaping.
An impedance peak can be manifested in the response curve, if the crossover isn't designed properly. But it doesn't have to create a response peak, if the crossover is properly designed. There are a handful of ways to provide impedance compensation, but some work better than others, depending on the circuit topology. So depending on the implementation, impedance peaks may be reflected in the response curve, but they don't necessarily have to be. An impedance peak can be troublesome, but it shouldn't be, as it is relatively easy to mitigate. An impedance peak certainly isn't a deal breaker.
Response peaks that occur even when the compression driver is connected directly to the voltage source are another matter entirely. If the horn creates ripple even when connected directly to the amp, impedance compensation won't help. Notch filters used for this purpose are response shaping circuits. To me, those are a deal breaker, and I do not like to use a horn that creates excessive ripple. In my experience, they sound unnatural even when used with response-shaping notch filters to bring the response curve flat.
As for the frequency of the horn/driver's impedance peak, it is dependent on the acoustic load provided by the horn. In my experience, an exponential radial horn shifts this peak about 20% lower than a comparably sized conical or catenary horn with the same horizontal coverage angle.
Of course you would want a damper to provide impedance compensation if the impedance peak is in the passband. I'm not particularly concerned with what kind of damper is used for impedance compensation. Just as long as the impedance peak is brought in line, it's fine.
What I am concerned about, are horns that create response ripple even when connected to a voltage source, e.g. output impedance very low. In this case, we're not talking about impedance peaks potentially interacting with the crossover, we're talking about a resonant horn.
I should probably make a distinction between impedance peaks and response peaks. This is important when talking about electrical filters, whether they are used for impedance compensation or for response shaping.
An impedance peak can be manifested in the response curve, if the crossover isn't designed properly. But it doesn't have to create a response peak, if the crossover is properly designed. There are a handful of ways to provide impedance compensation, but some work better than others, depending on the circuit topology. So depending on the implementation, impedance peaks may be reflected in the response curve, but they don't necessarily have to be. An impedance peak can be troublesome, but it shouldn't be, as it is relatively easy to mitigate. An impedance peak certainly isn't a deal breaker.
Response peaks that occur even when the compression driver is connected directly to the voltage source are another matter entirely. If the horn creates ripple even when connected directly to the amp, impedance compensation won't help. Notch filters used for this purpose are response shaping circuits. To me, those are a deal breaker, and I do not like to use a horn that creates excessive ripple. In my experience, they sound unnatural even when used with response-shaping notch filters to bring the response curve flat.
Last edited:
A damper isn't used to lower the resonant frequency. It is used to damp the resonance, to reduce the amplitude of the impedance peak(s).
If the peaks are right down at the lower end of the passband, you can use a series coil and resistor, applied in shunt across the driver terminals. This allows a lower value resistor to be used and yet remains non-resonant. An even simpler method is to just use a snubber resistor of about 2x the DCR of the voice coil. This limits Zmax without reducing Zmin very much.
A snubber resistor also tends to damp the minor resonances at higher frequency, usually those caused by diaphragm breakup. Then again, the higher frequency resonances tend to show up in the response much more than they do in the impedance curve. So where that is found, impedance compensation can't do very much because the impedance curve isn't very anomalous to begin with. In that case, it's pretty much a wash.
Sometimes you'll see 5dB (or greater) response ripple accompanied with less than a couple ohms impedance ripple. This is often the case with higher frequency horn resonances, which are what I tend to avoid - Can't do anything with them using a snubber or any other kind of electrical damper. Some designers are tempted to try and correct that using notch filters as response modifiers (as opposed to impedance modifiers). Again, I find that practice to be less than satisfactory. Some people use drivers with a ton of breakup and try and tame it with notch filters, others use peaky horns and try to mitigate them the same way. I'd suggest this is the wrong way to go - get a better driver and/or horn instead.
If the peaks are right down at the lower end of the passband, you can use a series coil and resistor, applied in shunt across the driver terminals. This allows a lower value resistor to be used and yet remains non-resonant. An even simpler method is to just use a snubber resistor of about 2x the DCR of the voice coil. This limits Zmax without reducing Zmin very much.
A snubber resistor also tends to damp the minor resonances at higher frequency, usually those caused by diaphragm breakup. Then again, the higher frequency resonances tend to show up in the response much more than they do in the impedance curve. So where that is found, impedance compensation can't do very much because the impedance curve isn't very anomalous to begin with. In that case, it's pretty much a wash.
Sometimes you'll see 5dB (or greater) response ripple accompanied with less than a couple ohms impedance ripple. This is often the case with higher frequency horn resonances, which are what I tend to avoid - Can't do anything with them using a snubber or any other kind of electrical damper. Some designers are tempted to try and correct that using notch filters as response modifiers (as opposed to impedance modifiers). Again, I find that practice to be less than satisfactory. Some people use drivers with a ton of breakup and try and tame it with notch filters, others use peaky horns and try to mitigate them the same way. I'd suggest this is the wrong way to go - get a better driver and/or horn instead.
- Speaker motors and passive crossover filters (A study of the performance of loudspeakers in the presence of other reactive components)
- Crossover Electronics 101 (Seminar Handout, study guide and helpful formulas)
Last edited:
I misspoke, it should read: "but it will NOT lower the amplitude of the drivers resonance."
A parallel resistor will not change the drivers electromechanical response to any appreciable degree.
Earl - Of course it will not modify the response of the driver. That's precisely what I've been saying the last several posts.
The last couple of pages have been about the differences between filters used for impedance compensation and versus those used for response shaping. I have made a pretty clear case that dampers are for impedance mitigation and not to modify response.
That's why I didn't follow you. You appear to be parroting me, but phrasing it as an opposing argument.
Impedance modifiers change the transfer function of the crossover. That's quite a different matter than modifying the horn's response, the way it would sound driven with a voltage source. Kinda been my point all along.
The last couple of pages have been about the differences between filters used for impedance compensation and versus those used for response shaping. I have made a pretty clear case that dampers are for impedance mitigation and not to modify response.
That's why I didn't follow you. You appear to be parroting me, but phrasing it as an opposing argument.
Impedance modifiers change the transfer function of the crossover. That's quite a different matter than modifying the horn's response, the way it would sound driven with a voltage source. Kinda been my point all along.
Wayne
You claimed that tank circuits in a crossover were a mistake. That is false and as I explained why you kept backing away into some nebulous and unsubstantiated argument that did not make sense to me. If you are now saying that tank circuits are fine for lowering the response peaks in compression driver then, hey, we are OK.
You claimed that tank circuits in a crossover were a mistake. That is false and as I explained why you kept backing away into some nebulous and unsubstantiated argument that did not make sense to me. If you are now saying that tank circuits are fine for lowering the response peaks in compression driver then, hey, we are OK.
No, Earl, you might try reading what I've said. I'm not being unclear, I'm distinguishing between electrical filters used for response shaping and those used for impedance compensation.
In my opinion, filters used for impedance compensation are fine, and in fact, often required. I don't care what type of filter is used for this purpose, although I generally prefer a non-resonant type.
I do not care for notch filters used for response shaping. I think if a horn needs response shaping to mitigate ripple, it is best used as an ash tray. Not very good for that either, 'cause it has a hole in it but better an ash tray than a sound source.
In my opinion, filters used for impedance compensation are fine, and in fact, often required. I don't care what type of filter is used for this purpose, although I generally prefer a non-resonant type.
I do not care for notch filters used for response shaping. I think if a horn needs response shaping to mitigate ripple, it is best used as an ash tray. Not very good for that either, 'cause it has a hole in it but better an ash tray than a sound source.
Last edited:
Well we fundamentally disagree here and you have offered no rational for not using a resonant filter for response shaping except to say that it does not track changes in the driver which I thought that we both agreed were minimal for a compression driver. I don't think that your objections have any merit.
Earl,
It seems that Wayne is saying that he will only use a compression driver with no anomalies in the pass band and perfect waveguide or horn that has zero ripple anywhere from the crossover point up. I would like to know what combination that is as I have never seen a perfect combination that can do that?
It seems that Wayne is saying that he will only use a compression driver with no anomalies in the pass band and perfect waveguide or horn that has zero ripple anywhere from the crossover point up. I would like to know what combination that is as I have never seen a perfect combination that can do that?
I don't mind 2dB ripple, and find that is a reasonable expectation. There are plenty of horn profiles that are capable of reaching that goal.
What I find unacceptable is using notch filters to knock off the peaks of a resonant horn, one that has large peaks and dips in its passband. Lots of constant directivity horns are like that, and no amount of response shaping makes them sound good, in my opinion.
What I find unacceptable is using notch filters to knock off the peaks of a resonant horn, one that has large peaks and dips in its passband. Lots of constant directivity horns are like that, and no amount of response shaping makes them sound good, in my opinion.
Last edited:
This is a change in what you said. You discounted resonant filters, period. Not just "to knock off the peaks of a resonant horn". I stated that there were resonances in the driver itself and that resonant filters were the only way to smooth out these resonances. My waveguides have no resonances, but ALL compression drivers have at least two and at least one of those virtually always needs to be smoothed and ONLY a resonant filter can do that.
Its kind of hard keeping track of what it is you are saying - it seems to change.
My position hasn't changed in years. In fact, I wrote the same things in the PDF files I linked earlier in this thread, and I wrote those documents over a decade ago.
I think you may be confusing a filter used for impedance compensation with a filter used for response shaping. Those two impedace peaks you're talking about will modify the transfer function of a passive crossover if they are in the passband, especially if padding is used.
I think you may be confusing a filter used for impedance compensation with a filter used for response shaping. Those two impedace peaks you're talking about will modify the transfer function of a passive crossover if they are in the passband, especially if padding is used.
Argumentum Ad Hominem ...
… simply does damage to its author's credibility only. Religious parallels have been drawn, yet again, to somehow mask the issues of acoustics as well as psychoacoustics. While classic measures of distortion do not track the perception of it well, that fact is certainly not a new discovery. Thanks to the work of Klippel, Voishvillo and you as well, a regimin can now be tailored that gets us closer to an optimum design, were a prescribed level of acoustic performance can be delivered at an obtainable minimum cost and expenditure of time.
WHG
… simply does damage to its author's credibility only. Religious parallels have been drawn, yet again, to somehow mask the issues of acoustics as well as psychoacoustics. While classic measures of distortion do not track the perception of it well, that fact is certainly not a new discovery. Thanks to the work of Klippel, Voishvillo and you as well, a regimin can now be tailored that gets us closer to an optimum design, were a prescribed level of acoustic performance can be delivered at an obtainable minimum cost and expenditure of time.
WHG
I am not confused, I am pretty sure I know what I am doing.
I don't distinguish "a filter used for impedance compensation" from "a filter used for response shaping". As far as I am concerned there is only one goal and that is delivering a desired acoustic response. There is a filter, period. This filter must work with the driver/waveguide combination to achieve the goal. Resonant legs can be part of this filter with no ensuing problems, but acoustic resonances in the horn should be avoided because they, unlike the two principle driver related ones, cannot be dealt with in an effective manner with a simple electrical circuit.
There is a fundamental difference between a filter used for impedance compensation and another used for response shaping. You've alluded to the difference in your last statement, saying "acoustic resonances in the horn should be avoided because they, unlike the two principle driver related ones, cannot be dealt with in an effective manner with a simple electrical circuit."
Every horn will have peaks in its electrical impededance curve at the low end of its passband. However, a horn with good acoustic loading will not have corresponding peaks in its response curve when provided a drive signal from a voltage source, e.g. one that has nearly zero ohms output impedance. So the behavior of a horn with good acoustic loading is fairly different than one with poor acoustic loading.
This is germane to the discussion about impedance compensation versus response shaping. A horn with good acoustic loading generally needs no response shaping to deal with ripple, but probably needs impedance compensation if used with a passive crossover. However, a horn with poor acoustic loading may have excessive ripple, and impedance compensation won't help it. This is also true if the mouth isn't sized right, or if the flare profile or any number of other attributes cause it to have ripple. If a horn creates ripple in its passband even when driven by a voltage source, no amount of impedance compensation will help because that's not the problem. Such a horn is a resonant device, and the only way to smooth its response is with some form of equalization.
But even a horn that presents a good acoustic load will still have peaks in its electrical impedance curve. They are usually lower and of different shape than a horn that presents a poor acoustic load, but all horns create these impedance peaks at some level. These peaks are large enough to be reflected in the response curve when driven with a current source. But when a horn is driven by a voltage source, then the horn/driver's impedance peaks are not reflected into the response curve.
So impedance compensation must be used on a horn that has impedance peaks in the passband, esepcially if there is resistance in tweeter circuit, like what is used for padding. This impedance compensation is provided by some form of damping, and can be integral to the padding network or can be an additional shunt leg, used specifically for impedance compnesation.
Impedance compensation will modify the transfer function, but what sets it apart from a network used for response shaping is the damper circuit is really only needed to keep the load right for the filters. Again, it is to make the transfer function more like what it would be if the source impedance were zero.
On the other hand, a horn that has ripple in its response curve even when connected to a voltage source cannot be corrected solely with impedance compensation. Impedance compensation may be also required, but by itself, that's not all that is needed. A horn that produces peaks even when driven directly by an amplifier requires response modification, some form of equalization. This is a different matter entirely. You could connect such a horn directly to an amplifier, where no impedance modification is needed, but it would still be peaky and would still need equalization.
As for the two peaks at low frequency, not all horns exhibit this in their response curve. Exponential horns, for example, provide enough acoustic loading to provide smooth response even at low frequencies. Conical horns and waveguides provide much poorer acoustic loading, so they tend to have peaks at low frequency, but some are better than others.
I think it is good to say this again for emphasis: Not all horns create response peaks at low frequency. So to say it is a trait of the compression driver is somewhat misleading. It is a trait of the system, but not purely the compression driver. Just like the resonant peak of a direct radiator is modified by the box it is installed in, so is the resonant peak of a compression driver. The behavior is set by the system, not just one of its parts.
Some horns generate a nearly flat response curve, from their lower cutoff point all the way up to their upper cutoff. Some horns have a single peak at cutoff, followed by a relatively flat response. Others have more peaks, and some of the worst have periodic ripple all the way through the passband. So but my point is that not all horns create peaks at the lower end of their response curves. I would argue that the amount of ripple is one of the quality factors of the horn.
Every horn will have peaks in its electrical impededance curve at the low end of its passband. However, a horn with good acoustic loading will not have corresponding peaks in its response curve when provided a drive signal from a voltage source, e.g. one that has nearly zero ohms output impedance. So the behavior of a horn with good acoustic loading is fairly different than one with poor acoustic loading.
This is germane to the discussion about impedance compensation versus response shaping. A horn with good acoustic loading generally needs no response shaping to deal with ripple, but probably needs impedance compensation if used with a passive crossover. However, a horn with poor acoustic loading may have excessive ripple, and impedance compensation won't help it. This is also true if the mouth isn't sized right, or if the flare profile or any number of other attributes cause it to have ripple. If a horn creates ripple in its passband even when driven by a voltage source, no amount of impedance compensation will help because that's not the problem. Such a horn is a resonant device, and the only way to smooth its response is with some form of equalization.
But even a horn that presents a good acoustic load will still have peaks in its electrical impedance curve. They are usually lower and of different shape than a horn that presents a poor acoustic load, but all horns create these impedance peaks at some level. These peaks are large enough to be reflected in the response curve when driven with a current source. But when a horn is driven by a voltage source, then the horn/driver's impedance peaks are not reflected into the response curve.
So impedance compensation must be used on a horn that has impedance peaks in the passband, esepcially if there is resistance in tweeter circuit, like what is used for padding. This impedance compensation is provided by some form of damping, and can be integral to the padding network or can be an additional shunt leg, used specifically for impedance compnesation.
Impedance compensation will modify the transfer function, but what sets it apart from a network used for response shaping is the damper circuit is really only needed to keep the load right for the filters. Again, it is to make the transfer function more like what it would be if the source impedance were zero.
On the other hand, a horn that has ripple in its response curve even when connected to a voltage source cannot be corrected solely with impedance compensation. Impedance compensation may be also required, but by itself, that's not all that is needed. A horn that produces peaks even when driven directly by an amplifier requires response modification, some form of equalization. This is a different matter entirely. You could connect such a horn directly to an amplifier, where no impedance modification is needed, but it would still be peaky and would still need equalization.
As for the two peaks at low frequency, not all horns exhibit this in their response curve. Exponential horns, for example, provide enough acoustic loading to provide smooth response even at low frequencies. Conical horns and waveguides provide much poorer acoustic loading, so they tend to have peaks at low frequency, but some are better than others.
I think it is good to say this again for emphasis: Not all horns create response peaks at low frequency. So to say it is a trait of the compression driver is somewhat misleading. It is a trait of the system, but not purely the compression driver. Just like the resonant peak of a direct radiator is modified by the box it is installed in, so is the resonant peak of a compression driver. The behavior is set by the system, not just one of its parts.
Some horns generate a nearly flat response curve, from their lower cutoff point all the way up to their upper cutoff. Some horns have a single peak at cutoff, followed by a relatively flat response. Others have more peaks, and some of the worst have periodic ripple all the way through the passband. So but my point is that not all horns create peaks at the lower end of their response curves. I would argue that the amount of ripple is one of the quality factors of the horn.
Wayne - the issue is your criticizing Bill Waslo for using notch filters.
He and I both use them and there is nothing wrong with that. If I didn't need to use them, I wouldn't (why do something that is unnecessary?) You seem to be implying that you can get an ideal response without them and I don't see how. All of this "response shaping" versus "impedance shaping" is just talking around the issue.
He and I both use them and there is nothing wrong with that. If I didn't need to use them, I wouldn't (why do something that is unnecessary?) You seem to be implying that you can get an ideal response without them and I don't see how. All of this "response shaping" versus "impedance shaping" is just talking around the issue.
Maybe I need to reemphasis that all horn/driver combinations as a "system" will have peaks at low frequencies - even if the "horn" does not.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.