advanatge of 1s order over 2nd or 3rd order?

[Professor smith]

whgeiger you didnt address my question and where has it been shown that multi pole crossover can produce phase coherence? i need a clear explanation of that.

 

[whgeiger]

WHG, I guess I misunderstood what you were saying. I thought you were saying that a single electrical pole plus the poles of the drivers' natural roll-off yield multiple poles in the acoustic response. I certainly agree with that but....
Filters to me implies electrical response. A single cap may give a desired 3rd or 4th order acoustic response.
A single cap can provide plenty of protection if done right. See my earlier WinISD sim.

Isues recast as follows from previous posts:

1) In the first instance all polls, electrical and acoustical, were considered as the resulting loudspeaker system was being characterized as 1st. order.

2) In the second instance polls were being added in the shallow filter stop band to steepen it; nevertheless that action changes the filter's order as well.

3) As I prefer to do all signal tailoring in the digital domain, DC blocking capacitors remain in place on the analog side for MF and HF driver protection.

4) Even though acoustic response is rolling off as frequency declines, HF and MF driver excursions continue to double per octave when the driving signal is only being attenuated 6 db/oct. While the attenuated frequencies are inaudible, primarily due to small diaphragm size, these excursions impose artifacts on the higher frequencies as audible IM distortion. In the case of horn loaded compression drivers, that have phasing plugs in close proximity to their diaphragms, the power handling capacity of the entire system becomes constrained by the reduced excursion limits due entirely to the lack of an adaquate high pass filter slope.

Regards,

WHG

 

[pheonix358]

As I fed 200Watts of music into my 1st order 3 ways (2 coils, 2 caps 2 resistors) today I thought of your posts. Damn I am proud of myself. I am capable of doing the impossible! I said to my speakers "How are you feeling guys." They let loose with 'Another brick in the wall.' Yea man, that's what I'm talking about!

Morel 2x MW267, 2x MW144, 2x MDT337 (+/- 2dB 35Hz - 18KhZ) Got to love the impossible!

 

[whgeiger]

Voicing

whgeiger,

But what about the issue of timbre? By using a high order crossover does this not change the relative harmonics? That timing is what characterises the timbre and sound of an instrument or voice. Can this be sacrificed?

The frequencies that determine timber of a musical instrument typically are found at frequencies well above the range of its note fundamentals and will be reproduced by the HF driver alone. For this purpose a three way system with decade frequency band assignments. Low: 20-200Hz, Mid: 200-2000 Hz High 2000-20,000 Hz should minimize the the time smear between timber producing frequency components irrespective of filter order employed. Furthermore, it has been clearly demonstrated in this thread [1], as well as elsewhere [2], that it is not a prerequisite that phase coherence need be sacrificed when multi-poll filter sections are deployed in a crossover network. Alternatively, significant improvements may be realized in the performance of any driver assemblage when all passive components, including inductors, resistors and shunt capacitors, are removed from their low impedance drive circuits. This of course introduces power amplifiers dedicated to each driver type with all signal conditioning and separation being performed beforehand in either the digital or analog voltage domains where a high constant impedance is present. Here filters may be tailored to driver requirements with relative ease, not the other way around. In a DIY setting such arrangements provide indispensable flexibility in perfecting a system design even though the cost of additional power amplifiers must be sustained.

Regards,

WHG

References

[1] In this Link

http://www.diyaudio.com/forums/mult...s-order-over-2nd-3rd-order-8.html#post2304019

http://www.diyaudio.com/forums/mult...s-order-over-2nd-3rd-order-8.html#post2304197

[2] Examples from Elsewhere

Crossovers

(((acourate)))® - Room Correction, Speaker Optimization and Sound Improvement

Classical & Linear Phase Crossover | Lab.gruppen

 

[whgeiger]

Voicing

whgeiger,

But what about the issue of timbre? By using a high order crossover does this not change the relative harmonics? That timing is what characterises the timbre and sound of an instrument or voice. Can this be sacrificed?

The frequencies that determine timber of a musical instrument typically are found at frequencies well above the range of its note fundamentals and will be reproduced by the HF driver alone. For this purpose a three way system with decade frequency band assignments. Low: 20-200Hz, Mid: 200-2000 Hz High 2000-20,000 Hz should minimize the the time smear between timber producing frequency components irrespective of filter order employed. Furthermore, it has been clearly demonstrated in this thread [1], as well as elsewhere [2], that it is not a prerequisite that phase coherence need be sacrificed when multi-poll filter sections are deployed in a crossover network. Alternatively, significant improvements may be realized in the performance of any driver assemblage when all passive components, including inductors, resistors and shunt capacitors, are removed from their low impedance drive circuits. This of course introduces power amplifiers dedicated to each driver type with all signal conditioning and separation being performed beforehand in either the digital or analog voltage domains where a high constant impedance is present. Here filters may be tailored to driver requirements with relative ease, not the other way around. In a DIY setting such arrangements provide indispensable flexibility in perfecting a system design even though the cost of additional power amplifiers must be sustained.

Regards,

WHG

References

[1] In this Link

http://www.diyaudio.com/forums/mult...s-order-over-2nd-3rd-order-8.html#post2304019

http://www.diyaudio.com/forums/mult...s-order-over-2nd-3rd-order-8.html#post2304197

[2] Examples from Elsewhere

Crossovers

(((acourate)))® - Room Correction, Speaker Optimization and Sound Improvement

Classical & Linear Phase Crossover | Lab.gruppen

 

[catapult]

4) Even though acoustic response is rolling off as frequency declines, HF and MF driver excursions continue to double per octave when the driving signal is only being attenuated 6 db/oct.

You keep repeating that but it's simply not true. Per my previous post, here's the cone excursion of a mid tuned to Q=.5 at 105 with and without a 6dB-500 highpass. The excursion peaks at the 105 resonance and goes down below there. Below resonance, the SPL is rolling off at 18dB, 12dB from the box and 6dB from the filter.

 

[whgeiger]

No Surprise

The LF energy is being dissipated in the voice coil instead of moving the diaphragm. This is not a surprise. See my post:
advanatge of 1s order over 2nd or 3rd order? - Page 6 - diyAudio
Note that this is not a 1st. order response for the driver in question.
Regards,
WHG

 

[catapult]

Note that this is not a 1st. order response for the driver in question.

But it's a first order electrical filter which you have steadfastly proclaimed could not protect the driver from over-excursion. It does.

 

[pheonix358]

But it's a first order electrical filter which you have steadfastly proclaimed could not protect the driver from over-excursion. It does.

Of course it does, that's why there are heaps of successful and very fine examples in the real world which is where most of us live!

 

[whgeiger]

Sans Caveat

But it's a first order electrical filter which you have steadfastly proclaimed could not protect the driver from over-excursion. It does.

I should have included this caveat in all my posts but I did not do so. I will repeat it here again from the earlier referenced post. "Please note that when a signal is feed to a driver, and it does not respond proportionally to it, the difference gets translated into additional heat in the voice coil. If it does respond, then at a minimum, increased amounts of IM distortion will be introduced into the output due to the Doppler effect and operation of the driver suspension in its non-linear range."

Regards,

WHG

 

[Curmudgeon]

Re catapult's nice simulation; actually, it looks as if the excursion is ~.2 at500 Hz, and ~.4 at 250, so doubling in an octave. Thereafter, of course, the box reduces the excursion.

Certainly the final combined acoustic and electrical rolloffs exceed first order when looked at over the full range of each driver. It's generally understood that "first order" applies to the area around the xover region(s), where it really matters.

There have been several design houses that have successfully sold very good nominally first order systems; so whether or not it's possible should be a moot point...

Finally, and I hope not too unkindly, I have quite enjoyed the image of polls and Xeroxes gambling on the imaginary timbered plain...😛

 

[whgeiger]

Q & A

whgeiger you didnt address my question and where has it been shown that multi pole crossover can produce phase coherence? i need a clear explanation of that.

I answered this question in the response to your earlier question.

Regards,

WHG

 

[dlr]

If a question was asked by me, it was rhetorical and not a solicition for a response. If you consider responding to my posts a waste of time, then simply move. You were not asked to do so. In the meantime, just call it what it is. The number of polls you use equals the electrical filter order. The frequency limits of each drivers acoustical response will approximate some order of high and low pass filter as well, and they certainly will not be 1st. order as much as you might like them to be. As the text in this thread clearly demonstrates, phase coherence can be successfully approached using multi-poll analog filters. End of story.
Regards,
WHG

Crossovers for well implemented first order acoustic responses can be rather complex and are used to reduce the response to first order to some desired point into the stop band. As far as poles are concerned, the only issue is the poles of the acoustic response, as tailoring the acoustic response is the consideration. As long as the frequency limits of the final acoustic response matches the target to a reasonable point in the stop band, it is still effectively a system as described, outside of academic discussion, since all systems have tradeoffs, even yours. You just disagree with those of first order acoustic.

You may not prefer a system that others find satisfactory, that's fine, that is your choice. I note your lack of comment regarding John Dunlavy's many first order systems. Professional in the field, manufacturer, background in waveguide and electromagnetic theory with a history in antenna design. Yet he called them first order. Many found them to be among the best systems made at the time. That alone is sufficient to support the idea that a first order acoustic system can be quite satisfactory and successful commercially. There are others, of course, such Thiel. Very successful commercially. We're not talking pro audio and most here are into DIY for home use such as the two commercial manufacturers just mentioned.

Below is an interesting quote from an interview found here (highlights mine). Note that he is not referring to the electrical crossover, but the acoustic crossover, though he uses the term network.

Dunlavy: That's right. You can go one way but not the other. Flat on-axis response provides very poor correlation, in general, with what you hear. On the other hand, if you have near-perfect impulse and step responses, it follows that you must also then have from that a near-perfect frequency response on-axis. And using a first-order crossover network is the only way you can achieve accurate impulse and step responses. As soon as you go to a second-order crossover, the impulse response is hideous.

Whether you agree or not, even the industry uses this reference. That is not likely to change, much to your chagrin I'm sure. People usually understand the limits in the stop band, yet the industry still calls it first order.

I would also ask how you define phase coherence?

Dave

 

[dlr]

I missed the time limit on editing above, so I'll add the quote from Thiel's white paper here:

Complete phase coherence (the elimination of phase distortion) achieved by the use of first-order crossovers.

Generally accepted term in the industry. Not about to change, I'm sure.

Dave

 

[speaker dave]

You keep repeating that but it's simply not true. Per my previous post, here's the cone excursion of a mid tuned to Q=.5 at 105 with and without a 6dB-500 highpass. The excursion peaks at the 105 resonance and goes down below there. Below resonance, the SPL is rolling off at 18dB, 12dB from the box and 6dB from the filter.

In its passband where a driver is flat its excursion will increase increase 12dB per octave as you descend in frequency. That is, 4 times the excursion for half the frequency. Either electrical rolloff or acoustical/mechanical rolloff below resonance will decrease the excursion. If the intention is to maintain a first order acoustical rolloff then you will still have excursion doubling for every Octave you go down. You need greater than 12 dB/octave rolloff (either acoustical or electrical) to see excursion decrease at all for lower frequencies.

Clearly if you have the available excursion or don't mind giving up some power handling then you may be able to get by with a first order network. Still, higher orders will always reduce out of band excursion.

 

[Professor smith]

Furthermore, it has been clearly demonstrated in this thread [1], as well as elsewhere [2], that it is not a prerequisite that phase coherence need be sacrificed when multi-poll filter sections are deployed in a crossover network.

Trouble is this is at odds with what professor Dunlavy has stated in this interview:

Stereophile: Loudspeaker designer John Dunlavy: By the Numbers...

Quote: ''And using a first-order crossover network is the only way you can achieve accurate impulse and step responses. As soon as you go to a second-order crossover, the impulse response is hideous. ''

What is your response to that?

 

[catapult]

Man, I'm getting tired of repeating myself in all your threads. 🙂

HTGuide Forum - Duelund meets Dunlavy (aka Duelund meets transient perfect)

The response of this system is theoretically better than a pure first-order acoustic target for a multi-way because it maintains the phase relationships of all the drivers all the way from the lowest frequencies to the highest. While the slope is similar to a first-order system at the nominal XO frequencies, it gets steeper later. It's a 5-way system:

Tweeter: 4th order HP (acoustic response)
Mid-tweeter: 3rd order HP and 1st order LP (acoustic response)
Mid: 2nd order HP and 2nd order LP (acoustic response)
Mid-woofer: 1st order HP and 3rd order LP (acoustic response)
Woofer: 4th order LP (acoustic response)

 

[whgeiger]

Trouble is this is at odds with what professor Dunlavy has stated in this interview:

Stereophile: Loudspeaker designer John Dunlavy: By the Numbers...

Quote: ''And using a first-order crossover network is the only way you can achieve accurate impulse and step responses. As soon as you go to a second-order crossover, the impulse response is hideous. ''

What is your response to that?

I have provided references that clearly demonstrate phase coherence (between input and output signal) being delivered through multiple-poll networks. I will leave it to the author of the referenced quote to defend his position, (if he so chooses) as I have enough grief just defending my statements here. Sometimes the context specifics change, and what was true for one set is false for another.

Regards,

WHG

 

[whgeiger]

MFR

Crossovers for well implemented first order acoustic responses can be rather complex and are used to reduce the response to first order to some desired point into the stop band. As far as poles are concerned, the only issue is the poles of the acoustic response, as tailoring the acoustic response is the consideration. As long as the frequency limits of the final acoustic response matches the target to a reasonable point in the stop band, it is still effectively a system as described, outside of academic discussion, since all systems have tradeoffs, even yours. You just disagree with those of first order acoustic.

You may not prefer a system that others find satisfactory, that's fine, that is your choice. I note your lack of comment regarding John Dunlavy's many first order systems. Professional in the field, manufacturer, background in waveguide and electromagnetic theory with a history in antenna design. Yet he called them first order. Many found them to be among the best systems made at the time. That alone is sufficient to support the idea that a first order acoustic system can be quite satisfactory and successful commercially. There are others, of course, such Thiel. Very successful commercially. We're not talking pro audio and most here are into DIY for home use such as the two commercial manufacturers just mentioned.

Below is an interesting quote from an interview found here (highlights mine). Note that he is not referring to the electrical crossover, but the acoustic crossover, though he uses the term network.



Whether you agree or not, even the industry uses this reference. That is not likely to change, much to your chagrin I'm sure. People usually understand the limits in the stop band, yet the industry still calls it first order.

I would also ask how you define phase coherence?

Dave

Coming from you this is obviously a rhetorical question, but I will answer it anyway for the purposes of a MFR.

If signal phase remains materially unchanged between the input to, and the output from, a network, then such a network may be characterized as phase coherent.

Many others in the "industry" claim that the phase deviations found in typical (properly designed) multi-poll crossovers are inaudible. So I suspect the debate will continue.

My position remains, that if there is more than one filter poll in the signal path, the network it is not 1st. order; it is higher, irrespective of whether it is phase coherent or not.

Also the drivers that are being fed filtered signals are not phase coherent as well, nor can they be accurately modeled as single pole, 1st. order networks.

When the voice coil of a driver receives a signal that is of a frequency below the drivers pass band, that condition may give rise to the generation of distortion products within its pass band that are audible or worse.

While I am quite prepared to argue the facts surrounding an issue, I will not be drawn into a critique of the work of others. That is not the purpose of this forum, at least as I understand its mission here.

Regards,

WHG

 

[dlr]

While I am quite prepared to argue the facts surrounding an issue, I will not be drawn into a critique of the work of others. That is not the purpose of this forum, at least as I understand its mission here.

Regards,

WHG

My understanding of this forum is that anyone may discuss any aspect of any multi-way loudspeaker system at whatever level one wishes to include debate on previous work. The only limitation is appropriately "Conventional loudspeakers with crossovers".

When the voice coil of a driver receives a signal that is of a frequency below the drivers pass band, that condition may give rise to the generation of distortion products within its pass band that are audible or worse.

The operative word is "may". Many problems may arise. The key is designing to minimize such for the goal desired. Things that may arise do not preclude a design that minimizes them, to include out-of-band signal issues. Even first order if done right. Then the question becomes is it audible during the intended usage. All systems reach a level at which they will distort.

Critique is quite common. Besides, you have already indirectly, though significantly, critiqued the basis of the works of others such as Dunlavy without reservation.

Dave