Identifying Legitimately High Fidelity Loudspeakers: Myths & Facts about Crossovers

[agarwalro]

Sound power is an anechoic, imaginary, and maybe measured but mostly "calculated" concept. Think of it as the sound being radiated into all 360 spherical directions... averaged.

Power response is typically done anechoic in 10-15deg increments spherically around the speaker. The averaged results are then weighted according to the proportional area of the sphere represented by each measurement.

Thanks for the clarification. I should have goggled it before posting.

Balderdash, says the far-field power response crowd. The room reflections etc. are so close in time to the first arrival that they swamp the first arrival completely, leaving the overall far-field energy output of the speaker—the so-called “power response”—as what you hear to determine the tonal quality of that speaker.

It still reads like an in-room measurement, but that is a mental block on my side. But, it is impossible to be in-room measurement based for the very reason I was confused (that there is no control and therefore no reference).

 

[rushwj]

i am far from an engineer of any kind, much less an electrical engineer, so much of this has flown past me - but i've picked up some good kernels of info and enjoyed the detailed and scientific read involved.

interestingly, something "clicked" as i read this article just now. this may explain a phenomenon my wife and i noticed last night. i just switched from onix strata mini's (folded (i assume for better dispersion??) ribbon tweeter and planar midrange) to paradigm 60 v5 (1" metal dome tweeter, 2.5 way speaker, all cone drivers) and seem to notice a much more enveloping sound with the paradigms, especially in the upper frequencies. this effect was so much that my wife, while we were listening to 2-channel music) asked if the surrounds were on. this is in a fairly large room, open on one side to the kitchen, so not many reflection points. not to get into too much of a discussion of one brand vs. another (i enjoy both in different ways, but very glad for the current change for multiple reasons), but more the type of speaker/drivers used and the sound produced. of note, have not set up audyssey yet.

gene, does this make sense? am i missing the boat here or imagining it? how much of this is likely to change w/ audyssey? thanks.
-bill

 

[PaulF]

Good article. Just wanted to mention that there are examples of single order HP crossovers that are driven by design and not necessarily cost reduction.

http://bwgroup-support.com/downloads/techmanuals/bw/800D-TM.pdf
http://bwgroup-support.com/downloads/techmanuals/bw/802D-TM.pdf
Sumiko Audio : Sonus faber
Sumiko Audio : Sonus faber

 

[gene]

Good article. Just wanted to mention that there are examples of single order HP crossovers that are driven by design and not necessarily cost reduction.

http://bwgroup-support.com/downloads/techmanuals/bw/800D-TM.pdf
http://bwgroup-support.com/downloads/techmanuals/bw/802D-TM.pdf
Sumiko Audio : Sonus faber
Sumiko Audio : Sonus faber

Without seeing the schematic, the 1st example looks like 2nd order networks to me. Each driver has an LC element connected in the crossover though I may be mistaken.

Some companies will claim to use 1st order networks but upon closer inspection they are often higher order or most complex networks than the consumer is lead to believe.

In most circumstances there are no inherent benefits for using 1st order networks.

Correction, I see the schematic now and only the tweet has a 1st order network, notice the woofer has a much higher order more complex network however...

 

[MinusTheBear]

Here is some information on the crossover design of the Salon2 and what I believe to be Harman's take on good crossover design for a no compromise loudspeaker.


Sumit: You mentioned high-order crossovers a while back. Revel speakers have always used high-order crossovers. Several companies take the opposite approach i.e. they use low-order (first and second order) crossovers. What are the disadvantages of using low-order crossover designs? Are there any advantages to using low-order crossovers?

Kevin: Well, we were fortunate enough to have done research that has allowed us to know, to really understand, what are the characteristics that are important to sound quality and what are the characteristics that have some value but less, and those characteristics that don't have any sonic value and that allow us to make the optimum choices in the design. That all points very, very strongly to high-order crossovers because high-order crossovers are necessary in order to have low distortion which is way up there on the list of important sonic qualities.

High-order crossovers are important to have good dynamic capability without compression. It would really shock audiophiles to see how much the response of most high-end loudspeakers changes at different volume levels. They are like completely different loudspeakers when played even at moderate levels, and it is something that is very directly measurable. So we really focus on making sure that not only is the timbre really accurate, but that it changes as little as possible over a very wide dynamic range. Plus the distortion is below the audible threshold; resonances are below the audible threshold because our research has shown those are really important things.

If we used first-order crossovers, we would degrade the off-axis response, and therefore the timbre, we would completely degrade the distortion characteristics, we would loose our dynamic capability, our freedom from compression because tweeters and mid-ranges are then getting signals that are outside the frequency range that they are really designed to handle. So it's really mostly heat, and that heat makes the voice coil impedance go up, and as a result of that the filter network is mis-terminated because it's not seeing the termination impedance it expects to see, and then the response of the crossover is impacted.

So what that means is that when speakers heat up, voice coils heat up, the crossover networks don't work right anymore and you get peaks and dips in the response, several dB peaks and dips in the response. So with high-order crossovers and with all the things that we are doing in the transducer design to keep the voice coils cool which means we are generally using a very large voice coils which spreads out the heat over a large area, and we're using in some cases multiple woofers to further spread out the heat. We are using all of these techniques including the way we vent them which forces the air through the gap at the same time, the vents are designed in such a way so that they don't create noises. It's a very sophisticated approach to solving the problem with heat. But a big part of that is the high-order crossovers. It's an essential part of it.

Sumit: What crossover slopes are typically employed in Revel designs?

Kevin: Our networks are always tweaked to result in the smoothest possible transition between transducers. Because of that, their electrical characteristics don't meet the textbook definition of classical filters, which is why we don't specify them in our literature. However, the resulting response of the transducer/enclosure acoustic and the filter's electrical response is close to a 4th-order (24dB/octave) Linkwitz-Riley characteristic.

Sumit: Have you ever heard a good loudspeaker that uses a low-order crossover?

Kevin: I have heard good loudspeakers that use low-order crossovers. I haven't heard great loudspeakers that use low-order crossovers. They run into these problems. It's inevitable.

Sumit: Do you feel that time-coherence is an important property of a loudspeaker?

Kevin: No. We've done a lot of research in the area. In fact when I made my first loudspeakers at Symdex in 1976, I thought that it was important because that was the fad at the time. With some people, it's still a fad. But if you look at the way the human hearing functions, you'll see that time-coherence isn't important. I actually credit Dr. Stanley Lipshitz at the University of Waterloo with bringing this to my attention, to put it mildly, in the 70's at Mirage. He and Dr. Vanderkooy came up with a box that would let you alter phase response without altering amplitude response, and by using it you could do listening tests to determine the audibility of phase errors, or time-coherence. It was quite evident then that if you are in an anechoic chamber or you are using earphones, you can detect the difference especially with special clicks that are made to hear it. You can't always say which is right or better, and as soon as you introduce the room it's 100% inaudible. The importance of phase is in the crossover region because that's an indicator of the blending of the transducers that are being crossed over. So in and of itself it's not a significant thing. And in order to try to optimize it you really sell your soul in terms of things we know are really, really important to sound quality. That is a key fact. It's not like it's cost free, audibly, to optimize that parameter.

Feature Article

Hey Gene, you have to get a pair of these ultra high end speakers. You know you won't have a problem with the Salon2's bottoming out at 75db at 12 feet away!

 

[gene]

Here is some information on the crossover design of the Salon2 and what I believe to be Harman's take on good crossover design for a no compromise loudspeaker.


Sumit: You mentioned high-order crossovers a while back. Revel speakers have always used high-order crossovers. Several companies take the opposite approach i.e. they use low-order (first and second order) crossovers. What are the disadvantages of using low-order crossover designs? Are there any advantages to using low-order crossovers?

Kevin: Well, we were fortunate enough to have done research that has allowed us to know, to really understand, what are the characteristics that are important to sound quality and what are the characteristics that have some value but less, and those characteristics that don't have any sonic value and that allow us to make the optimum choices in the design. That all points very, very strongly to high-order crossovers because high-order crossovers are necessary in order to have low distortion which is way up there on the list of important sonic qualities.

High-order crossovers are important to have good dynamic capability without compression. It would really shock audiophiles to see how much the response of most high-end loudspeakers changes at different volume levels. They are like completely different loudspeakers when played even at moderate levels, and it is something that is very directly measurable. So we really focus on making sure that not only is the timbre really accurate, but that it changes as little as possible over a very wide dynamic range. Plus the distortion is below the audible threshold; resonances are below the audible threshold because our research has shown those are really important things.

If we used first-order crossovers, we would degrade the off-axis response, and therefore the timbre, we would completely degrade the distortion characteristics, we would loose our dynamic capability, our freedom from compression because tweeters and mid-ranges are then getting signals that are outside the frequency range that they are really designed to handle. So it's really mostly heat, and that heat makes the voice coil impedance go up, and as a result of that the filter network is mis-terminated because it's not seeing the termination impedance it expects to see, and then the response of the crossover is impacted.

So what that means is that when speakers heat up, voice coils heat up, the crossover networks don't work right anymore and you get peaks and dips in the response, several dB peaks and dips in the response. So with high-order crossovers and with all the things that we are doing in the transducer design to keep the voice coils cool which means we are generally using a very large voice coils which spreads out the heat over a large area, and we're using in some cases multiple woofers to further spread out the heat. We are using all of these techniques including the way we vent them which forces the air through the gap at the same time, the vents are designed in such a way so that they don't create noises. It's a very sophisticated approach to solving the problem with heat. But a big part of that is the high-order crossovers. It's an essential part of it.

Sumit: What crossover slopes are typically employed in Revel designs?

Kevin: Our networks are always tweaked to result in the smoothest possible transition between transducers. Because of that, their electrical characteristics don't meet the textbook definition of classical filters, which is why we don't specify them in our literature. However, the resulting response of the transducer/enclosure acoustic and the filter's electrical response is close to a 4th-order (24dB/octave) Linkwitz-Riley characteristic.

Sumit: Have you ever heard a good loudspeaker that uses a low-order crossover?

Kevin: I have heard good loudspeakers that use low-order crossovers. I haven't heard great loudspeakers that use low-order crossovers. They run into these problems. It's inevitable.

Sumit: Do you feel that time-coherence is an important property of a loudspeaker?

Kevin: No. We've done a lot of research in the area. In fact when I made my first loudspeakers at Symdex in 1976, I thought that it was important because that was the fad at the time. With some people, it's still a fad. But if you look at the way the human hearing functions, you'll see that time-coherence isn't important. I actually credit Dr. Stanley Lipshitz at the University of Waterloo with bringing this to my attention, to put it mildly, in the 70's at Mirage. He and Dr. Vanderkooy came up with a box that would let you alter phase response without altering amplitude response, and by using it you could do listening tests to determine the audibility of phase errors, or time-coherence. It was quite evident then that if you are in an anechoic chamber or you are using earphones, you can detect the difference especially with special clicks that are made to hear it. You can't always say which is right or better, and as soon as you introduce the room it's 100% inaudible. The importance of phase is in the crossover region because that's an indicator of the blending of the transducers that are being crossed over. So in and of itself it's not a significant thing. And in order to try to optimize it you really sell your soul in terms of things we know are really, really important to sound quality. That is a key fact. It's not like it's cost free, audibly, to optimize that parameter.

Feature Article

Hey Gene, you have to get a pair of these ultra high end speakers. You know you won't have a problem with the Salon2's bottoming out a 75db at 12 feet away!

Harman is right on the money with this! I don't buy into "we use 1st order networks or NO network at all to yield better phase response, or keep in the magic pixie dust to better preserve the fidelity

That being said, you can use lower than 4th order networks if the drivers are robust enough and have a smooth natural rolloff. Many companies employ 2nd order networks and rely on the rolloff of the drivers to achieve a 3rd or 4th order overall response. This is fine.

However, running a midrange fullrange in a 3-way system IMO is NOT! Most designers will agree and very few loudspeaker companies do this unless they are simply trying to save cost for budget designs.

 

[PaulF]

Yep, I was just referring to the tweet. My understanding is that B&W's diamond driver was able to lower the resonant frequency so they didn't require as aggressive a filter slope.

While many of the proponents of single order designs that I have seen (Thiel, Vandersteen, Duntech/Dunlavy) are (or were) attempting to achieve perfect phase/time alignment between drivers (which I am neither for or against), one could assume that someone trying to achieve that level of perfection in a speaker would be inclined not to use a first order filter if such a design had other deleterious effects on the speakers overall performance... but who can be sure

 

[gene]

Yep, I was just referring to the tweet. My understanding is that B&W's diamond driver was able to lower the resonant frequency so they didn't require as aggressive a filter slope.

While many of the proponents of single order designs that I have seen (Thiel, Vandersteen, Duntech/Dunlavy) are (or were) attempting to achieve perfect phase/time alignment between drivers (which I am neither for or against), one could assume that someone trying to achieve that level of perfection in a speaker would be inclined not to use a first order filter if such a design had other deleterious effects on the speakers overall performance... but who can be sure

The Scan Speak 9500 has one of the lowest resonance frequencies I've seen in a tweeter (around 600Hz I believe) and yet nobody I know of uses a first order filter on it. But if that is B&W's claim then so be it. It's weird that I see no resistor pad down on their tweet in the schematic. I need to get a set of these to listen/examine. The B&Ws at our 2007 SOTU event sounded very good but they did sound strained at high volume levels as many of our listeners noted in the feedback sheets they filled out after the demos. However, it was a pretty large room the speakers were trying to fill.

Thiel crossover networks are a lot more complex than 1st order if you take a closer look. Also, they typically don't do as well as other similarly priced speakers, at least from our testing, under heavy load conditions.

 

[AcuDefTechGuy]

So B&W, Thiel, and Vandersteen don't use 4th order crossovers?

From B&W:

" We carry out exhaustive listening tests rigorously assessing the performance of each component until we find the optimum component for each position in the circuit. Fine-tuning by ear is only possible if the crossover is simple and the section of the crossover that perhaps benefits most from our policy of listen-and-learn is the part handling the signal for the tweeter. In most Bowers & Wilkins speakers, it is carried by a single, ear-chosen component that preserves the very finest detail."

800 Series Diamond | Technology - Bowers & Wilkins | B&W Speakers

 

[GranteedEV]

Yep, I was just referring to the tweet. My understanding is that B&W's diamond driver was able to lower the resonant frequency so they didn't require as aggressive a filter slope.

While many of the proponents of single order designs that I have seen (Thiel, Vandersteen, Duntech/Dunlavy) are (or were) attempting to achieve perfect phase/time alignment between drivers (which I am neither for or against), one could assume that someone trying to achieve that level of perfection in a speaker would be inclined not to use a first order filter if such a design had other deleterious effects on the speakers overall performance... but who can be sure

Id imagine The key to successful first order crossovers would be to cross them well above their natural rolloff... but how high is high enough???

I think crossovers that high in frequency (4-5khz) are problematic with most drivers. The off axis response suffers. Like Kevin Voecks said, we listen in rooms not anechoic chambers!!

The better way to get good phase tracking is with active speakers.

IE http://www.grimmaudio.com/whitepapers/speakers.pdf

 

[PaulF]

B&W wrote a white paper that outlined their design approach to the 800D series speakers (the generation prior to the current Diamond series), including why they went with certain choices for the LF-MF and MF-HF crossovers.

The paper can still be found at the following link (warning, it's a little slow loading).

http://www.fransvaneeckhout.be/bw/bw_images/800_Development_Paper.pdf

 

[AcuDefTechGuy]

So B&W uses the 2nd order Linkwitz-Riley crossover in their 800D?

 

[Nuance AH]

I have to wonder if something was wrong with the speaker measured under Myth 4... Here is a link to a post of the Ellis 1801's impedance plot. It's obviously different from the one in the article. I am curious as to why...

http://forums.audioholics.com/forums/853737-post744.html

 

[gene]

I have to wonder if something was wrong with the speaker measured under Myth 4... Here is a link to a post of the Ellis 1801's impedance plot. It's obviously different from the one in the article. I am curious as to why...

http://forums.audioholics.com/forums/853737-post744.html

I didn't want to mention names but the speaker in question is no longer in production after I took a look at it and found the impedance issue. I measured both speakers and they both measured the same so it was likely a design issue. The speakers sounded very good but they just couldn't play very loud.

 

[Nuance AH]

I didn't want to mention names but the speaker in question is no longer in production after I took a look at it and found the impedance issue. I measured both speakers and they both measured the same so it was likely a design issue. The speakers sounded very good but they just couldn't play very loud.

Ah, the name is on the graph, Gene... It's a little late for that.

 

[gene]

Ah, the name is on the graph, Gene... It's a little late for that.

Crap I replaced the trace with "Sweep" but didn't see it where the filename called it out. I will change out the image. It wasn't my intent to give Salk a black eye, especially since they no longer make that speaker.

[edit] I fixed the image and my apologies to Salk.

 

[Nuance AH]

It's all good, Gene. I spoke to Dennis and he said he's pretty sure the crossover guy in the Salk facility wired these speakers incorrectly. Fortunately you found the issue through testing, though. I don't know if any production models were accidentally wired backwards as well or if it was just this pair, but the problem has been resolved. And I do believe Salk still offers the Ellis 1801 as a current model (it is still on their website). Of course, I am sure it's a properly wired version.

 

[lsiberian]

I agree with Kevin on using L-R 4th order networks, but it isn't always necessary if the power requirements are low. Certainly curving a driver off well before it bottoms out drastically will result in better overall sound if implemented properly. I also agree with curving off the top in most situations too.

 

[GranteedEV]

It's all good, Gene. I spoke to Dennis and he said he's pretty sure the crossover guy in the Salk facility wired these speakers incorrectly. Fortunately you found the issue through testing, though. I don't know if any production models were accidentally wired backwards as well or if it was just this pair, but the problem has been resolved.

It has to be something along those lines because the graphs don't seem to match up above 1khz. The impedance peak on Dennis' graph is half an octave higher with higher magnitude, and there's no impedance dip. The behavior near tuning is different too, but I think the boxes used are not the same (one being a bookshelf (?) fand the other being a more damped tower (?))

It sounds like someone misread the schematic.

 

[gene]

It has to be something along those lines because the graphs don't seem to match up above 1khz. The impedance peak on Dennis' graph is half an octave higher with higher magnitude, and there's no impedance dip. The behavior near tuning is different too, but I think the boxes used are not the same (one being a bookshelf (?) fand the other being a more damped tower (?))

It sounds like someone misread the schematic.

Jim told me the crossover design was not his own and after I found this issue he said he was discontinuing the product. Perhaps he has a redesign now. I have no way of knowing. I tried to review 2 pairs of speakers from Salk and both pairs had issues and there is no room in my schedule at this time for a 3rd try. Perhaps later in the year we can try again as I still have high hopes for his products.