Dolphin ... add John Bau (Spica) to that list.
- ...
- 155 posts total
hello, perhaps this is somewhat of an off topic question, concerning crossovers I have been under the impression that a 4th order linkwitz-riley electronic crossover was a far superior choice for auido production/reproduction of music, especially due to phase and time matching between adjacent drivers. For an excellent source of info on this: http://www.rane.com/note107.html I realise that the thread has been mainly about less complex systems, such as a simple pair of commercially available speakers, or the same with a sub/s. But even a 5 way crossed over system comprised of 2 stacks of loaded cabs is really nothing but a complex 5 way speaker system. Any comments will be well appreciated, thanks |
4th order LR does have the benefits described. It should be noted that your particular link seems to be talking mainly about ACTIVE crossovers, meaning (typically) op-amp based electronic filters placed upstream of the power amps. This is an absolute piece of cake compared to designing proper passive crossovers, but very few speakers on the market go this route as it requires an amp for every driver, and makes the overall package pretty spendy. ATC does a great job at this, but their market share is pretty low overall because most people want to be able to mix-n-match their amps and speakers (I'm not saying this is a better approach, quite the opposite, merely that it's what most people seem to want). Implementing an ACCURATE 4th order L-R in a passive crossover with real-world drivers and all their problems is a giant headache, to say the least. Lots of manufacturers will claim to be using this type, and they are in a theoretical sense, but in fact the end result isn't a truly accurate 4th order L-R due to the frequency response/ impedance/ phase variations in the drivers themselves. They can sometimes come close, but the end result is never ideal like it can be when using op-amps. This is why active crossovers are such a piece of cake by comparison. |
Dolphin, you asked The detractors of time coherent designs almost always mention: dispersion characteristics, smoothness of power response, distortion, wave interference, off-axis lobing, and compression. I think you touched on wave interference and off axis lobing, how about the rest?See below... Can a 1st order crossover based speaker be good in these areas or are they mutually exclusive?As good in some, better in most. Can a 1st order crossover based speaker compete with the best 4th order based speakers in these parameters?As good or better- ask any of our owners or dealers. Someone in this thread mentioned that crossovers, 1st order and others, can be implemented in series or parallel. Can you talk a bit about the pro's and con's of either implementation in a 1st order crossover?After considering a response for quite a while, I should not do the series-designers any favors by explaining the results of our research. We use parallel circuits. Think about damping factor and also the distortions passed on, to begin with. So, to answer your original questions, with respect to first-order crossovers- Dispersion characteristics: No problem with us. This does become a complicated issue when cabinet reflections are considered, which we avoid. Do note, however, you have never read of dispersion problems with most any 1st-order speaker design. The math is `way too involved to show why here, but there will be info on this on our website. Smoothness of power response: About the same, although this is usually botched by choice of crossover point (any style crossover) and by using spaced, double drivers in one frequency range. Biggest deviation we see when "power response" is poor, is a hollow-sounding voice range past about 30 degrees off axis to the sides. For those not familiar with this term, it was coined to describe how it might be good in some circumstances for a speaker to put out a "smooth amount" of acoustic power per frequency into the room- pretty vague, considering the "results" were an integration of the output at various angles over a complete hemisphere, which could be skewed by having a tweeter very bright on-axis and dull elsewhere- just to mention one of the flaws in "integrating". This method was championed first by the AR LST, Design Acoustics, and the Walsh driver, and now the current omni designs. It is better to say that we want a speaker to have a smooth dispersion w/frequency off to the sides (no holes), tilted downwards in the highs so we don't send too many highest-highs to the sidewalls or wall behind the speakers. Distortion: Harmonic- depends on the design of the drivers. The best drivers have NO problem in any type of living-room use, with an appropriate crossover point that respects the dispersion pattern and the radiation impedance seen by each driver. IM distortion- depends on the drivers again AND also the crossover points AND the woofer excursion allowed below 50Hz. Wave interference: Here's a concept- THERE ARE NO WAVES. We heard sound only AT our ears; an air pressure fluctuation- rising and falling minutely, UNPREDICTABLY. Unless you listen to pure, single tones, sustained, like from a tuning fork- then the wave concept is useful. But only as a solution to that very simple, eighth-grade wave math. It does not describe very much about how we will hear music. - Speakers designed only via sine wave analysis sound radically different from each other, and do not measure like they sound, because of their designers' preferences and interpretations about what sine-wave measurements mean. - Speakers designed via the time domain approach (uses extraordinarily difficult math which can keep track of the music signal's demands), include sinewave analysis automatically (the converse is never true). And guess what- these designs sound `way more similar than different, and their measurements- no matter how performed, consistently more correlate with what we hear on music. Off-axis lobing: Audible only on selected sine waves. You must realize, of course, that cancellation arguments depend on "relative distance to the drivers is now different when you are standing up". And thus to get a cancellation of a particular sinewave, you must be exactly a half-wavelength farther away/closer to the tweeter compared to the mid. Which is 180 degrees. Which is 4.5" closer/farther at 1.5kHz, and 2.25" at 3kHz, and 1.5" at 4.5kHz. So pick your frequency for cancellation. If you are standing up, remaining motionless at one spot, there is only one distance difference, say 1.5", which would then put a null on sine waves at 4.5kHz, 2.25kHz, and 9kHz. And also create partial nulls beginning within +/- 20% of that primary 4.5kHz frequency (as the distance difference reaches less/more than that 180 degrees). Which means a general dip from 3600Hz to 5400Hz. Which is less than a half octave- a few notes on the piano- only its harmonics go that high. A dip which could be "covered up" (usually is) by tweeter "splash" off a flat cabinet face. So then move around the room a little (why else are you standing?), and the null frequencies move to different tone ranges- usually higher as you move away. So you hear a different tonality/tone balance/timbre in the treble. Is it unpleasant because of the transient distortion? (see below) Yeah, if you play it at >95dB on music with a lot of treble information. You did want complete honesty, right?? The sine wave math for these nulls is not inaccurate- it is just useful for sine waves and on pink noise. What we hear from first-order speaker designs on music with its varied tones and timbres (i.e., no sustained single, pure no-harmonic-content tones) is a reduction in the treble, a compression of the depth of the image, an accentuation of the leading-edge of the low-treble sibilants (`cause tweeter is closer), a blurring of the dynamic contrasts, and a reduction in the clarity of separate performers singing the same line (think chorale and massed strings). All because of the time delay imposed by being at that "1.5-inch" distance offset. Which is a constant 1/2250 second of time DELAY (= 4.5kHz half-period = .11 milliseconds ). Contrast that with the constantly-changing HUNDREDS of degrees of time DELAY that the higher-order designs impose, no matter where you sit or stand. Ten to several hundred times longer time delays than the .11msec above!!! Delay times that also VARY with EACH frequency no matter where you sit or stand, unlike first-order designs which give you only that ONE, constant, time delay at every frequency when you stand up. And this gross amount of varying time delay creates far worse distortions of the same kind mentioned in the paragraph above. Not to mention that some of those designs (many) also invert the polarity of the mid vs. the tweeter and woofer, so the initial transients are also warped by one driver sucking in, while the others push out. This is a POLARITY INVERSION, which many try to tell you "well that is just 180 degrees". Yeah- on sine waves. Tell the drummer pounding outwards on the kick-drum skin that you are going to make his snare drum whack SUCK IN, and also that his kick drum will get there THREE FEET late because the speaker has the woofers around the side of the cabinet, time delayed even more by the crossover. And then try to explain that NONE of that kick drum's harmonics will be arriving three feet late- only the lowest fundamental. The higher tones will arrive sooner, so his pulsing rhythm will sound lagging, and less powerful. And that his "sucking snare" will likely sound hollow. And the crack of his stick on the snare head will be of positive polarity, AND arriving a few inches sooner than the sound from the sucking-in skin... Thus, I do not see the point in warping the time-domain for critical home or studio listening, especially since, for the last 15 years, we have had drivers that will handle the power and excursions required. So, it is (not only) my humble opinion, that high-order crossovers screw up the music's timbre, dynamics, rhythm, transients and imaging, because they warp the time domain so grossly, and differently, at each and every frequency. And so on them, certainly it does not matter much where you sit or stand, or measure- it is always "out of phase", far more than standing up on a first-order speaker. To demonstrate this, play a particularly poor recording on high-order speakers vs. 1st-order speakers. And then hear it over decent headphones- which also have little time-domain distortion (better to call it that than "phase shift"). And finally compression: Not a problem for home or studio nearfield with the best drivers out there. Most drivers are not very linear in terms of power compression (from voice-coil temperature increases and from magnetic-field non-linearites vs. stroke). And a high-order crossover protects those drivers, and sounds high tech, and needs to be "computer designed" for the "best" results. Which is also good for advertising. And for which is easy to present the "benefits" via sine-waves. Karls and Vettemanbc- You are both correct, on all points you make. Those crossovers are the proper way to go if you cannot take the 1st-order route, such as in pro-sound, because they screw things up AT THE CROSSOVER POINT by injecting EXACTLY 360 degrees of delay at THAT frequency. Which means they "sound OK for PA speakers". And "EXACT" is best achieved via electronic crossovers for the reasons Karls states. However, Vettemanbc, you say "especially due to phase and time matching between adjacent drivers". Phase, yes. Time- no. A common mis-interpretation, based again on sine wave analysis. If Rane said it wrong, shame on them (of course maybe that is what they wanted you to believe). Seandtaylor99- Sorry Spicas, although easy to listen to, are not time coherent. They are instead smoothly time-delayed as the music moves into the treble- but this does compress the image from front-to-rear, and make for laid-back dynamics. It was indeed a higher-order circuit, necessary to protect the drivers he had available back then. The circuit he used warps the time-domain to much less degree and more gradually from frequency-to-frequency than the highest-order crossovers. He was among the first to do that. Celestion did it some in their old Ditton 33 10" 3-way bookshelf model from the mid `70's. Kef is trying that again, I believe in their new series. Will any designer out there show me where I am off base? I am not about attacking them- I would like them to justify why they believe we cannot hear time-domain distortion. The numbers I give above are not in dispute- just their audibility and the need for "waveform fidelity" (Technics, 1976). In fact, I will absolutely refrain from any comment or question until others have posed their questions to those designers and had them answered. Hope this helps. I cannot seem to explain the benefits of a time-coherent approach to design any more easily. Tried many times. Best regards, Roy Johnson Green Mountain Audio greenmountainaudio.com |
- 155 posts total