Reversing the polarity of a mid in a 3-way or a tweeter in a 2-way is just a result of correcting for the phase shift of the crossover. The exception being 4th order. It's about getting the drivers to sum correctly. If they were 180 degrees out of phase, they cancel around the crossover point. In fact, one of the best ways to confirm the drivers are in phase is measuring the reverse null.
Now, if you want to talk about absolute phase and the half a millisecond @ 2000 Hz between cycles and whether that's audible, that's a different story. I don't know the audibility threshold up there but down around 100 Hz, it's around 30 milliseconds and several cycles for our ears to respond. Less for the trailing edge. |
In speaker designs that I've only simulated for now, sloping the baffle 10 degrees allowed a slightly higher crossover with the same drivers, which is not so arbitrary. In a 3-way, it also allowed symmetric slope crossovers. In a 2-way, adjusted for BSC, it ended up asymmetric similar to the way I'ld do a flat baffle. Since a normal TM tends to lobe downwards, the sloped baffle tends to correct that and even allow a shorter height. The downside was that it was sensitive to listener height and distance. |
The best answer I've heard for using a sloped baffle came from Paul Hales,... "It looks cool"...
http://www.stereophile.com/floorloudspeakers/699hales/ |
http://www.audioholics.com/room-acoustics/human-hearing-phase-distortion-audibility-part-2
That's bound to be misunderstood. Related articles in links.
The 30 milliseconds I mentioned previously came from a U-tubed Earl Geddes seminar on multiple subwoofers in a small room. |
Bomb,
That was not my intended meaning. Simply put, we're human.
Obviously you have your opinion and I'm not trying to argue, passive aggresively or otherwise. I may have reacted a bit sarcastic when I thought you overstated.
Perhaps you can explain something to me. In GMA's specs, they state phase shift acoustically over given frequencies. Does that mean impedance phase (reactance) or total phase? Either way, impressive. |
Bif,
Don't get me wrong. Phase is taken into consideration on any competently designed speakers. My first didn't but I wasn't particularly competent then. Not yet either. Still, with four 6-1/2's per side, they sounded more dynamic than any "coherent" speakers I've heard short of the Dunlavy SCIVa. Never considered the two were related.
Time coherence implies that the first cycle of a tone is more important than the second or third. That, I'm unsure of. Really, I don't know. Low order crossovers have advantages, including transient behaviour. I like them because they're simpler and cheaper but not for all drivers and circumstances.
Here's the test... Listen to any 2.5 way speaker. By design, the woofers roll off at different frequencies and must have increasing phase separation as frequencies increase. 45 degrees by 1000 Hz is fairly typical. According to the article, where our ears are most likely to detect. The .5 woofer will be lower SPL but still audible. Can you hear any phase distortion?
Problem is that we don't hear distortion as distortion, not unless you have the trained hearing of a professional under controlled conditions. It can be measured by instruments but human brains interpret those as frequency differences and there could be any number of explanations including comb filtering, diffraction, or the driver. Almost as easily, they can be masked.
So, forget the test, it's useless. More likely that someone will misinterpret that as all 2.5's are .... |
I have no doubt Roy is sincere and his efforts are genuine. I don't have enough information to intelligently debate his views either way. Beep is an interesting phenomena in itself. |
The DIY forums and this one have something in common. Very rarely does anybody listen to any advice or criticism that they don't want to hear. |
For individual drivers, cone woofers have voice coils and are inductive. So, yes, they do have phase shift as frequencies increase. Some are more inductive than others. Even dome tweeters have some degree of phase shift.
A first order, parallel low pass is an inductor coil with phase shift, typically 90 degrees in the pass band and more beyond. They're cumulative and that's called acoustic slope. In a 2-way, there's also baffle step compensation, which inolves a bigger inductor well into the pass band, causing even more phase shift, maybe another 90 degrees more or less. And that's just first order. Add another 90 degrees for every order over that. Basics 101.
In the next class, we'll discuss capacitors, high pass filters, zobels, notch and contour filters, all involving various degrees of phase shift. Then, on to impedance phase and reactance. Your homework is expected and there will be a test. |
@Bifwynne
For the first part of your question, you misunderstand. Pass band is the part of the frequency the driver is covering, unattenuated, within the filter. Actually, I used the term technically incorrectly in the BSC context since that is attenuated long before the crossover point. Driver rolloff caused by inductance usually occurs out of the pass band but is still important. If a driver could, realistically cover from 35 to 20 KHz, than it would require very little inductance. There are drivers with little inductance, relatively, like the Satori MP16, but the numbers you mention are bordering on some AVR brochures :O
The second part is beyond me, even if I could understand the question. |
How fast is the train travelling? |
Those S5 plots are actually quite smooth. A sign of well matched, quality drivers, a good cabinet and unobtrusive crossovers. That phase dip in the bass is fairly standard and hard to avoid. It's from the woofer, not the xover. Other than the low impedance, it would be a relatively easy load. Of course, there's always ways to make those plots look better but it might not sound any better. |
Even with DSP, bet there's still a market for $300 Revelators vs. a $20 Silver Flute and vice versa. |
To answer Al's rephrased question or Bruces's origional, I'll give the example of the speakers I'm working with lately. The midwoofer is an inexpensive 5" that has a fairly nasty and fairly normal breakup around 5KHz and rolls off rapidly after that. The design I'm using it in is basically a flat baffle MTM (+++) with a low crossover point of about 1500 Hz. BSC is handled separately, but we won't get into that because that gets complex. The low pass is a first order (electrical) with an inductor.
By itself, no crossover, the woofer shows 30 degrees of phase shift by 1500 Hz even though it is nowhere close to rolling off. With the crossover, it's 120 degrees at the same point and nearly 180 degrees by 3000 Hz. It's the combined acoustic slope that matters and that's measured in Hz and dB. Phase is along for the ride.
Things get a bit more complicated. To attenuate the cone breakup 5K, which would still be audible, I added a "tweeked" Zobel. By that, I mean I oversized the cap and undersized the resistor so that it falls somewhere between a filter and impedance compensation. It also comes in handy to get phase dialed in. Tried but couldn't get a notch filter to work well in this case. Got it about 20 dB down.
As some of you might have guessed, for a tweeter to cross that low, it has to be particularly rugged and there's only a few I know that capable, particularly with only a second order high pass. Didn't want the crossover that low but that's where the combination wanted to be. It's already 6 dB down by the crossover point, which gets summed back, when drivers are in phase. The tweeter, with crossover, has begun rolling off from around 5000 Hz. Another "trick" was utilized to round the knee. By the tweeter's Fs (resonant frequency) it's down 20 dB. The tweeter's phase shift from 1500 Hz to 20K, before any baffle diffraction and with crossover, is only 60 degrees. Essentially, little to no phase shift without crossover. If you're still paying attention, you might think something's wrong with my math. Shouldn't second order shift 180 degrees? For a high pass, the phase shift is caused by capacitance, not inductance.
In order to get the driver's phase aligned, I needed to invert the polarity of the tweeter. By the way, this sims out to a 45 dB reverse null at 2m, so I think it's pretty much on target. Nicer is that it's consistent over a wide vertical and horizontal range and listening distances. Gently sloped plateau on the impedance phase to +30 (inductive) degrees maximum, which by most standards, is quite good.
That's a very simple example, even for a two-way. You should see what happens with real woofers. Remember the old spinning plates act, where a guy balanced plates on poles and ran around to keep them going while he added more plates? Now, tie the poles together with strings and springs and rods and hinges and that's speakers. |
Holm Acoustics also offers a similar product but it's more expensive. Without the preamp feature, there's miniDSP, Ground Sound modules and what I consider the best of the 'pro' units, Xilica, which is also used on the top-of-the-line Legacy speakers. Almost forgot the McIntosh MEN220.
You still gotta know what you're doin', so they're really no easier to use than designing a passive system. |
The Holm site has an interesting, albeit monotone, video tutorial how these things work. Specifically his, but generally for all. |
Thanks Roy. Again, you beeped. I'll counter with pop, bang, whack, pow, and all the other 60's Batman fight words that seem to have something else in common. Not arguing you're goal of recreating as accurately as possible but the words describe multiple, changing tones that appear to define our hearing ability more than the actual sound. So, my rhetorical question becomes: Is the b and p really in beep or is that our imagination? |
Looked up quadrature and unless you're masochisticly inclined to imaginary vectors, this might be easier...
http://sound.westhost.com/ptd.htm |
