sfogg Posted November 23, 2003 Share Posted November 23, 2003 Above is the ES600 as a two way again but this time using the tilt EQ function of my processor. Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 23, 2003 Share Posted November 23, 2003 ES600 again but this time adding 3dB to the high end to spruce up the Altec a touch. Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 23, 2003 Share Posted November 23, 2003 One of the things that was mentioned in the squaker measurement thread of a week or two ago was that running these tests at 1/24 octave make the curves look much worse compared to some you see elsewhere. The attached is the exact same measurement as the ES600 done above but rescaled to be the more common 1/3 octave RTA. Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 23, 2003 Share Posted November 23, 2003 As far as the listening to the LaScala used as a 2 way above I want to listen to a bit more to really nail down my impressions but a few things jump out. First is if you play pink noise and move around vertically in front of the one speaker with the ES600 the sound hardly changes at all as your position changes. This is due to the very narrow range of overlap between the two horns causing very little interference between them I really don't hear much of a difference until I get past the vertical dispersion limits of the 511b. If I do the same thing on my three way LaScala you the sound changes quite a bit. You get areas where parts of the noise gets louder and softer from the interference between the drivers. This of course affects the off axis vertical FR of the speaker which will effect the resultant sound the listener hears. Comparing one LaScala 3way vs. the other as a two way using the ES600 on drums and vocals there is little more of a hollow sound or smearing to them on the more traditional three way. I didn't really notice it till it was gone listening to the ES600 2 way LaScala. The comparison between the two is a little difficult since there position is different and the ES600 drops the level of the 2 way a couple of dB compared to the regular ALK. I tried to balance the level of the two (as SPL differenes play big games with perceived sound quality) and then fed each material mixed down to mono then just used the balance control to go from side to side. If I get a chance I want to built a crossover comparison rig which will let me switch between the ES600 2 way and the La Scala as a three way using the normal ALK type A at the touch of a button from across the room. Might be tough finding the time for that though right now. Shawn Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Author Share Posted November 23, 2003 Shawn, All your measurements look like what I expect to see and they look good. I think you are going to eventually want an equalizer to drop that hump I see around 150 Hz on all your plots. I suspect that's wall bounce, but I'm not sure. One thing though. I think you will find a difference in the pitch as you move vertically in front of you speaker. These horns are designed to have a collapsing vertical response to compensate for mass rolloff in the driver. You should look for smooth dispersion horizontally, not vertically. I would also like to see a pair of indentical plots blowing up the 600 Hz area with the driver in phase and out of phase. I did that here and will post the plots after you do it. I want be sure I have the + and - terminals marked properly. Al K. Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Author Share Posted November 23, 2003 Shawn, I forgot to mention a big point here that I was delited to hear you mention. When I preplaced the gradual slope squawker / tweeter crossover in my modified Belles with a 7500 Hz extreme-slope, I also noticed a major reduction in a smearing sound I have been listening to on every stereo I ever heard. This would show up mainly on strings in a symphony orchestra. I suspect it's becasue you hear everything twice in time when two driver that are not time aligned are playing simultaneously using a gentle slope crossover. The extreme-slope netwrok only lets one driver play each component of the complex waveforem depending on its frequency. Your ear only gets each component once. It's obvious on an oscilloscope, but I didn't you would be able to hear it. Apparently you can! Al K. Quote Link to comment Share on other sites More sharing options...
John Warren Posted November 23, 2003 Share Posted November 23, 2003 . Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Author Share Posted November 23, 2003 John, Yes, If you analyze your table I think you will see that the phase is just 45 degrees per element. This is for Butterworth response only though. When you get into Chebyshev response, it gets slightly different. Whne you start into filters with "notch" sections like this, you might as well forget about it! I wrote a program that actually summs up the amplitude and phase of two channel from analysis and displays them with any time delay to one channel I key in and either in or out of phase. The bottom lines is that it depends on the postion of the drivers with respect to each other in time if the cancell or add. I connected the ES600 network high and low outputs to my analyzer A and B inputs with it set to sum spectrum A+B. The results were a negative 1.5 dB summ. I suspect they are exactly in phase and having 7.5 dB loss. Exactly 6 dB loss would have summed up perfectly flat. I could do that by using larger wire in two of the coils, but it isn't worth it! The big point a lot of people forget about this is that component loss mess up the summation at the crossover. This makes the popular Linzowitz-Riley (spelling ?) a moot point in a passive network, at least with respect to the 6 dB crossover alignement. The ES600 is designed for 3 bB cross and depends on another 3 dB loss to make it summ flat. A Butterworth network that is supposed to summ up to a rise at the crossover actually summs closer to flat with losses! Al K. Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 23, 2003 Share Posted November 23, 2003 John, "Why would the overlap of any conventional filter other than two second orders (or 6ths) result in cancellation?" Take any two drivers that are reproducing the same material and change the distance between them relative to the listener. By changing the path length from one to the other (by either the drivers moving or the listener moving) and literally the phase (delay) of the each driver changes. You get points where material cancels and points where they sum better. This gets worse the more you get off axis and the shallower the slope is. Shawn Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 23, 2003 Share Posted November 23, 2003 Al, "I think you are going to eventually want an equalizer to drop that hump I see around 150 Hz on all your plots." My processor is *hopefully* going to have room EQ in it in the future so that will take care of that. "You should look for smooth dispersion horizontally, not vertically." Both are important. But in comparing the effects of the crossover slope you will notice bigger differences vertically then you do horizontally for vertically arranged drivers. If the drivers were arranged horizontally then the effects would be larger in the horizontal plane. I'll try and do in and out of phase for you around the crossover point. " Your ear only gets each component once. It's obvious on an oscilloscope, but I didn't you would be able to hear it." It is fairly obvious on drums. With the ES600 they have a more defined attack. Through the three way they aren't as crisp sounding. Shawn Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 23, 2003 Author Share Posted November 23, 2003 John, I forgot to directly answer your question but Shawn answered it exactly right. The bottom line is that it isn't very important. I don't think it can summ higher then +3 dB but it can come to a huge null. The good thing is that a 3 dB peak is minor and no matter how huge a null you get, your brain fills it in making it a moot point also. I sill think it's good to pick the polarity that summs flatest though. Why not do it? Al K. Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 23, 2003 Share Posted November 23, 2003 "Take any two drivers that are reproducing the same material and change the distance between them relative to the listener. " Also worth noting that since the path length from each driver is already different in a LaScala these even happens to a degree on axis. Shawn Quote Link to comment Share on other sites More sharing options...
John Warren Posted November 24, 2003 Share Posted November 24, 2003 ---------------- On 11/13/2003 11:16:34 PM Al Klappenberger wrote: I have come to believe group delay is not important anyhow. Al K. ---------------- Group delay is just the derivative of the phase vs. frequency plot and the phase vs. frequency plot is intimately tied to the frequency response by the Hilbert transform (assuming miminum-phase ). They are all functions of each other telling you the same thing in different domains. It is not a choice of which one is important, its a matter of which one is applicable. Quote Link to comment Share on other sites More sharing options...
John Warren Posted November 24, 2003 Share Posted November 24, 2003 Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 25, 2003 Author Share Posted November 25, 2003 John, Here's the computed group delay of the low channel of the ES600 network. It is assuming polypropylene caps and lossless inductors with the actual DC resitance of each one included. The amount of losses seem to have a major effect on the height of the group dealy peak right at the crossover, so it is hard to predict by computer analysis and I can't measure it accurately. I will let others decide how important it is. I say it is not important at all. You and I have been around and around on the group dealy issue and I think it's time to let the world decide about it. There is a long list of qualified people on both sides of the issue, so it will not be decided here! The group delay in the network I have in my Belles (described in the downloadable paper on my alkeng.com web site) is very similar to that of the ES600. The two designs are very similar except this one does not need the Zobel to give the filters a good termination. The woofer driver voice coil inductance is synthesized right into the low filter on the ES600. This one is a better design. Al K. Quote Link to comment Share on other sites More sharing options...
sunnysal Posted November 25, 2003 Share Posted November 25, 2003 Al, I am not sure if you are saying that you regard all group delay as unimportant (even 10ms or more) or just group delay that falls within the 3ms limit as described by B & L ? tony Quote Link to comment Share on other sites More sharing options...
Al Klappenberger Posted November 25, 2003 Author Share Posted November 25, 2003 Tony, The subject of group delay is a HUGE bag of worms I don't want to get into again and again. If your are interested in the subject, John and I debated it, at length, here on the BB a while back. A big part of it is the mass confusion between time delay and group delay. You have to start with that point and work forward. The two are NOT the same! Al K. Quote Link to comment Share on other sites More sharing options...
John Warren Posted November 25, 2003 Share Posted November 25, 2003 Quote Link to comment Share on other sites More sharing options...
sunnysal Posted November 25, 2003 Share Posted November 25, 2003 thanks guys for helping me calrify this issue. I will look up the older thread to see what I can find there. I may come back with a few questions. John Curl got all worked up about how k-horns "had time and phase issues" thta made him drop them years back, I have never totally recovered from his comments (I have a lot of respect for the guy and he has been very nice to me) and have struggled to try to understand what he was talking about, some things I get others still illude me. regards, tony Quote Link to comment Share on other sites More sharing options...
sfogg Posted November 26, 2003 Share Posted November 26, 2003 Just an update... I definitly am enjoying this combo (ES600 902-8B on 511b) and am going forward with getting more of the ES600s from Al so I can hear these in stereo and have a third for my center channel. The clarity of the speaker is pretty suprising compared to the 3 way. I think a lot of that is from the lack of overlap of the drivers from the steep slopes. The Altec 902 run wide range (instead of being rolled off below 6k like in the 3 way) is an impressive driver... esp. considering what I bought them for. The two way has a more extended top end then the 3 way with the t35. Shawn Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.