Question for the line array guys (1 Viewer)

[RichardHOS]

Really have no idea why this thought came up... probably because it's late and my brain is beginning to short circuit.

Anyway, I understand the physical basis of comb filtering. I also understand that for line arrays there typically seems to be a concerted effort to keep the center to center distances of the drivers within a certain percentage of the wavelength of the upper end of their desired response, say 25%. Not sure if a -30dB or -40dB down target is used before comb filtering begins, or if the target is right at the crossover frequency.

In any case, this is the question that at the moment I can't seem to answer for myself... why use the center to center distance at all? I know that for the wavelengths involved the driver will act like a point source due to its size, but in reality it isn't a point source. It has a distributed radiating area, and part of that area is adjacent to part of the next driver's area.

To illustrate better my question, imagine for a moment that the drivers were square. A single driver would still model and behave like a point source, but a line of them would have distributed radiating area of a nearly continuous surface from top to bottom of the line. I wouldn't expect comb filtering in the vertical direction in that case anymore than I would for large planar drivers with continuous diaphragms, except for very, very high frequencies.

So a line of round drivers isn't quite as continuous as a line of square drivers, but it's less broken than a string of point source radiators with center to center separation.

I suppose, what I'm really curious about, is whether someone has taken measurements of the vertical comb filtering of a line array at high frequencies and compared those results to what is predicted by a simplistic point source model.



Oh, and if comb filtering was really a concern for the XO point and drivers chosen, would a double column of drivers with one staggered such that the drivers overlapped ~20% in both the vertical and horizontal directions yield the behavior of essentially a continuous diaphragm? Higher driver cost, I suppose, but since when has a line array been about the simple and cheap way of doing something?

OK, time for bed. :b

 

[John E Janowitz]

I'm actually working on measurements like this right now. I can tell you that with a line of the RTW2 ribbon tweeters, comb filtering does look different than it would be predicted for a set of point sources about 6 5/8" apart.

I just did the measurements today and would have graphs to show, but when I pasted the screen captures, somehow everything above 15khz got cut off.

Here's a brief rundown of what I noticed. I used a line of 12 RTW2's and did measurements with from 1 to 12 drivers being wired up. I'll be posting the results on the thread on the line array project here tomorrow when I can get the graphs fixed.

http://forum.stryke.com/viewtopic.php?t=85

FTI, a line of ribbon tweeters ends up being pretty flat up to about 16KHz. At that point there tends to be a large dip, up to about 20dB or so, depending on the wiring. Power tapering as others suggested did very little to help with this dip.

John

 

[RichardHOS]

Interesting... and somewhat comforting to know I wasn't out in left field. I suppose real world measurements are sometimes hard to replace.

Keep us posted on how all the line array projects work out.

 

[John E Janowitz]

Another interesting option is a Bessel array. The Bessel array is typically used for pro audio to get the high efficiency of multiple drivers, with the dispersion characteristics of a single driver. It works well for the high frequency section in a 2 way line array actually, but also has it's own drawbacks. Mainly for me the drawback is that the patent doesn't run up for another 3 years yet.

John

 

[RichardHOS]

lol.

 

[Jim Griffin]

Richard and John,

Richard, if you haven't downloaded my latest line array white paper, you should as many of your questions are addressed within it. I don't have enough postings on this message board to include links but you can go to the audiodiycentral web site to down load a copy. That would be the usual:

www dot audiodiycentral dot com address

John, the tweeter line dip that you see around 16 kHz is evidence that you are probing too close to the array. What you are seeing is the frequency at which the distance between the active elements of the planar tweeters is two wavelengths--hence a cancellation or dip. If you probe the line at 2 meters distance or greater, then the dips should fill.

These planar tweeters--they are not true ribbons for what it is worth--have limited vertical dispersion beyond their active element area. Hence, you have limited sound overlap so their SPLs don't tend to add too much. However, there is enough vertical sound dispersion from the ends of the tweeters to fill-in the sound field as you move away from the array--especially if you are positioned at a normal listening position.

Power tapering is clearly not effective unless the sound fields from the sources overlap. As you have very little overlap of sound fields with these tweeters, then tapering is not going to be benficial. Power tapering would work best for cone type drivers that have wide dispersion so that their sound fields overlap. Power tapering would do nothing to fill-in the response dips that you are seeing in the tweeter line measurements that you are making. I use power tapering only for the woofer line in my arrays and my goal is to reduce the sound bloom as you go between lines. In the literature the previous work on power tapering has been focused on reduction of the off axis nulls but that is a far field effect and not a near field effect.

Bessel arrays have no application for an in-home line array as the array effect sums in the far field. One well known pro sound engineer has suggested a 30 times the array height listening distance for Bessel arrays which is clearly not useful for an in-home application. Bessel arrays do not increase the sensitivity of the drivers used in the array--power handling is clearly increased. Often the impedance of a Bessel array is too low to be useful. Bottom line is that Bessel arrays are a dead end for use in in-home line arrays and in fact are seldom used even in pro sound applications. All of the latest pro sound line arrays operate in both the near and far fields and do not use the Bessel array configuration.

Jim

 

[RichardHOS]

 

[John E Janowitz]

Jim,

I have done extensive measurements at 1m, 2m, and 3m distances from the line array. There will be dips/peaks at all of these distances from the array, but they vary depending on the distance. At 3m there is still a large dip just over 16khz at the center of each tweeter, but nearly flat summation between tweeters. This is what most people did not take into account, and the theory did not consider. At nearly all points away from the line, things summed relatively flat in the blank spaces between tweeters, but got bumpy at the middle of each tweeter. It is easy to make the response curve look flat if you measure at a height between ribbons.

I did measurements moving the mic up and down in 1" increments. Theory would state that they should "fill in" as you said and sum relatively flat the further you go back, but as I said, this happens only at the blank spaces between ribbon elements. At the height equal to the middle of any tweeter, you get some large dips. I am still attempting to determine exactly why these dips occur, and what distance they are attributed to. At 3m, the big dip at 16-17KHz is up to about 18dB or so. The strange thing is that this dip does not occur at 1m or 2m distances.

Power tapering helps only slightly, decreasing the magnitude of this dip by about 5dB, but also creates other problems lower in frequency.

It seems that tapering the response, and not the power is what is most beneficial. Above 8Khz, a pair of the ribbon tweeters sums very well up past 20khz, giving smooth response, and the least deviation between response at tweeter centers and space between tweeters. From 5Khz-8KHz, 4 or 6 tweeters being used gives the best response with least deviation. Below 5KHz, all 12 tweeters being used give only about 2dB of variation at all line heights.

Another strange thing that I noticed was when I put foam blocks between tweeters. I made some small foam blocks to put between the tweeters, hoping to absorb a little more of the already limited off axis response. Doing this actually increased the magnitude of the 16KHz dip even more.

In any case, there is a lot more work to do before I finalize this line array. For now, it looks like tapering the response as you go away from the middle of the line is the best option.

John

 

[Jim Griffin]

JOhn,

Thanks for the updates on your data collections. I haven't seen a dip at the center of my arrays but I looking at 6 to 8 planar tweeters in a line so according to your results that is likely not to generate a dip.

Again, I don't recommend tapering a planar or rbbon tweeter line as you have very little sound field overlap on these drivers. If you do power taper the woofer line, it is best to do it symmetrically about the center of the array (highest power output in the center of the array). My tapering is less than 2 to 1 power change from the ends to the center of the array--too much power tapering will reduce your large line array to a mini-monitor size which is not a good thing to do. The intent in my work is to reduce the sound bloom so that the sound between the two lines balances and is natural.

Let us hear how your efforts turn out.

Jim