Nice effort Eric! Great response. Hopefully the poster will accept your advice in the helpful spirit you intended.
bb
> 1. First thing you need to do is download
> Jim Griffin(PHD)'s white paper on nearfield
> line array design, which is the use of line
> arrays in rooms as opposed to concert halls.
> You can search on google by looking for
> "James Griffin" and "white
> paper" or I'll email a copy to you. If
> you are already into this you are ahead of
> the game. This is what we all use to define
> how line arrays have to work. Before Jim did
> his research no one knew the pitfalls for
> using a line array in a room.
> 2. The next step is that you need to leave
> an octave between the place that you cross
> if you use a 24 db per octave crossover.
> An octave is a doubling or half of the
> frequency. So if you cross at 3000, an
> octave down is 1500, 2 octaves down is 750.
> 24 db/octave is pretty much silent at those
> frequencies.
> So the the next step is the look at the
> frequency curve of the speaker. pushing hard
> more than 3db down is going to make a
> tweeter screech.
> Zaphaudio has a frequency response graph for
> this tweeter at:
>
> <A HREF="http://www.zaphaudio.com/tweetermishmash/fr.html">http://www.zaphaudio.com/tweetermishmash/fr.html</A>
> Now if you're going to use 35 of them they
> aren't going to be putting out much, so you
> can probably fudge the numbers and go down
> 1/2 an octave. But in any case you want to
> not go below the FS of 2200 when you are at
> least 3/4 of an octave down.
> Bottom line for me is that I doubt that you
> want to cross at below 3000hz for this
> little item, using 24 db/octave slope and 30
> of them.
> 3. The size of them. Assuming that you ****
> them right up against each other their
> center to center distance is 1.30 inches.
> This means using the formula 13560/1.30
> inches = 10430, which means that your comb
> filter distortion is going to start at
> 10,000+ hz. This is pretty low, and clearly
> audible. It means that the frequency
> response of the highs is going to
> deteriorate starting there and there is
> nothing you can do about it. For any example
> of a frequency response graph showing what
> happens to the highs in comb filtering go
> here. This is an example using a speaker
> that has no tweeters, where the distortion
> starts much lower, but you could expect the
> same slop of drop starting around 10,500 in
> your tweeters:
>
> <A HREF="http://www.audioroundtable.com/ArraySpeakers/messages/1512.html">http://www.audioroundtable.com/ArraySpeakers/messages/1512.html</A>
> For how comb filter distortion sound, here
> is a quote by Jim Griffin here on the PE
> forum:
> "How comb filtering sounds
> "Comb lining will dull the highs and
> will be most noticeable in the lack of air
> (dull highs) especially as you move off
> axis. Essentially, the high frequencies are
> rolled off as the ensemble of the tweeter
> outputs don't add up but start to cancel.
> Now the ear is less sensitive in the upper
> octave (10-20 kHz) you may not observe the
> full impact of this effect. But if you
> compare to an array that is properly
> designed, then there is no comparison as the
> proper array will yield a more airy (and
> accurate) sound.
> "Jim Griffin on PE 11-21-06"
> In my opinion, you would be much better off
> with the ND20TA's. Cut the flanges with a
> scroll saw and you can get Comb filter
> distortion up to above 14,000hz. I'll email
> you a tutorial on the cutting if you wish.
> In addition, you will be able to cross lower
> with them, and they won't screech as soon.
> Also compare the frequency response graph on
> the Zaph site between the ND20Ta and the 16
> and you can see which one is flatter.
> 4. The next question is what you are going
> to use for a mid/woofer with the ND16's. If
> you can't cross below 3000 with them, then
> the mid ranges have to be able to go up to
> 3000. This is pretty limiting, and it means
> that your center to center distance between
> the midrange speakers cannot exceed 4.5
> inches, which means that you are limited to
> midrange speakers of about 4 inches and
> down. Now this is easily possible the one of
> the HV models, which will run you about $10
> each. You'll need about 40 of them. But it
> also means that you will absolutely have to
> have some kind of woofer into the bottom
> end, since you little three or 4 inchers
> won't go below 100hz in the best of worlds.
> You have to avoid comb filtering distortion
> in the mid range too, which has much more
> serious effect on the sound even than
> tweeter comb filter distortion. It causes a
> warbling sound to the music, and serious
> lack of frequencies because of the
> cancellation of frequencies, in the very
> audible midrange areas.
> 5. Designing the crossover. Passive
> crossovers are difficult to design
> accurately. If you've followed this forum at
> all, you can see how this works with a fella
> named Paul N. O'Neal who has developed a
> terrific point source speaker, but who kept
> us informed across the board on the
> development of his crossover. Doing this for
> a line array is even harder, and you can't
> use something off the shelf. with all those
> speakers you are going to be pumping a lot
> of power into them, and you will also have
> likely a huge difference in the sensitivity
> between the mid and the tweets which have to
> be compensated for, and done so at high watt
> levels. Rick Craig will design one of these
> passive crossovers for you but his design
> work will cost you $300 and then you still
> have to buy the $200 bucks in parts. For all
> of these reasons and more, most line array
> designers use electronic crossovers, and bi-
> and tri- amping. You need to consider this
> too. I use a system that puts out a total of
> 1330 watts RMS and is similar to what you
> might be building. It has four power amps,
> one pre-amp, a Rane 24/db slope electronic
> crossover, and a 2/3 octave constant Q
> equalizer.
> ------------
> As you can see, the issue that you bring up
> is not simply the crossover for the tweeter
> but the integration of the whole system. in
> line arrays, the integration of the whole
> system is much more important and especially
> so considering the incredible amount of work
> involved in making them. It took two years
> for me to build mine.
> Here's an example of a discussion with a
> person who failed to pay clear attention to
> the integration of the system. His midranges
> turned out to be too big, and he had a
> serious drop out of frequencies between
> about 1600 and 2400:
>
> <A HREF="http://www.pesupport.com/cgi-bin/config.pl?read=340023">http://www.pesupport.com/cgi-bin/config.pl?read=340023</A>
> Hope this all helps, and didn't discourage
> you. But it would be a shame to spend all
> that money and the enormous time and energy
> building, only to have it turn out badly for
> you.
> kind regards,
> Marlboro