Crossover points and music power distribution

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2003
I'm working on plasma tweeters, and, assuming it all pans out, I'll need to find a proper crossover point. Of course, I'm limited by the hardware, but I also want to take into consideration what the best crossover point from the point of view of sonics is. The engineering limits of where I can crossover is perhaps even below 500 Hz, certainly no higher thank 1 kHz. With this in mind, what is recommended?

Additionally, so that I can calculate how much power the low-end frequencies amplifier (driving dynamic speaker) will need to get a matched setup, does anyone have any information on the approximate distribution of power vs frequency of various music types?
 
Well, the ear is most sensitive in the 1kHz to 4kHz range, so it would be desirable to cross over before that, with 500Hz being a very good choice. Crossing over even lower, say 300 Hz, has its attractive aspects as well, however, power handling requirements go up dramatically. The information I have shows that for typical orchestral music, the 250-500hz octave contains the most energy, on average, reaching 10 acoustic watts, peak. (vs. 4 peak watts in the next lower octave of 125-250, and 2 peak watts in the next higher octave, 500-1k). The classic crossover point of 500Hz seems to be the best balance, as well, since it equally shares the audio spectrum. 5 octaves to the bass transducer, and 5 octaves to the treble transducer.
 
The amplitude sensitivity is highest at 1-4k but I have not seen any studies showing that this goes for phase distoriton. Some authors have claimed high sensitivity below 1.5-2k, so crossing above that would be recommended according to that. I like to think that with the fundamental as a basis and starting point, the higher up the crossing the better. The first harmonics are much higher in amplitude normally and it seems they should be most important to be in correct phase relationship with the fundamental.

Power/peak power requirements,

think that 400Hz is the point for most music in general where the power is equal above and below. The amplitude demand is highest in the mids, almost as high in the tweeter range and the lowest in the bass.

Some modern synth/electric music may alter this thumb of rules though. So, a 200W amp to the mid, 100W amp to the tweeter and 50W amp to the (sub)woofer may not be as crazy as it sounds.

As always, it depends.. :)

/Peter
 
Hornlover said:
Well, the ear is most sensitive in the 1kHz to 4kHz range, so it would be desirable to cross over before that, with 500Hz being a very good choice. Crossing over even lower, say 300 Hz, has its attractive aspects as well, however, power handling requirements go up dramatically. The information I have shows that for typical orchestral music, the 250-500hz octave contains the most energy, on average, reaching 10 acoustic watts, peak. (vs. 4 peak watts in the next lower octave of 125-250, and 2 peak watts in the next higher octave, 500-1k). The classic crossover point of 500Hz seems to be the best balance, as well, since it equally shares the audio spectrum. 5 octaves to the bass transducer, and 5 octaves to the treble transducer.
500hz is right in the middle of the male voice and is not ideal for a crossover freq as such, lower than 300hz is much better to avoid the problem. It is much easier to create 10 acoustic watts at mid freqs than it is to make 4 watts at lower freqs, it takes many more electrical watts for the movement of enough air to get the same spl at 200hz as 1khz. I have read that it is actually 200hz for the 50/50 point in electrical power over the audio spectrum.
Plasma drivers have only been used for the very to end of the audio spectrum, how low do you think your going to get? Below 5khz is not going to happen! The amount of power(and voltage) needed to get below that in a plasma will be aproaching the limits of relativity and also make one helluva lot of ozone!
 
Disabled Account
Joined 2003
I'm not talking about corona discharge tweeters, but Hill Plasmatronics-style DC glow discharge tweeters. Originally crossed over at 1 kHz, I've been in correspondence with owner Jay Philippbar and he says crossing over at 500 Hz looses only a couple of dB. My design is derived from these and I'm expecting similar results.

Anyway, this has been discussed in a couple other threads.
 
Markp, you are corrent in that 500Hz is in the voice range, however, in many tests we have run, it has been found that the ear is very tolerant to a crossover point of 500Hz. It would be nice to have the luxury of a lower crossover point. We typically use 500 because we are using compression drivers for midrange use, and have limited availability of drivers or horns to cross over lower than 500 (there are always compromises, huh). I got my figure of the most energy in the 250-500Hz octave from information gathered by Altec-Lansing, showing power distribution of a typical symphony orchestra. Undoubtedly, various other music sources will have various other energy distribution spectrums. For rap, by what I hear in teenagers cars, it seems to be centered at about 70Hz. By using 500Hz as a good dividing point, 5 octaves for each, it was based more on distortion reasons, not energy distribution.
Obviously, choosing an optimum point is based on many factors, which is typical for real world problems, and solving them usually means a compormise somewhere. Good engineering tries to make the compromises livable.
 
I personally dont like crossovers that high. I can hear them too easily. That is why I dont want any crossover in the ears most sensitive range, between 1 and 4k. I prefer no crossover higher than 800, and preferably lower. Another at 5k or so, if needed, but keep that midrange intact! Also, any woofer good at decent bass response isnt any good passed about 500-800 Hz, in my opinion. The music is in the midrange.
 
Human voice on 1 speaker philosophy

Putting as much of the human vocal range on 1 speaker as possible has been my experimental philosophy.

Today there are speakers that can handle 80Hz to 1.5Khz if steep slopes are used. Original efforts with LR8 Xover on MTM with Scan Speak 5" Revelators and Seas Millennium sounded dull, but new 96db slope digital Xover that easily compensates for offset and phase have made these speakers a delight for vocals. I plan to build an isolated woofer box to extend to WMTMW with some Lambda TD15s.

To fit in with a wall mounted flat screen, I'm also experimenting with pushing these speakers against the wall and reducing/removing the baffle step compensation.

I would really like to believe HornLover's 500Hz xover could excell on human voice, but my experiments with Xovers near 500Hz have taken the charm out of Kathleen Battle's soprano voice. HornLover....what slopes and Xover topology works best on 500Hz 2-way??
 
tiroth said:
That's a very interesting article, but I don't understand where your conclusion came from.
They have shown that we are extremely sensitive to phase/delays for frequencies below 1K. That might be a surprise, but you can detect delay difference of about 13microsec at frequency of 500Hz. There goes your soundstage and imaging. Put XO right there? no thanks.

That sensitivity goes to nada near 1.5K. Beyond that we have increased sensitivity to amplitude variations and from around 3K we have most sensitivity for that. This leaves a gap between 1.5K and 3K where the "physiological crossover" is. Its the range where we are not good at either kind of localization and things are fuzzy. Fact that we have the most absolute sensitivity in 3K range isn't neccessarily meaning that amplitude linear crossover at 1.5K is audible.

Selection of 1.5K was my conclusion based on goal of placing any phase distortions of XO into range we are least sensitive to, at the same time avoiding most sensitivity to linear distortions at 3K.

If you observe equal loudness curves http://hyperphysics.phy-astr.gsu.edu/hbase/sound/eqloud.html then you can easily notice that ear is less sensitive between 1K and 1.5K than below and above and is basically flat from 500 to 1K. This further makes me prefer focusing on phase issues rather than be afraid of ear sensitivity at 3K.

My conclusion is that range near 1.5K is most forgiving for XO. YMMV.

PS. http://www.linkwitzlab.com/ has picked 1440Hz as XO, and although I don't know if his selection criteria has anything to do with the above, the Orion speakers have only received excellent feedback. Thats not to say that this proves my point, but only that it is possible to make flawless speakers with XO at 1.5K. (btw site worth reading cover2cover)

PPS. Also if you place XO too low, you will overload your tweater by spectral density.

Thats all IMO obviously.
 
As freq goes up wavelength goes down and it gets increasingly harder to localize sound sources above 7khz or so due to our physiological limitations. The distance between our ears makes it easiest to locate sounds above ~250hz and below 7khz where the wavelength gets too short for our heads to be 1/4 wavelength wide. Below ~250hz there is not enough difference in amplitude between the ears due to the very long wavelength for our brains to locate the sound. Phasing is of the utmost importance for localization and with a crossover at 500hz you are shooting yourself in the foot, speaking metaphorically.
 
I wonder the same as tiroth.

Remember that the article deals with the localization due to ITD and ILD, IOW differences between the two ears. This is not the same thing as the waveform distortion from a cross over.

Very good article nonetheless. Thanks for posting. I put it up on the Mad board also.

/Peter
 
So the reasoning is that a crossover in the 250-1.5kHz region will screw things up because we're especially sensitive to acoustic phase in this region, right?. OK, thanks for clarifying, that wasn't obvious to me.

Next question, does this really matter? If we choose to cross in a region where the phase of the drivers is not rotating, won't the phase of the speaker be relatively smooth? Looking at the graphs I have on hand, I can't detect the crossover point by looking at the acoustic phase...
 
A phase accurate/TP x-over side step the problem but there is still the issue with off axis dispersion, which may be one of the factors that makes a x-ing audible.

Again, remember that phase detection BETWEEN the two ears is not the same thing as phase distortion in the absolute sense.

/Peter
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.