I can't seem to find any useful info on the power consumption of individual speakers based on frequency in a 2 or 3 way. For example, How much power does a tweeter crossed above 3000 Hz use in comparison to a woofer crossed below 500 Hz?
Any formulas or links to this info. I appreciate anyone's help.
Any formulas or links to this info. I appreciate anyone's help.
Hi,
The 50% power point is ~ 350Hz for rock, a little higher for classical.
Treble wise peak and average powers are rarther different, average
levels above 3kHz are low, but peak levels not so much.
Mid/treble split around 3Khz of the remaining 50% is ~ 30%/20%.
For bi-amping with the same amplifiers active bass/mid and passive
mid/treble (possibly with active BSC) is an attractive option, see :
http://www.musicanddesign.com/HybridDesign.html
/sreten.
The 50% power point is ~ 350Hz for rock, a little higher for classical.
Treble wise peak and average powers are rarther different, average
levels above 3kHz are low, but peak levels not so much.
Mid/treble split around 3Khz of the remaining 50% is ~ 30%/20%.
For bi-amping with the same amplifiers active bass/mid and passive
mid/treble (possibly with active BSC) is an attractive option, see :
http://www.musicanddesign.com/HybridDesign.html
/sreten.
kucharcity said:
The reason that you can't find any information is because its a dificult problem. SO much depends on the spectrum of the music. One can talk about a "typical" selection, but then I could always find an atypical example that was completely different. When talking about power you have to remember to talk in linear frequency terms. In other words for a broadband noise spectrum there is as much "power" from 1000 Hz to 2000 Hz as there is from 0 Hz to 1000 Hz. Its just that most music tails off in spectrum at the top end, above about 1000 Hz.
When an amp clips it shifts the spectrum upwards and this changes the "typical" spectrum dramatically. This is why a short blast of a clipped signal can take out a tweeter almost instantly.
In real terms, tweeters take a lot more power than woofers. Consider a woofer from 0 - 1000 Hz. and a tweeter from 1000 Hz up to 20 kHz. The tweeter is getting 19 times more power than the woofer.
Thanks to both of you for your replies. They were both very helpful. While I don't quite understand why a tweeter would possibly use 19 x's the power of a woofer above 1000 Hz, it's more logical for me to see the the power to be 30/20 %(Mid/Tweet) above 350 Hz. However I'm not an expert and I respect all responses.
While I have not done a comprehensive study of this, when trying to answer the same question I have written a program in Matlab that takes a chunk of a song in, breaks it up into octave-sized frequency bands, and then computes the rms power in each of those bands. After sending in about a dozen "typical" songs (reasonably uncompressed rock songs), I found that the energy per octave is slightly downward-tilted in frequency and falls off pretty quick below 30 Hz. If I break the signal up into say, 20-80 Hz, 80-250 Hz, 250-1.5k Hz, and 1.5-20 kHz the power going into each of these bands is pretty close.
The EIA 426B standard noise is designed to approximate the power distribution in music. The frequency content of this is as such: 4th order BW highpass at 40 Hz, half order lowpass (-3 dB/oct) at 1 kHz. What I have observed is "on average" consistent with this.
The EIA 426B standard noise is designed to approximate the power distribution in music. The frequency content of this is as such: 4th order BW highpass at 40 Hz, half order lowpass (-3 dB/oct) at 1 kHz. What I have observed is "on average" consistent with this.
This is quite reasonable.
Had the other "gentleman" not been so rude, I would have pointed out that my claim was for noise as implied by the context of the note. With a -3 dB slope to the noise (i.e. pink) the power frraction above about 1 kHz will be about 1/2 the total. At -0dB it with be 19/20 the the total just as I claimed.
Had the other "gentleman" not been so rude, I would have pointed out that my claim was for noise as implied by the context of the note. With a -3 dB slope to the noise (i.e. pink) the power frraction above about 1 kHz will be about 1/2 the total. At -0dB it with be 19/20 the the total just as I claimed.
gedlee said:
Hi,
a) Is simply not true.
b) Is only true for white noise, constant energy per unit frequency.
This power spectrum is irrelevant when considering music.
(in other words forget about linear frequency power spectrums)
c) for pink noise, constant energy per octave, 50% is above/ below 1khz.
(for a symmetrical bandwith, say 10 octaves, 32Hz to 32Khz).
How this leads to 95% to a typical tweeter is beyond me.
A typical c/o point of 3KHz yields 65% bass/mid / 35% treble.
What 0dB has to do with the 95% claim is also beyond me.
/sreten.
Rybaudio said:
Correct, if 426b uses pink noise as a source, I was assuming white noise.
The thing that often gets missed is that the power spectrum shifts upwards with clipping going to pink and then more to white in character. Since clubs clip nearly all of the time, the power spectrums that you are assuming would be inadequite in this application and you will see HF failures in the field.
You haven't found this to be true? Because I have.
That is something I have never needed to look into. For my personal system, it isn't much of a concern, as it essentially never clips.
Don Keele created 426B. If I remember right, he arrived at that power distribution using a simple analysis somewhat like what I was talking about above. If that is the case, then nonlinearities weren't considered, so your point is very relevant. I have no idea of the magnitude of the effect due to clipping in a typical club situation, so I can't comment any further.
Don Keele created 426B. If I remember right, he arrived at that power distribution using a simple analysis somewhat like what I was talking about above. If that is the case, then nonlinearities weren't considered, so your point is very relevant. I have no idea of the magnitude of the effect due to clipping in a typical club situation, so I can't comment any further.
I did not write that spreadsheet. I have fished it at a point from the net. It says it is based on some 'best fit exponential curve' from some acoustics book. It asks for a password to unlock. So I cant extract the function for you. Its output just seems logical to me from practical experience.
P.S. Want me to email it to you, so you can find out more?
P.S. Want me to email it to you, so you can find out more?
salas said:
Thanks, but its not that important to me.
Just for information, we do our power testing with White noise. The reason is that we see field failures with the tweeter if we don't. I suspect that its the clipping issue. Sure its overkill in some respects, but it has aved us a lot of tweeters.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.