That, of course, is exactly the argument. To replace a passive system with an equally divided (by power) active system summing to the same total power (3 x 100W, 10 x 30W, etc.) will only be equivalent if the music is nicely divided amongst sections. Look at the spectrum of music during the loudest peaks and you will find that it is not a noise signal but is a complex tone. As such its distribution will not be wide and noise like, rather it will be concentrated somewhere.
I am not arguing that active designs have no benefit, rather that the sum of the amplifier powers will have to be greater to have parity all the time. In the limit your only guarantee of parity is to give each section the same power as the original passive system's amplifier.
Now, this is not the case if the passive network contains a lot of attenuation. We studied this at KEF and found some passive/active related systems (the LS5/1a and the LS5/1ac) where there was so much network attenuation of the drivers that the active counterpart had a clear advantage. This spured my development of a passive system with no passive EQ, the P60 Broadcast Monitor. The mid band sensitivity was allowed to rise about 6dB and this was actively EQed. Since the un-eqed response was close to the peak spectrum of music the headroom was well optimized. The later KEF 107 and other models used this approach with the KEF KUBE.
David S.
Last edited:
Hi,
I understand the parity argument, but it doesn't follow guaranteeing
parity is the best distribution of the total amplifier power available.
A typical 3 way active speaker will work better most of the
time with peak SPL in each band +3dB/0dB/-3dB compared
to 0dB/0dB/0dB with most quality programme material.
I accept for some material equal peak SPL would work best,
but for a lot of material it won't, due to the higher bass levels
(e.g. CD) and lower treble levels (e.g. vinyl) commonly seen.
If you have a passive 3-way driven by a 100W amplifier and
go active to 100W bass, 50W midrange and 30W treble,
although parity is not maintained, it will work just fine,
and go louder cleaner with all material with bass.
rgds, sreten.
Some music signals are noise like, e.g. drums and some are
clearly tone like, e.g. organ pedal notes, some are inbetween,
e.g. distorted rhythm guitar with noise like leading edges
resolving to to a more tone like spectrum for the chord.
Active speakers generally have better dynamics than passives
due to the handling of noise like transients, but you might not
be able to hear the point at all say with a solo choirboy.
There is no easy answer, its programme dependent, and common
sense should prevail, e.g. a 3 way x/o'd at 300Hz and 3kHz clearly
needs a different amplifier regimen to one x/o'd at 3KHz and 10KHz.
I understand the parity argument, but it doesn't follow guaranteeing
parity is the best distribution of the total amplifier power available.
A typical 3 way active speaker will work better most of the
time with peak SPL in each band +3dB/0dB/-3dB compared
to 0dB/0dB/0dB with most quality programme material.
I accept for some material equal peak SPL would work best,
but for a lot of material it won't, due to the higher bass levels
(e.g. CD) and lower treble levels (e.g. vinyl) commonly seen.
If you have a passive 3-way driven by a 100W amplifier and
go active to 100W bass, 50W midrange and 30W treble,
although parity is not maintained, it will work just fine,
and go louder cleaner with all material with bass.
rgds, sreten.
Some music signals are noise like, e.g. drums and some are
clearly tone like, e.g. organ pedal notes, some are inbetween,
e.g. distorted rhythm guitar with noise like leading edges
resolving to to a more tone like spectrum for the chord.
Active speakers generally have better dynamics than passives
due to the handling of noise like transients, but you might not
be able to hear the point at all say with a solo choirboy.
There is no easy answer, its programme dependent, and common
sense should prevail, e.g. a 3 way x/o'd at 300Hz and 3kHz clearly
needs a different amplifier regimen to one x/o'd at 3KHz and 10KHz.
Last edited:
Hi, my electronics knowledge is not all that great unfortunately. What is the limiting factor for amplifiers? I always assumed it was the maximum undistorted output voltage - is this incorrect?
So how does a 20W amplifier differ from a 100W amplifier? One of them will do 12.65V and the other will do 28.28V - or are there other differences as well?
So how does a 20W amplifier differ from a 100W amplifier? One of them will do 12.65V and the other will do 28.28V - or are there other differences as well?
The second paragraph omits the vital information of the load impedance. The values of voltage you quote are applicable to 8 ohm resistive loading.
So the difference between a 20 and 100 watt amp involves a little more than just the output voltage swing. Load impedance of real speakers varies enourmously. A 8 ohm 3 way speaker for example can present an impedance of anywhere between 3 ohms (or less) to over 30 ohms depending on frequency of applied signal.
So when you initially say,
"I always assumed it was the maximum undistorted output voltage - is this incorrect?"
that statement is correct provided you qualify it with a load impedance. Real speakers also complicate the matter by having a "reactive" component to them although this is often greatly overestimated. A capacitor for example could be chosen to provide the "equivalent" of say 8 ohms at 300 hz. The problem with that is the fact that the current and voltage are 90 degrees out of phase in what is now a purely reactive load. That would be very tough on the amp.
AC capacitor circuits : REACTANCE AND IMPEDANCE -- CAPACITIVE
In practice most speakers are fairly benign in the "reactive" component of the impedance.
Then I guess your previous, harshly worded and absolute response was a bit uncalled for?
David S.
Coincidentally, I am currently considering a project just as you describe.
Because of the mechanical nonlinearities and inefficiencies inherent in bass generation I've quite accepted the need for a larger amp.
The question for me has been just where to xo. At what point does 'bass' become 'mid'? Obviously if I am using a tube amp I want as much of that sound as possible while at the same time relieving it of the burdensome lower registers. A post above describes 200 Hz as 'mid' - at what point does bass end and mid actually begin?
Sreten,
Average power is certainly less in the HF range for most music, but assuming equal driver sensitivity, and the desire to reproduce transients at the same SPL, any combination of crossover points require exactly the same peak power for each band.
That said, LF peaks are of a much longer duration than HF peaks, so generally take a more robust amplifier design, but without going too far down that rabbit hole, my previous statement stands.
Art Welter
(Regarding midrange) Depends on who you ask and the context of the question.
A midrange crossover point may be in the 400 to 800 range. With a 2-way system and typical 2-3k crossover, the woofer is really the midrange as well.
The middle of a piano keyboard includes 262Hz (middle C) or 440 (A above middle C).
The half way point in spectral energy is around 200 (equal average power above and below that frequency) and that is why the number has been used in this discussion.
David S.
A midrange crossover point may be in the 400 to 800 range. With a 2-way system and typical 2-3k crossover, the woofer is really the midrange as well.
The middle of a piano keyboard includes 262Hz (middle C) or 440 (A above middle C).
The half way point in spectral energy is around 200 (equal average power above and below that frequency) and that is why the number has been used in this discussion.
David S.
Last edited:
Using 20-20KHz as the range of music, a three way split using an equal reproduction range puts the mid range around 200 to 2500.
If we divide the ranges further into four, low is up to around 120 Hz, low mid to 800, high mid is 800 to 4000 Hz, and highs above 4K.
That said, it is much easier to communicate precisely using frequency range to avoid confusion, in a two way system "high" could mean frequencies anywhere from 100 to 5000 Hz.
Art
I've always considered "bass" to be frequencies you can feel and hear at the same time, midrange to be between that and where the ear is most sensitive, then highs.
I guess that would put me in the 0-150 (bass) 150-4000 (mid) 4000-20000 (treble) camp.
Combining the term "midbass" is usually only done, at least as far as I understand, in order to combine the duty - and then accurately refer to a "midbass" unit. It reaches a little deeper than a typical midrange driver so it's tough to call it that, but that's really what it is, a midrange driver with decent motor strength and excursion to get it to reach low enough and then mate to a typical subwoofer.
I blame the "midbass" term on the "mono subwoofer" market, where they manufacture subwoofers to only cover below the localization threshold, leaving us with a need for midrange+midbass combo drivers to pick up where they leave off.
Holywood! .|..
Years ago I worked in broadcast FM. Just as the RIAA has pre/de empasis, bradcasting has NAB pre-emphasis. Built into our limiter in the audio chain was the pre-emphasis curve. This curve was some attempt at limiting only the components that might cause overmodulation. I know we did not quite have the variation of music styles and their dynamics of the selection today, but, that curve is somewhat modeled after the necessary power-bandwith typical in audio/music. Was it not?
I know we could find that curve in a mater of minutes... Hmmm
I know we could find that curve in a mater of minutes... Hmmm
I was told that as a rule of thumb the equal power point of average music (which is largely similar to pink noise aka 1/f noise) is around 350Hz.
In other words if we have a 2way active speaker with a 350Hz xover the two amps would need to be of equal power provided the two drivers are of the same sensitivity.
Here is how wiki describes 1/f noise:
"There is equal energy in all octaves (or similar log bundles). In terms of power at a constant bandwidth, 1/ƒ noise falls off at 3 dB per octave."
How correct any of this is I do not know.
However my own active 4ways are crossed at 250Hz, 1200Hz and 15 000Hz.
The woofer has a sensitivity of 91dB, the mid 92dB. The woofer always takes more power than the mid regardless of type of music, not even with music which hardly has any low bass components.
In other words if we have a 2way active speaker with a 350Hz xover the two amps would need to be of equal power provided the two drivers are of the same sensitivity.
Here is how wiki describes 1/f noise:
"There is equal energy in all octaves (or similar log bundles). In terms of power at a constant bandwidth, 1/ƒ noise falls off at 3 dB per octave."
How correct any of this is I do not know.
However my own active 4ways are crossed at 250Hz, 1200Hz and 15 000Hz.
The woofer has a sensitivity of 91dB, the mid 92dB. The woofer always takes more power than the mid regardless of type of music, not even with music which hardly has any low bass components.
They are just defining what we commonly call "pink" noise. It is equal energy per some percentage bandwidth (as in "per 1/3rd Octave") rather than per linear bandwidth (as with white noise). If that were the case then the half power point would be exactly half way from 20 to 20,000 Hz on a log scale or about 640 Hz.
This is much less accurate than looking at the average spectrum of music. Most published averages peak at a lower frequency and roll off to both sides.
I did a web search and couldn't find much on the typical music power spectrum (I have it in books) but there was a little info on it here:
Loudspeaker Sensitivity – What's a Watt, Anyway? Synergetic Audio Concepts
...with one good spectrum of voice that peaked at 315 Hz. I would expect music to be wider in bandwidth, at least at the low end.
Regards,
David S.
Attachments
![VOL37_APR09_Loudspeaker_Sensitivity_Fig3[1].jpg](/community/data/attachments/286/286810-aa3cce42ec78ef7a8ccc63aee8e570b8.jpg)
And here are some typical loudspeaker test signals that were designed to mimic the long term spectra of music.
Speaker power handling < Pro-Audio References
David S.
Speaker power handling < Pro-Audio References
David S.
Attachments
Originally there was no eq or level settings on the bass at all but I had to move the speakers into corners after moving house.
Since then I had to knock the bass back by about 6dB to get a level output and still the bass takes more power than the mid on the vast majority of music I play.
There is some eq on the tweeter, a shallow notch around 2.5k and 6dB/oct boost above 5k.
My ribbon supertweeter (>15k, no eq but level reduced by 4dB) only takes 5Wrms before melting but after many a party it is still fine, even after the woofer got the full 400W available to it. I drive it with a 56Wrms amp.
You mentioned spectrum power. Long ago (1998?), I wrote a C program that did windowed FFT's on any track/song from my CD collection (I converted all 500+ CD's to digital files of the WAV format type for safe storage by using EAC). Window type and FFT size were user specified. And, long ago, PC's back then couldn't do the math quick enough. But now they can.
If I were to mathematically sum certain frequency bands ("bins" in the program), would that obtain a spectrum power reading that would be accurate for a given range?
After reading some books on DSP (again, back in around 1998), I wrote that program to see if I could digitally filter out anything I wanted to. This was for the fun/challenge of it because in theory, a digital bandpass filter that goes from say 5433Hz to 10238Hz EXACTLY could be done with enough processing (convolution was the method? I cannot recall). And, it did work. The GUI was crude but worked. Also, a digital track converted to a frequency spectrum over time could be stored and then reconverted to a time domain signal perfectly.
As for the spectrum power, each frequency bin does produce a number which represented its dB value accurately as related to the dB range possible using 16-bits (and then visualized via the crude GUI). So, a mathematical summation of the desired frequency range should be possible?
I'm just trying to simplify things via some program and math I learned long ago. I always wanted the program to do more but the processors could not handle it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.