How much comparable power do we need for different frequencies?

[Jumu]

Does anyone know how much power we need for each frequency? I mean, I wanna know, how much relative power do I need for 20hz, 30hz, 50, 100 ...

Lets say; at 20hz, for each additional 3 db, I have to double the power comparing 30hz

for 30hz for each additional 3db, I have to double the power comparing 40hz.
..... etc.

I hope I could explain

 

[JeremyFife]

There is some related info and discussion here https://audiosciencereview.com/foru...-fletcher-munson-loudness-compensation.29543/

 

[AnalogSteph]

I swear we had a discussion of music power spectra just the other day, but don't ask me where that was. Average power can be quite "pink", but peak power may be a lot flatter.

 

[Jumu]

Thank you for replies.

You know all sensitivity measures of speakers are taken differently. What i am wondering is a general rule without counting on speaker’s specifications.

What I am trying to understand is what is the rational relationship between power and frequencies?

For example, let’s say we have a 90db/1mt sensitivity speaker. So;
How much power do we need to get
90db/1mt and 100db/1mt at 20hz?
90 and 100db at 30hz?
90 and 100db at 40hz?
…etc.

 

[Jumu]

I swear we had a discussion of music power spectra just the other day, but don't ask me where that was. Average power can be quite "pink", but peak power may be a lot flatter.

Music power spectrum? Maybe this is what I need

 

[NTK]

For a standard surrogate test signal that mimics music, you can download the m-noise (AES 75) and run your spectrum analyses.

AES75 Users and Resources

The AES Standards Committee is the organization responsible for the standards programme of the Audio Engineering Society. It publishes a number of technical standards, information documents and technical reports.

www.aes.org

 

[Jumu]

For a standard surrogate test signal that mimics music, you can download the m-noise (AES 75) and run your spectrum analyses.

AES75 Users and Resources

The AES Standards Committee is the organization responsible for the standards programme of the Audio Engineering Society. It publishes a number of technical standards, information documents and technical reports.

www.aes.org

Thank you. I will check it

 

[Jumu]

Let’s say, both tweeter and subwoofer has 90db sensitivity. But these sensitivity numbers are measured in different frequencies. At least it’s not the sensitivity of all frequencies. It’s generally sensitivity of 1khz for many speakers.

If we had a one single full range driver and if we knew the different sensitivity values of each frequencies, (not an average in 1khz), then we could have known the answer

 

[NTK]

It is much less confusing when looking at the voltage of the driving signal instead of power.

Speaker output is proportional to voltage, and amplifiers amplify voltage (output voltage = input voltage × gain). The standard voltage level for speaker measurement is 2.83 Vrms. What you get from the measurement, using a frequency sweep at 2.83 Vrms, is the standard frequency response of the speaker. You can back calculate the power required for that frequency response curve if you have the impedance curve of the speaker.

In other words: Get from the standard frequency response curve the dB SPL level at your frequency of interest, get the speaker impedance at that frequency. The apparent power (in unit of VA) will be: 2.83^2 / | Z_speaker |. The real power (in watts) will be the apparent power multiplied by the cosine of the impedance phase ( = angle(Z_speaker) ).

I will say in closing that SPL vs power is a pretty useless metric (unless you are concerned about your electricity bill due to your music listening habit). What is important is that your amplifier can supply the required current per Ohm's law at the required output voltage (= input voltage × gain per the volume knob setting).

 

[Chris A]

Does anyone know how much power we need for each frequency? I mean, I wanna know, how much relative power do I need for 20hz, 30hz, 50, 100 ...

Here is a cumulative spectrogram ("plot spectrum") from Audacity of Steely Dan's Reelin' in the Years (demastered and upwards expanded to approximately studio recorded dynamics):

Does this answer your question?

Chris

 

[Hayabusa]

Here is a cumulative spectrogram ("plot spectrum") from Audacity of Steely Dan's Reelin' in the Years (demastered and upwards expanded to approximately studio recorded dynamics):

View attachment 402502

Does this answer your question?

Chris

This is averaged. You miss the peak values, very valuable for knowing the power needed.

 

[JeremyFife]

I'm asking now, not trying to explain or make a statement: isn't the point that a well designed speaker - anechoically flat, or with the FR response graph sloping down in a good room - is outputting the same energy at each frequency? (edit: for any given input power)

 

[Chris A]

Here is a snippet from Michael Hedges' album Aerial Boundaries, the track is "Bensusan", a solo acoustic guitar album. This is one of the most dynamic albums that I have in CD format (crest factor of 18 dB):

Chris

 

[Hayabusa]

Here is a snippet from Michael Hedges' album Aerial Boundaries, the track is "Bensusan", a solo acoustic guitar album. This is one of the most dynamic albums that I have in CD format (crest factor of 18 dB):

View attachment 402511

Chris

again, does not say much a about the power needed, I will check if I can find a plugin for audacity that tracks the peaks in the spectrum..

 

[Chris A]

I can do this all day long (trust me). You can't do a spectral conversion on a zero-length sample, so any snippet will be finite length.

Besides, I think your objection is definitely moot on the particular point.

The Bensusan spectrum is dominated by high-frequency percussive effects of Mr. Hedges' playing for that short snippet (shown in the last spectral plot).

If you don't ask what kind of instrument or instruments/voice are playing, and how they're being played, then the spectrum empirically is -15.5 dB/decade downslope. Try it (using Audacity) on any music that your ears can stand to listen to for more than a few seconds, i.e., music that sounds balanced in any way (not pure percussion).

Chris

 

[Hayabusa]

I can do this all day long (trust me). You can't do a spectral conversion on a zero-length sample, so any snippet will be finite length.

Besides, I think your objection is definitely moot on the particular point.

The Bensusan spectrum is dominated by high-frequency percussive effects of Mr. Hedges' playing for that short snippet (shown in the last spectral plot).

If you don't ask what kind of instrument is playing, and how it's being played, then the spectrum empirically is -15.5 dB/decade downslope. Try it (using Audacity) on any music that your ears can stand to listen to for more than a few seconds, i.e., music that sounds balanced in any way (not pure percussion).

Chris

ok, lets agree to disagree

 

[Chris A]

Actually, I don't... Think about the spectral balance of the "peaks" as you call them. They aren't measurably different than the average. Only the transient attacks on percussive instruments (or string/wind instruments played percussively) have enhanced higher frequency content, and those output levels are boundable in terms of their inherent power.

Mastering guys always seek to take away those big transients anyway. You have to search for high dynamic range recordings to experience that regime--and those recordings I find are actually pretty rare. Most loudspeakers that are direct radiating are also significantly distorting in compression or thermal compression during loud passages.

Also note that what is important is the average values of spectral power, since just about any amplifier worth its weight possesses dynamic power output capability for those short transients.

Trying to read between the lines of what the OP (@Jumu) is really asking, I know of no one that uses amplifiers or loudspeakers with near-zero undistorted headroom if they typically play a variety of music styles--unless perhaps they're into chip or perhaps very low power tube amplifiers producing lots of compression distortion and second or third harmonic distortion while playing typical music at "reference" levels (i.e., extremely loud).

Chris

 

[Hayabusa]

Actually, I don't... Think about the spectral balance of the "peaks" as you call them. They aren't measurably different than the average. Only the transient attacks on percussive instruments (or string/wind instruments played percussively) have enhance higher frequency content.

Also note that what is important is the average values of spectral power, since just about any amplifier worth its weight possesses dynamic power output capability for those short transients.

Trying to read between the lines of what the OP (@Jumu) is really asking, I know of no one that uses amplifiers or loudspeakers with near-zero undistorted headroom if they typically play a variety of music styles--unless perhaps they're into chip or perhaps tube amplifiers producing lots of compression distortion while playing typical music at "reference" levels (i.e., extremely loud). .

Chris

Ok lets give it an other try: The OP asks about the amplifier power needed for different parts of the spectrum.
Assuming the goal is to have the amplifier handling this without clipping, it needs to be able to handle the peaks in the selected parts of the spectrum.
This is very different from the accumulated spectrum which gives an average per frequency component.
For instance the whole spectrum of Steely Dan (Normal CD):

and one part at 48.3 sec :

As you can see there is a peak at 200Hz at -19dB
in the accumulated spectrum this down by more than 10 dB.

His amp surely needs to be able to handle that -19dB peak and not the level in the accumulated plot

 

[Chris A]

...in the accumulated spectrum this down by more than 10 dB.

What I see is the instruments in the midrange have momentarily dropped out for that snippet...

I can do this all day long. Perhaps the OP can shed more light on his actual question...(?)

Chris

 

[StigErik]

What you could to is to play a few favorite tracks at your preferred listening level, and record the actual signal level (voltage) that goes to each loudspeaker driver using a sound card and some DAW software like Audacity. You may need to pad down the signal level, and also find and set a reference level. As an example you could set 4V to equal -20 dBFS. Then 0 dBFS will equal 40V, or 200W @ 8 ohms.

I've done this, and my findings were quite interesting... Just look at the very large peaks on my tweeter! Top of the scale is 100V. The tweeter's highest peak is at ~9 dB, which is ~36V, or ~450W @ 3 ohms !! Poor tweeter...

Some context is needed:
Woofer : XO at 60 / 450 Hz. Sensitivity ~105dB (four 15")
Lower mid: XO at 450 / 3kHz. Sensitivity ~90 dB (Magnepan 3.7)
Upper mid: XO at 3k / 6 kHz. Sensitivity ~80 dB (Magnepan 3.7)
Tweeter: XO at 6 kHz. Sensitivity ~80 dB (Magnepan 3.7)

Subwoofers not included here.

Music was Yello - Junior B played back with the volume control cranked up to eleven.