3 way speaker system

[xtp]

The input data (what I know ;D):
-Subwoofer: 20Hz-200Hz
-Midrange: 200Hz-4000Hz
-Tweeter:4000Hz-20000Hz
The output data (what I'd like to know ???):
-Subwoofer amp power
-Midrange amp power
-Tweeter amp power

I'm using 36 dB/octave active band-pass filters to separate the frequency domains for each speaker.

 

[audioguru2]

Hi Xtp,
Of course the power requirement is determined by how loud you want the sound to be. Since your hearing's response to loudness is logarithmic (so you can hear a pin drop and also hear a gun blast), then 100W sounds only twice as loud as 10W. Doubling the power sounds only a little louder. Half the power sounds just a little less.

The output power of the amps is also determined by the power rating  and efficiency specs of the speakers. A subwoofer has a low efficiency and needs more power to play music and movie effects. A tweeter is efficient and fragile and there isn't much power in high frequency music and effects anyway.

36dB/oct is an even multiple which causes phase cancellation at the crossover with Butterworth filters. It is 6-poles which is extreme. Usually a 3 or 4-pole (18db/oct or 24db/oct) Linkwitz-Riley crossover filter is used which is simpler and doesn't cause phase cancellation. ;D 

 

[xtp]

The filters are (theoretically : - I haven't built them yet) six 6dB/octave high pass and six 6dB/octave low pass in series, each using a resistor and a capacitor.

One high pass filter and one low pass filter in series form a band pass filter.
Six 6dB/octave band pass filters in series should form a 36dB/octave band pass filter :-\.

I simulated 6, 12, 18, 24 and 36dB/octave band pass filters using a computer program and the last one was the best, it sounded better than the others ;D.

If I wanted to make the building of this sound system easier I would have chosen a single amp per channel and separate the frequencies with passive filters (inductors and capacitors) but I don't think it would sound better than "one amp per speaker".

 

[Mosh1]

hi, so you want to built a 3 way system with active crossover, that means, you separate the frequencies before the amps?
well it is the best way but the more expensive, you need to have 3 diferent amps per side, in my work we generaly use k3900 (3900w RMS) for bass, k2400(2400w rms) for mids, all class H, and a k1200 for the hi (class ab) all in bridge mode,
for claculating the powers u need check your speakers and see what is the rms power of them, example:
300w rms LF driver i will put a 300w amp, and so on...

regards

 

[audioguru2]

The filters are six 6dB/octave high pass and six 6dB/octave low pass in series, each using a resistor and a capacitor.

Connecting filters in series results in a response that is very droopy, attenuating frequencies that you need that are far from the desired crossover frequency and even attenuating the crossover frequency too much. Butterworth and Linkwitz-Riley filters use a circuit with feedback to result in a very sharp cutoff at the calculated crossover frequency.

One high pass filter and one low pass filter in series form a band pass filter.
Six 6dB/octave band pass filters in series should form a 36dB/octave band pass filter :-\.

Without feedback, a 6db bandpass filter filter has very gradual cutoff slopes that actually attenuate frequencies in the bandpass. Adding them in series makes the attenuation worse. Feedback is needed in a multi-stage bandpass filter to result in sharp slopes so that the bandpass isn't attenuated.

I simulated 6, 12, 18, 24 and 36dB/octave band pass filters using a computer program and the last one was the best, it sounded better than the others

Simulate a properly designed Butterworth or Linkwitz-Riley filter and you will see their very sharp slopes. Listen to one and it will sound fantastic.

If I wanted to make the building of this sound system easier I would have chosen a single amp per channel and separate the frequencies with passive filters (inductors and capacitors) but I don't think it would sound better than "one amp per speaker".

Using a filter and amp for each speaker is much better than using a single amp, because clipping in a lower frequency amp doesn't affect the higher frequency amps. A filter ahead of an amp is better than an inductive-capacitive filter feeding a speaker because the resistance of the inductor and high-capacitance required by the capacitor ruins the damping and response for the speaker. ;D

 

[audioguru2]

for calculating the powers u need check your speakers and see what is the rms power of them, example:
300w rms LF driver i will put a 300w amp, and so on...

If your speakers and amps have true RMS specs then an amp with a power rating of 3 to 10 times the speaker's rating can be used since music or speech is never continuously at max power. The extra power is needed for the speakers to produce momentary peaks in the music or speech without having the amp clipping. The speakers must have properly designed enclosures so that the extra power doesn't cause them to fail mechanically. In a properly designed enclosure, the air inside or inside a vent is used to control its excursion. ;D

 

[Alun]

What about about the extra power this could make the speakers themselves start clipping?

 

[audioguru2]

Hi Alun,
The speakers shouldn't run into their mechanical limit if the enclosure is properly designed, until the momentary input power far exceeds their continuous max limit caused by heating.

A sealed or completely open enclosure (mounted on a wall between rooms, speakers on a car shelf and using the trunk (boot?) as the enclosure are half-way between) causes a speaker to reach its mechanical limit at the resonant frequency of the system when the cone is moving with long strokes. A vented enclosure has the vent doing the work of producing sound at the resonant frequency of the system (the cone doesn't visibly move), but reaches severe mechanical limit problems below system resonance when it isn't producing sound anyway. Therefore high-power vented speaker systems should have a highpass filter to attenuate frequencies below system resonance.

I had a sealed speaker system from RadioShack with an actual speaker inside stamped "5W" but RadioShack rated the speaker system at 40W for each speaker. On my 70W stereo it played very well but without much bass. I vented it with a long, bent tube inside so its resonant frequency was much lower and it sounded awesome, but the wind from its vent scared my dog. It never reached its mechanical limit but it eventually fried. :'(

 

[xtp]

OK... :-\

So separating the frequencies for the three amps won't be a big problem.

If I choose a 10W amp (per channel) for the subwoofer (20Hz-200Hz) then how powerful should the midrange speaker (200Hz-4kHz) amp and the tweeter (4kHz-20kHz) amp be?

 

[audioguru2]

So separating the frequencies for the three amps won't be a big problem.

Designing the crossover filtering is your biggestproblem.
Are you still going to cascade six 6dB/octave RC bandpass filters without using a Butterworth or Linkwitz-Riley design? Then the response of the filters will be very droopy and the system won't work.
For example, look at the bandpass filter for the 200Hz to 4kHz midrange:
1) You use an RC network for the highpass filter to have its -3dB (half power) frequency at 200Hz. Then its response at 2kHz will be almost at maximum.
2) You buffer each of the six filters with an opamp so they don't affect each other then their total  response at 200Hz will be -18dB, the response at 100Hz will be almost -36dB and the response at 2kHz wll be down a few dB.
3} You use an RC network for the lowpass filter to have its -3dB (half power) frequency at 4kHz. Then its response at 400Hz will be almost at maximum.
4) You buffer each of the six filters with an opamp so they don't affect each other then their total  response at 4kHz will be -18dB, the response at 8kHz will be almost -36dB and the response at 400Hz wll be down a few dB.
The result is a terrible bandpass filter. If you use the same type of filters for the woofer and tweeter then the overall result will have almost no sound at frequencies anywhere near the crossover frequencies.
If you use a Butterworth or Linkwitz-Riley bandpass filter design then the overall result will have a flat frequency response to very close to the crossover frequencies then sharp dropoff beyond, where the next bandpass filter takes over. 

If I choose a 10W amp (per channel) for the subwoofer (20Hz-200Hz) then how powerful should the midrange speaker (200Hz-4kHz) amp and the tweeter (4kHz-20kHz) amp be?

With such a low amount of power then it doesn't matter. Use a 5W amp for the mid-range and for the tweeter. ;D

 

[xtp]

Designing the crossover filtering is your biggest problem.

Not yet.
My 3 way speaker system project is two years old now and since then it's been permanently changing, so I'm not sure what kind of filters I'm going to use for the crossover. This project is still on the "drawing board".

My biggest problem is how much power should each speaker get.
I'd choose 10W (RMS) for the subwoofer, 10W (RMS) for the midrange and 5W (RMS) for the tweeter and that's because I want my system to be able to withstand music played with instruments, not made by electronic means.
The 10W subwoofer amp will have a peak power of 20W, so it'll be perfect for my 8 inch kevlar cone subwoofers (it says that their maximum power is 120W but I don't believe it ).

... : Can you show me some schematics of those Butterworth or Linkwitz-Riley bandpass filters?

 

[audioguru2]

Hi Xtp,
10W per channel, per speaker driver isn't enough for a good music playing system. You will be tempted to "turn up the volume" causing the amplifiers to clip, which will destroy the tweeters (and your hearing). Use much more power and don't worry about "blowing-out" the speakers unless they are really cheap. The sounds in music don't play continuously for any amount of time to overpower speakers.
When I worked for a sound system equipment manufacturer, one dealer brought back some of our cheapest speakers for replacement under warranty. He installed them in a bowling alley and they were literally burnt and even smelled burnt. The problem was caused by the low amplifier power he used. At heavy amplifier clipping the speakers received a continuous DC voltage that alternated as AC. We refused warranty replacement for the speakers so he bought more and a much higher-power amplifier for them. Then the system worked fine.
I burnt a little speaker that was rated at 40W. I had it operating for years from my 70W amplifier. When I removed the burnt 4" "woofer" I noticed that it was stamped 5W. The replacement was also stamped "5W" and also worked well for years.

Here is a good audio site with some reading for you:
http://www.sound.westhost.com/index2.html
http://www.sound.westhost.com/biamp-vs-passive.htm
http://www.sound.westhost.com/bi-amp.htm
http://www.sound.westhost.com/bi-amp2.htm
http://www.sound.westhost.com/clipping.htm
http://www.sound.westhost.com/tweeters.htm
http://www.sound.westhost.com/sp-boxes.htm
http://www.sound.westhost.com/sp-boxes2.htm
http://www.sound.westhost.com/project09.htm

 

[xtp]

Hi Xtp,
10W per channel, per speaker driver isn't enough for a good music playing system. You will be tempted to "turn up the volume" causing the amplifiers to clip, which will destroy the tweeters (and your hearing). Use much more power and don't worry about "blowing-out" the speakers unless they are really cheap. The sounds in music don't play continuously for any amount of time to overpower speakers.

I don't think we're talking about the same kind of watts :-\.
If you look at same types of speakers you'll se that the power stamped on each speaker is different from one manufacturer to another.
For example 10" car subwoofers can be found with a power rating of 180W to 1000W , even if the 1000W subwoofer is less powerful than the 180W subwoofer, but their real power is probably around 40W .
A lot of car audio systems (that include a radio and a CD player) have their power amp built with TDA 1554 IC that has a maximum output power of 4x11W single ended or 2x22W bridged, but all manufacturers write 4x45W or 4x60W on them.
My 8" 8ohm subwoofers are stamped "80W RMS continous power" and "120W RMS peak power". When I tested them in properly designed enclosures with a 2x35W/4ohms or 20W/8ohms power amp they almost reached their mechanical limit below the amp's maximum power .

When I worked for a sound system equipment manufacturer, one dealer brought back some of our cheapest speakers for replacement under warranty. He installed them in a bowling alley and they were literally burnt and even smelled burnt. The problem was caused by the low amplifier power he used. At heavy amplifier clipping the speakers received a continuous DC voltage that alternated as AC. We refused warranty replacement for the speakers so he bought more and a much higher-power amplifier for them. Then the system worked fine.
I burnt a little speaker that was rated at 40W. I had it operating for years from my 70W amplifier. When I removed the burnt 4" "woofer" I noticed that it was stamped 5W. The replacement was also stamped "5W" and also worked well for years.

My amps will have overcurrent protection ;D, so they won't allow any dangerous DC to the speakers.

Here is a good audio site with some reading for you:
http://www.sound.westhost.com/index2.html
http://www.sound.westhost.com/biamp-vs-passive.htm
http://www.sound.westhost.com/bi-amp.htm
http://www.sound.westhost.com/bi-amp2.htm
http://www.sound.westhost.com/clipping.htm
http://www.sound.westhost.com/tweeters.htm
http://www.sound.westhost.com/sp-boxes.htm
http://www.sound.westhost.com/sp-boxes2.htm
http://www.sound.westhost.com/project09.htm

It seems funny but I know the Linkwitz-Riley filters from an old magazine ;D (20 years old...), it wasn't written anywhere that those filters were Linkwitz-Riley filters.
Thanks anyway.

 

[audioguru2]

Hi Xtp,
I was talking about real RMS-rated watts, not the phoney numbers by car radio lies.
1) Car radio amps and speakers are rated momentarily which is called "music power" so the internal supply voltage does't have time to drop to its operating value and so that they don't melt.
2) Phoney power rating numbers rarely mention the horrible distortion caused by overdrive during the rated power.
3) "Peak" or "maximum" power numbers are simply the real power rating doubled, since the peak voltage of a sine-wave is its RMS value times the root of two and so is the current.
4) Car amps rated at 200W are 4 bridged amps that have a real output of 14W each into 4 ohms at just before clipping with a real supply voltage of 13.2V. The TDA1554 rating is the same, its "22W or 44W peak" is really only 14W for each of its two bridged channels.

Your 8" sub-woofer that reaches its mechanical limit with only 20W is either really cheap or is in an enclosure that doesn't damp its motion properly. I have an 18" sub-woofer that weighs a tonne and I've heard (and felt) one being driven very well in an enclosure designed for it with a 1500W (real watts) amp at clipping. It ain't cheap! Mine is still brand new in its box, I've never tried it. I'll sell it at a car SPL competition but I wish its spec's mentioned "peak music power".

When I mentioned DC on fried speakers I was talking about the under-powered amp that was driven well into clipping so its output was square-waves that alternated between its positive DC voltage and its negative DC voltage.

I hope you understand that crossover filters in the links I provided are designed for a flat passband (unlike yours) and no cancelling phase-shift at each crossover frequency. ;D 
 

 

[xtp]

Hello Audioguru,

Your 8" sub-woofer that reaches its mechanical limit with only 20W is either really cheap or is in an enclosure that doesn't damp its motion properly.

I didn't say that my 8" subwoofers reached their mechanical limit and 20W was probably the average power delivered to the speakers and they ain't cheap either. I ment that I wouldn't turn the volume up more, everything in the room was shaking, I probably scared my neighbours too . The enclosure could support 12" subwoofer at full power ;D.

For the subwoofer and the midrange I think I'm going to use something like the 10W amp in the link below but without the bass boost (for each speaker - that means four amps).
http://www.redcircuits.com/Page61.htm

I hope you understand that crossover filters in the links I provided are designed for a flat passband (unlike yours) and no cancelling phase-shift at each crossover frequency. ;D

 

[audioguru2]

I didn't say that my 8" subwoofers reached their mechanical limit at 20W

Sorry, you said something like that: "When I tested them in properly designed enclosures with a 2x35W/4ohms or 20W/8ohms power amp they almost reached their mechanical limit below the amp's maximum power".

The enclosure could support 12" subwoofer at full power.

Then its volume is too big for an 8" subwoofer.

For the subwoofer and the midrange I think I'm going to use something like the 10W amp in the link below but without the bass boost (for each speaker - that means four amps).
http://www.redcircuits.com/Page61.htm

It isn't much power but maybe your room is small.

I found that the filters that I was talking about (and you were saying they won't work) were Butterworth filters.

A Butterworth filter has positive feedback and a flat response. The multiple 6dB RC filters you talked about aren't Butterworth and won't work.

A Butterworth crossover with an even number of RC stages creates an acoustic notch at the crossover frequency when the speakers are wired in-phase. If the speakers are wired out-of-phase they add and create a boost at the crossover frequency. An odd number of RC stages creates a flat response at the crossover frequency. Linkwitz-Riley filters create a flat acoustic response at the crossover frequency regardless of the number of RC stages. ;D

 

[xtp]

Sorry, you said something like that: "When I tested them in properly designed enclosures with a 2x35W/4ohms or 20W/8ohms power amp they almost reached their mechanical limit below the amp's maximum power".

I didn't want to turn the volume up more.

Then its volume is too big for an 8" subwoofer.

No...
I found the size of the enclosure in a book (1980s).

It isn't much power but maybe your room is small.

Yup. 3m/4m/2.4m.

A Butterworth filter has positive feedback and a flat response. The multiple 6dB RC filters you talked about aren't Butterworth and won't work.
A Butterworth crossover with an even number of RC stages creates an acoustic notch at the crossover frequency when the speakers are wired in-phase. If the speakers are wired out-of-phase they add and create a boost at the crossover frequency. An odd number of RC stages creates a flat response at the crossover frequency. Linkwitz-Riley filters create a flat acoustic response at the crossover frequency regardless of the number of RC stages. ;D

Ok...

 

[audioguru2]

Hi Xtp,
I am fixing the JVC two-way 8" speakers my son got from E-bay. Their pretty big enclosures were glued together with hot-melt wax, and were falling apart and buzzing. After picking away the wax, I am gluing them properly with wood-glue and wood blocks.

They have built-in overload protection: a light bulb. It looks like a car's long and narrow dome lightbulb. The very low cold resistance of the lightbulb increases about 10 times when it gets hot, which limits the current to the speaker. My amp is powerful so I will look for light in the dark (through the vent) and hear if the lightbulb causes a delayed compression of loud "beat" sounds.

 

[xtp]

Hi Audioguru,
My enclosures are not glued, I used 4cm long skrews (about 50 of them for each enclosure)