Mostly in the past I've only been bulding 2-ways (and doing some crossovers for 2,5-ways) so I decided skipping the 3-way to try a 4-way design. 😱 The reason for this is not entirely clear 😱 but it felt right at the time so that's what I did.
Since I've gotten good results with aluminum drivers and waveguides I would be using them this time too, complicating the passive crossover a bit (but as it turns out not that much).
Also the entire speaker should be sealed with separate enclosures for the different frequency bands. Material is only single 12mm MDF with internal braces to keep the weight down because these are quite big considering most people hide their miniature plastic speakers behind the couch. 🙂
The bass is handled by a 10" Scan speak 26w-8534g00 8-ohms in a 70L closed cabinet and mounted on the side wall.
Lower midrange an 7" Visaton AL170 8ohms 10L closed.
Upper midrange a 5" Visaton AL130 8ohms 5L closed.
And the tweeter a Seas H1499-06 27TBCD/GB-DXT 6ohm with bult in diffraction waveguide and rear chamber.
So how did they turn out? Pretty good I think. They are the best sounding speakers I've bult by a small margin. What is improved over the earlier 2-ways is the timing and definition in the bass region. Attack and sustain is shorter and single notes are more separated on the bass guitar and drums. A bit more stable imaging and treble and more effortless dynamics.
The crossovers weigh in about 2kgs each and component count is 25, slightly fewer than the well known 4-way NHT 3.3 with 31 elements.
Crossover network schematic
Spliced measurements (looks weird because of recalculation of points)
Only gated measurements in simulation 0, 30 and 60deg off axis
Simulated impedance and measured right/left impedance
In room measurements 50cm distance (100cm from back wall and 80cm to the floor giving the suckout at 170Hz coinciding with the 200cm half wavelength point). No microphone compensation: should be a bit less at 10kHz and a bit more at 20kHz
Finished speaker
Crossover mounted and soldered 🙂
Feel free to ask any questions about the design if you're interested.
As promised some experiences during the build:
Building a 4-way is alot more work, obviously. Especially soldering the crossover, cables to the driver terminals and fitting the parts inside the enclosure.
There is a lot of holes to saw through the internal braces. Sawing round holes without burning the material takes practice and time.
Measurements are more complicated and there is alot more to consider when it comes to the interaction with the room att lower frequencies. Spliceing measurements means you have to do the measurements with the splicing in mind and carefully choosing splicing points and recalculating levels. In the end it comes down to pure judgment calls what is the better representation of reality. Do the measurement-work carefully enough and the result will benefit from it alot reducing the need for any excess iterations of the crossover.
Keeping the impedance high in the mid bass-region is very challenging. With drivers of 8ohms you probably have to settle for a 4 ohms minimum impedance. A nice trick is to pair the series and parallel elements separately rather than to connect highpass and lowpass in cascade for the bandpass filter. This has both effect on the impedance and the frequency response. Things are a lot more dependent when making crossovers closer in frequency to each other.
Gluing all the parts together with all the internal walls in the right places, and making everything air tight is a handful. I ended up with a couple of minor surprises like the cutout of the lower bass speaker ending up towards the front and not the back as first intended. Not so much difference but annoing just the same.
Sealed boxes need a lot more stuffing than I expected. To get the impedance peak in the bass down I had to put in a couple of polyester blankets in addition to the wool/polyester sheets to begin with. Same goes for the mid bass which at first had a resonance at 325Hz visible in the impedance curve. Lots of damping material needed to get that tamed.
What I discovered was that sealed boxes is pretty much the way to go to get the natural definition in the bass region and separation of different tones and a quicker attack/decay. Even when the room adds many times more reverb to the sound at low frequencies what is happening at the sound source matters a lot when it comes to perceived timing and detail. It's easier for the ears to do the math subtracting the room when the sound sources are pulsing out the right vibes into the room to begin with. That is my experience building these speakers.
Since I've gotten good results with aluminum drivers and waveguides I would be using them this time too, complicating the passive crossover a bit (but as it turns out not that much).
Also the entire speaker should be sealed with separate enclosures for the different frequency bands. Material is only single 12mm MDF with internal braces to keep the weight down because these are quite big considering most people hide their miniature plastic speakers behind the couch. 🙂
The bass is handled by a 10" Scan speak 26w-8534g00 8-ohms in a 70L closed cabinet and mounted on the side wall.
Lower midrange an 7" Visaton AL170 8ohms 10L closed.
Upper midrange a 5" Visaton AL130 8ohms 5L closed.
And the tweeter a Seas H1499-06 27TBCD/GB-DXT 6ohm with bult in diffraction waveguide and rear chamber.
So how did they turn out? Pretty good I think. They are the best sounding speakers I've bult by a small margin. What is improved over the earlier 2-ways is the timing and definition in the bass region. Attack and sustain is shorter and single notes are more separated on the bass guitar and drums. A bit more stable imaging and treble and more effortless dynamics.
The crossovers weigh in about 2kgs each and component count is 25, slightly fewer than the well known 4-way NHT 3.3 with 31 elements.
An externally hosted image should be here but it was not working when we last tested it.
Crossover network schematic
An externally hosted image should be here but it was not working when we last tested it.
Spliced measurements (looks weird because of recalculation of points)
An externally hosted image should be here but it was not working when we last tested it.
Only gated measurements in simulation 0, 30 and 60deg off axis
An externally hosted image should be here but it was not working when we last tested it.
Simulated impedance and measured right/left impedance
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
In room measurements 50cm distance (100cm from back wall and 80cm to the floor giving the suckout at 170Hz coinciding with the 200cm half wavelength point). No microphone compensation: should be a bit less at 10kHz and a bit more at 20kHz
An externally hosted image should be here but it was not working when we last tested it.
Finished speaker
An externally hosted image should be here but it was not working when we last tested it.
Crossover mounted and soldered 🙂
Feel free to ask any questions about the design if you're interested.
As promised some experiences during the build:
Building a 4-way is alot more work, obviously. Especially soldering the crossover, cables to the driver terminals and fitting the parts inside the enclosure.
There is a lot of holes to saw through the internal braces. Sawing round holes without burning the material takes practice and time.
Measurements are more complicated and there is alot more to consider when it comes to the interaction with the room att lower frequencies. Spliceing measurements means you have to do the measurements with the splicing in mind and carefully choosing splicing points and recalculating levels. In the end it comes down to pure judgment calls what is the better representation of reality. Do the measurement-work carefully enough and the result will benefit from it alot reducing the need for any excess iterations of the crossover.
Keeping the impedance high in the mid bass-region is very challenging. With drivers of 8ohms you probably have to settle for a 4 ohms minimum impedance. A nice trick is to pair the series and parallel elements separately rather than to connect highpass and lowpass in cascade for the bandpass filter. This has both effect on the impedance and the frequency response. Things are a lot more dependent when making crossovers closer in frequency to each other.
Gluing all the parts together with all the internal walls in the right places, and making everything air tight is a handful. I ended up with a couple of minor surprises like the cutout of the lower bass speaker ending up towards the front and not the back as first intended. Not so much difference but annoing just the same.
Sealed boxes need a lot more stuffing than I expected. To get the impedance peak in the bass down I had to put in a couple of polyester blankets in addition to the wool/polyester sheets to begin with. Same goes for the mid bass which at first had a resonance at 325Hz visible in the impedance curve. Lots of damping material needed to get that tamed.
What I discovered was that sealed boxes is pretty much the way to go to get the natural definition in the bass region and separation of different tones and a quicker attack/decay. Even when the room adds many times more reverb to the sound at low frequencies what is happening at the sound source matters a lot when it comes to perceived timing and detail. It's easier for the ears to do the math subtracting the room when the sound sources are pulsing out the right vibes into the room to begin with. That is my experience building these speakers.
Last edited:
Why? Not having enough challenge in building a seamless 2 way?
Like too many gears in a transmission it's the constant shifting that kills you in the end
Like too many gears in a transmission it's the constant shifting that kills you in the end
Open them in another window and they'll appear.
Why 4-way? Well, multiple drivers sharing the frequencies reduces the distortion.
Using a sealed enclosure for the bass means you'll need a larger driver to acheive the same range/sensitivity/SPL. A larger driver will beam earlier at higher frequencies and make the baffle larger. So letting a midbass take the range from 150Hz-700Hz and a upper midrange between 700-3000Hz will greatly reduce distortion. Also enabling the option of a smaller waveguide for the tweeter with better top end behaviour and still matching the dispersion of the midrange.
Either using 5" or 7" wouldn't do both.
I'm building a 4 ways too (2x10"+6.5"+tweeter+piezo for each loudspeaker) but I chose a different approach:
- woofers (2 Dayton DC-250 in a reflex box) are crossed with the midrange (around 280 Hz) through an active crossover so that I can avoid inductors. They will be driven by a 200 W amp (P101 by Rod Elliott)
- midrange-->tweeter and piezo are filtered through a passive crossover and driven by a modified Technics SUV-5 (the preamp section has been disabled so it works just as a "pure" power amp).
It is still under developing.
Regards,
Roberto
- woofers (2 Dayton DC-250 in a reflex box) are crossed with the midrange (around 280 Hz) through an active crossover so that I can avoid inductors. They will be driven by a 200 W amp (P101 by Rod Elliott)
- midrange-->tweeter and piezo are filtered through a passive crossover and driven by a modified Technics SUV-5 (the preamp section has been disabled so it works just as a "pure" power amp).
It is still under developing.
Regards,
Roberto
I see this:
Forbidden
You don't have permission to access /Martin/Filter4way.JPG on this server.
With the picture in a separate page, F5 doesn't do anything constructive nor does pressing the reload button but pressing return in the address box works. The link to the crossover seems to work. It looks like the html in Martin's post is asking the diyaudio server here or the one in Sweden to do something it doesn't want to do.
Martin, how did you create the links to your images?
Sorry about the pictures guys! Copying the url and pasting it into a new window worked on windows/opera.
Good question! 🙂 Simply because i knew that driver had very good midrange and matched the tweeters dispersion and wave front center*.
*Wave front center or wave front radius is the point in space which sound appears to arrive from with the same phase shift. Not to be confused with acoustic center which is the point behind the driver which the sound appears to arrive from in time (0 msec when measuring an impulse). Wave front center can in some cases even be in front of the membrane. It's a bit complicated but the main thing is to match the wavefront of the tweeter and midrange so that they are always in phase along the horizontal plane.
Anyway these two drivers have the same wave front center mounted on a plane baffle so there is no growing suck out off axis.
Good question! 🙂 Simply because i knew that driver had very good midrange and matched the tweeters dispersion and wave front center*.
*Wave front center or wave front radius is the point in space which sound appears to arrive from with the same phase shift. Not to be confused with acoustic center which is the point behind the driver which the sound appears to arrive from in time (0 msec when measuring an impulse). Wave front center can in some cases even be in front of the membrane. It's a bit complicated but the main thing is to match the wavefront of the tweeter and midrange so that they are always in phase along the horizontal plane.
Anyway these two drivers have the same wave front center mounted on a plane baffle so there is no growing suck out off axis.
Nice to see someone else using speaker workshop for crossover simulation! Very good correlation of the simmed impedance to the actual 🙂
Passive crossover at 150Hz too! and the inductor is air cored 😱
DCR of that 8mH must be pretty high?
Tony.
Passive crossover at 150Hz too! and the inductor is air cored 😱
DCR of that 8mH must be pretty high?
Tony.
Pressing enter works after opening them in a new tab. It's the server that only approves linking from the mother website.
In my opinion, though I've never used that particular driver, dome mids and especially soft dome types have properties that aren't suitable integrating them to a tweeter. Wobbling vc, no real control over the larger dome with break up modes that have to be heavily coated and the similar off axis roll off characteristics of a dome tweeter without waveguide and at lower frequency making it more audible. The only real benefit over a cone the way I see it is the naturally limited bandwidth helping the high-pass in the crossover section.
🙂
The cone midrange is easier to match to a tweeter with the help of a waveguide because the waveguide/tweeter have similar dispersion above the crossover point (shape is similar but without the "cone" moving). So that the response can be linear but tilting slightly off axis all the way up to 20kHz. The dome midrange will have too much energy in the middle of it's band and too little at the top.
Thanks! Speakerworkshop is powerful but tricky to work with.
DCR of the coil is about 1ohm and accounted for in the design. Weighs 1.3 kg each. 😀
In my opinion, though I've never used that particular driver, dome mids and especially soft dome types have properties that aren't suitable integrating them to a tweeter. Wobbling vc, no real control over the larger dome with break up modes that have to be heavily coated and the similar off axis roll off characteristics of a dome tweeter without waveguide and at lower frequency making it more audible. The only real benefit over a cone the way I see it is the naturally limited bandwidth helping the high-pass in the crossover section.
🙂The cone midrange is easier to match to a tweeter with the help of a waveguide because the waveguide/tweeter have similar dispersion above the crossover point (shape is similar but without the "cone" moving). So that the response can be linear but tilting slightly off axis all the way up to 20kHz. The dome midrange will have too much energy in the middle of it's band and too little at the top.
Thanks! Speakerworkshop is powerful but tricky to work with.
DCR of the coil is about 1ohm and accounted for in the design. Weighs 1.3 kg each. 😀
Oh, about that. I later discovered that the right speaker had a leakage between the compartments explaining the difference in the bass, that's sealed now.
I measure the caps and use small values of MKT-caps (not showing in picture) in parallel to match the values. I think that should be beneficial for the imaging matching the crossovers so they behave electrically the same.
Nice rationalizing, Martin! I still think that eg. Visaton Ti100 (4") would have been a better choice. Is the tweeter 2-3dB too hot?
Last edited:
You are probably right, I was safe:ing a bit going with drivers known to work together. Yes the tweeter is the only thing I had to attenuate with a physical resistor. The cone-drivers all have their final level-adjustments built in in the dcr of the series inductors.
This is essentially a one shot crossover design, only adjustment made is the shunt series resonance LC of the midrange low-pass improving the phase matching.
edit: If you were referring to the room-measurements they're not compensated for the microphone response which have a 2dB bump around 7-10KHz and a roll-off of 3dB towards 20kHz.
Last edited:
- Status
- Not open for further replies.