3-way active, time aligned, constrained layer construction

Status
Not open for further replies.
definitely special 🙂, but some studies indicate the rounded chamfer even better than straight chamfer.

i cannot give you reference for that unfortunately.
The same Olson paper showed that the sphere was the best of all but impractical. The chamfered rectangular prism was shown to be very close to the sphere and practically feasible.
 
Hi Bon, greetings from Brisbane. Love the cabinets.
Interested in your understanding of the x-overs and phase. Can you explain your thinking on the type; i.e. why Bessel?
Is this to minimize phase issues within the power amps or to achieve other goals?
Did you arrive at these decisions subjectively or from a theoretical basis?
I know some people are getting good results by pushing the mid/tweeter cross over point up past 6kHz, presumably for phase related questions. Any thoughts?

Cheers, Jonathan
 
Interested in your understanding of the x-overs and phase. Can you explain your thinking on the type; i.e. why Bessel?
Is this to minimize phase issues within the power amps or to achieve other goals?
Did you arrive at these decisions subjectively or from a theoretical basis?
Hi Jonathan. The standard filter topology is Butterworth, L-R is just a cascaded pair of Butterworth filters. Important considerations for choosing filters include 1) No amplitude ripples in the passband, 2) Utimate amplitude roll-off rate, 3) Roll off rate through the x-over region, 4) Phase behaviour through the x-over frequency transition region. The Butterworth, L-R, Bessel all behave the same way for 1) and 2). They differ markedly in 3) and 4). It comes down to a matter of priorities to trade off between 3) and 4). The L-R has faster initial amplitude roll-off rate but the phase goes through rapid non-linear change through the x-over frequency region. By comparison the Bessels initial roll-off rate is less steep but the phase is much more gradual and linear through the x-over region. Since I time and phase align my mid-tweeter and bass-mid at the selected x-over, it is convenient to choose the filter topology with the best phase behaviour through x-over i.e. Bessel. That's the theory. The practice has shown that it satifies my expectations. What I notice is the precision of the soundstage, especially for extreme R/L positions. Some recording producers really can play with your mind in this regards. Some writers have expressed the opinion that extreme outside-the-speaker imaging is a result of accidental room reflections. I totally disagree. If the producer/engineer knows his business, it's a dead easy. For example, the Eurythmics "I love you like a ball and chain" LP, has been in my collection for 25 years. It was only after, understanding and implementing time and phase alignment that I heard after the solo, the synthesizer swirling was a precise ellipse passing around both speakers as the foci. Dave Stewart was playing a joke, because naturally a ball and chain gets swung around your head. Sorry to go on a bit, but you did ask.



I know some people are getting good results by pushing the mid/tweeter cross over point up past 6kHz, presumably for phase related questions. Any thoughts?
The problem is getting a mid-range that is comfortable up there. It is a juggling act. All cone 5.5-6.5 inch mid ranges start to struggle above 2 kHz. The directivity narrows, break up occurs too close to x-over. The best are pretty good though. Another issue is that aligning the drivers at higher x-over frequencies becomes more difficult. Small changes in distance can have dramatic effects on alignment. For example, small head movements.
 
Thanks Bon for the explanation on why you chose Bessel. I am brand new to an active setup and have yet to tackle phase and time alignment (still need to understand the software better, and understand how to properly change it). With so many options it is hard to know when to use what.
 
I'm big on active too.. 009.jpg

012.JPG
Notice my kids old skateboard wheels? Good for the circulation. "Dad, Where are my wheels?" Huh?Wheels? What wheels?
Computer based system consisting of;
Computer
Squeezebox touch
Behringer DEQ 2496, Behringer SRC 2496, and two Behringer DCX 2496's(One per channel).
Signal stays in the digital domain until the DCX outputs..
Conrad Johnson MV 50 for planar magnetic mid-tweet drivers
Three Behringer A500's for woofers
500w Bash amp
Dayton cabs
Dayton reference 12" subwoofer
Peerless XXLS's
Peerless HDS Exclusives
BG Neo 10's planar magnetics in sealed cabinets
Xover points(4th order); 40 80 160 320 (Planar magnetic dsp'd flat from 320hz to 20khz).
 
Last edited:
I'm big on active too.

Behringer DEQ 2496, Behringer SRC 2496, and two Behringer DCX 2496's(One per channel).
Signal stays in the digital domain until the DCX outputs..
Three Behringer A500's for woofers
I've got an A500 and three DCX2496. I hot-rodded the A500 power supply by piggy-backing an extra 20,000 uF capacitance per channel onto the stock (a puny 8000 uF or so). I also ungraded the electrolytics where they wre accesible. Can't do much more because of the piddly SMD used. They are good value. Why do you need the SRC 2496?
 
I've got an A500 and three DCX2496. I hot-rodded the A500 power supply by piggy-backing an extra 20,000 uF capacitance per channel onto the stock (a puny 8000 uF or so). I also ungraded the electrolytics where they wre accesible. Can't do much more because of the piddly SMD used. They are good value. Why do you need the SRC 2496?
It does a couple of really neat things(It actually does much more than that). First, it asynchronously re-clocks digital data from the computer. Second, it upsamples digital data to 24/96. I listen mainly to MOG(Only available in the U. S., I think). This is a digital radio service that has 16 million songs in album form transmitted at 320kbps. Upsampling makes it sound much more liquid and listenable. The SRC is basically a poor man's DCS Purcell or Debussy. I got mine for $175.00. Here is a good article..

hifi-advice.com - Behringer Ultramatch upsampler review
 
Last edited:
It does a couple of really neat things(It actually does much more than that). First, it asynchronously re-clocks digital data from the computer. Second, it upsamples digital data to 24/96. I listen mainly to MOG(Only available in the U. S., I think). This is a digital radio service that has 16 million songs in album form transmitted at 320kbps. Upsampling makes it sound much more liquid and listenable. The SRC is basically a poor man's DCS Purcell or Debussy. I got mine for $175.00. Here is a good article..

hifi-advice.com - Behringer Ultramatch upsampler review
Another interesting article. This one on the A500 amp..
HTTP Error 403
 
Don't for a moment think I am putting Bon's efforts down, but really I think anyone can produce the same kind of results... I am not particularly gifted but I always produce excellent results for one very simple reason... I have copious amounts of patience, and provided you think out every detail, to the most minuteaspect and you don't allow yourself to be rushed, there is no reason to believe you couldn't do the same.

But most people want results yesterday and that is the difference between great and ok.
 
I agree. Anyone CAN do as good a job as Bon did. As long as they are extremely talented! There are so many variables involved with a cabinet design like this, it's not just about good woodworking skills. You need a deep understanding of the physics involved, and a large amount of time invested in this wonderful hobby before you can pull off a design like that. Constrained layer damping? You need to really, really, appreciate things like that in order to succeed in making them work..
 
Last edited:
I'm not necessarily saying everyone could think it up either, but now that the idea is out there, Bon has provided a good template for anyone with the dedication and persistance to follow 🙂

Most people are content to throw six MDF sides up and call it a cabinet (not quite that simplistic but you get my drift). The level of thought that has gone into every layer of this project is where I see the difference, not so much the cabinet work.

Again the short cutter gets out the circular saw and cuts a fairly straight line and file / sands the difference, the dedicated person builds a fence for their workbench so they know the line is perfectly straight.

I still think preperation and just taking your time can make an awful lot of quality improvement, years of experience or not.
 
I'm not necessarily saying everyone could think it up either, but now that the idea is out there, Bon has provided a good template for anyone with the dedication and persistance to follow 🙂

Most people are content to throw six MDF sides up and call it a cabinet (not quite that simplistic but you get my drift). The level of thought that has gone into every layer of this project is where I see the difference, not so much the cabinet work.

Again the short cutter gets out the circular saw and cuts a fairly straight line and file / sands the difference, the dedicated person builds a fence for their workbench so they know the line is perfectly straight.

I still think preperation and just taking your time can make an awful lot of quality improvement, years of experience or not.
I agree. Way more planning and forethought than actual physical work.
 
Thanks Bon re; cross over type explanation.
I am thinking along similar lines (but much less money and FAR less mass!!!) and starting to think seriously about phase.
At the moment I am thinking 4th order but a "hybrid" 2nd order low level active and 2nd order high level passive. This will achieve some "optimised compromises" (if you'll excuse the expression) for a clutch of single rail Class A JLH's..
I was interested to see comments about filter slope altering the step response of the power amps......Bessel obviously being least intrusive in this respect.
Also interested in the process of aligning the drivers.
Saw a Danish site that was speaking seriously about single millimeters being important in driver spacing! Is that your experience?

Jonathan
 
At the moment I am thinking 4th order but a "hybrid" 2nd order low level active and 2nd order high level passive. This will achieve some "optimised compromises" (if you'll excuse the expression) for a clutch of single rail Class A JLH's.
I have settled on 4th order Bessel for mid-tweeter (24 dB/octave ultimate slope) and 4th order L-R for bass-mid. This might seem in conflict with my previous explanation of the Bessels superiority until you consider that the chosen bass-mid x-over frequency is 150 Hz where phase effects are much less noticeable. In fact at this frequency the steeper initial filter slopes are useful for keeping high level bass away from the mid range. The 150 Hz is not a random choice BTW. Research has shown that below 150 Hz, the human ear/brain is unable to locate the position of an acoustic source. I don't get the same buzz out of building electronics so I am happy to use what is affordable, reliable, and sounds good. In fact, I have found that when the amps are not asked to deal with a complex passive x-over, they are more than capable of driving the naked speakers.
I was interested to see comments about filter slope altering the step response of the power amps......
Jonathan
This must be for passive x-overs. I can't see how an active x-over that is before the power amp can modify the power amp step response. Whereas a nasty passive can make even a good power amp freak out.

Also interested in the process of aligning the drivers.
Saw a Danish site that was speaking seriously about single millimeters being important in driver spacing! Is that your experience?
Jonathan
Try and find a copy of a Technical Note by Rane Acoustics. It is intended for pro audio but it gives a really clear set up procedure for phase alignment of drivers at the listening position. It is a good starting place. My set up procedure is a modified version but the principle is the same. In short, for each enclosure, play a sine wave at the chosen x-over frequency, set the mid and treble levels separately to be as close to identical as measured with a measurement mic at the listening position. With the sine applied to both drivers, reverse the phase of one driver and adjust the phase and delay until the mic reading bottoms out. I can easily get a 30-40 dB null. With a little practice this can be improved.
Then put the drivers back in phase. They should now be time aligned. With only two driver active the level should rise 3 dB over one driver. Reversing the drivers and repeating the level should drop 30-40 dB. Play pink noise and observe the frequency response. It should be smooth and flat through the x-over region. With one driver reversed, there should be a deep notch in the frequency response at the x-over frequency. In practice, it is at least a couple of hours work. Wear ear protection.
Yes, the DCX2496 allows delay adjustment increments of 1 mS (2 mm). I find the best null require this level of adjustment.
 
I agree. Way more planning and forethought than actual physical work.
Hi. Thanks for all the nice comments. Look, I am a competent woodworker but I really needed to ramp up facilities, materials and skills for this project. As Silent Screamer has stated, it is mainly a matter of doggedness, careful thought, planning and practice. For every crucial step, I stepped back and slept on the procedure, sometimes for a couple of days. I had plenty of scrap pieces to practice special saw cuts and routing. I didn't proceed until I had practiced and understood how it could go wrong. I forgot to mention the months refining the design. A few things only came to fruition during the build. A few moments of ... "what was I thinking?". So it was not a smooth linear process but a lumpy, jerky one. I did have a bit of previous experience. In fact I have made it hard for myself because my audio buddies love my current speakers, a decade old effort. They mostly ask "why are you even bothering?".
 
Status
Not open for further replies.