@krivium: Thanks for the suggestion. I will definitely try out the calculator.
Speaker placement-wise, they will eventually move to stands like this
With that the drivers are about 35cm away from back wall. The CPU will move down to the floor on a trolley.
With this the measurement comparison between yesterday (V1) and today (V2) looks like this
I need to listen more to decide which sounds better 🙂
And then move to the next placement iteration.
Speaker placement-wise, they will eventually move to stands like this
With that the drivers are about 35cm away from back wall. The CPU will move down to the floor on a trolley.
With this the measurement comparison between yesterday (V1) and today (V2) looks like this
I need to listen more to decide which sounds better 🙂
And then move to the next placement iteration.
Hmm. I see a case where i would try 'shave' everything 'protubering' above 60db spl to have a straight line from 45hz to 10khz... and then implement a straight downwarding slope from 1khz ( -6db circa 20khz) for a try at least.
Oh no! I start to have reflex/behave like a room corection software! Lol
Oh no! I start to have reflex/behave like a room corection software! Lol
If you like them, just go crazy: implement a sub/low way 'a la D&D' as a stand ( a pair of sub in sealed cab (height making them at correct height and using LT transform to adapt to volume ) boundary loaded by the wall), make the bandwith up to 100hz and transform the whole damn thing into a 4way... xover circa 100hz, 400hz, 2,5khz. 🙂
It would be able to produce really high spl and make for a nice tower able to adapt to a lot of rooms imho.
Could be a nice option if you decide to make the design accessible to others as a more 'general speaker' than your own target ( with same advantage as D&D 8c).
@krivium: I like the looks of the Kii3 bass towers. So that kind of cabinet/stand might look nice with this one I think.. 🙂
But my wife will kill me if I buy more drivers immediately.. 😛 😀
So, I will at least hold off on driver/speaker purchases for a while.. 😀
At the moment, I am just trying to use all the drivers I have been having in storage for a while 🙂
I will share all the CAD files of this project soon for others to adapt to their drivers 🙂
But my wife will kill me if I buy more drivers immediately.. 😛 😀
So, I will at least hold off on driver/speaker purchases for a while.. 😀
At the moment, I am just trying to use all the drivers I have been having in storage for a while 🙂
I will share all the CAD files of this project soon for others to adapt to their drivers 🙂
Vineeth i wasn't thinkin about an implementation of active cardioid but your attraction about Kii might be only on the look i don't know?
Anyway for what i had in mind you won't see the drivers**, they should face the front wall ( from listener pov) with a 5 to 10 cm gap at max.
The point is to really make the wall as a giant reflector. Given the freq involved and size of drivers you'll end up with hemispherical* directivity pattern.
The delay introduced ( because the drivers are located on the rear face + gap needed to the wall) can easily be compensated through delay and D&D prooved merging this directivity pattern to cardioid can be successful.
Martijn is a very, very clever designer and there is a lot of example of this in the 8C, they are well worth a bit of retro engineering or at least studying the implementation of solutions choosen.
* maybe not hemispherical but in the spirit anyway: as a stand you could locate one of the driver close to floor and that way this one could see it's own ouput boosted by +6db ( cause two boundaries involved) and as such it's own directivity would not be hemispherical anymore as it would involve only one boundary to reach such a pattern... but if you use another one closer to the loudspeaker this one would be more hemispherical as less loaded by the floor... you got the idea. Of course if you make a tower of multiple drivers things get more complicated ( in BXT kii's spirit) but i'm not sure it would be needed anyway ( it would rize other concerns as volume required for multiple drivers as well as low freq extension ( in comparison to less but bigger diameter drivers).
** and as they are hidden you could use 'car subs' which can be ugly looking but with parameters fullfiling design requirement in a better way ( parameters optimised for sealed in low volume and high xmax which would make LT requirements less demanding overall) and much lower price to invest. Anyway this is just words in the air and i'm sure you already have your own idea about this.
Anyway for what i had in mind you won't see the drivers**, they should face the front wall ( from listener pov) with a 5 to 10 cm gap at max.
The point is to really make the wall as a giant reflector. Given the freq involved and size of drivers you'll end up with hemispherical* directivity pattern.
The delay introduced ( because the drivers are located on the rear face + gap needed to the wall) can easily be compensated through delay and D&D prooved merging this directivity pattern to cardioid can be successful.
Martijn is a very, very clever designer and there is a lot of example of this in the 8C, they are well worth a bit of retro engineering or at least studying the implementation of solutions choosen.
* maybe not hemispherical but in the spirit anyway: as a stand you could locate one of the driver close to floor and that way this one could see it's own ouput boosted by +6db ( cause two boundaries involved) and as such it's own directivity would not be hemispherical anymore as it would involve only one boundary to reach such a pattern... but if you use another one closer to the loudspeaker this one would be more hemispherical as less loaded by the floor... you got the idea. Of course if you make a tower of multiple drivers things get more complicated ( in BXT kii's spirit) but i'm not sure it would be needed anyway ( it would rize other concerns as volume required for multiple drivers as well as low freq extension ( in comparison to less but bigger diameter drivers).
** and as they are hidden you could use 'car subs' which can be ugly looking but with parameters fullfiling design requirement in a better way ( parameters optimised for sealed in low volume and high xmax which would make LT requirements less demanding overall) and much lower price to invest. Anyway this is just words in the air and i'm sure you already have your own idea about this.
I calibrated SPL on my measurement mic using a minidsp UMIK-1+SPL meter in REW. Here is the result with the current crossover in place (It is 86dB SPL at 1m at 1kHz)
This itself was uncomfortably loud for me and others at home and for the listening situation of this speaker, which is mid to nearfield.
So I won't take a measurement at any more louder SPL 🙂
This itself was uncomfortably loud for me and others at home and for the listening situation of this speaker, which is mid to nearfield.
So I won't take a measurement at any more louder SPL 🙂
So great, we are all 101% clear that everyone will benefit from a teated room, including me and including @vineethkumar01 . Now back to the more interesting parts of this thread...
@vineethkumar01 - I am a little surprised that the cardiod behavior of the coax mid goes as low as it does. In your opinion, how important is it that the melamine foam slows the speed of sound through the 55mm thickness in addition to absorption?
j.
@hifijim: My knowledge about the whole passive cardioid pattern formation is limited. All my experience about this comes from ATH based BEM sims that we had done in the past and the experiments I did at the time. That is how I arrived at this 50ish mm thickness. Just to post some graphs from the experiments,
With no damping, this is what we got
Response is more dipole due to the cancellation from backwave
Slot using 25mm thick melamine foam. Note that 25mm thick area at the back is open before the backwaves hit the damping material. So some amount of dipole pattern is sort of guaranteed
And that is what we see in the measurements as well. But now we have cardioidish pattern above 600Hz.
With 50mm thick damping, the following is what we see.
I think we got reasonably good cardioidish behaviour now. Looking at the group delay plot above, we see constant ish group delay till around 3kHz
I think it tells us something. But I need to think about it more to wrap my head around it.
One thing I am confident about is without that relative shift in the phase of the backwave, the proper cancellation for this polar pattern will not happen. So it is important to have that melamine foam layer create that phase shift. Beyond this, I think others with more knowledge about the whole matter can tell us..
Requesting @tmuikku, @fluid
With no damping, this is what we got
Response is more dipole due to the cancellation from backwave
Slot using 25mm thick melamine foam. Note that 25mm thick area at the back is open before the backwaves hit the damping material. So some amount of dipole pattern is sort of guaranteed
And that is what we see in the measurements as well. But now we have cardioidish pattern above 600Hz.
With 50mm thick damping, the following is what we see.
I think we got reasonably good cardioidish behaviour now. Looking at the group delay plot above, we see constant ish group delay till around 3kHz
I think it tells us something. But I need to think about it more to wrap my head around it.
One thing I am confident about is without that relative shift in the phase of the backwave, the proper cancellation for this polar pattern will not happen. So it is important to have that melamine foam layer create that phase shift. Beyond this, I think others with more knowledge about the whole matter can tell us..
Requesting @tmuikku, @fluid
There are studies that indicate the speed of sound in absorbing material can drop to about 100m/sec , so about 1/3. I do not find a study specific with melamine foam, nor one that sort of correlates with the slot approach. Perhaps studies for anechoic room application of melamine foam?
About this,I calibrated SPL on my measurement mic using a minidsp UMIK-1+SPL meter in REW. Here is the result with the current crossover in place (It is 86dB SPL at 1m at 1kHz)
This itself was uncomfortably loud for me and others at home and for the listening situation of this speaker, which is mid to nearfield.
So I won't take a measurement at any more louder SPL 🙂
I had found that two of the output channels connected to the woofer driver on my audio interface have gone bad.. 😀
It has constant 2% THD across full spectrum. This might definitely have some effect on that 300Hz bump & beyond.
Need to get a better interface/DSP solution first to take this kind of measurement...
This is a great result. Really good stuff. Amazing and astonishing that it was so °simple“ to achieve.
I assume you filled the whole volume in the "slit" around the driver basket and magnet with basotect?
It looks like filling the gap with basotect fixed the 1.4 kHz directivity (and FR) anomaly nearly completely. Maybe a little bit more damping (extending to the outside or compressed a little bit) could be even better as there is still a dip in power response in this region. Did you think about that?
@olieb: Actually I planned to have a bit more larger radius ring to hold the coaxial driver with more rounding of the edges that begin right at the driver frame. This would have allowed me to have even more basotect into the space behind the driver.
However, due to some last minute changes in the cabinet manufacturing, I couldn't get it and hence went with the current ring size... Due to aesthetic reasons, I just stopped with the current amount of basotect..
As you said, more might have given a better directivity curve I think
However, due to some last minute changes in the cabinet manufacturing, I couldn't get it and hence went with the current ring size... Due to aesthetic reasons, I just stopped with the current amount of basotect..
As you said, more might have given a better directivity curve I think
I understand, at some point decisions have to be made and compromises have to be taken.
It would still be interesting to see what happens to the directivity when you place a layer of the felt stuff you showed around the basotect.
I have a pair of Kef coax from R300 laying around. Maybe I can get around to try your geometry with these, but I have no place to measure anything meaningful I am afraid. And then there is the time problem ;(
It would still be interesting to see what happens to the directivity when you place a layer of the felt stuff you showed around the basotect.
I have a pair of Kef coax from R300 laying around. Maybe I can get around to try your geometry with these, but I have no place to measure anything meaningful I am afraid. And then there is the time problem ;(
It is a pretty reasonable directivity response considering the simplicity of the construction. Significant group delay changes will be from excess phase, crossover or non minimum phase behaviour. There is an acoustic filter that is causing a steeper slope at the same place there is a change in group delay. It is not made easier to see what the phase is doing, because the measurement still includes time of flight fixed delay.
To get perfect cardioid behaviour the leak from the side has to coincide with the time it takes for the sound from the front to makes it's way around to where the sound is leaking out. Passively this is hard to do over a wide bandwidth. A cardioid pattern can be created by combining a dipole with a monopole and that can be happening in some of the measurements up above with little to no damping. As the coax becomes more forward radiating less sound is directed to the sides so there is less to cancel, you can see where this is starting to happen when the directivity begins to rise.
You can also get some idea of what is happening from where the nulls are positioned. At 50Hz there is a pair of nulls quite close to 90 degrees, there the pattern will be quite dipolar. When the nulls are in the 90 to 160 degree range you begin to enter some sort of cardioid, like super or hyper. The DI is a bit lumpy because the cancellation is not completely even with frequency.
A hybrid passive+active crossover for 4 output channel DSP (instead of current 6 channels): crossover V1
Trying to refine this more.. 🙂
Trying to refine this more.. 🙂