High all!
I am testing a lot this here:
and this here:
the first idea to use this speaker chassis came from here:
Troels Gravensen The Loudspeaker II
I have made a lot of listening tests with acoustical LR2 LR4 and LR8 filters and stay now with LR8 - a little bit like what the guys from Gauder Akustik in Germany is doing:
Gauder Acoustic English Web Site under construction...
But of course not with high order passive crossovers, very difficult to develop, with miniDSP and a lot of listening tests for hobby diy audio doable.
Crossover frequencies:
SUB - bass mid woofer 150 Hz
bass mid - mid high horn 1200 Hz
mid high - super tweeter 7000 Hz
I am testing a lot this here:
and this here:
the first idea to use this speaker chassis came from here:
Troels Gravensen The Loudspeaker II
I have made a lot of listening tests with acoustical LR2 LR4 and LR8 filters and stay now with LR8 - a little bit like what the guys from Gauder Akustik in Germany is doing:
Gauder Acoustic English Web Site under construction...
But of course not with high order passive crossovers, very difficult to develop, with miniDSP and a lot of listening tests for hobby diy audio doable.
Crossover frequencies:
SUB - bass mid woofer 150 Hz
bass mid - mid high horn 1200 Hz
mid high - super tweeter 7000 Hz
Last edited by a moderator:
Just wonder if you are aware of peaking group delay problems with filter of order higher than 4, e.g. LR4. This arises because the filter has high q poles located near the cutoff frequency. The higher the order the higher the Q value must be. The high Q poles cause rapid phase shifts, which are the same as peaking group delay. Only FIR filters can be design without the phase/delay issues, e.g. to have linear phase (which is the same as constant group delay).
On the other hand, the output of any IIR type filter has a non-flat delay versus frequency and this is a source of waveform distortion, if you care about that sort of thing. When the difference between the nominal passband delay and the peak delay is large you can get effects similar to ringing in the time domain, and this can be audible for some frequencies (e.g. near 1kHz).
Further reading:
https://www.linkwitzlab.com/frontiers.htm#F
On the other hand, the output of any IIR type filter has a non-flat delay versus frequency and this is a source of waveform distortion, if you care about that sort of thing. When the difference between the nominal passband delay and the peak delay is large you can get effects similar to ringing in the time domain, and this can be audible for some frequencies (e.g. near 1kHz).
Further reading:
https://www.linkwitzlab.com/frontiers.htm#F
As far as i understand it depends very much from the type of speakers used.
My assumption is that with the horn types i am using the pros for steep filters are more than the cons.
With the speaker setup below LR 8 did not make it, with the KEF Coax Mid High i was undecided between LR 2 and LR 4.
This is an old picture with partly passive crossovers before the purchase of the miniDSPs.
The thing is that with the KEF Coax LR 2 or LR 4 make the listening tests and not LR 8.
My assumption is that with the horn types i am using the pros for steep filters are more than the cons.
With the speaker setup below LR 8 did not make it, with the KEF Coax Mid High i was undecided between LR 2 and LR 4.
This is an old picture with partly passive crossovers before the purchase of the miniDSPs.
The thing is that with the KEF Coax LR 2 or LR 4 make the listening tests and not LR 8.
Attachments
Last edited by a moderator:
The placement of the drivers is going to cause some response problems. Do not put speakers so close to the side walls. This seems like a pretty narrow room with blank, reflective walls - you might start there, with some wall treatment and rearrangement of the speakers. Also (not in picture) it helps if the wall behind you (the listener) is acoustically dead. These changes will likely open up the soundstage in this space a bit. Then the loudspeaker should be re-positioned - there seems to be several possibilities. Your could rearrange the middle parts (with the black rectange, below the black components) to the outside and put the speakers in their place, next to the subs, You could also place the subs at the walls (assuming there is no neighbor on the other side) and move the rest inwards, or place the subs below the other driver to lift them up a bit (the 3-way seems a bit short, can't quite tell) with the stack moved inwards away from the side walls. Since the drivers are small, they will radiate in all directions (not just in a straight line) up to relatively high frequencies and even if the speaker is away from the wall and toed in, you will get a reflection. A diffuser and possibly a little adsorption where the reflection would occur is really needed. You can find the spot using a mirror placed on the wall and where you can see the speaker in the mirror from your listening chair will be where the reflection source will be strongest and where you need to locate some treatment.
Regarding filter type - it doesn't matte the type of driver/speaker, high order analog or IIR digital filters will have peaking group delay that will be a detriment to the sound. Whether you can "hear" it or notice it is up to you and depends on how well the rest of your system is set up (e.g. other problems could dominate).
Are you making these comparisons only using your ears or also using measurements or some other less subjective means? If by ear, how can you possibly know what is going on? Apart from the filter "shape" there are also phase issues and driver integration around each crossover point that is very important and the ear is just not a good tool for this stage of crossover development. With a multiway system it is nearly impossible to get it right if you are just twiddling between a few different crossover options and frequencies using your miniDSPs. At one time many years ago I was trying to do that very same thing using a Behringer DCX2496 until I realized it was futile and started getting into proper crossover design.
Unfortunately the living room itself is so far ready to use as it is, my wife is living there too. I am happy so far that i am allowed to use this side entirely for the large open baffle sub with the two 15 inch drive units and the smaller satellite cabinets for the stereo setup. One day or another there will be another room available under the roof, but this will be another story for a large home theatre.
I would put your subs under your mains. Much easier to deal with the phase and with the high crossover, will blend easier. You might even be able to bring it up a bit more.
Personally, I try not to exceed LR2 acoustic, which often is lower order electrical.
Personally, I try not to exceed LR2 acoustic, which often is lower order electrical.
Personally I never use less than LR4, but also never more than that either. I use DSP exclusively and don't fret about "acoustic" slopes like many passive crossover design folks who are hoping to use as few components as possible/necessary. As long as the phase lines up nicely through the crossover region I am feeling satisfied and 4th order gives much better passband control and out of band rejection. LR2 is just wayyyyyy to shallow in most cases, for multiway speakers more than 2 way. IMO of course. YMMV.
Very interesting setup,
I use RME soundcard with convolver vst and rephase or isotope ozone, maybe this will interest you...
You have more computing power and more possibility and flexibility. Of course it has it cons (pc)...
I use RME soundcard with convolver vst and rephase or isotope ozone, maybe this will interest you...
You have more computing power and more possibility and flexibility. Of course it has it cons (pc)...
Last edited:
Here the list for the driver units used:
- open baffle subwoofer with Sonido SCW400 woofer (not sold anymore unfortunately)
- kick bass mid range woofer custom Sonido SWR 250 with no whizzer cone in closed box
- mid range high horn and compression driver Eighteensound XT1086 + NSD1095N
- super tweeter Visaton TL16H
Hi forum members!
In Germany one of the most popular DIY speaker shops has adopted the steep crossover design, here:
Thalion Acoustic
The fact that one of the biggest shops in Germany offers speaker kits with this steep crossovers approach shows me that depending from the drive units used the pros for a steep crossover are more than the cons.
Looking at this kit here:
Accutop 36 dB
indicates that the pros are more with the mid range - high range crossover section where the group delay issues are less severe and with high end drive units with ceramic extremely stiff loudspeaker membranes where the resonances of the stiff material without the steep filters would need to be filtered out with absorption circuits anyway. My own listening tests with my DSP crossovers showed so far that high Q PEQ items can severely compromise the sound quality of the output especially in the midrange frequencies between 400 Hz and 4000 kHz. More disadvantages than the advantage to get a more linear on axis frequency response with some loudspeaker membrane resonances filtered out. My last findings are that the steep filtering is no good for the subwoofer to kick bass crossover about 150 Hz , here the last tests with less steep acoustic LR 2 filtering showed the best result.
BTW i am not relying on my ears only but doing a first proof of concept with a loudspeaker simulation software. Without a first start with a simulation i would be completely lost.
In Germany one of the most popular DIY speaker shops has adopted the steep crossover design, here:
Thalion Acoustic
The fact that one of the biggest shops in Germany offers speaker kits with this steep crossovers approach shows me that depending from the drive units used the pros for a steep crossover are more than the cons.
Looking at this kit here:
Accutop 36 dB
indicates that the pros are more with the mid range - high range crossover section where the group delay issues are less severe and with high end drive units with ceramic extremely stiff loudspeaker membranes where the resonances of the stiff material without the steep filters would need to be filtered out with absorption circuits anyway. My own listening tests with my DSP crossovers showed so far that high Q PEQ items can severely compromise the sound quality of the output especially in the midrange frequencies between 400 Hz and 4000 kHz. More disadvantages than the advantage to get a more linear on axis frequency response with some loudspeaker membrane resonances filtered out. My last findings are that the steep filtering is no good for the subwoofer to kick bass crossover about 150 Hz , here the last tests with less steep acoustic LR 2 filtering showed the best result.
BTW i am not relying on my ears only but doing a first proof of concept with a loudspeaker simulation software. Without a first start with a simulation i would be completely lost.
Last edited by a moderator:
MEGB1262, unless you are able to determine the acoustic requirements for crossing then I can understand you wanting to cross steep. It is a tempting option given the issues that abound. It is complex work to learn any better about the situation. In some cases a steep cross looks like the only way, sometimes because the system was put together randomly and sometimes not. Designing it to be otherwise requries the knowledge of why.
Then there are mechanical limits. It's always disappoijting to make such compromises for simple reasons like this, but sometimes we do.
Then there are mechanical limits. It's always disappoijting to make such compromises for simple reasons like this, but sometimes we do.
As a person who has always strived to not let that class of limitation stand in the way of what I'm doing, I look at this and wonder if you aren't indeed doing the same?
Hello,
Megb, I think you are also experimenting problems related to the DSP and the processing power. High slope low frequency crossover require a huge amount of CPU and taps to have good results. They also generate a good amount of delay...this is why I was suggesting to go with a computer and an rme soundcard...
Like big horns and big compression drivers fans we agree here that the critical voice area around 300 and 3000 should be covered with the same driver... compressions users have to put steep filters because they are already at the limits of the components. That's a sufficient reason but I don't think it is the only one. That would be the same for a dome tweeter playing the lowest reasonable frequencies...
I personally don't like having female voice harmonics coming from the woofer. When I listen to it alone it sounds bad...
If you like 150hz lr2 you should try a higher and steeper cutoff...
Megb, I think you are also experimenting problems related to the DSP and the processing power. High slope low frequency crossover require a huge amount of CPU and taps to have good results. They also generate a good amount of delay...this is why I was suggesting to go with a computer and an rme soundcard...
Like big horns and big compression drivers fans we agree here that the critical voice area around 300 and 3000 should be covered with the same driver... compressions users have to put steep filters because they are already at the limits of the components. That's a sufficient reason but I don't think it is the only one. That would be the same for a dome tweeter playing the lowest reasonable frequencies...
I personally don't like having female voice harmonics coming from the woofer. When I listen to it alone it sounds bad...
If you like 150hz lr2 you should try a higher and steeper cutoff...