The other thing you need to understand is that to get your target LR2 or LR4 (or whatever acoustic target you want) you may not use that at all as far as the filters you select in the DSP program. Eg an LR2 filter in conjunction with your drivers natural rolloff, could possibly give you a LR4 acoustic rolloff.
It is not uncommon to have different order filters on the woofer and the tweeter say for instance 2nd order on the woofer and third order on the tweeter, to get a 4th order acoustic rolloff. It all depends on how close to the individual drivers natural roll-offs you are. The flatter the driver is at (and extending beyond) the crossover point, the more likely you are to hit the acoustic target, with the same electrical filter order.
Tony.
It is not uncommon to have different order filters on the woofer and the tweeter say for instance 2nd order on the woofer and third order on the tweeter, to get a 4th order acoustic rolloff. It all depends on how close to the individual drivers natural roll-offs you are. The flatter the driver is at (and extending beyond) the crossover point, the more likely you are to hit the acoustic target, with the same electrical filter order.
Tony.
The flattening logic is that you get textbook response whatever crossover filter you apply, so you can experiment with different slopes and points in the future very easily. If you apply a 3 kHz LR2 filter you get a 3 kHz LR2 response (on that axis) and so on.
But some says, it really doesn't matter what slopes you use and doesn't need to use textbook curves. Try it yourself.
Btw I think for LR4 slopes, 1 octave from the xo point is enough to be flat if the farther response doesn't peaks severely. For LR2, 1.5-2 octave is recommended. But others opinions may differ.
Yes, a low-shelf for tweeter for example. You need experimenting here, but don't do it loud.
Last edited:
Change Q in the Hypex software and observe the shape of the filter flanks in the graphical interface... you will see how Q affects a filter. Per definition, the XO Fs is not changed - even if it might look like that
//
This is perfectly clear!
However, building a 3rd order LR is impossible, but you are speaking of another type of filter I assume.
I'll start working with an LR4 filter (or I will try building one) because it seems easier and better in terms of tweeter protection. I'll flatten the frequency respone (natural) for the drivers as follows:
Bass (xo 350 Hz): 700Hz and below.
Mid (xo 350 / 2000 Hz): 175 Hz and upwards up to 4000 Hz.
Tweet (xo 2000 Hz): 1000 Hz and opwards.
Is this correct?
Given I'll build a 4th order (acoustic) LR filter, do any of you dare to lower the tweeter xo even further down?
I'll use the flattening approach to make building filters easier. However, would a sound be 'better' if it's easier made? Meaning: using the natural response in building the filter? Or does this not matter (in real life)?
Always worth reading, especially page 6: https://www.grimmaudio.com/wordpress/wp-content/uploads/speakers.pdf
"Finally, the icing on the cake. The sum of an ideal LR4 system is a second order all-pass with a Q of 0.7. In order to avoid the problems associated with correcting phase exactly, build an inverse all-pass filter based on the theoretical ideal. This filter will be non-causal so there’s a good reason for using FIR.
The step response of this system is precisely what we wanted: no pre-echo, nearly minimum phase."
Can anyone explain this step?
The step response of this system is precisely what we wanted: no pre-echo, nearly minimum phase."
Can anyone explain this step?
And done! The result is soundwise very impressive. I used the steps you all suggested (and the Grimm audio pdf).
I'll post photos soon but I'll enjoy the sound for some time (it took me four hours).
I ended op crossing at 420 and 2500 Hz because those frequencies were best for the mids natural response.
I'll post photos soon but I'll enjoy the sound for some time (it took me four hours).
I ended op crossing at 420 and 2500 Hz because those frequencies were best for the mids natural response.
Last edited:
I am simply astonished by the result.
The bass is already incredible and will be better controlled using on a few resonance frequencies using DIRAC live (on board my c658) but the mid and tweet are unlike anything I've heard so far in my home. Incredibly clear sound stage.
It feels like being at the recording studio.
Thanks guys.
The bass is already incredible and will be better controlled using on a few resonance frequencies using DIRAC live (on board my c658) but the mid and tweet are unlike anything I've heard so far in my home. Incredibly clear sound stage.
It feels like being at the recording studio.
Thanks guys.
Fantastic I think this thread has been a great example of why you still need to consider the fundamentals when using active crossovers! Many people do as Geertidow initially did, because "active makes things easy". Once you understand what is going on though, getting a really good result becomes possible.
Just found this post which I thought I'd post here https://www.diyaudio.com/forums/mul...ats-difference-whats-favorite.html#post488920 as it describes things in an easy to understand way. Normally any articles on filters require digestion of large amounts of theory and math!
Tony.
I think this thread has been a great example of why you still need to consider the fundamentals when using active crossovers! Many people do as Geertidow initially did, because "active makes things easy". Once you understand what is going on though, getting a really good result becomes possible. Just found this post which I thought I'd post here https://www.diyaudio.com/forums/mul...ats-difference-whats-favorite.html#post488920 as it describes things in an easy to understand way. Normally any articles on filters require digestion of large amounts of theory and math!
Tony.
I was searching everywhere for a simple explanation of how to cascade firts and second order filters to get different results. I know I've read them before, but could not find any. That was the best I found. Oh well
The math is always a mind bender for me. Thanks for pointing out the issues Alle. You have touched on the next aspect (3d radiation) that we haven't really got into in this thread.
Tony.
The math is always a mind bender for me. Thanks for pointing out the issues Alle. You have touched on the next aspect (3d radiation) that we haven't really got into in this thread.
Tony.
Some photos of REW as promised.
The first measurements (to flatten frequency fresponse) were taken 20-30 cm from the driver, facing eacht driver directly. For the bass (two parallel drivers) I measured in the middle. After applying some filters, these were my flattened responses:
Overlay of flattened responses
20191223-Flattened-Drivers-OVerlay — imgbb.com
Then, I decided to build filters (acoustical LR4 filters with xo at 420 and 2500 Hz.
After some fiddling with the q's I ended up with best results when choosing a normal Q of 0.71. These measurements were made at approximately 30 cm away at tweeter (listening) level.
The xo point is at -6dB from baseline (approximately)
Overlay frequency response bass, mid, driver (LR4 420 Hz and 2500 Hz).
20191223-LR44202500q71overlay — imgbb.com
Then, I took a full sweep at 150 cm from the tweeter:
20191223-Full-Sweep-Finalfrom150cmtweetlevel — imgbb.com
There are some room effects in this sweep.
I aligned timing for the tweeter using 'acousting timing reference' with a sweep starting at 1000 Hz with either bass/tweet on, mid only on and tweeter only on.
It seemed that the bass (bass sweep from 200 Hz) and tweet arrived 150 us faster at the position at 150 us. But I doubted if the bass would actually arrive later (mid and bas conus are same position). So I build 3 presets on the Hypex (bass & tweet + 150 us delay, tweet 150 us delay, no delays).
After careful listening the tweeter delay only seems to sound best, but it is really difficult to be sure.
The next steps are:
1. (optional, not sure of benefit): allign drivers in time from listening position. However, I'm not sure what the best method is.
2. measure DIRAC live and controle the room effects (probably for up to 1000 Hz)
The first measurements (to flatten frequency fresponse) were taken 20-30 cm from the driver, facing eacht driver directly. For the bass (two parallel drivers) I measured in the middle. After applying some filters, these were my flattened responses:
Overlay of flattened responses
20191223-Flattened-Drivers-OVerlay — imgbb.com
Then, I decided to build filters (acoustical LR4 filters with xo at 420 and 2500 Hz.
After some fiddling with the q's I ended up with best results when choosing a normal Q of 0.71. These measurements were made at approximately 30 cm away at tweeter (listening) level.
The xo point is at -6dB from baseline (approximately)
Overlay frequency response bass, mid, driver (LR4 420 Hz and 2500 Hz).
20191223-LR44202500q71overlay — imgbb.com
Then, I took a full sweep at 150 cm from the tweeter:
20191223-Full-Sweep-Finalfrom150cmtweetlevel — imgbb.com
There are some room effects in this sweep.
I aligned timing for the tweeter using 'acousting timing reference' with a sweep starting at 1000 Hz with either bass/tweet on, mid only on and tweeter only on.
It seemed that the bass (bass sweep from 200 Hz) and tweet arrived 150 us faster at the position at 150 us. But I doubted if the bass would actually arrive later (mid and bas conus are same position). So I build 3 presets on the Hypex (bass & tweet + 150 us delay, tweet 150 us delay, no delays).
After careful listening the tweeter delay only seems to sound best, but it is really difficult to be sure.
The next steps are:
1. (optional, not sure of benefit): allign drivers in time from listening position. However, I'm not sure what the best method is.
2. measure DIRAC live and controle the room effects (probably for up to 1000 Hz)
Last edited:
You should turn on the phase plot for each driver. It isn't necessary to guess on the delay.
Can you manage the timing? It is tricky ensuring each mesurement is timed the same as each other both electrically, and physically.
I timed the drivers using Acoustic Timing Reference in REW and sweeping from 200 Hz for the mid and bass (first a sweep with mid driver muted, then one with bass driver muted), and afterwards a sweep from 1000 Hz with bass and tweeter muted and after bass and mid muted. The measurements consistently showed a delay of the mid driver of 180 us.
I used various methods described on the forum but this is the only one that gives me results I understand.
However, if you know a good method to time the drivers then feel free to share. Based on the position of the drivers the tweeter should be delayed approximately 50 us.
I used various methods described on the forum but this is the only one that gives me results I understand.
However, if you know a good method to time the drivers then feel free to share. Based on the position of the drivers the tweeter should be delayed approximately 50 us.