Halo!
So yeah, my 4-way active monitor speakers are ready to launch, but I have some questions regarding proper tune of FIR crossover.
My project is aimed to provide studio-quality transparent and analityc sound like a hi-graded studio monitors. I gonna work on mixing and mastering, so any hi-fi colorings are not interested for me.
My questions:
1) Is there any sense to use such a crazy 90 db/oct filter slope, or "just 60 db/oct" will be enough? (There is enough nominal power in speakers and amplifiers, and they're located as close as possible.)
2) Reasonable FFT size? 16k, 8k, 4k?
I'm attaching some photos of my system and screen of current settings I was using for 3-way (Until subs arrived).
So yeah, my 4-way active monitor speakers are ready to launch, but I have some questions regarding proper tune of FIR crossover.
My project is aimed to provide studio-quality transparent and analityc sound like a hi-graded studio monitors. I gonna work on mixing and mastering, so any hi-fi colorings are not interested for me.
My questions:
1) Is there any sense to use such a crazy 90 db/oct filter slope, or "just 60 db/oct" will be enough? (There is enough nominal power in speakers and amplifiers, and they're located as close as possible.)
2) Reasonable FFT size? 16k, 8k, 4k?
I'm attaching some photos of my system and screen of current settings I was using for 3-way (Until subs arrived).
Attachments
Your different bands will have different directivities. A high order of filter may exacerbate that overall.
That driver layout might look "cool" but it's not gonna produce a smooth sound due to interference and diffraction - something that can be very hard to cure with EQ... I think you need to reconsider - study some great monitors just how their drivers are laid out to beginn with?
//
//
Piling on here: TNT is correct, maybe. It depends on the particular frequency bands chosen for the crossover.
The offender is (are) the two midwoofers with yellow (kevlar?) cones, side by side. If these are used to high enough frequency, the different pathlengths to the listener will cause interference as you move off to the side. If you only listen directly in front of the speaker, this will not cause any problem whatsoever. It's exactly the same issue as with an MTM, except an MTM is typically oriented with the two midwoofers separated vertically. Since you head is usually situated in the same vertical position (plus or minus a foot) when you listen, you just don't typically get to high enough off axis angles with an MTM for this to be a big deal. But when you have the drivers side by side you might easily listen by several feet off axis.
The magnitude of the problem totally depends on how high in frequency the drivers are used. The problem gets worse at higher frequencies (e.g. above 1kHz) because the higher you go, the smaller the wavelength and the more phase angle that a given pathlength difference will cause. Once the phase angle reaches 180deg you get 100% cancellation, a null, off axis while on axis there is 100% output. That causes some problems with the sound of the loudspeaker in the room.
The only "fix" for this issue that has been created with the physical design you have chosen (the layout of the drivers) is to use as low a crossover point as possible from the two midwoofers to the next higher driver (the top one?) to minimize the off axis problems. If you can keep that crossover point well below 1kHz, e.g. 500-700Hz, then everything will be just fine. By inspection of your FIR design program, I think you are using around a 500Hz crossover point, so you should be OK. Lower is better.
As far as the slope goes, once you reach 60dB/oct you are getting to the point of diminishing returns in term of the slope so I would not worry about it too much.
The offender is (are) the two midwoofers with yellow (kevlar?) cones, side by side. If these are used to high enough frequency, the different pathlengths to the listener will cause interference as you move off to the side. If you only listen directly in front of the speaker, this will not cause any problem whatsoever. It's exactly the same issue as with an MTM, except an MTM is typically oriented with the two midwoofers separated vertically. Since you head is usually situated in the same vertical position (plus or minus a foot) when you listen, you just don't typically get to high enough off axis angles with an MTM for this to be a big deal. But when you have the drivers side by side you might easily listen by several feet off axis.
The magnitude of the problem totally depends on how high in frequency the drivers are used. The problem gets worse at higher frequencies (e.g. above 1kHz) because the higher you go, the smaller the wavelength and the more phase angle that a given pathlength difference will cause. Once the phase angle reaches 180deg you get 100% cancellation, a null, off axis while on axis there is 100% output. That causes some problems with the sound of the loudspeaker in the room.
The only "fix" for this issue that has been created with the physical design you have chosen (the layout of the drivers) is to use as low a crossover point as possible from the two midwoofers to the next higher driver (the top one?) to minimize the off axis problems. If you can keep that crossover point well below 1kHz, e.g. 500-700Hz, then everything will be just fine. By inspection of your FIR design program, I think you are using around a 500Hz crossover point, so you should be OK. Lower is better.
As far as the slope goes, once you reach 60dB/oct you are getting to the point of diminishing returns in term of the slope so I would not worry about it too much.
This seems to conflict with what you say in the next paragraph where you say it causes some problems.
OP plans to cross those 6.5" midwoofers at 300Hz, so I doubt centre to centre spacing is going to be an issue when the wavelength of 300Hz is over 1metre long.
There's no real advantage to go much steeper than 24dB/octave. I believe some people have proposed that phase shift is audible in extreme cases (>48dB/octave).
There's no real advantage to go much steeper than 24dB/octave. I believe some people have proposed that phase shift is audible in extreme cases (>48dB/octave).
Hi,
1) I'd say just try it...since it's so easy to build FIR files with different xover slopes.
Or I should say so easy if you are using complementary linear phase xovers, which i kinda assume you are since you're using FIR.
A trade-off with steeper, as Allen said, is that if you have different band directivity's, steeper can make them stand out more.
However, in my experience it's usually easier with steeper to move xover points to where directivity's match better, because steeper gives more latitude with safer excursion levels at bandwidth ends.
The other really big plus with steeper xovers, is lobing reduction. No trade-offs here that I know of.
And of course if you're using linear phase xovers, there's no increase in phase rotation or group delay from steeper.
2) Rephase defaults to a FFT of 32k; my go-to Fir generator 16k, but has the option to raise to 32 or 64k.
This is not taps you're speaking of, right? Just what the program uses to make calculations, right?
Taps are a whole 'nuther animal, huh 🙂
Gentlemen, thank you so much for such a deployed answers! That's a pleasure to have this support through my beginner's DIY audio steps.
Let's go one by one.
Maybe that's a bit step out from what you're trying to say - but good news that I decided to cover the front surface of speakers by a 1/4"+ (even 1/2" on subs) thickness damping material (cotton blanket and a black felt). With a quiet close laid of drivers and, as small as possible, the size of the front panel, I hope to have minimal impact from cabinet-reflections.
Next band with a Coral MD-5, 3" aluminum dome, is 330Hz HPF
90dB/oct and it's as low as possible for that driver.
In the future, I will research one more plugin which I can use to correct the phase response of my systems with an all-pass filter. (And yeah, for sure I already set the delay of every speaker relatively to my subs).
I will be grateful, if you can introduce decent software (or maybe even guide🙂 especially for creating FIR-filters (I'm ok with loading IR in convolution players).
FIR Designer from Eclipse audio seems very fine, but too expensive for my case🙁(
P.S: I'm not finally sure, that FFT Size and Taps are the same options, but that's what I have in my ReaFIR VST, they call it FFT Size🙂 And yeah, I have there 32k as the maximum setting.
Any guidance whenever I'm doing something wrong... Welcome!
One more time - thank you all! I think you can imagine how far the newbie can be lost in a such difficult world of DIYAudio...
Let's go one by one.
Yeah, I already know about this issue - traditional d"appolito concept "Sub - LF - MF - T - MF - LF - Sub" was too expensive and excessive for my goals, so I decided to step further with what I currently have...
No doubt, you're right.
Maybe that's a bit step out from what you're trying to say - but good news that I decided to cover the front surface of speakers by a 1/4"+ (even 1/2" on subs) thickness damping material (cotton blanket and a black felt). With a quiet close laid of drivers and, as small as possible, the size of the front panel, I hope to have minimal impact from cabinet-reflections.
Correct! Eton Adventure A1-160.
2x6.5" 380Hz LPF 90 dB/oct. Not as high to have interaction between driver's even I'm moving my head in a small range around the sweet spot.
Next band with a Coral MD-5, 3" aluminum dome, is 330Hz HPF
90dB/oct and it's as low as possible for that driver.
I prefer to agree. Tomorrow I will xover my system again, and this time I'll stick to 60 dB/oct slopes.
I don't expect that much from the off-axis sound of my speakers. And I don't think that in terms of "room reflections and reverb" this different band directivity will be crucial. So, as long as my mids are below 400-500 Hz I will be just calm 🙂
I'm newbie to FIR too, so, I'm using Reaper DAW like a host for all xovers, and built-in VST plugins, ReaFIR exactly (you can see them in a first message). Probably - not the best solution, because I'm not correcting the phase of speakers, only xoving them and trying to make amplitude response as flat as it is reasonable in the context of room response I'm catching in my mic during measurements (you can see a few EQ points throughout the bands).
In the future, I will research one more plugin which I can use to correct the phase response of my systems with an all-pass filter. (And yeah, for sure I already set the delay of every speaker relatively to my subs).
I will be grateful, if you can introduce decent software (or maybe even guide🙂 especially for creating FIR-filters (I'm ok with loading IR in convolution players).
FIR Designer from Eclipse audio seems very fine, but too expensive for my case🙁(
P.S: I'm not finally sure, that FFT Size and Taps are the same options, but that's what I have in my ReaFIR VST, they call it FFT Size🙂 And yeah, I have there 32k as the maximum setting.
Any guidance whenever I'm doing something wrong... Welcome!
One more time - thank you all! I think you can imagine how far the newbie can be lost in a such difficult world of DIYAudio...
There are commercial FIR designers like Acourate and Audiolense that can do what what you are asking for. But they are also in a similar price range as Eclipse audio...
There are a couple of guides I wrote using them:
Advanced Acourate Digital XO Time Alignment Driver Linearization Walkthrough - CA Academy - Audiophile Style
Integrating Subwoofers with Stereo Mains using Audiolense
Should give you some idea of the process.
Good luck!
There are a couple of guides I wrote using them:
Advanced Acourate Digital XO Time Alignment Driver Linearization Walkthrough - CA Academy - Audiophile Style
Integrating Subwoofers with Stereo Mains using Audiolense
Should give you some idea of the process.
Good luck!
Well, OK, I did re-write some of my post to be less critical once I realized that Enotovvod was using a 300Hz crossover point and I probably should have modified the "next" paragraph as well.
I think the speaker will be just fine. I'll leave it at that.
Hi Enotovvod,
Sorry if i misled you with questions from my previous reply...they just were meant to see if you were talking about number of taps (coefficients) or the FFT size used by what FIR generator is making the coefficients.
You are correct that they are not the same thing.
I highly recommend rePhase for learning how to build FIR files.
Since all steps are manual, it teaches what to use when....what is the right tool for the task....and in what order.
For a start, load in a single driver's response, and use the Minimum Phase EQ tab to make adjustments until you like the response. (keeping polarity +)
Then simply add linear phase xovers, and generate.
There is nothing left to do except match driver levels, and put in delays to offset physical acoustic center differences.
My guess is, those simple steps will probably exceed 90% of DIY tuning efforts.
Once you get the handle of all that, here's a tip for determining correct timing to low frequency (sub) acoustic centers, to help get delays spot on.
Every measurement program out there has trouble with this, when a sub low pass filter is in place.
In whatever FIR generator you are using, note the latency of the number of taps you are using. In rePhase it is the 'impulse delay' in the box below the Generate button. Same thing as the FIR filter latency.
Bypass the FIR filter and measure loopback time, or time of flight, etc....whatever is the same way you measure the distance of the other drivers (which don't usually vary).
The turn the FIR filter on, and add the FIR latency, to get time distance to sub.
You get an exact time that can only be off by any small amount the sub's acoustic center might change with frequency.
This has a very small error potential compared to the variations in measurements, when trying to measure a sub with a xover in place.
I figure you may need this trick since you said 4-way.
Good luck !
Oh ps, in rePhase you can see the different fields for FFT size and taps..