That is lobing caused by the drivers being non coincident, happens in almost every speaker to a greater or lesser extent. A minimum phase filter used carefully can add delay to a driver to bring them into alignment when delay per channel isn't available.
If you don't take that into consideration then a mess is what you get
Using a linear phase crossover and electronic delay on each channel makes the job of getting the crossover reasonable much easier but it comes at it's own cost so no free lunch either way.
Good point, rather I should just send analog to the Hypex....I was focused on the removal of a conversion I guess
The signal is still going to get converted to digital in the Hypex isn't it....I guess a reasonably priced external sound card with digital out is a better solution.
then again they don't have FIR.....I've some thinking to do lol
Last edited:
Why wouldn't you be able to use FIR (inside the computer) running the external sound card?
That is lobing caused by the drivers being non coincident, happens in almost every speaker to a greater or lesser extent. A minimum phase filter used carefully can add delay to a driver to bring them into alignment when delay per channel isn't available.
If you don't take that into consideration then a mess is what you get
Yes exactly! The side lobes (and who dare go into beaming) and iir crossovers and speakers that are not on the same axis and are barley if even 1/4 wave apart is way more disastrous then an fir ..... the fir seems to really clean up polars of the loudspeaker as whole in this scenario. Although yes your right.... no free lunch.... lol I like that, otoh instead of McDonalds your getting prime rib. Yeah at a cost.... however , for me in a car it sure seems to be free lunch coupons at every channel.
There’s nothing more annoying then a nicely centered stage except one frequency in the vocals gets pulled one way or other or becomes completely diffuse, I like my vocal to come from between the speaker (the middle of the dashboard) at all the frequencies that Are in the vocal range..... with iir crossovers and a 4way it’s very very difficult with all the close reflections and horn loading because of 1:4 space...... I’m just sayin..... maybe on a home system it’s different as everything is uniform as far as arrival times and acoustic centers , in a car where 1 to 2 ms of delay on one side isn’t unheard of or 6db differences between left and right.
The fir crossovers made linear sure massively clean all this up..... the reflections (sometimes from side lobes) and direct sound all come from a linear time..... so dealing with the reflections is nothing more then that, I don’t have all the phase twisting causing all this time issues...
Anyway that was the only thing I could think of when he said someone told him to do an fir at each channel because of “cancellations”
I think I was correct
maybe .... Not what I meant. Avoiding the conversion can make some sense and also avoids another analogue gain structure component.
There is no need for such an expensive DAC if you are going to send the signal digitally many cheaper devices can do that however going multichannel inside the PC and outputting to a multichannel soundcard makes a lot of sense as it gives a lot of options per channel. It makes less sense if using the Hypex plate amps with DSP. If your Crown amps are class AB they may well be ideal and the fans can be replaced with quiet ones as they are usually cheap noisy ones. Noctua or similar PC case fans are much quieter even at the same voltage. The Hypex software to control the DSP on the plate amps is functional but woeful in comparison to anything else I have used.
Equalizer APO can do multichannel if that is the direction you want to go
Equalizer APO / Documentation Wiki / Configuration reference
Create yourself a CD with sine sweeps at different frequencies (bottom of page in the link). Listen to that on your corrected system in your car. Want to bet there are frequencies that do sound diffuse? And in a car you're in a luxury position. As the center should be in the middle of the dash, this means both ears get a slightly different timing due to sitting off axis.
In our home, where we sit dead center, these diffusive frequencies would be even more obvious. Its just part of Stereo and having two ears. Crosstalk is bound to happen.
I've played with Car Audio for quite a while. Because of the results I got, I just had to upgrade my home system.
Last edited:
I repost this, 4 days left before the release, feedback and bug reports welcome.
It has now been more than two years since the 1.4.3 version was released, and although I started implementing new features and corrected a few bugs in the meantime, I did not feel like anything was ready enough for a release. Then I had to concentrate on other things and never got around to actually finish it...
So, here is a development release. I tried to clean things up a bit and removed features that were too far off.
The only new feature remaining is a loudness compensation functionality, that could for example let you build "quiet" or "concert level" presets while preserving the same perceived tonal balance as your reference level. It is not exactly where I wanted it to be, but it is functional and probably good enough for the matter, and after 2 years without working on it I think it is better to release it as-is, at least for for now.
Other modifications are listed in the "what's new" entry in the Help menu and include adjustments and bug corrections.
Please note this is a development version. As such it has several limitations, to wit:
So please do not replace your current rephase version with this one, this is just for evaluation and bug reports, in preparation for the stable version.
- probably a few bugs
- listening compensation parameters are not saved in settings
- expiration in 10 days to avoid having unfinished development versions living their life in the field. The stable 1.5.0 version should have been released by then.
With all that said, here is the link: rePhase 1.5.0 DEV
Your feedback is welcome.
So one thing I’ve noticed in a car , and yeah simple signal delay get you there to get a good center....as long as both channels are doing the same thing.... I’ve done tones and sweeps before...... my car is a little crazy so I actually do have 250hz to 2.5khz nailed down as I now have 3ways in my a pillars ..... so I’m spoiled and I’m not trying to get loud. , but as far as off axis and on axis go....I have always every time (with direct radiators) had bad sound and a collapsed diffuse stage with the fronts “towed in” even when sitting at close distances.... I had all my mains pointing forward and let me be slightly off axis.... that’s where it’s at!
Huge huge wide massive soundstage that is stable and a center that just doesn’t quit..... I see all this hi end home audio and very respectful ppl doing tow in.... I just never understood it... I’ve listened to a ton of home speakers ( I’ve spent the last 30yrs in pro audio as a job and in car audio) and to this day I stand firm.....
Slightly off axis for the listener is where it’s at.... and transfer functions that match (acoustic ones) so point them both in same direction and make them exactly the same
Is that how you do yours?
That’s why I think you’re actually in the luxury position because you can have off access listening and have no come filtering issues caused by use of signal delay and your listening room isn’t a metal box and you’re not sitting in the corner
Last edited:
As I still fighting with my active speaker filters pre-ringing I had questions.
What type of linear-phase filters, available in rePhase, with at least 24 dB/octave slope give less (in amplitude and time) pre-ringing?
What is better if I want less pre-ringing:
1. to use for speaker bands filtering minimum phase filters, correct phase of filters with "Filter Linearization" and adjust rest phase manually.
2. to use for speaker bands filtering linear phase filters and adjust rest phase manually
What king of phase equalizer adjustments must be avoided not to get audible pre-ringing?
What type of linear-phase filters, available in rePhase, with at least 24 dB/octave slope give less (in amplitude and time) pre-ringing?
What is better if I want less pre-ringing:
1. to use for speaker bands filtering minimum phase filters, correct phase of filters with "Filter Linearization" and adjust rest phase manually.
2. to use for speaker bands filtering linear phase filters and adjust rest phase manually
What king of phase equalizer adjustments must be avoided not to get audible pre-ringing?
Last edited:
A complementary crossover will have no pre (or post) ringing whatsoever, as the summation will simply cancel any ringing of the low and hi pass filters. The challenge lies in obtaining said complementary crossover. In practice that implies taking into account the natural response of the driver (making good use of the "compensate" minimum-phase filter), and having good directivity matching and the small distance between drivers to maintain that good summation over a wide angle.
The sharper the filter slope, the more ringing will "lick out" in case of bad summation.
The best way to avoid pre ringing is to only use minimum-phase corrections (including aforementioned compensate filters), and keep linear-phase for crossovers.
The sharper the filter slope, the more ringing will "lick out" in case of bad summation.
The best way to avoid pre ringing is to only use minimum-phase corrections (including aforementioned compensate filters), and keep linear-phase for crossovers.
Can I get the benefit of complementary crossover ringing canceling in 3 way speakers? The mid driver had there actually 2 filters, LP and HP?
Must then both filters have exactly same numbers of taps?
How this complementary crossover ringing canceling is possible in real speaker when speaker elements had different directivity and personal/not same frequency response without filters? Direct sound can have working complementary crossover ringing canceling, but room reflected sound can have doubled ringing amplitudes?
Must then both filters have exactly same numbers of taps?
How this complementary crossover ringing canceling is possible in real speaker when speaker elements had different directivity and personal/not same frequency response without filters? Direct sound can have working complementary crossover ringing canceling, but room reflected sound can have doubled ringing amplitudes?
Last edited:
Yes you can.
If you go for a minimum-phase crossovers you will have to take into account the phase shifts implied by the other crossover point.
For example if you have a 24dB/oct LR crossover at 300Hz, and another one at 1kHz, you should also add a 300Hz 24dB/oct LR HP filter to your upper driver, and a 1kHz 24dB/oct LR LP filter to your lower driver, in order to get them were they are supposed to be in terms of phase shifts. This also guaranties that you will be able to linearize their phase using the intended "filter linearization" entries.
As a side note, same should be done for the general HP and LP of the system (although the LP can generally be safely ignored there).
For example if you are crossing over to a sub at 80Hz, you should not try to get a perfect 80Hz HP filter, but one that also encompass the HP filter of your sub (eq a bass reflex design at 30Hz).
Now the good news: if all of your components have a linear phase (ie made acoustically linear phase in their active filtering settings), then summing is easy as you don't need to take any upper or lower phase shift into account. Hurray for linear phase
(you still have to take the system's HP into account if you don't intend to linearize its phase though).
I cannot stress enough how important "compensate" filters are for obtaining a proper result there.
Not necessarily. Just make sure the result curve is close enought to the target one, and you compensate the delay difference (as well as potential processing/buffering differences in case of a time-domain convolution).
One exception to this is the use of brickwall filters, as they rely solely on the number of taps and windowing algorithm used to define their response, and cannot be complementary if the HP and LP taps and windows are different.
(and even then: any EQ point in the pass band might throw that complementary away. Better avoid them...).
Yes, and this is also true for "classic" minimum phase filters, where you get some (arguably less offensive) post ringing when/where the summation is not correct.
In any case, minimum or linear phase, a proper system requires selecting and implementing drivers based on their directivity behavior, taking the natural response of the drivers into account (including boxes, baffles, horns, you name it...) and build coherent acoustical filters from that.
Linear-phase behavior can theoretically make any bad summation more harmful, but it also makes them easier to avoid in the first place.
Last edited:
POS ....
So that was interesting....
So, I have to ask ...
The part you were telling him about inherent phase shift on minimum phase crossovers...
like let’s say one uses a 12db LR Filters on all drivers in a 3 way , the BP filter the low pass and high pass are out of phase with one either the LP or the HP all depending if you go for the invert switch on the BP....
I understand that completely, and adding the inherent out of band filter solves that....
So what if in the case of a 1/2 linear phase filter and 1/2 linearized filter..... there both that same thing am I not right?
Once linearized ..... there no need for inherent filters to be added?
Is that right or wrong
Once linear , does the impulse get moved, like the out of band stuff still shift even tho it’s flat phase response is it still shifted ? Like a minimum phase crossover?
I’m probably not saying this right, hope you understand
And my understanding is depending if it’s a LP or. HP (leading or following) the in band or stop band magnitude will shift also..... is that the case with linear phase , or linearized ?
(Two different ways of asking same thing)
Thanks
So that was interesting....
So, I have to ask ...
The part you were telling him about inherent phase shift on minimum phase crossovers...
like let’s say one uses a 12db LR Filters on all drivers in a 3 way , the BP filter the low pass and high pass are out of phase with one either the LP or the HP all depending if you go for the invert switch on the BP....
I understand that completely, and adding the inherent out of band filter solves that....
So what if in the case of a 1/2 linear phase filter and 1/2 linearized filter..... there both that same thing am I not right?
Once linearized ..... there no need for inherent filters to be added?
Is that right or wrong
Once linear , does the impulse get moved, like the out of band stuff still shift even tho it’s flat phase response is it still shifted ? Like a minimum phase crossover?
I’m probably not saying this right, hope you understand
And my understanding is depending if it’s a LP or. HP (leading or following) the in band or stop band magnitude will shift also..... is that the case with linear phase , or linearized ?
(Two different ways of asking same thing)
Thanks
Last edited:
Not sure I exactly get what you mean, but looking at the frequency response and the impulse give you the same information. You can disregard the impulse in most cases, the frequency response being much more revealing and less confusing: simply try to get matching/tracking phase shift between drivers, and ignore the impulse altogether.
If you (virtually) get each driver's phase and magnitude flat in and around the passband (using compensate filters), and add linear-phase complementary filters, then summation is as easy as dialing the proper delay.
That is the method I recommend: https://www.diyaudio.com/forums/mul...ion-eq-fir-filtering-tool-68.html#post4322701
This is *much* easier than building a minimum-phase filter, especially if more than 2 ways are to be used, but of course this requires having access to a multiway convolution engine...
If you (virtually) get each driver's phase and magnitude flat in and around the passband (using compensate filters), and add linear-phase complementary filters, then summation is as easy as dialing the proper delay.
That is the method I recommend: https://www.diyaudio.com/forums/mul...ion-eq-fir-filtering-tool-68.html#post4322701
This is *much* easier than building a minimum-phase filter, especially if more than 2 ways are to be used, but of course this requires having access to a multiway convolution engine...
Hi Thierry!
There are other functionalities I would like to add to that tab in the future, related to the listening environment or listening preferences.
I searched information about two impulse response convolution and if I understand it right, complementary crossover ringing canceling between 2 speaker elements can work only if both elements has exactly same impulse response. Otherwise summary elements impulses (with filter impulse added) are differently summed and cancelling will not work. Usually it make no sense to use same elements for mid and tweeter and then filter them to 2 bands.
My understanding is that better is to avoid any ringing (also post-ringing) at all.
So question remains:
What type of linear-phase filters, available in rePhase, with at least 24 dB/octave slope give less (in amplitude and time) pre-ringing?
Can with rePhase be generated higher order Bessel and Gaussian filters (they have very little pre and post ringing)?
I understand 2nd order is possible, but higher orders?
My understanding is that better is to avoid any ringing (also post-ringing) at all.
So question remains:
What type of linear-phase filters, available in rePhase, with at least 24 dB/octave slope give less (in amplitude and time) pre-ringing?
Can with rePhase be generated higher order Bessel and Gaussian filters (they have very little pre and post ringing)?
I understand 2nd order is possible, but higher orders?
I think you are missing the forest for the trees here. It is not all or nothing. The closer you get the two sides of the acoustic slope to match and be complementary the better the cancellation will be up to the theoretical ideal of complete cancellation of pre and post ringing. There is no need to use the same elements but matching the acoustic slopes are directivities makes it work better and that also happens to be good design in general so a win win. If you make a sloppy job of it and use wildly varying directivities and mismatched slopes then it will not be as good as it can be.
There is a time frequency trade off in filters Gaussian has no ringing but is a hopeless crossover filter as a result.