Welcome, Guest
Username: Password: Remember me
  • Page:
  • 1
  • 2

TOPIC: Lin phase Eq

Lin phase Eq 3 years 11 months ago #27359

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
Hi Pos, or anyone that would know

I had a question regarding linear phase Eq
What conditions would this be of use?

Seems regular PEQ flattens phase and magnitude together, that combined with some linear phase crossovers should yield a linear phase responce (granted you did your RTA work correctly;) )

So why would we need a linear phase Eq?
Is it for room modes? Or is it for room correction? Comb filtering?

Seems like it works pretty good if I have a peak from a reflection and use linear phase Eq to ajust the responce with the reflection in the responce. Of course my rta doesn't show that however the IR does.

Could you point me in the right direction on when to use linear phase Eq ?

I also tried it on some higher order modes caused from a horn and it seemd to do some wired stuff but also in some cases made it sound better .....

Thanks in advance

Andrew
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27360

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hi Andrew,

As you point out minimum-phase EQ does flatten the phase when you flatten the magnitude, so that is the type of EQ you want to use for sound reproduction (sound production is a different story).

In the case you are describing with the reflection, you should not try to EQ it at all, as it will obviously change in drastic ways from one listening position to another.

One possible scenario where linear phase EQ can be useful is when you want to first linearize the phase, and then tackle magnitude correction without messing up what you have done before.
Of course it is much more logical to do it the other way, ie start with magnitude corrections using minimum-phase EQ, and then focus on phase.

Another potential use could be loudness compensation (is the brain minimum-phase? :huh: ), although I doubt it will make any difference in the end...
https//wavetracing.com | rephase.org
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27362

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
Maybe using phase Eq , one could match the Q or use .7as Q and raise the phase on a LR filter by 3db and basically make a butterworth alignment. Or is that wrong?

Basically if I wanted to linearize a butterworth filter I could go filter linearization , linearize whatever Bw filter slope I'm wanting to use, than goto phase Eq and center it at crossover , and raise the phase by 3db with a Q of .7 and basically make a linearized butterworth filter?

Yeah it might have some minor ripples at crossover but shouldn't hear that no?

I mean , someone could get trick with it and use adjacent sliders around that one phase ajustment and go a bit above and below crossover and tweak the phase responce so that it matches a butterworth filter exactly, no?
Last Edit: 3 years 11 months ago by denver8me4dinner.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27369

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
If you wanted to do that you would use a minimum-phase EQ and then cancel it magnitude using linear-phase EQ.

But anyway, why would you want to adress a butterworth crossover in phase linearization?
If it is an even order one then if will likely get turned into a LR with EQ (-3dB at crossover point and voilĂ ), and if it is an odd order one (the only one that should be used) then it can be linearized using a LR linearization of the order-1.
https//wavetracing.com | rephase.org
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27374

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
pos wrote:
If you wanted to do that you would use a minimum-phase EQ and then cancel it magnitude using linear-phase EQ.

But anyway, why would you want to adress a butterworth crossover in phase linearization?
If it is an even order one then if will likely get turned into a LR with EQ (-3dB at crossover point and voilĂ ), and if it is an odd order one (the only one that should be used) then it can be linearized using a LR linearization of the order-1.

Only reason is right now I'm going analog four way out of a dexp99 car head unit. It's four way active built in and 30band and TA. Was just curious as to a few experiments. I wouldn't run a BW filter for keeps though, and I wanted to tickle your feathers a bit :silly:

So , I super appreciate your help.
Now that I'm getting the hang of it your software really works very well. I will add a donation at surge force on payday. I appreciate your time.

On the more serious side,
I was also wondering abut tap optimization.
If I may;
Right now I using two- 2x4hds total of 8192 taps/4-2048 blocks all together.

I have sub/horns on one , and mid/midrange on second . I'm sure you know why I did that, I ended up with 1688taps on sub and 300 on horns and on other dsp 1024 on mids and midbass.

I noticed that once I go past a certain number of taps the fft length changes. Example diffrance between 511 taps and 512 taps changes the fft length to next size block.

With that said, obviously a longer fft uses more taps, so would you know if using the horn and sub 1 tap below the next size block make it more efficient?

Kinda hard to explain I'll try again ;)

How should I go about maximizing taps?
Should I just make the horn filter the way I want with the least # of taps I can and whatever's left make the sub with? That's sorta what I did , but if there's a better way to get more taps to the horn and use less of sub or vice versa. I was wondering how you do it and what you think I should try with my 2048 taps per unit.

It also seems like filter linearization uses less taps than making the filter in the Fir, is that so?

What kind of combination with the iir crossovers, PEQ and rephase combo would you suggest I try with a pair of the 2x4hds to make best use of it. Would I gain more if I just linearize all the filters, I love the sound of the horbach-keele filter on sub, but I won't miss it if you have a more efficient idea.

Thanks, Andrew
Last Edit: 3 years 11 months ago by denver8me4dinner.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27407

  • WaveFlex
  • WaveFlex's Avatar
  • Offline
  • Fresh Boarder
  • Posts: 13
  • Karma: 0
Hi Andrew,

For caraudio application, one solution to make a linear phase filtered system is to :
1) measure each speaker separately and tune it to desired target response with IIR PEQ and XO directly on minidsp soft.
2) by channel linearize in rephase the filter used in the minidsp soft (you are right, linearize the phase use less taps than filtering)
3) TA the filtered/EQ front speakers of each side to tune crossover points
4) add the same delay to alls the speakers of a side to make centered stereo soundfield.
5) switch on the sub and align them to the frontstage.

You have a linear phase filtered loudspeaker system, for sure regardless of phase behaviour of the passenger compartment.

Hope this helps,
Pierre
Last Edit: 3 years 11 months ago by WaveFlex.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27446

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
Thanks Pierre!
That's basically what I have done, and it's good! It measures well. I also have some APF in the midbass to get both front seats in phase.

Sometimes , it seems as if the delay from a crossover helps the overall sound. I'm not sure if I've gone crazy but in the minimum phase part of the midrange especially. The linearize crossover sounds good, just the IIR crossover only the HP side of the midrange only seems to sound better with just IIR. Obviously the sub and midbass are a wreck without FIR. So I was thinking I don't have something set right or there's a better way. That's why I asked.

I have the midrange on the glass next to the horn so there's lots of reflection

Andrew






Last Edit: 3 years 11 months ago by denver8me4dinner.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27447

  • WaveFlex
  • WaveFlex's Avatar
  • Offline
  • Fresh Boarder
  • Posts: 13
  • Karma: 0
Maybe you can remove the "added phase" to the more phase turning seat to match the other side instead of using an APF who kill group delay ;)
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27449

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
WaveFlex wrote:
Maybe you can remove the "added phase" to the more phase turning seat to match the other side instead of using an APF who kill group delay ;)

I'm not sure I following what you mean.
Take out the APF, and do what?

Sorry I'm kinda still new at rephasing
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27450

  • WaveFlex
  • WaveFlex's Avatar
  • Offline
  • Fresh Boarder
  • Posts: 13
  • Karma: 0
In rephase, do the inverse of your APF on the speaker of the other side.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27451

  • dreite
  • dreite's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 1397
  • Thank you received: 658
  • Karma: 83
An inverse of an all-pass filter??

Dave.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27452

  • WaveFlex
  • WaveFlex's Avatar
  • Offline
  • Fresh Boarder
  • Posts: 13
  • Karma: 0
It is not what we are doing when we linearize a second order filter ?

In a car, the two side doesn't have the same phase.
When you suppress the amount of phase of one speaker to match the other side in FIR (to have the best stereo summation), it is better for the impulse than matching the phase of a speaker by adding an APF on the "best" speaker.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27453

  • WaveFlex
  • WaveFlex's Avatar
  • Offline
  • Fresh Boarder
  • Posts: 13
  • Karma: 0
A soft like Filtershop can simulate an inverse all pass filter.
But in REW, with the EQ menu you can simulate the good 2nd order APF(s) (Freq/Q) for making the inverse correction in rephase.
To see the phase of the desired APF in rephase, just add the corresponding filters in the minimum phase tab to move the phase like the APF, and then linearize this phase. Suppress the minimum phase filter firstly added and you obtain the inverse correction of the APF to feed the FIR bank of the opposite speaker to suppress the amount of phase compared of the other side.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27454

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
WaveFlex wrote:
It is not what we are doing when we linearize a second order filter ?

In a car, the two side doesn't have the same phase.
When you suppress the amount of phase of one speaker to match the other side in FIR (to have the best stereo summation), it is better for the impulse than matching the phase of a speaker by adding an APF on the "best" speaker.

Okay, so ditch the APF right ?; than make a slider in phase Eq and sim the APF in phase Eq and apply the inverse of that?

What difference does it make though besides that I will come back up to 0phase where a APF eventually leads to -180 when I have the crossover right in front of the APF center frequency should it not matter

The speaker with APF plays 40-280hz , there 10" beyma 10G40s under the back seat firing forward. The passenger side I used a 250hz 2nd order APF with a Q of 1. It really worked well I think. But the midrange in dash which is a 6ND430 plays 280-630 and no APF just linearized crossovers.
The 6NDs sound better with IIR LR24 when I linearize it it causes problems between the midbass and mid. I also used Delay to offset the distance between the front mids and rear midbass. I am not using any delay between left and right because I want it to have good imaging in both seats of the car, if I use delay I can make one seat sound very good at the cost of stage width and ambiance. So imaging is good in both seats and a happy medium for two listeners.

There's also a dip at 250 caused by driver interaction and that dip is why I have been using IIR on the HP of the midrange. Seems like the delay caused by the filter fixes my null. I would like to fix it another way if I can. And that's basically it. I know it's a bit overwhelming being in a car but I'm almost to nirvana :woohoo:
Last Edit: 3 years 11 months ago by denver8me4dinner.
The administrator has disabled public write access.

Lin phase Eq 3 years 11 months ago #27456

  • denver8me4dinner
  • denver8me4dinner's Avatar
  • Offline
  • Gold Boarder
  • 3038564518
  • Posts: 261
  • Thank you received: 52
  • Karma: 6
Okay I treys it . It does work better with no APF

Thanks again Pierre .

I think I got it as good as it's going to be with two sets speakers coming from different locations .

Phase looks flat ish now both sides ,
Before had some hard notches in midbass

I was only able to get the phase Eq to go so far because of run out taps And tried every window found one that looked decent , think it was a bartlett-hahn window. Don't know if using different Windows across channels is bad but it is a bit better
Last Edit: 3 years 11 months ago by denver8me4dinner.
The administrator has disabled public write access.
  • Page:
  • 1
  • 2
Moderators: devteam