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

TOPIC: How to find out my active speaker's Xover details?

How to find out my active speaker's Xover details? 6 years 5 months ago #12811

  • phranky
  • phranky's Avatar
  • Offline
  • Junior Boarder
  • Posts: 25
  • Karma: 0
Oh sorry - now I finally got, that the extreme phase shifts were caused by my phase range setting. With -180/180 it looks like on your screenshot.
--> So all I have to do now, is to flatten the FR with the Paragraphic Gain EQ and do some finetuning with the Paragraphic Phase EQ, right? And then generate the impulse to import it to the FIR on my OpenDRC, yes?

But just for understanding, to do the correct settings also for the right channel:
- Which measurement value defines the -36 µsec timeoffset and if "invert polarity" is activated or not?
- How did you amount to the "vented hight Q" box setting at 40 Hz and the subsonic setting "butt 12dB/oct" at 35Hz?
The administrator has disabled public write access.

How to find out my active speaker's Xover details? 6 years 5 months ago #12819

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
phranky wrote:
Oh sorry - now I finally got, that the extreme phase shifts were caused by my phase range setting. With -180/180 it looks like on your screenshot.
Yes, that is a bug I will have to fix.
--> So all I have to do now, is to flatten the FR with the Paragraphic Gain EQ and do some finetuning with the Paragraphic Phase EQ, right? And then generate the impulse to import it to the FIR on my OpenDRC, yes?
Correct, but if you want to do EQs I would suggest taking several measurements at different positions in order to avoid correcting things that only exists at one point in space (diffraction, specular reflexions, ...) and use smoothing/gating/caution (!). I would suggest staying away of high Q corrections unless you are really sure you know what you are doing. In general it is safer to only correct the speaker itself, not the effect of its environment.
In any case, you should do the amplitude corrections first, using phase-minimum EQs, and only then use separate phase EQ (with greate care and even more caution than amplitude EQ).
But just for understanding, to do the correct settings also for the right channel
I would suggest using the *exact* same setting (exact same impulse in fact) for both channels. If ever you need to use different corrections (they must be really close, or something is wrong somewhere), set the centering to "middle" to ensure both corrections will imply the same delay (you should keep the "float" centering option though, as its impact on delay is negligible), or be prepared to compensate any difference within the openDRC...
- Which measurement value defines the -36 µsec timeoffset and if "invert polarity" is activated or not?
This one is not easy to answer...
With your speakers polarity will always be reversed, but time offset might have to be reset from one measurement to another, depending on the way REW does its impulse centering.
Let's see how it does, and if necessary I will explain you how to proceed by looking at the impulse (easiest way) or at the phase frequency response.
- How did you amount to the "vented hight Q" box setting at 40 Hz and the subsonic setting "butt 12dB/oct" at 35Hz?
Partly guess, partly trial and error.
https//wavetracing.com | rephase.org
Last Edit: 6 years 5 months ago by pos.
The administrator has disabled public write access.
The following user(s) said Thank You: phranky

How to find out my active speaker's Xover details? 6 years 5 months ago #12845

  • phranky
  • phranky's Avatar
  • Offline
  • Junior Boarder
  • Posts: 25
  • Karma: 0
Thank you so much for taking time to answer my questions, pos! You really help a lot...
pos wrote:
In general it is safer to only correct the speaker itself, not the effect of its environment.
Why that? Actually all I want is to clear the room's effect.
pos wrote:
In any case, you should do the amplitude corrections first, using phase-minimum EQs, and only then use separate phase EQ (with greate care and even more caution than amplitude EQ).
Okay, let me guess the reason: because of the "pre-ringing" of the FIRs? I just tested it with some extreme linear-phase EQ settings (just to hear the difference for the first time) and it sounds horrible. So minimum-phase EQs won't introduce the pre-ringing?
pos wrote:
I would suggest using the *exact* same setting (exact same impulse in fact) for both channels.
Hmmm.. My speakers have pretty different responses, especially in the bass area. So why not correct them individually?

Just to explain, if it matters: I don't have a real dedicated studio, I'm about to achieve a passably linear sound in my home studio for music creation and mixing (not mastering). I have a pretty symmetric setting with wideband absorbers in front & back end and on the ceiling. But I'm not able to install basstraps or superchunks here. And on the left side there are shelves, while I have 2 windows on the right side. I know, that's not ideal.. just want to optimize it as far as possible with room EQing.
The administrator has disabled public write access.

How to find out my active speaker's Xover details? 6 years 5 months ago #12925

  • phranky
  • phranky's Avatar
  • Offline
  • Junior Boarder
  • Posts: 25
  • Karma: 0
Just to show you my differences in the bass and why I think, there're separate EQs on the left and right channel necessary:
Attachments:
The administrator has disabled public write access.

How to find out my active speaker's Xover details? 6 years 5 months ago #12969

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hello Phranky, sorry for the delay.
phranky wrote:
pos wrote:
In general it is safer to only correct the speaker itself, not the effect of its environment.
Why that? Actually all I want is to clear the room's effect.
Just to explain, if it matters: I don't have a real dedicated studio, I'm about to achieve a passably linear sound in my home studio for music creation and mixing (not mastering). I have a pretty symmetric setting with wideband absorbers in front & back end and on the ceiling. But I'm not able to install basstraps or superchunks here. And on the left side there are shelves, while I have 2 windows on the right side. I know, that's not ideal.. just want to optimize it as far as possible with room EQing.
This is a complicated matter. The danger with "room" correction is that if you can correct things that are only valid at one location (where the measurement was taken), such as diffractions, specular reflections, room modes buildups or suckouts (*never* compensate modal suckouts!), etc.
Also, you should try to only compensate the direct sound (and close reflexions if you cannot avoid them, say up to a few ms), not the power response. This is probably different in the bass where our "integration window" is longer.
Short answer is you should never use a single measurement for that kind of correction. You should either do multiple measurement at different positions inside your the listening window and go through them one by one in rephase to see the effect of your correction (using drag and drop), or better still do an averaging of said measurements (I think REW can do that). Also use gating and smoothing. All these techniques have their merits and should be used together, with multiple measurement and analysis....
In my opinion the only differences in corrections between channels should be there to compensate for driver or passive filters differences, or in the bass if you have symmetry issues with your room or placement. Above a few hundreds of Hz well placed passive solutions like the panels you are using should be able to take care of any symmetry issue.

If what you are looking for is room correction you should probably check a dedicated software like Align2 (see the dedicated section on this forum), that will take all the (hopefully good) decisions for you.
Jean-Luc is an expert in these things B)

pos wrote:
In any case, you should do the amplitude corrections first, using phase-minimum EQs, and only then use separate phase EQ (with greate care and even more caution than amplitude EQ).
Okay, let me guess the reason: because of the "pre-ringing" of the FIRs? I just tested it with some extreme linear-phase EQ settings (just to hear the difference for the first time) and it sounds horrible. So minimum-phase EQs won't introduce the pre-ringing?
If you use minimum-phase EQs on a minimum-phase system (which a loudspeaker driver is), flattening the amplitude does also flatten the phase. Minimum-phase EQ is what you should use 99% of the time.
Phase-only EQ should only be used with for low Q correction, to help with crossover phase linearization.
pos wrote:
I would suggest using the *exact* same setting (exact same impulse in fact) for both channels.
Hmmm.. My speakers have pretty different responses, especially in the bass area. So why not correct them individually?
You mean your room? If your speakers themselves have differences that large, by all mean change them :P . By the way, you should try to measure them at the same location, to see what is position-related, and what is speaker-related.
Anyway, if you do want to apply different corrections for the two channels, please note that in that case letting rePhase do an automated centering based on energy (default) will most probably lead to different implied delays for the two channels (see the indication given by rePhase in samples and ms after generation), which *will* sound bad. So in that case you should either use the "middle" centering, enforcing the same (suboptimal) centering on both channels, or compensate for the delay difference in the openDRC.
https//wavetracing.com | rephase.org
Last Edit: 6 years 5 months ago by pos.
The administrator has disabled public write access.
The following user(s) said Thank You: phranky

How to find out my active speaker's Xover details? 6 years 5 months ago #13021

  • phranky
  • phranky's Avatar
  • Offline
  • Junior Boarder
  • Posts: 25
  • Karma: 0
No problem pos, I'm very thankful for so much detailed information!
The danger with "room" correction is that if you can correct things that are only valid at one location (where the measurement was taken), such as diffractions, specular reflections, room modes buildups or suckouts
My head / listening position really just moves in a circle of max. 20cm, but to play safe I'll do three measurements for each speaker/channel with different head locations. Then I'll average them in REW.
--> Is there a way to average the phase of these measurements?
*never* compensate modal suckouts!
Okay, so I'll only reduce gains, but won't boost any :)
Also use gating and smoothing.
Where is gating applied to the measurement in REW and what does it mean?
If you use minimum-phase EQs on a minimum-phase system (which a loudspeaker driver is), flattening the amplitude does also flatten the phase. Minimum-phase EQ is what you should use 99% of the time. Phase-only EQ should only be used with for low Q correction, to help with crossover phase linearization.
So I'll only use minimum-phase filters and then do rough adjustments with the Phase EQ. But won't use linear-phase filters. Thus I won't hear any pre-ringing, right?
if you do want to apply different corrections for the two channels, please note that in that case letting rePhase do an automated centering based on energy (default) will most probably lead to different implied delays for the two channels (see the indication given by rePhase in samples and ms after generation), which *will* sound bad. So in that case you should either use the "middle" centering, enforcing the same (suboptimal) centering on both channels, or compensate for the delay difference in the openDRC.
So is this the right conclusion: I would use the "energy" centering and then add the both channels' delay difference to the less delayed channel in the Open DRC?
Which measurement value defines the -36 µsec timeoffset and if "invert polarity" is activated or not?
if necessary I will explain you how to proceed by looking at the impulse (easiest way) or at the phase frequency response.
When I'm doing new measurements now, I worry that I need to know now, how to find out the correct time offset value :S

I hope I'm not digging to deep, also in your free time!! Hopefully these will be the last points to finally find the right approach for creating a linear reference sound for my music productions. Thank you so much!
The administrator has disabled public write access.

How to find out my active speaker's Xover details? 6 years 5 months ago #13040

  • pos
  • pos's Avatar
  • Offline
  • Platinum Boarder
  • Posts: 350
  • Thank you received: 196
  • Karma: 39
Hi Phranky
phranky wrote:
--> Is there a way to average the phase of these measurements?
No, not with this method at least. You should rely on a single measurement for (gentle and low Q) phase corrections.
phranky wrote:
Where is gating applied to the measurement in REW and what does it mean?
I don't know how you apply it in REW, but it is certainly there somewhere.
Gating is a windowing of the impulse that will attenuate and reject things that happen after a given delay from the main peak.
The idea is to reject reflexions and only keep the main signal:
The closer the mic from the source and the farther the boundaries from the source/mic path, the longer the delay between main signal and 1st reflexion, and the longer the gating can be. And of course the longer the gating the more precision you can have in your measurement, especially in the lows: gating provoke a frequency smoothing phenomenon that varies with frequency.
If you use minimum-phase EQs on a minimum-phase system (which a loudspeaker driver is), flattening the amplitude does also flatten the phase. Minimum-phase EQ is what you should use 99% of the time. Phase-only EQ should only be used with for low Q correction, to help with crossover phase linearization.
So I'll only use minimum-phase filters and then do rough adjustments with the Phase EQ. But won't use linear-phase filters. Thus I won't hear any pre-ringing, right?
Any adjustment will either cause ringing or correct it: if your (minimum-phase) system has a hole or peak in its frequency response, it also has ringing there, and flattening the aberration with minimum-phase EQ will remove that ringing.
Wrong minimum-phase correction will cause post ringing (exactly like if this aberation came from the speaker itself) whereas linear-phase correction will cause symmetrical ringing (acausal, pre and post, and of course pre rining is more harmful as it is less likely to be masked with musical signal). And of course a correction can be right at one position (where you took your measurement), and completely wrong a few centimeters away. The higher the Q the more ringing you get, *and* in the same the more likely your correction is to be wrong at some places.
Whenever you manipulate amplitude and phase independently (ie linear-phase amplitude EQ, or phase-only EQ) you take the risk of having pre ringing if (where) your correction gets wrong.

To prevent problems you should first use minimum-phase EQ (the more confident you are with your (averaged) measurement(s) the more high Q EQs you can do, but beware nonetheless...), apply the correction, then measure your speaker again (with correction engaged, is convolution with the openDRC in the signal path), use some smoothing, and apply phase linearization using the "filter linearization" tab, and only use additional phase EQ if you really feels you need to, always with low Q corrections.
So is this the right conclusion: I would use the "energy" centering and then add the both channels' delay difference to the less delayed channel in the Open DRC?
Yes! Do not worry with fractional samples though...
Which measurement value defines the -36 µsec timeoffset and if "invert polarity" is activated or not?
if necessary I will explain you how to proceed by looking at the impulse (easiest way) or at the phase frequency response.
When I'm doing new measurements now, I worry that I need to know now, how to find out the correct time offset value :S
The easiest way to do this is by looking at the impulse and manually placing the "t=0" marker (don't know how it is called in REW...) at the first peak of the impulse.
If this peak is negative then you should also reverse polarity. When this is done you should see your phase curve gently and asymptotically reaching 0° (or a 360° multiple if looking at an unwrapped curve) at the Nyquist frequency (this means the global LP filter of the system is ignored, but that is what we want).
You should then export your amplitude and phase curves (with or without smoothing) and load the in rephase, and not having to adjust anything there (maybe just amplitude offset...).
Another possibility is to let your measurement software place its t=0 marker automatically and modify it in rephase, as I did with your measurement. Doing so by just looking at the phase curve is a bit tricky and requires some practice (detecting reversed phase, etc.). However chances are that REW will always place its t=0 marker the same way (HOLM tends to do so, setting t=0 on first positive peak), which means it will always remove the same amount of excess phase regardless of the measurement distance, and in that case the "-36 µsec and invert polarity" setting in rephase should work for all your measurements...
Just try it: if several measurement give the same curve up high with these settings (phase down low can vary a lot) then you are good to go.
I hope I'm not digging to deep, also in your free time!! Hopefully these will be the last points to finally find the right approach for creating a linear reference sound for my music productions. Thank you so much!
no problem, do not hesitate to ask further questions.
https//wavetracing.com | rephase.org
The administrator has disabled public write access.
  • Page:
  • 1
  • 2
Moderators: devteam