It is normal to invert polarity of speaker pairs to get crossover point to sum flat. I do this on my midbass drivers.

Ge0

 

But mine is Right channels only... if I switch both R/L inverted, dip.... just only left normal and right inverted to make in phase and linear curve for crossover point.

 

What type and slope of crossovers are you using?

 

Butterworth with -24dB slop at all

 

ya it a normal phenomenon. you will be in phase at the xover. but out pretty much everywhere else. on my phone so i cant post a link but andy does a very good job of explaining it on the audiofrog website.

 

would you send me the links .... please I would learn more thanks so much

 

As far as Left to Right phase is concerned. The most important point here is to maintain a coherent center image. Behavior at crossover points should be secondary. I believe you need to explain your system a little better for us to help you further. For example:

Speakers
Speaker locations
active or passive (or both)
crossover set points
equipment used (example DSP or no DSP)
measurement method
software used.
etc...

Ge0

 

Thank you Galan. That gives us a better picture.

Can you post an image of your response concerns when L&R side midrange speakers are in and out of phase from each other? Screen captures from REW are easy enough to take.

Also, please clarify what you said in your first post. What did you mean "I tune phase angle to made a linear curve for the crossover point around +6dB for R/L channels between MidRange to Bass with Polarity Normal for all of them" What do you mean you tune phase angle? What setting did you use?

Ge0

Sure, I will replicate it tomorrow, capture it and send you then... because I was watching the live measurement during the tuning and not every response I captured and save as record.

For the phase angle, i captured the screen here to shown you what I did

 

Thank you Galan. That gives us a better picture.

Can you post an image of your response concerns when L&R side midrange speakers are in and out of phase from each other? Screen captures from REW are easy enough to take.

Also, please clarify what you said in your first post. What did you mean "I tune phase angle to made a linear curve for the crossover point around +6dB for R/L channels between MidRange to Bass with Polarity Normal for all of them" What do you mean you tune phase angle? What setting did you use?

Ge0

 

OK, so you are just adjusting the phase angle of your sub. This shouldn't impact midrange behavior. At least I think...

If you turn on only your midrange channels C and D. Do you get better center coherence with them in phase with each other or out of phase with each other?

Also, 250Hz seems a little low to cross over a 3" midrange. It's Fs is close to 200Hz. The pod might be influencing midrange frequency response as well. Maybe your dip is not being caused by a phase issue but rather a low frequency roll-off issue. Something to think about. Try changing midbass to midrange crossover to Linkwitz Riley 24dB/octave and moving it up to 400Hz. If for anything just to see what happens.

Maybe the boost at the midrange crossover point with Right side speaker polarity inverted is due to constructive interference between the midrange drivers at the lower frequencies of the drivers pass band. Thus, masking an issue with low frequency roll off. However, this may also adversely affect your staging in the important midrange spectrum.

We are just guessing and trying experiments at this time. But, we will help you find a solution.

Ge0

 

OK, so you are just adjusting the phase angle of your sub. This shouldn't impact midrange behavior. At least I think...

If you turn on only your midrange channels C and D. Do you get better center coherence with them in phase with each other or out of phase with each other?

Also, 250Hz seems a little low to cross over a 3" midrange. It's Fs is close to 200Hz. The pod might be influencing midrange frequency response as well. Maybe your dip is not being caused by a phase issue but rather a low frequency roll-off issue. Something to think about. Try changing midbass to midrange crossover to Linkwitz Riley 24dB/octave and moving it up to 400Hz. If for anything just to see what happens.

Maybe the boost at the midrange crossover point with Right side speaker polarity inverted is due to constructive interference between the midrange drivers at the lower frequencies of the drivers pass band. Thus, masking an issue with low frequency roll off. However, this may also adversely affect your staging in the important midrange spectrum.

We are just guessing and trying experiments at this time. But, we will help you find a solution.

Ge0

First of all, thanks you very much stay with me... much appreciated
May be typo/bad memory, the inverted polarity is Left channel instead of Right, sorry!!!
Again, I did few approaches already, and I mentioned one of them above, I will reveal all to you now as I'm afraid no one interesting to read for a page of message at beginning.
Mine is RHD right-hand traffic.

You left me few questions and I would answer here one by one.
1- the screen capture is just shown you what you question me about the angle only. it's deployed some angle for most of the driver... I will disclosed detail after i answer to you question.

2- It hasn't been comparing each other as very time consuming in making linear curve and I spent all of time to try with (polarity; angle and allpass 2nd order) to seek any combination get better linear curve or which is going worse. But I will test once I got both setting (all normal vs L-Inverted + R-Normal) with a close linear EQ, and perform analog test by my ear with some vocal solo song, i think that it would be better to discriminate which one is better for center coherence. But in theory, if a way to get linear curve with less EQ, in both individual channel or stereo also, this should be better center coherence, right?

2- For the FS of Midrange is 188Hz from spec. I also measured the Midrange with Bypass xover, it's OK can be down to 200Hz for sure.

For your suggestion, HP from 250 >> 400 Hz for Midrange with Linkwitz Riley 24dB/octave, I will test and get back you then. For this test, do i need to play with bass for xover performance as well? if yes, do I need to change "Linkwitz Riley 24dB/octave" LP to Bass ?

OK, I can categorize 2 approaches I did so far, when I got dip at xover point during four drivers playing together (of course, it's completed the xover by mono channel without issue and they are in pair with Normal Polarity (The phase angle is 0 degree to all of them. I haven't touched for this part at the moment)

The first, earlier one is,
I just changed 'inverted' polarity to left (not right) of midrange and resolve the problem of dip at xover point. It dip again at xover point, if I changed inverted polarity to pair of midrange.
This action helps for xover point in pair, but made the linear curve to dip and peak much for total frequency of Midrange right now. Again, I got different measurement result during few days of testing with same EQ same setting for just this Midrange performance. Thus, i try another way out...

The second, later one (currently) is,
I reset the polarity value same for pair of Bass and Midrange, I touch with phase angle now, try to just move the phase angle for both bass + Midrange (mono) to make a linear curve at xover point, and play them all then. So far, it's stable and I deployed it to analog test, the center coherence is OK, bass and treble on the dashboard, of course, I would get few more options to compare each other. Here is the setting for four drivers below

MidRange HP: 262Hz LP: 3657Hz both Butterworth -24dB/octave
Right Ch - Polarity Inverted: 163.125 degree
Left Ch - Polarity Inverted: 0 degree

Bass HP: 85Hz LP: 240Hz both Butterworth -24dB/octave
Right & Left Ch are Polarity Normal and 28.125 degree

I captured you three screen shot for the measurement with above setting.

I would repeat my question, my case is it Normal? my approach for correction is appropriate? Of course, the aim is getting center coherence.

 

I can't tell you if your approach above is normal. I've never experimented with shifting phase angles of my midbass or midrange drivers since I have no way to measure what it is actually doing to overall system phase. But, it seems to work for you.

I'm still curious about why inverting phase of one of your midranges makes the problem go away. Did you try the test I asked you to? Turn off all speakers except for your midranges. Listen to a track with someone speaking. Does the persons voice seem more clear and centered with the two speakers in phase (0,0) or out of phase (180,0) with each other?

Ge0

 

Refer to your suggestion, I have compared it with current preset:
MidRange HP: 262Hz LP: 3657Hz both Butterworth -24dB/octave
Right Ch - Polarity Inverted: 163.125 degree
Left Ch - Polarity Inverted: 0 degree

To mute all of drivers except the midrange, play with a solo vocal and a testing file (with in phase out phase) to compare current setting (0,163.125) ; another (0,0); another other (0,180) (180,0) . I consist my current setting is a very little (very little) better than (0,0) or (0,180) (180,0), they are very closed to each other. What a bit different much easy to differentiate is during I was sliding the bar to switching the angle.

This is a test with created another preset to my DSP. with all xcover with "Linkwitz Riley -24dB/octave" also followed your suggestion for Bass - LP extended to 364Hz and Midrange - HP to 380Hz. (becoz the bass happening big dip around 246Hz, must roll-off at around 350Hz)

For the xover point with is still dip as picture shown, this is Left channel.
I tried with slide the phase angle with this kind of slop, it's not working. No matter way is helpful for the dip at xover point. But I have tested Right Channel with "jtrosky" as suggested allpast function, it looks work. I will post the detail as replied "jtrosky"

 

Keep in mind that adjusting the phase of a "set" (L/R pair) of drivers may mess up the phase at the crossover between two "different" drivers (midbass and mids, for example) - and vice-versa. The goal is get proper summation for both L+R pairs and at the crossovers.

Personally, what I would do is try switching to linkwitz-riley 24dB acoustic crossovers on all speakers, set all speakers to "normal" polarity and reset what you are calling the "phase angle" back to "0" for all drivers. Then if you have any dps in the combined L+R response of any driver "set" or improper summation at any crossovers between two "different" drivers (midbass and mids, for example), try addressing them with allpass filters.

Allpass filters are awesome and something that I just recently learned how to use. They solved so many phase-related issues that I had with speaker pairs and also fixed poor summation at crossovers. Not only do measurements look 100% better, but the system sounds more "cohesive" when all speakers are playing.

I'll take some measurements of my current system both with and without the allpass filters enabled to show the difference that they can make. If I'm understanding your issue properly, I think an allpass filter or two may be the solution you are looking for.

From what I remember reading, I think that the "Phase" slider in the Helix is basically just a super-fine time-alignment adjustment - so it adjust the phase for the entire frequency response for that channel - unlike an Allpass filter, which adjusts the phase for a very small portion of the frequency response.

 

Keep in mind that adjusting the phase of a "set" (L/R pair) of drivers may mess up the phase at the crossover between two "different" drivers (midbass and mids, for example) - and vice-versa. The goal is get proper summation for both L+R pairs and at the crossovers.

Personally, what I would do is try switching to linkwitz-riley 24dB acoustic crossovers on all speakers, set all speakers to "normal" polarity and reset what you are calling the "phase angle" back to "0" for all drivers. Then if you have any dps in the combined L+R response of any driver "set" or improper summation at any crossovers between two "different" drivers (midbass and mids, for example), try addressing them with allpass filters.

Allpass filters are awesome and something that I just recently learned how to use. They solved so many phase-related issues that I had with speaker pairs and also fixed poor summation at crossovers. Not only do measurements look 100% better, but the system sounds more "cohesive" when all speakers are playing.

I'll take some measurements of my current system both with and without the allpass filters enabled to show the difference that they can make. If I'm understanding your issue properly, I think an allpass filter or two may be the solution you are looking for.

From what I remember reading, I think that the "Phase" slider in the Helix is basically just a super-fine time-alignment adjustment - so it adjust the phase for the entire frequency response for that channel - unlike an Allpass filter, which adjusts the phase for a very small portion of the frequency response.

thanks you for your words... and I would say "it works"... but I have few questions that I don't know how to control it... the allpast filter.

I would show you the first picture with right channel of bass + midrange, also the dip at xover point for polarity normal / inverted combination setting as shown blue and red lines.

So, the second picture shown what I choose for red line (midrange is inverted polarity) and the dip is around 502Hz, and I select one band of midrange driver and enter the 502 at allpass filter, it goes worse...

Next, I tried with normal polarity for both bass and midrange, and setting the allpast filter to 345Hz, it's ALL UP. As shown you below, there are 3 purple lines which I captured with Q5; Q3 and Q1 of allpass filter to 345Hz, the summation is very closed to each other, unless, I adjusted the 345 Hz with -6dB at parametic EQ (the orange curve)

So my question is, the Q is not impact to how much raise right?

 

I am not an expert on allpass filters, I'm pretty new to them as well. Make sure you are using 2nd Order Allpass filters though. Personally, I haven't yet figured out when 1st order allpass filters are usefull. I do know that 1st order allpass filters seem to affect areas outside of the freuqnecy/Q combination that you specify though. I've had pretty good luck using 2nd order allpass filters, setting the frequency to the "center" of the dip you are looking to correct and setting the Q to the approximate "width" of the dip that you are looking to correct.

In your case, I'd probably start with a 2nd order allpass filter at around 400hz with a Q of 2 or 3. But I don't really have any "scientific" way to figuring out those values. I've just been starting with what I think looks right and then adjusting as needed until I got the results I was looking for. You may have to use more than one allpass filter at different locations to get things right as well.

Also, once you get the crossover area summation fixed up, you need to make sure you check the L+R combined speaker pair summation again as well - becuase an allpas filter for the crossover area may have caused issues with the L+R combined summation! I've found that it takes a little time to get the right combination of allpass filters to where crossver and L+R combined summation all work out. You may not get it perfect either, but they definitely help. You can also experiment with putting the allpass filter on different drivers. For exmaple, in my case, if I couldn't get the results I wanted by putting the allpass fitler on the left midbass, I would try putting it on the left dash speaker instead (around the same freq). Sometimes, it works out better on one driver than the other.

Again, since I'm not very experienced with allpass filters yet, I've had to spend a lot of time experimenting with different combinations. I probably have a total of 10 allpass filters defined right now between my door midbass and my dash widebands. I needed this many to get everything summing properly at both the crossovers and the L+R combined all at the same time. No idea if 10 allpass filters is considered "excessive" or not, but it seems ot work in my case without any "cons".

 

I agree that the response looks strange after the allpass filter - it seems to have boosted things way more than I would have expected... Did you use 1st or 2nd-order allpass filters?

As a "for example", here are some measurements that I took today - the graph includes individual speakers, L+R combined, Left Side, Right Side and All.

 

The bigger issue here is you are using an all pass filter in a crossover region to fix a big phase issue in the same place... this is a sure fire way to mess with the imaging and mess up the stage

a phase issue is when the sound from the driver also has a reflection from the width or length of the car which is then dominant, the best way to deal with this is move the crossover up on the midbass if the phase issue is with the mid, or down if it’s with the midbass, basically adjust the crossover to avoid the phase issue, that is the best way to deal with this situation

phase issues in crossover regions will negatively effect the blend between drivers also, the brain will also potentially hear two locations for the sound at that frequency, the mic doesn’t pick up everything we hear as it’s a point source and is simple, the brain is very clever, you may not need to do anything other than a little crossover adjustment to fix this issue

 

thank you for join... I thought the point you raised as well... but stop it without execute... as the xcover point setup will be different L & R.... so it's OK? It's because that the dip at crossover point just happening to one side only...

 

I don't know, I used some allpass filters in the crossover region and it worked very well. In fact, I learned how to do it from one of Nicks video on the very subject (using an allpass filter in the crossover region to fix poor summation in the crossover region)...

I have had absolutely no ill side effects from doing so - it made things better, not worse. In my case, I had similar summation-related issues at the crossover region even when using multiple different crossover values (400hz or 500hz, for example).

How-To: Verify Phase Cohesion Through A Crossover

 

I don't know, I used some allpass filters in the crossover region and it worked very well. In fact, I learned how to do it from one of Nicks video on the very subject (using an allpass filter in the crossover region to fix poor summation in the crossover region)...

I have had absolutely no ill side effects from doing so - it made things better, not worse. In my case, I had similar summation-related issues at the crossover region even when using multiple different crossover values (400hz or 500hz, for example).

How-To: Verify Phase Cohesion Through A Crossover

putting a band aid on something Is not a good solution if you can avoid the phase issue in the first place, whether you can hear it or not you now have two lots of sound still arriving at the listening position at the same time, if you can adjust the crossover by 50hz and not need the all pass then that’s a holy better solution

Just because you can do something doesn’t mean it’s the right thing to do in every situation 👍🏼
Also please note that video and that method just gets phase cohesion through a crossover... it doesn’t mean that the phase elsewhere is correct and matching, for example if time alignment isn’t correct etc, I can get good phase cohesion with incorrect time alignment settings for example... it’s not right, but checking it so simply can be misleading without knowing the bigger picture

 

thanks you all support and advise. you all are pro and willing to share experience. of course, you all have a lot of experience and knowledge than me, anyone who seeking advise from here...

I would say, the solution can be more than one and you guys got different approach as you handle with different environment, equipment and personal character, preference as well.

I'm going to finish the 2nd preset with you guys suggested, with allpass filter (2nd order. To: jtrosky, I did 2nd order allpass filter for the picture I sent above) and I will compare with what I did with different phase angle.

Also, I will try with switching the xover point instead of allpass filter deployment, but I'm doubt whether 50 Hz is workable solution to me... let's see. the bass driver is max to around 380Hz ....

 

ok, I have compared with analog test for my previous with different phase angle for midrange against a newly one with 0 degree of phase angle to entire drivers with xcover type: Linkwitz -24dB/Oct.

The result is "the newly one" win....

 

Hey - that looks pretty good! How does it sound?

 

I played 3 songs, and comparing 2 presets with every song. I pick mine original and new one then. The new one is a bit better center coherence and sound field a bit wider...

But I also tried with allpass filter (2nd order) but it's not work to me. You can see below three curves as measured one by one, blue is first one with 452Hz HP of Midrange to Bass, a little bit at 450Hz. the orange curve is allpass filter 2nd order applied to 450Hz Q3, the dip switch to 388Hz. the red one is allpass filter 2nd order applied to 388Hz Q3... never ending...