Tweeter's inverted phase.

[MIsha]

Hello.
I kindly ask the respected community to help me understand what has happened and to recommend possible solutions.

I would prefer not to name the loudspeaker system in question. I don’t want to cast a shadow on the “sun.” However, the problem is specifically a lack of clarity and brightness in the sound.

So, the room is 5.4 × 4 m with a 2.5 m ceiling. The speakers are placed symmetrically along the long wall, about 2.7 m apart.

As I mentioned, I am missing sparkle, clarity, and energy in the high frequencies. Cymbals and similar instruments are formally present, but in reality, they are barely there. Everything else I like very much. If you look at the measurement graphs provided, you won’t find any issues in the HF region. On the graph everything looks perfect — but in real life it is not.

No EQ adjustment in the Wiim Ultra has produced any effect. The dull, grainy hiss just becomes louder. An attempt to add a boosting network in the HF crossover gave me +4 dB above 8 kHz, but this also didn’t deliver the expected result.

I then dared to reverse the tweeter polarity. As expected, this produced a dip at the crossover point. Surprisingly, this had no negative impact on the sonic image. On the contrary — soundstage depth appeared, a lot of “dirt” disappeared, and real metallic tone began to come through in cymbals. Now I can actually listen and hear them.

Unfortunately, the result is still far from what I want. That is why I am asking knowledgeable people for advice and explanation.
Thank you.

 

[kyuu]

I mean, the most obvious explanation is that you are used to a tilted-up treble response and the relatively neutral one in your measurements sounds "dull" to you. Actually, there's a bit of a fall-off after 10kHz, although if those are far-field measurements it seems like they might already be a bit bright do to not trending down as frequency increases and the overall V-shape.

If you have the ability, I would try just applying a high-shelf boost of a couple or three dB and see if that works for you.

If there's something actually wrong in the measurements, I'll leave that for others better qualified.

 

[Penelinfi]

Well, you haven't actually stated what you want, so how will you know if you've achieved what you want?

Perhaps the issue lies elsewhere than some single frequency measurement.

Could be the type of drivers, or the layout used, even the room. The dispersion. Maybe they are using a full range woofer (eg Zu) or not very well filtered one (some B&W) and it's contributing to the so called grainy treble.

I feel like I come across a lot of people asking questions these days with one single piece of information and a whole heap of "don't want to or can't do this and that"

 

[DVDdoug]

It looks like there is a broad crossover range (not steep slopes making a "notch") and you've approximated the famous smiley face EQ with cut-mids or relatively-boosted lows & highs.

That may make the sound more pleasing and you can do that if you want, but after awhile some recordings may not "sound right" and it's probably not a good idea to "permanently" EQ that way.

On the contrary — soundstage depth appeared,

Did you invert the tweeter in both speakers? If not, left & right being out-of-phase that can create a "widening" effect which might be "fun" but it's not an "accurate" image. (If you invert one woofer or the whole speaker, the bass will mostly cancel.)

Twitter's inverted phase

Tweeter's.

 

[Arindal]

As I mentioned, I am missing sparkle, clarity, and energy in the high frequencies. Cymbals and similar instruments are formally present, but in reality, they are barely there. Everything else I like very much. If you look at the measurement graphs provided, you won’t find any issues in the HF region. On the graph everything looks perfect — but in real life it is not.

I am not particularly sure what you mean by ´grainy hiss´ or ´dirt´, but the whole situation sounds to me as a problem with the speakers interacting with the room in an unwanted manner.

You mentioned a base of 2.7m between the speakers, Assuming you use an equidistant triangle, this means a similar listening distance which might be way too far for many speakers, resulting in a dominance of reflections in the room and dependence of the actual sound quality on the speaker´s off-axis behavior (which might be poor).

To verify that, I would recommend to drastically reduce the size of the stereo triangle and listen in a true nearfield setup with the listener like 1 or 1.2m away from the speakers. If you experience clarity and transparency in this case, you can be pretty sure that in your setup the room reflections pose the main problem.

I then dared to reverse the tweeter polarity. As expected, this produced a dip at the crossover point. Surprisingly, this had no negative impact on the sonic image. On the contrary — soundstage depth appeared, a lot of “dirt” disappeared, and real metallic tone began to come through in cymbals. Now I can actually listen and hear them.

That is indicative of a massive problem in the transitional band between midrange and tweeter. It would be helpful to know which speakers you are using, or at least know some details about driver/baffle geometry, crossover frequency and topology.

 

[Sokel]

Take another set of measurements at a meter or so (more if its a big-ish 3-way) but this time go to IR Windows tab and tick the MTW box.
This will rule out some of the room for highs and we will get a clearer picture.

 

[Heinrich]

No EQ adjustment in the Wiim Ultra has produced any effect. The dull, grainy hiss just becomes louder. An attempt to add a boosting network in the HF crossover gave me +4 dB above 8 kHz, but this also didn’t deliver the expected result.

I then dared to reverse the tweeter polarity. ...

First of all, the cymbals and such won't ever sound as crisp as they do in real life, esp not with humble two-way household speakers. Even if those were mothers's finest. They get tamed on the recording with limiters, or added reverb, and available playback volumes won't be real either.

Your measurements show a dip in a certain range that is actually connected to the impression of powerful treble. 2kHz namely - all the brilliance ist not situated as high as many people think. In order to keep on guessing you could elevate the level the with a broad e/q setting Q=0.7, +4dB 2.3kHz.

Additionally you might distinguish between near-field measurement and far field. Your report and plots indicate, that what we see is something in between.

ps: second is reverted plus shelf +4dB for upper treble, right? As I said, not to teach you, but a more systematic approach ...

 

[MIsha]

I thank everyone who replied.

First of all, I don’t want to create any intrigue. But I do feel a serious conflict between my own opinion regarding the loudspeaker and the opinion of many experienced people here. I also feel a conflict with the measurements and interpretations of the measurements made by respected specialists. Therefore, at least for now, I will stay with my own position.

That said, we are talking about a modern three-way loudspeaker system from the premium segment. I have attached the crossover schematic. The design includes a dome tweeter with a very specific character and an even more unusual midrange driver. No, it was not made during a full moon out of rice paper. Quite the opposite – it is engineering in the best sense.
It is a magnificent piece of work and the best I have ever heard. Perhaps my expectations are set too high. But so far, I have not been disappointed, and I hope to find the right balance.

The balance issue shows up in the fact that samples of cymbals or triangles sound magnificent. At the listening position, I see almost 107 dB peak SPL, and it is pure pleasure. I know the sound of cymbals very well, and this is exactly it. Yes, I allow for some sound degradation in recordings and understand the recording process. (By the way, please recommend a few exemplary tracks with reference-quality cymbal recordings.)



As I mentioned – equalization has no real effect. That is the problem. A +4 dB HF boost in the crossover had no desired effect at all. Yes, on the graph it looks like a mix of reverse polarity and HF boost. I think the problem lies in the time domain.
Now regarding the “smile” of the equalizer. In this case, it turned out to be something similar to the BBC dip. In the far field, this dip is much less visible on the frequency response.

Yes, both tweeters are wired in reverse.

Nearfield listening was also done, and no fundamental differences were found. Definitely, more attention will be given to room acoustic treatment in the near future, but I am 50+++ years old and I understand the limits of what is possible.

I found MTW box, thanks. Later I will be able to repeat the nearfield and listening-position measurements.

What do I want? I want cymbals to have metal and power, so that they stop sounding like a pencil sketch. I want clarity and energy in the high-frequency range. All of this is possible. And I understand very well where the boundary of realism lies. What I want is close. I am just missing some elusive subtlety.

Finally, I would like to ask for at least a tentative answer to my question: Why did reversing the tweeter achieve something that no equalization could?

 

[Keith_W]

I am not particularly sure what you mean by ´grainy hiss´ or ´dirt´, but the whole situation sounds to me as a problem with the speakers interacting with the room in an unwanted manner.

You mentioned a base of 2.7m between the speakers, Assuming you use an equidistant triangle, this means a similar listening distance which might be way too far for many speakers, resulting in a dominance of reflections in the room and dependence of the actual sound quality on the speaker´s off-axis behavior (which might be poor).

It isn't due to room reflections:

The ETC is excellent. Although having said that, the OP only took measurements from 300Hz - 20kHz so goodness knows what the low frequencies are doing. Here is a more zoomed out view of both measurements:

I presume that the red measurement has the tweeter polarity inverted. OP: you should always sweep the full frequency range from 20Hz to 20kHz. Sometimes, if you have too much bass / midbass, it can make the top end sound muted by comparison. I can't read Cyrillic, but I am guessing one measurement was taken on 18th Apr and the other on 17th Mar ... in other words, they were taken months ago and the OP pulled them out of his collection.

The other interesting thing to see is that despite both measurements being nominally at the same SPL, the THD of the polarity inverted measurement is sky high - up to 10% at 3kHz! The only good explanation is that the two measurements were taken at different SPL's and then normalized in REW afterwards.

The way to check this is to extend the waterfall out to 1000ms and look at the noise floor. Look at that, the measurement on the left (tweeter inverted) is 50dB above the noise floor, and the one on the right is 70dB. This is the opposite of what I would expect - I would expect the measurement with the higher distortion to be the louder measurement. But it isn't. Other possible causes for the very low noise floor on the measurement on the right is a very long sweep time. Sometimes, this can heat up voice coils and produce nonlinear measurements (i.e. an artefact) ... but again, this measurement looks better.

Anyway, I am not sure if the OP took his measurements correctly. Certainly, the two measurements are NOT COMPARABLE because one of them clearly shows a lot of distortion producing the rising treble, and the other one does not. Furthermore, the pattern of reflections in the ETC in both measurements suggests the microphone was moved between measurements.

Please read this thread, and repeat the measurement making sure that (1) the microphone is not moved (2) the measurements are at the same SPL, (3) sweep from 20Hz to 20kHz. If you want us to compare the effect of the tweeter polarity inversion vs. non-inversion, take two sets of measurements please. One with polarity inverted and the other not inverted. Clearly label the graphs so that we know, and we don't have to guess.

 

[Heinrich]

I would prefer not to name the loudspeaker system in question. I don’t want to cast a shadow on the “sun.”

Might be an obstacle.

On the graph everything looks perfect — but in real life it is not.

It doesn't look good in the first, the second represents a defect.

That is why I am asking knowledgeable people for advice and explanation.

The input is a bit sparse.

I thank everyone who replied.

First of all, I don’t want to create any intrigue. But I do feel a serious conflict between my own opinion ... Therefore, at least for now, I will stay with my own position.

o/k?

That said, we are talking about a modern three-way loudspeaker system from the premium segment. I have attached the crossover schematic. The design includes a dome tweeter with a very specific character and an even more unusual midrange driver.

Don't you think that a reverted tweeter shall be a problem then? Regardless how special from what segment it is, but what do I know (from the schematics)? Wilson Audio comes to mind, as they seem to deliberately take advantage of irregular phase in the x/over region. Some pay buckeloads of dollar for the resulting odd sound signature as a speciality. Maybe it's a Wilson?

 

[sceptical1]

Philharmonic BMR?
If you want to be taken seriously tells us what speakers you have.

 

[Soandso]

There are certain speaker manufacturers who deliberately invert their tweeters' phase. They consider it worthwhile for how you hear their speakers playing music.

My purpose in commenting is to mostly point out that the technical considerations related to such tweeter phase inversions can be searched for on-line and readily found in multiple on-line discussions. I never got around to making my own speakers and my own notes of that inversion tactic got misplaced in my last move.

 

[OldHvyMec]

Wiring out of phase is nothing new. JBL has been doing it for years, as has Wilson, and if you look, so have several manufacturers that use multiple drivers. Usually the mids. I personally won't use speakers built that way. JBL actually reverses the neg/pos post so people won't change the way the bass section can be wired.

A clue is if there is a foam added to either side of any of the drivers (usually the tweeter or tweeters) they are wired in phase and need diffusion off axis.

I'll go as far as to say people KILL the way a great driver combination can sound, sacrificing a flatter response where padding the driver works a lot better.
Usually, what is behind the driver is what makes the difference, not in front, but as I said if you wire out of phase you my get a better overall response, but at the expense of better overall sound.

Here is something you might not like to hear. Have your ears checked and add an HF limiter to the walls in the form of simple pleated curtains and you'll be able to enjoy the tweeter in a phase correct wiring. I personally learned that little trick from a guy named Brian Cheney. He NEVER wired drivers out of phase and always used the least number of passive components he possibly could.

BTW, he was never a fan of JBL early monitors or Wilson from their very start. Certain drivers are meant to be wired either way. Monsoon push/pull small planars drivers are one that you can. I've NEVER seen a dome that sounds or works correctly if you do. If that were the case, OEMs wouldn't mark the post neg/pos, would they?

I also want to add this view of speaker wiring, "in phase" is more of a philosophy than a fact. I'm sure many may disagree, but then I won't use their speakers either. LOL

Infinity did a bit of smoothing, wiring out of phase; they also built passive crossovers with component numbers in the hundreds, too. RS2.5 and RS4.5 were a DIYer's nightmare.

Polk was another who did all kinds of weird $hit to smooth a response at the expense of clarity or precision sound. After it is all said and done, they do have to pass a listening test. If a person is prone to HF boil, (like many young people), you can expect some manufacturers to wire out of phase. It can really add to fatigue, but unlike fatigue resulting from sealed rooms and sub/bass bloat. It takes me a day to recover from that nonsense vs 20 minutes from HF boil.

Those ears of yours and a little diffusion in the room may work wonders.

Best of luck.

Regards

 

[Heinrich]

There are certain speaker manufacturers who deliberately invert their tweeters' phase. They consider it worthwhile for how you hear their speakers playing music.

Without a background in speaker (crossover) tech/ it is easy to raise suspicions like feeling the music cut into halves, some part reverted back in time?

Wiring out of phase is nothing new. JBL has been doing it for years, as has Wilson, ... Those ears of yours and a little diffusion in the room may work wonders.

To name the two in one sentence is an art in itself. The post otherwise exemplifies on what can be gained when deliberately not accepting the engineering basics as a given truth. Times I argue w/ engineering dogma myself, but from the contrary 'side'/'camp' as you may put it.

Curious if the o/p would agree or not?

 

[Arindal]

There are certain speaker manufacturers who deliberately invert their tweeters' phase. They consider it worthwhile for how you hear their speakers playing music.

It is not about worthwhile, it is usually used in case the crossover filters apply phase shift in the region of 180deg relative to each other, like it is the case with two 2nd order Butterworth or LR. Instead of going full scale into cancellation on axis, inverting one of the drivers leads to nice summation on axis.

I personally won't use speakers built that way.

So you prefer to have the same phase shift behavior solely caused by the crossover, plus massive cancellation?

De gustibus non est disputandum...

He NEVER wired drivers out of phase and always used the least number of passive components he possibly could.

The latter is a recipe for acoustic disaster. Particularly with modern drivers which tend to be more broadband and designed having steep topologies in mind.

That said, there are cases in which it is not necessary to invert one driver as the relative phase shift caused by the crossover is below 90deg or closer to 360deg. 1st order or 4th filters might fall into this category.

I've NEVER seen a dome that sounds or works correctly if you do

Which means you can hear the absolute acoustic polarity of a midrange driver or tweeter? I doubt that. Which experiment did you do to verify this? And no, just inverting one driver does not count, as the impact on the transitional bands is audible.

If that were the case, OEMs wouldn't mark the post neg/pos, would they?

Yes, they would, as polarity is important for the sound pressure level particularly in transitional bands, where phase shift and inverted polarity can lead to cancellation, lobing and kinked directivity.

That does not mean the absolute acoustic polarity was audible. What you hear is the tonal outcome, i.e. either cancellation or addition. Extreme example: If you construct a line source, and half of drivers has the pins marked in the wrong way, in theory you could end up with hearing (almost) nothing - full cancellation!

 

[MIsha]

In the attachment there are two files. Each file contains measurements with MTW enabled and without it. My listening position (with the equalizer enabled) and one channel at a distance of 1 m (with the equalizer enabled and disabled).


I fully understand the “incorrectness,” but this way I like it much more than with the correct connection. And I absolutely do not understand why. Moreover, some recent changes in the amplification stage have brought me even closer to the desired sound. I would very much like to understand the nature of this phenomenon. Thank you.

 

[Thomas_A]

It is not about worthwhile, it is usually used in case the crossover filters apply phase shift in the region of 180deg relative to each other, like it is the case with two 2nd order Butterworth or LR. Instead of going full scale into cancellation on axis, inverting one of the drivers leads to nice summation on axis.



So you prefer to have the same phase shift behavior solely caused by the crossover, plus massive cancellation?

De gustibus non est disputandum...



The latter is a recipe for acoustic disaster. Particularly with modern drivers which tend to be more broadband and designed having steep topologies in mind.

That said, there are cases in which it is not necessary to invert one driver as the relative phase shift caused by the crossover is below 90deg or closer to 360deg. 1st order or 4th filters might fall into this category.



Which means you can hear the absolute acoustic polarity of a midrange driver or tweeter? I doubt that. Which experiment did you do to verify this? And no, just inverting one driver does not count, as the impact on the transitional bands is audible.



Yes, they would, as polarity is important for the sound pressure level particularly in transitional bands, where phase shift and inverted polarity can lead to cancellation, lobing and kinked directivity.

That does not mean the absolute acoustic polarity was audible. What you hear is the tonal outcome, i.e. either cancellation or addition. Extreme example: If you construct a line source, and half of drivers has the pins marked in the wrong way, in theory you could end up with hearing (almost) nothing - full cancellation!

It’s not really part of this discussion but flipping a signal 180 degrees can be audible, if the waveform is asymmetric. Been demonstrated in several ABX on this site. Which has nothing to do with the crossover phase

 

[Salt]

I thank everyone who replied.

First of all, I don’t want to create any intrigue. But I do feel a serious conflict between my own opinion regarding the loudspeaker and the opinion of many experienced people here. I also feel a conflict with the measurements and interpretations of the measurements made by respected specialists. Therefore, at least for now, I will stay with my own position.

That said, we are talking about a modern three-way loudspeaker system from the premium segment. I have attached the crossover schematic. The design includes a dome tweeter with a very specific character and an even more unusual midrange driver. No, it was not made during a full moon out of rice paper. Quite the opposite – it is engineering in the best sense.
It is a magnificent piece of work and the best I have ever heard. Perhaps my expectations are set too high. But so far, I have not been disappointed, and I hope to find the right balance.

The balance issue shows up in the fact that samples of cymbals or triangles sound magnificent. At the listening position, I see almost 107 dB peak SPL, and it is pure pleasure. I know the sound of cymbals very well, and this is exactly it. Yes, I allow for some sound degradation in recordings and understand the recording process. (By the way, please recommend a few exemplary tracks with reference-quality cymbal recordings.)



As I mentioned – equalization has no real effect. That is the problem. A +4 dB HF boost in the crossover had no desired effect at all. Yes, on the graph it looks like a mix of reverse polarity and HF boost. I think the problem lies in the time domain.
Now regarding the “smile” of the equalizer. In this case, it turned out to be something similar to the BBC dip. In the far field, this dip is much less visible on the frequency response.

Yes, both tweeters are wired in reverse.

Nearfield listening was also done, and no fundamental differences were found. Definitely, more attention will be given to room acoustic treatment in the near future, but I am 50+++ years old and I understand the limits of what is possible.

I found MTW box, thanks. Later I will be able to repeat the nearfield and listening-position measurements.

What do I want? I want cymbals to have metal and power, so that they stop sounding like a pencil sketch. I want clarity and energy in the high-frequency range. All of this is possible. And I understand very well where the boundary of realism lies. What I want is close. I am just missing some elusive subtlety.

Finally, I would like to ask for at least a tentative answer to my question: Why did reversing the tweeter achieve something that no equalization could?

A child, 3 month old, would view the attached image correctly downside up ... ?

 

[Salt]

Moreover, some recent changes in the amplification stage have brought me even closer to the desired sound. I would very much like to understand the nature of this phenomenon. Thank you.

If you really appreciate what you're listening to, regarding any genre you prefer, why not go for it (even if it's not the 'accurate' way)?, but don't ask for absolution in this here community.

 

[OldHvyMec]

If you construct a line source, and half of drivers has the pins marked in the wrong way, in theory you could end up with hearing (almost) nothing - full cancellation!

If I assembled an LS array the FIRST thing I would do, is what I always do, listen to each individual driver and do a simple SPL check. I'm not an engineer, I'm a simple retired master mechanic with common sense. BTW, I've built over 100 pairs of speakers over the last 45+ years. Mainly LS or Hybrid LS.

I did find a whole box of 8" cone drivers wired incorrectly. Polk OEMs, and it wasn't theoretical.

Monsoon Neo 8, BG Neo 8, Strathearns 24" ribbons, and AC ribbons were the staple of my builds. I recently purchased over 60 GSR planars and ribbons for a few prototypes. I haven't started any new builds, but hopefully, pretty soon. Great build quality and reasonable pricing considering BG's crazy price hikes over the last 10 years. AC is even worse.

My choice of drivers is very limited because I know how the drivers I use work and how they sound. I'm not sure if the drivers I use are considered modern, but they were considered consistent by their own standards.

The only cone drivers I've used in the last 30 years were Dayton RS series and Seas on occasion, if I could find a sale in bulk, overstock or going out of business. I love Seas, I just don't care for the price. Considering I've used mainly RS Dayton's, I couldn't find a reason to continue to use Seas.

I've had a few requests, but after comparisons and side-by-side listening sessions, they have always gone for Dayton, especially after the price comparison.

In my youth, I repaired speakers for Infinity, VMPS, and Polk as a sideline and an authorized repair person. I was one of the few who actually did road calls, mainly on the RS1 A/B and the QLS1. I did a few Polk SDA or their larger cabinets. I also did upgrades to the RS1 B, QLS1, and II, adding Infinities servo bass systems.

As I said, I pick and choose what I worked on, and like to repair. I didn't need to understand why I didn't care how a speaker did or didn't sound. I simply didn't work on them. I fixed a few Wilsons, too, that came to a screeching halt. 700.00 USD for a pair of tweeters, and they wanted to pay less than the fuel cost for the service call. 1990 or close.

Simply said I didn't need to educate myself any further than was needed unless I desired to do it. My hands were plenty full from reading tech manuals from Cat, Deutz, Detroit, Cummins, Grey Ford Marine, and half a dozen hydraulic pump, MCV, and RCV manufacturers.

The speaker gig was for the extra money and a way to fund most of my equipment/speaker ideas. I also like doing cabinet work. Anything to get away from 90-weight gear oil and using a cutting torch was preferable after 49 years in the business.

BTW, thanks for the input, I never turn away information even if it's not needed for what I'm doing.

Regards