KEF Q350 Speaker Review

[QMuse]

Remember that Stereophile's off-axis plots are normalized to the on-axis response. (I think that's the wrong way to do it.) So what a dip in their on axis response and a spike at the same band means is the speaker has a small hole in the on axis response. Odd to see it here. Normally it really presents on coaxes and other round waveguides.

It was a joke, I'm aware it is because of normalization.

 

[Victor Affonso Damas]

That distortion plot looks bad. So if I sell my q350s what else can I buy for 400 euros?

That is not very bad when you listen to it, it is all second order distortion which is not bad to the ears. It happens because this type of rubber surround does not have a symmetrical compliance.

 

[amirm]

Quite the opposite. Different speaker designs need different placement to measure and sound optimal. A good reason to question Harman's tests.

If science could so easily be questioned with "I don't think so," we would have no science left!

Room impact on a speaker in low frequencies is immense so obviously, if you don't use good tools like equalization, you will have a lot of work on your hand to optimize bass.

If you are screwing around with placement of speakers due to their flaws in on-axis or off-axis response, then a speaker that doesn't need that will by definition be superior. Why create work for the customer because the designer didn't know what he was doing?

Importantly, if your statement is based on anecdotal testing, then it has no value. You need to bring here controlled testing that demonstrates how you managed to get better sound doing whatever you were doing. It is entirely possible that moving things around did not result in better sound than another speaker that didn't require it.

It is a common retort that bad speakers sound better with better placement. Come back with proper research and then we can discuss the merits.

 

[napilopez]

Yep, you should be - it would be interesting to hear with what theory would @napilopez come with in case he wouldn't notice it so quickly as you did.

You can't fool me! Jk good thing I didn't read your post earlier

Completely agree with all you say

My point is, though, that we just don't seem to have the crucial piece of evidence tying wide horizontal directivity to listener preference.

Now I do think that, looking at a variety of different data as a whole, it is reasonable to speculate that wide horizontal directivity is a factor (independent of dispersion generally) in listener preference. In particular, I believe that we have evidence tying wide horizontal directivity to perceived spaciousness, and some additional evidence correlating spaciousness with listener preference.

However, I don't think we've got those key pieces of evidence that would connect these dots and turn speculation into observation.

Until them, I guess I'm just more reluctant than others here to declare unequivocally that wide-horizontal-directivity designs tend to be more preferred than medium- or narrow-directivity ones.

Totally fair. Part of me feeling more strongly about that then you has to do as well with the research on preference for more sidewall reflections. But at the same time, I'm a very strong believer that directivity is a matter of preference, so just because a majority might prefer it, including myself doesn't mean it should necessarily be a design goal.

Anway, I know we're pretty much on the same page, but for anyone else following along, just wanted to drive home the point of how misleading the ER and ERDI curves can be with regards to horizontal directivity and, consequently, soundstage performance. Keep in mind I'm using the ER Fix discussed in this thread.

Here's the Q Acoustics 3030i's Early Reflections and ERDI curve compared to the Total Horizontal Reflections (also defined in the CTA-2034A, basically the ER without the vertical Ceiling and Floor Bounce) and Horizontal DI, which is the difference between the listening window and Horizontal Reflections. To be clear, AFAIK nobody calculates a horizontal DI curve and it's not in any standard, I just decided to take give it a try since it was a simple alternative.

You can see the ERDI and HRDI give quite different impressions of soundstage performance, imo. The ERDI curve would probably have you think the soundstage performance isn't decent, but the 3K peak and 5K dips are caused purely by vertical artifacts. And as I think most of us know, evidence suggests the primary effect of vertical reflections is to affect timbre/tonality rather than soundstage/imaging.

The same can be seen with the Focal Chora 806, another wide directivity design:

By contrast, here is the Dutch and Dutch 8c, a narrower design which claims to be constant directivity down to 100Hz. I cut off my polar measurements at 200Hz due to limited quasi-anechoic resolution, but still, you can clearly see the constant directivity nature far more in the HR and HRDI curves:

The vertical influence on the ERDI means you don't see the effect of the cardioid horizontal directivity quite as much, but the Horizontal DI is practically perfect -- almost a straight line throughout the measured frequency range. Indeed: constant directivity down to at least 200Hz.

Lastly, a comparison among the three speakers:

Again, the ERDI doesn't tell us all that much about what to expect regarding soundstage performance, but the Horizontal DI makes the differences much more clear.

For comparison purposes it might be useful to use a taller aspect ratio:


While I still prefer looking at detailed SPL plots, and others prefer polar maps, if we had to boil down soundstage performance to a single curve IMO the above approach would be better. ERDI is still more useful for telling us about how EQable a speaker is though.

Anyway, sorry for the tangent. Since directivity always comes up in review threads, I thought I'd talk more about how I go about interpreting it.

 

[amirm]

If I'm not mistaken Amir is only performing a quick tasting session, not a thorough listening assessment.
And there is a chance that his taste doesn't match that of Science. Mine doesn't.

You don't know if your taste matches the science or not, until you participate in the same way the science was developed, i.e. in their controlled double blind tests.

I have done that twice. In both occasions, I voted the same as the science. And so did the majority of people that took the test with me (high-end audio dealers and acousticians). Granted, there were one or two people that voted differently but you don't know if are one of them or not.

When taking the test, the tonality of each speaker hugely stands out. It is that which sets preference in the testing. This is why directivity doesn't place a statistically significant position. It is the low order bit as we say in computer science.

The test of two speakers with identical tonality but different directivity is not there due to difficulty of creating such a test. Until then, the effect is hard to isolate. I know I did not pay attention to it at all in the two tests I took.

 

[amirm]

On a separate note, we tend to talk more about Paper II because it's the one that led the generalized model, but it I think it might be worth looking a bit more closely at Paper I, as it goes into much more detail about the speakers. Though there aren't any names, we can probably figure out more about each speaker with some sleuthing as they were top ranked speakers by consumer reports.

I actually read every word of the first paper and indeed, some key facts come out of it which you hit on. I have been meaning to write an article on it but have not had time.

For now, the testing involved using the same speakers consumer reports picked to test some 20 years ago now. Their spin data shows massive differences between them and hugely varying frequency response.

Two decades later, the industry has learned to incorporate the lessons of that research. So most of the speaker we test are far closer to each other than what they researched. This makes the job of analyzing the results both objective and subjectively very hard. On the later domain, it is fun to read the comments in the paper where the same speaker was both voted bright and dull by different listeners.

To some extent, we are plowing ahead where the research stopped.

 

[QMuse]

I actually read every word of the first paper and indeed, some key facts come out of it which you hit on. I have been meaning to write an article on it but have not had time.

For now, the testing involved using the same speakers consumer reports picked to test some 20 years ago now. Their spin data shows massive differences between them and hugely varying frequency response.

Two decades later, the industry has learned to incorporate the lessons of that research. So most of the speaker we test are far closer to each other than what they researched. This makes the job of analyzing the results both objective and subjectively very hard. On the later domain, it is fun to read the comments in the paper where the same speaker was both voted bright and dull by different listeners.

To some extent, we are plowing ahead where the research stopped.

What is your opinion as to why your listening impressions frequently don't match the preference rating?

 

[73hadd]

WOW! - When searching for information on another topic, I found this, which states that when measurements would indicate a high preference score, but was not realized in listening tests, the culprit is IMD:

https://www.avsforum.com/forum/89-speakers/710918-revel-owners-thread-399.html#post56105466

ETA: I am not sure who the "Harman source" is, that seems to be referenced in the specific IMD comments.

 

[QMuse]

WOW! - When searching for information on another topic, I found this, which states that when measurements would indicate a high preference score, but was not realized in listening tests, the culprit is IMD:

https://www.avsforum.com/forum/89-speakers/710918-revel-owners-thread-399.html#post56105466

ETA: I am not sure who the "Harman source" is, that seems to be referenced in the specific IMD comments.

This part is quite interesting:

"The other factor with the KEF uniQ designs has to do with off-axis directivity. According to their own published spinorama on a current floor stander - I forget the model no. - the off axis output declines at a faster rate than the Revels. It is smooth but downward tilted. That will not flatter the sound in normally reflective rooms. "

 

[napilopez]

WOW! - When searching for information on another topic, I found this, which states that when measurements would indicate a high preference score, but was not realized in listening tests, the culprit is IMD:

https://www.avsforum.com/forum/89-speakers/710918-revel-owners-thread-399.html#post56105466

ETA: I am not sure who the "Harman source" is, that seems to be referenced in the specific IMD comments.

Thanks for this! If that source is reliable, that is a pretty remarkable observation that might validate @amirm 's thoughts. Though at the same time I think many of us know that if two speakers perform well in the important metrics, then other factors come into play, such as distortion.

But it's worth re-iterating the Harman source is mostly talking about KEF's two-way designs.

What's perhaps most insightful about the Harman source's comment in that thread is that it implies IMD may play a more negative role than vertical lobing does in our impressions of a speaker, once again implying people should prioritize horizontal directivity over vertical directivity.

 

[tecnogadget]

Having good and smooth Vertical directivity is really neat, trust me.
You can always listen while sitting on the couch but when you get up all walk by your room it keeps sounding amazing, not a small feat, and quite an anti-vice solution.

Especially good when you have guests and want to be engaged in a conversation standing with a few drinks haha.

 

[napilopez]

This part is quite interesting:

"The other factor with the KEF uniQ designs has to do with off-axis directivity. According to their own published spinorama on a current floor stander - I forget the model no. - the off axis output declines at a faster rate than the Revels. It is smooth but downward tilted. That will not flatter the sound in normally reflective rooms. "

Given the published date of that post (April 2018) and that the R3 series didn't launch until late 2018, they are almost certainly talking about the Reference series speakers. As far as I know, those are the only ones with published spins other than the R series. This is the spin of the Reference 5 prototype in KEF's whitepaper:

Indeed, that slope is quite negative. The listening window also suggests the speaker is tuned somewhat dark.

 

[QMuse]

Thanks for this! If that source is reliable, that is a pretty remarkable observation that might validate @amirm 's thoughts. Though at the same time I think many of us know that if two speakers perform well in the important metrics, then other factors come into play, such as distortion.

But it's worth re-iterating the Harman source is mostly talking about KEF's two-way designs.

What's perhaps most insightful about the Harman source's comment in that thread is that it implies IMD may play a more negative role than vertical lobing does in our impressions of a speaker, once again implying people should prioritize horizontal directivity over vertical directivity.

"the off axis output declines at a faster rate than the Revels"

This is about ER curve, right? Unfortunately boht, horizontal and vertical off-axis output are baked into that one.

 

[napilopez]

Having good and smooth Vertical directivity is really neat, trust me.
You can always listen while sitting on the couch but when you get up all walk by your room it keeps sounding amazing, not a small feat, and quite an anti-vice solution.

I like good vertical directivity as much as anyone, but standing up vs sitting down will almost never be more than 10-15 degrees unless you are very close to the speaker. The vertical issues we normally talk about are at steeper angles.

 

[QMuse]

I like good vertical directivity as much as anyone, but standing up vs sitting down will almost never be more than 10-15 degrees unless you are very close to the speaker. The vertical issues we normally talk about are at steeper angles.

I don't relly care how it sounds when I stand up.

Would you say that early reflections from side walls, ceiling and floor will hit you at app 45 deg angle (more or less)? If that is the case shouldn't that be considered important?

 

[napilopez]

I don't relly care how it sounds when I stand up.

Would you say that early reflections from side walls, ceiling and floor will hit you at app 45 deg angle (more or less)? If that is the case shouldn't that be considered important?

Sorry, I'm not sure I follow. Are you saying that we should prioritize 45 degree angles for reflections?

 

[tuga]

You don't know if your taste matches the science or not, until you participate in the same way the science was developed, i.e. in their controlled double blind tests.

I have done that twice. In both occasions, I voted the same as the science. And so did the majority of people that took the test with me (high-end audio dealers and acousticians). Granted, there were one or two people that voted differently but you don't know if are one of them or not.

When taking the test, the tonality of each speaker hugely stands out. It is that which sets preference in the testing. This is why directivity doesn't place a statistically significant position. It is the low order bit as we say in computer science.

The test of two speakers with identical tonality but different directivity is not there due to difficulty of creating such a test. Until then, the effect is hard to isolate. I know I did not pay attention to it at all in the two tests I took.

I wasn't criticising you assessment method, only highlighting its limitations.

It's good to know that T&O's preference polls focused on tonal balance and not directivity.
In that case it is likely that my preference does indeed match that of the majority (flat on-axis); in terms of in-room balance I find the B&K curve more to my liking.

 

[QMuse]

Sorry, I'm not sure I follow. Are you saying that we should prioritize 45 degree angles for reflections?

Yes. If typical range of angles they are arriving from can be identified wouldn't you say that reflections coming from these angles are more important then the others?

 

[tuga]

This part is quite interesting:

"The other factor with the KEF uniQ designs has to do with off-axis directivity. According to their own published spinorama on a current floor stander - I forget the model no. - the off axis output declines at a faster rate than the Revels. It is smooth but downward tilted. That will not flatter the sound in normally reflective rooms. "

I think that whoever wrote that got it backwards as it is the room that "flatters", not the speakers.
The speaker designer may or may not decide to use flattering reflections by controlling directivity, just like an electronic equipment may or may not choose to spice things up with a touch of 2nd order harmonic distortion (i.e. Ayre).

In my opinion early reflections are not unlike a form of euphonic (or flattering) distortion, pleasing in some ways but not without its downsides.

 

[tuga]

Given the published date of that post (April 2018) and that the R3 series didn't launch until late 2018, they are almost certainly talking about the Reference series speakers. As far as I know, those are the only ones with published spins other than the R series. This is the spin of the Reference 5 prototype in KEF's whitepaper:

View attachment 64691

Indeed, that slope is quite negative. The listening window also suggests the speaker is tuned somewhat dark.

Is there a convention for the tilt of a "dark/bright" speaker?

I tend to find that there are more bright speakers than there are neutral ones and very few are actually dark.
I also think that many 20 year old speakers would be judged "dark" by today's standards; things like "soundstage" and "air" have taken significant (and in my view undeserved) relevance.

Could this be another grey zone, a matter of personal taste?