Loudspeaker Measurement Question?

[speakerman39]

Folks, I have a question regarding loudspeaker measurement. It appears that most designers design their speakers for a flat on and off axis frequency response. I get that, but can someone explain to me why some such as B&W are known to design speakers that do not measure very well. But, yet still sound very good.

For example, below is a 5-measurement average (NRC Canada 2007) of the B&W CM1 S1's that I used to own. As you can see for yourself, they are far from flat. However, in my room they sounded superb to me. Can anyone here explain to me why this is so? If a flat frequency response is the end goal, then why does the CM1 S1's still sound so good to me?

How can I tell by looking at this graph what areas would sound bad? Perhaps, some of what I am seeing here is not audible? How do I differentiate that? What can someone gain by looking at this graph before listening to the CM1 S1's? Just trying to understand measurements better. Any and all help would be greatly appreciated. Thanks!

http://www.soundstagenetwork.com/measurements/speakers/bw_cm1/



Cheers,

Phil

 

[speakerman39]

I see that Sensitivity was measured at 81dB which is very low. It looks like the CM1 S1's would benefit from a more beefy amp, no? Also, does it really make a difference in the frequency response graph if measurements are taken at say 1 m vs. 2 m's? If measurements are taken at the MLP would the graph look nearly the same as compared to 1 m or 2 m's? Just curious. Thanks!


Cheers,

Phil

 

[shadyJ]

Folks, I have a question regarding loudspeaker measurement. It appears that most designers design their speakers for a flat on and off axis frequency response. I get that, but can someone explain to me why some such as B&W are known to design speakers that do not measure very well. But, yet still sound very good.

For example, below is a 5-measurement average (NRC Canada 2007) of the B&W CM1 S1's that I used to own. As you can see for yourself, they are far from flat. However, in my room they sounded superb to me. Can anyone here explain to me why this is so? If a flat frequency response is the end goal, then why does the CM1 S1's still sound so good to me?

How can I tell by looking at this graph what areas would sound bad? Perhaps, some of what I am seeing here is not audible? How do I differentiate that? What can someone gain by looking at this graph before listening to the CM1 S1's? Just trying to understand measurements better. Any and all help would be greatly appreciated. Thanks!

Many speakers are not designed for a purely flat response but instead have a 'voiced' response that the designers think will sound better for various reasons. One possible reason is that a speaker that has hotter treble will sound more detailed, and this can help sell the speaker when listening to it side-by-side vs a more neutral speaker on a sales floor. Same thing goes for a speaker with a bump in the bass, it can sound 'fuller' than a neutral bass response. In fact, that could be what is happening with those B&W speakers. In pop music and rock music this sort of response makes the speaker stand out more, however in other types of recordings suck as orchestral, it can make the performance sound unnatural.

But as for 'good' or 'bad', that is subjective. If you like the voicing of these speakers, than for you, they are good. They might sound perfectly natural to you, but to really understand their character, you need to do an A/B comparison with fast switching against other speakers, preferably neutral ones.

Something else to keep in mind that that some flaws in the response are a lot more audible than others.

 

[shadyJ]

I see that Sensitivity was measured at 81dB which is very low. It looks like the CM1 S1's would benefit from a more beefy amp, no? Also, does it really make a difference in the frequency response graph if measurements are taken at say 1 m vs. 2 m's? If measurements are taken at the MLP would the graph look nearly the same as compared to 1 m or 2 m's?

81dB sensitivity is certainly on the low side, but the requirements for a beefier amp depend on how loud you listen. If these are situated on a desktop, than a lower powered amp will likely be just fine, even with a low sensitivity. For a medium to large room, however, yeah, you will want a bigger amp, since a low-power amp will likely run into clipping pretty easily in a scenario like that.

As for measurements at different distances, it can make a difference, but it depends on the speaker. For a large tower speaker, 1 meter is nowhere near good enough to capture the response, and even 2 meters may still be too close. However, for a small near-field monitor, the curve at 1 meter should pretty much look like the curve at 2 meters, so a 1 meter measurement distance is fine. It depends on the distance that the sound from the drivers are intended to integrate.

 

[speakerman39]

Thanks Shady. The CM1 S1's did sound very good to me on everything I threw at them. My musical tastes are rather eclectic to say the least. But, I did mainly listen to them in the near-field. The one thing that really impressed me was that playing the CM1 S1's had me check and re-check to make sure my sub was NOT on. Found that to be absolutely amazing.

I really liked the overall balance as nothing sounded over pronounced. Did not feel the need to tow them in very much. Both on and off axis sounded very good to me. Of course, my room is rather small albeit 12' X 15' and I do not listen at absurd levels. When compared to my Wharfedale Dentons the mids stood out more on the Dentons. Hence, vocals were more full and warm sounding. But, the low end was easily won by the CM1 S1's.

I hope to learn how to take and interpret loudspeaker measurements in my very near future. Once again, hope to get more familiar w/REW and/or the miniDSP as well. Will be sure to get me a CSL UMik 1 when the times comes. It will be a while as I have a very bumpy road ahead of me. Thanks!


Cheers,

Phil

 

[speakerman39]

Below is the measurement of the BMR's. Talking about a ruler-flat frequency response both on and off axis. As such, the idea is if a loudspeaker measures flat it will need less EQ when put in a typical room. Conversely, a loudspeaker that measures not so flat will need much more work to get it as flat as possible. So, I get that as it makes perfect sense to me. So then, "voicing" results in a non-flat response? That is just merely manipulating the frequency response which seems counter intuitive to me. It sounds like "voicing" can get rather complicated. Much easier to design for a flat frequency response, no?

Cheers,

Phil

 

[Swerd]

What you see in the frequency response of the B&W CM1, where there is a lowered response in the upper mid range, is often called the "BBC dip". According to the lore, it was developed in an earlier British speaker design, and was often imitated by other commercial British speaker companies.

"According to Harbeth's founder, who worked at the BBC during the time that this psychoacoustic effect was being explored, the primary benefit this little dip gave was in masking of defects in the early plastic cone drive units available in the 1960's. A spin-off benefit was that it appeared to move the sound stage backwards away from the studio manager who was sitting rather closer to the speakers in the cramped control room than he would ideally wish for. (See also Designer's Notebook Chapter 7). The depth of this depression was set by 'over-equalisation' in the crossover by about 3dB or so, which is an extreme amount for general home listening. We have never applied this selective dip but have taken care to carefully contour the response right across the frequency spectrum for a correctly balanced sound. Although as numbers, 1kHz and 4kHz sound almost adjacent in an audio spectrum of 20Hz to 20kHz, the way we perceive energy changes at 1kHz or 4kHz has a very different psychoacoustic effect: lifting the 1kHz region adds presence (this is used to good effect in the LS3/5a) to the sound, but the 4kHz region adds 'bite' - a cutting incisiveness which if over-done is very unpleasant and irritating."​

Search Google for "BBC dip". Here are two hits from that search worth reading.

https://hydrogenaud.io/index.php/topic,75195.0.html (I found the paragraph I quoted above in this thread.)

https://forums.audioholics.com/forums/threads/should-speakers-be-designed-to-have-a-flat-spl.86954/
Note that Dennis Murphy has much to say, in this thread, on the subject of flat vs. BBC dip in loudspeakers. Hint, he doesn't like the BBC dip.

 

[shadyJ]

Below is the measurement of the BMR's. Talking about a ruler-flat frequency response both on and off axis. As such, the idea is if a loudspeaker measure flat it will need less EQ when put in a typical room. However, a loudspeaker that measures not so flat will need much more work to get it as flat as possible. So, I get that as it makes perfect sense to me. So then, "voicing" results in a non-flat response? That is just merely manipulating the frequency response which seems counteractive to me. It sounds like "voicing" can get rather complicated. Much easier to design for a flat frequency response, no?

The BMR curve is very good, but it might not appear to be as flat as it seems. The X axis is more than twice as wide as the Y axis, and that will do a lot to smooth out that curve. Also different measurement techniques can produce different results; where was the mic situated in that measurement? Not to say that Dennis is using tricks to get unrealistic results, but you can't really compare different measurement sets unless you know how those measurements were done.

As for voicing, like I said, there are reasons why designers voice the speakers the way they do. Sometimes they have good purpose, and sometimes its just a bad idea.

 

[PENG]

Phil, I would also add that B&W speakers typically have low distortions (relatively speaking). I think that's one reason why they could sound good to the right ears without ruler flat graphs. On the frequency response side, NRC did their measurements in a real anechoic chamber so you will have very different response in your room. I've measured the responses of my KEF speakers in my room. Their FRs (based on NRC's) should be as good as the BMR's, assuming the BMR's were also measured in an anechoic chamber or equivalent. Yet the FRs of both my LS50 and R900 look extremely terrible (relatively)in my room, even when measured at 1 meter. To me, the going by the ear thing sounds good, but I prefer to go with my eyes more (i.e. measurements), pick out the good ones (distortions), then use my ears, or go by ears, then eyes, always both.

 

[Winkleswizard]

Folks, I have a question regarding loudspeaker measurement. It appears that most designers design their speakers for a flat on and off axis frequency response. I get that, but can someone explain to me why some such as B&W are known to design speakers that do not measure very well. But, yet still sound very good.

For example, below is a 5-measurement average (NRC Canada 2007) of the B&W CM1 S1's that I used to own. As you can see for yourself, they are far from flat. However, in my room they sounded superb to me. Can anyone here explain to me why this is so? If a flat frequency response is the end goal, then why does the CM1 S1's still sound so good to me?

How can I tell by looking at this graph what areas would sound bad? Perhaps, some of what I am seeing here is not audible? How do I differentiate that? What can someone gain by looking at this graph before listening to the CM1 S1's? Just trying to understand measurements better. Any and all help would be greatly appreciated. Thanks!

http://www.soundstagenetwork.com/measurements/speakers/bw_cm1/



Cheers,

Phil

Have had comparable experiences but have not been able to do even a reasonable side by side comparision. Have sometimes wondered how changes in hearing with age affect as well.

If you have auto-eq, do you like the sound better unequalized or equalized?

As Swerd mentions... (retraction on my part). My recollection on the BBC dip was incorrect. After rereading Swerd’s post and some of the references, I updated my original post. Linkwitz does mention something about micing techniques that I thought was the reason for the BBC dip. Please excuse the dust.

Hoping all goes well with your surgery!

Ww

 

[speakerman39]

What you see in the frequency response of the B&W CM1, where there is a lowered response in the upper mid range, is often called the "BBC dip". According to the lore, it was developed in an earlier British speaker design, and was often imitated by other commercial British speaker companies.

"According to Harbeth's founder, who worked at the BBC during the time that this psychoacoustic effect was being explored, the primary benefit this little dip gave was in masking of defects in the early plastic cone drive units available in the 1960's. A spin-off benefit was that it appeared to move the sound stage backwards away from the studio manager who was sitting rather closer to the speakers in the cramped control room than he would ideally wish for. (See also Designer's Notebook Chapter 7). The depth of this depression was set by 'over-equalisation' in the crossover by about 3dB or so, which is an extreme amount for general home listening. We have never applied this selective dip but have taken care to carefully contour the response right across the frequency spectrum for a correctly balanced sound. Although as numbers, 1kHz and 4kHz sound almost adjacent in an audio spectrum of 20Hz to 20kHz, the way we perceive energy changes at 1kHz or 4kHz has a very different psychoacoustic effect: lifting the 1kHz region adds presence (this is used to good effect in the LS3/5a) to the sound, but the 4kHz region adds 'bite' - a cutting incisiveness which if over-done is very unpleasant and irritating."​

Search Google for "BBC dip". Here are two hits from that search worth reading.

https://hydrogenaud.io/index.php/topic,75195.0.html (I found the paragraph I quoted above in this thread.)

https://forums.audioholics.com/forums/threads/should-speakers-be-designed-to-have-a-flat-spl.86954/
Note that Dennis Murphy has much to say, in this thread, on the subject of flat vs. BBC dip in loudspeakers. Hint, he doesn't like the BBC dip.

Thanks Richard for this info. It does explain some things to me that I have been wanting to read up on. Will take a closer look at the links that you posted. Very curious to see what Dennis had to say about the BBC Dip. Dealing w/a bit of nausea at the moment.


Cheers,

Phil

 

[speakerman39]

The BMR curve is very good, but it might not appear to be as flat as it seems. The X axis is more than twice as wide as the Y axis, and that will do a lot to smooth out that curve. Also different measurement techniques can produce different results; where was the mic situated in that measurement? Not to say that Dennis is using tricks to get unrealistic results, but you can't really compare different measurement sets unless you know how those measurements were done.

As for voicing, like I said, there are reasons why designers voice the speakers the way they do. Sometimes they have good purpose, and sometimes its just a bad idea.

I am sure some smoothing is going on. But, I am also certain that Dennis would never do anything to try to trick someone into thinking his speakers are not what they seem to be. Honestly, I am hoping to save up some money so that I can opt for the BMR's. My new room is going to be larger and I will not use them in the near-field. That is the plan, but got a move and surgery coming up right around the corner so to speak.


Moving is very expensive. Neck surgery is no fun. Got a rather bumpy road ahead of me. As such, it is going to be a good while before I can opt for anything. Thus, to me the BMR's are a textbook example of just how ruler flat a frequency response should be. The Phil 3's would be even better, but a bit more than I can spend. Sure never hurts to dream though.


Cheers,

Phil

 

[speakerman39]

Phil, I would also add that B&W speakers typically have low distortions (relatively speaking). I think that's one reason why they could sound good to the right ears without ruler flat graphs. On the frequency response side, NRC did their measurements in a real anechoic chamber so you will have very different response in your room. I've measured the responses of my KEF speakers in my room. Their FRs (based on NRC's) should be as good as the BMR's, assuming the BMR's were also measured in an anechoic chamber or equivalent. Yet the FRs of both my LS50 and R900 look extremely terrible (relatively)in my room, even when measured at 1 meter. To me, the going by the ear thing sounds good, but I prefer to go with my eyes more (i.e. measurements), pick out the good ones (distortions), then use my ears, or go by ears, then eyes, always both.

Peng, I think that I read that Dennis did have the BMR's measured at the NRC. But, I could be wrong. When you speak of distortions, can you be a bit more specific? How do you spot distortions from the FEQ response graph? What do I need to look for? I mean, lower distortion is always the best way to go. I get that. But, how do I interpret that from a graph?


Cheers,

Phil

 

[speakerman39]

Have had comparable experiences but could not readily compare them. Have sometimes wondered how changes in hearing with age affect as well...

If you have auto-eq, do you like sound better unequalized or equalized?

As Swerd mentions, should consider the recording source too. The BBC dip seems to assume a certain bias that would be present in all recordings. Might have been true at one point in time in a certain region, but would think that B&W is past the day where they would continue to voice their speakers for that reason.

Ww

I do think our hearing changes as we age. Just 15-years ago I could not stand to listen to any Klipsch speaker at all. However, my local BB always has the Reference Series on display. After getting back into the hobby nearly 3-years ago, I actually kind of liked what I heard. Then, one day they had a set of the PR bookshelfs and I got to demo them. Holy smokes I was blown away. That is, the PR Series bookshelfs were NOT harsh at all and very detailed.

Most of the time when available I do prefer to use EQ. Audyssey has always giving me good results. Of course, I usually have to go in and do some tweaking to my tastes. However, YPAO has done an admiral job as well.

I think that the BBC Dip that I have experienced was much more present in my Wharfedale Dentons. Honestly, I really liked what I heard. More so than the B&W CM1 S1's. Vocals really stood out like never before.

However, I did try them in a as FL and FR while sitting about 8' away in my small room. Not exactly sure what was off, but the Dentons did NOT make for good fronts when watching a BR. It was so bad, I had to stop watching my movie. I think that the Dentons were designed more for 2 CH music with the aide of a capable sub. At least, that was my experience.


Cheers,

Phil

 

[Swerd]

Peng… When you speak of distortions, can you be a bit more specific? How do you spot distortions from the FEQ response graph? What do I need to look for? I mean, lower distortion is always the best way to go. I get that. But, how do I interpret that from a graph?

Here is an example of the distortion for the B&W CM1 that the NRC publishes. It shows two traces. The upper trace is the same speaker frequency response curve that you showed in your first post, but it's loudness is raised to about 90dB. The lower trace is Total Harmonic Distortion + Noise (THD+N) measured over a range beginning as low as 50Hz and higher depending on the speaker being tested.

At any frequency where both traces are visible, you can see how much louder the response (upper trace) is compared to the THD+N (lower trace) by the vertical difference between the two traces. At 200-500 Hz where THD+N is low, the difference is about 40 dB (very good), and at 100 Hz the difference is about 22 dB (not bad). When THD+N is at least 25 dB quieter that the response, you essentially do not hear the distortion.

Dealing w/a bit of nausea at the moment.

Sorry to hear about the surgery. I hope you get some relief afterwards.

 

[Winkleswizard]

Have always liked the dynamics of speakers like Klipsch. Your hearing change is likely due to drop off in high frequency hearing as happens when most of us age. The Klipsch have an upward tilt that would offset.

While I have some of that too, have also had some occassional pressure issues in one ear that seem to affect localization. Am still trying to reproduce the flyover I heard from the opening of Star Wars when I was younger. It was before any discrete surround, in a room that was rather small but the speakers were B&W 801s (and some other very nice electronics).

My main focus now is trying to get sound that I am happy to take into retirement (in a smaller residence). Previously did not have a real appreciation for the compromises of smaller rooms, but when I visited a hifi shop in London, I noticed they had no large speakers. Since the shop was not that small, I later realized that most of their market was for those who lived in much smaller spaces.

Anyway, have always thought highly of B&W, so not surprised you like them regardless. If I did not get into designing speakers, would probably own some myself!

Ww

 

[Swerd]

Very curious to see what Dennis had to say about the BBC Dip.

I'm not Dennis, but I am familiar with what he thinks about any dips or raised responses across the mid range. He does not like them one bit. If a speaker he hears has the dreaded BBC dip, or a raised mid range as with inadequate baffle step compensation, it takes him only a few seconds before he recognizes it.

Depending on who's speaker it is, he might say something like "Oh, that speaker must be British – it sounds so very polite". That's if the owner is present and he doesn't want to offend him.

If the owner is absent, he might say something a bit more direct, such as, "That's quite enough of that. Turn it off please."

I don't believe I had ever heard a speaker with a truly flat response across the mid range until I first heard one of Dennis's DIY designs. At first listen, I thought it sounded bright, but oddly compelling. After listening for some time (at least ½ hour) I quickly got used to it, and my old speakers had suddenly become too warm sounding. I can absolutely hear this difference best when listening to voices. Some people adapt slower or faster than I did, and others never adapt, preferring other types of voicing.

 

[Dennis Murphy]

The BMR curve is very good, but it might not appear to be as flat as it seems. The X axis is more than twice as wide as the Y axis, and that will do a lot to smooth out that curve. Also different measurement techniques can produce different results; where was the mic situated in that measurement? Not to say that Dennis is using tricks to get unrealistic results, but you can't really compare different measurement sets unless you know how those measurements were done.

As for voicing, like I said, there are reasons why designers voice the speakers the way they do. Sometimes they have good purpose, and sometimes its just a bad idea.

The two aspect ratios aren't all that different once you adjust for the 10 dB vertical markings vs. the 5 dB in my graph. In any event, I did have the BMR measured at NRC, and the main difference is greater resolution, resulting in more tiny ups and downs along the trend. The other difference, also apparent on the CM1 plot, is a peak in the 100 Hz region that's an artifact of the NRC's anechoic chamber, which for reasons I won't go into, doesn't deal well with rear-ported speakers. As for the BBC dip issue, what you see on the CM1 plot is more like a BBC cavern. A true BBC dip is much more subtle and resembles a shallow U shape across a wider frequency range. A classic current example would be the SVS Prime monitor. The CM1 plot is showing a narrower dip of almost 8 dB. That would certainly be audible, and the effect would range from distracting on well-recorded complex music with lots of energy in that region, to perhaps pleasant on certain other material. Any DIY'er has probably attended get-togethers where a particular speaker impressed people as non-irritating and easy to listen to, only to discover during a follow-up measuring session that the tweeter had been wired backwards, causing a big dip at the crossover region.

 

[Winkleswizard]

Agree, sharp dips or peaks where our hearing is most sensitive is just bad design.

Our hearing adapting to sound we like is expected behavior. The question is whether it is accurate or not. But, think this goes to the original question as well. No doubt it is essential to be able to discern how well a speaker matches the real world. I think most of us know there is more to being accurate than just flat frequency response too.

Only the designer really knows if they traded off one thing for another. For example, I can readily eq to flat, but the resulting sound may have rather high distortion. Distorted is not accurate, but the FR graph sure looks good!

With there being comparable tradeoffs in recordings and other parts of the reproduction chain, I can see why many audio designers may go for what sounds good over “accurate”...

Ww

 

[PENG]

Peng, I think that I read that Dennis did have the BMR's measured at the NRC. But, I could be wrong. When you speak of distortions, can you be a bit more specific? How do you spot distortions from the FEQ response graph? What do I need to look for? I mean, lower distortion is always the best way to go. I get that. But, how do I interpret that from a graph?


Cheers,

Phil

I just go by published specs and available bench test measurements. Many speakers manufacturers don't even publish distortion specs, but B&M does.