It makes sense. We should start by dropping the listening window for a moment and replacing it with the listening axis, it's just confusing matters.
Next let's assume we equalise for a flat response on axis for any speaker. (Even though we don't always do that, it helps when making a point like in this particular comparison.)
Then we measure power.. which then, as I implied above, gets normalised to the axial response... The smoothness of the resulting power curve represents the performance of the speaker on the whole insofar as it represents the character of a particular design. As sheeple points out, a controlled directivity design is going to have a different character to a typical three way.
Next let's assume we equalise for a flat response on axis for any speaker. (Even though we don't always do that, it helps when making a point like in this particular comparison.)
Then we measure power.. which then, as I implied above, gets normalised to the axial response... The smoothness of the resulting power curve represents the performance of the speaker on the whole insofar as it represents the character of a particular design. As sheeple points out, a controlled directivity design is going to have a different character to a typical three way.
As sound power is the sum of all measurements weighted by the portion of a sphere it represents it is always the same and cannot be normalized to anything. It does not change regardless of the listening axis.
The number of ways seems irrelevant to me. The hypothetical contrast is between a speaker with different DI's. When you add the word "typical" in front of 3 way it implies rising DI, but 3 way on it's own does not.
It was suggested that DI/SP are related to another. If smooth and gradually sloping early reflections is the goal, this is not what is baked into the directivity of a controlled directivity two way speaker, as opposed to a many three way speakers with rising DI. Now even when the listening axis is EQ’ed with a sloping shelf, what has not changed is DI. So as DI is baked into the speaker physically, for a given flat listening axis there is a given slope. This cannot be changed. The speaker that creates the tilt of early reflections by a sloping of the listening axis will not sound the same as a speaker that has a tilting due to rising directivity.
I could have replaced that entire paragraph with - "Look at what the DI tells you"... except that I was trying to explain what it was and why it matters in this case. What do you think?
Partly agree, partly not sure of your point? When I said typical I meant so in the same sense as Dr Geddes meant when referring to a 'typical' 2 way in this paper - http://gedlee.azurewebsites.net/Papers/directivity.pdf
That the paragraph was in response to your assertion that power could be normalized based on listening axis.
This entire line of discussion stemmed from me pointing out that the directivity is what causes the in room and power responses to behave as they do.
That directivity is not tied to the number of ways in a system.
Now this i grasp. Directivity is result of physical design, and impacts the direct to reverb relationship at the listening position in given listening room situation. In my situation the listening position is ~ 3m (living room 5.2 x 7.5m) , therefore in the reverb field. Therefore i need quite directive speakers. How much and flat or rising DI is a choice i am working on (within the physical limits of course ;-))
Last edited:
Fluid. Ohms law also has three dependent elements. Each is no more important than the others and each can be used as a point of reference.
Yes, but I suspect you're splitting hairs and missing the point.
They are related, once the crossover is fixed the DI is fixed but the sound power can be varied by altering the on axis response with EQ.
That is not a bad idea but it is not the only way to good sound.
The DI cannot be changed but the slope of the in room and power response can be altered with EQ.
If someone measures a CD speaker in room and tries to make that measurement match a rising DI in room then the sound will be dull like ra7 mentioned earlier as they will slope the on axis down a lot, but if the slope is slight and done with Anechoic measurements then a middle ground can be reached. Earl did it with his speakers, Martijn did similar with the 8C.
No just trying to be clear. When you insert words that were not in the original premise and mean things that you do not explicitly say, confusion results.
As it should be, and also as we should be expecting that sheeple meant if we are to give credence to the point being made, which is that speakers with different DIs won't be the same even if we try to compensate.
No, so I qualified with the word 'typical' and verified that with a reference to a paper...
The underlying point hasn't changed.
Regarding the Sointuva - I would love to see the spinorama data for the tweeter alone. I am curious as to how much baffle effect is present in the waveguide tweeter response.
If it were a flat-flange tweeter, we could make a reasonable simulation with VituixCad. The speaker is 240 mm wide by 450 mm tall, and I estimate the edge radius to be 25 mm. I guesstimated the position of the tweeter. With a normal dome tweeter, we would see an on-axis baffle gain (diffraction hump) of about 2 dB from 800 to 2k, centered at about 1.3k. However, with the waveguide, I am certain the response would be quite different. I speculate that the influence of the baffle size/shape is less important. But how much so?
j.
If it were a flat-flange tweeter, we could make a reasonable simulation with VituixCad. The speaker is 240 mm wide by 450 mm tall, and I estimate the edge radius to be 25 mm. I guesstimated the position of the tweeter. With a normal dome tweeter, we would see an on-axis baffle gain (diffraction hump) of about 2 dB from 800 to 2k, centered at about 1.3k. However, with the waveguide, I am certain the response would be quite different. I speculate that the influence of the baffle size/shape is less important. But how much so?
j.
Attachments
Sheeple> when you say "controlled directivity" do you mean constant directivity? As in a perfect dipole, or some MEH I've seen, where entire pass band of the speaker drops as you move off axis? Your comments make sense to me in that context but if others are thinking controlled directivity, that might be why you guys are talking past each other? A controlled directivity speaker has a sloping DI also, just at a higher DI (and hopefully smoother) value than typical hifi speaker. A prefect constant directivity (to me anyway) would have a flat DI, but also at a higher value than a typical hifi speaker.
I was purely being descriptive and pointing at the fact that a controlled directivity speaker’s signature (augerpro: most HiFi 2-ways Fall in this category, I’d even think JBL M2 is not a constant directivity speaker, although close) cannot be 'tamed' with EQ to resemble a typical 3-way slope. But I was not assuming you didn’t knew.
I was not discussing whether this or that is better. The method to EQ the listening axis/LW has limits that correlate with DI, and this is a property of the radiators and enclosure.
As much as it is banal and known, and we have seen some discussion of the sonic presentation of controlled directivity speakers, I think there is more to learn about it, especially: What is good filters with a low DI, plateauing?
There is very little diffractive ripple in the measurements just a bit as shown here but that is below the crossover frequency. The waveguide and physical tweeter size make it become quite directional beyond 3K. Not much gets to the edge.
To be clear I am on the same page, the speaker as a whole, I'd agree. But the HF portion, I'd disagree. Above 800Hz to about 8kHz, DI is practically ruler flat. See image below.
FWIW, my anecdotal, personal experience is that - when a speaker is designed for flat on-axis response & has flat DI (in the treble, at least) that it will sound bright in-room and would benefit from sidewall absorption. Or, you can use EQ to shelve it down if that's easier.
Yes, and keep in mind that this data is 1/20th-octave. Most backyard, gated measurements aren't going to show that. Even if they are gated to 5ms (which is on the higher end of the "typical" 3-5ms), that means you'd have only a data point at 1000/1500/2000Hz. Those little ripples aren't going to show up. Certainly not to the degree that we see them here. Better yet, they're practically constant. A small little filter here or there will iron those out. But, I'm not even sure it's necessary, all things considered.
This is exactly the point I made earlier and Sheeple has expanded and tried to explain. The bright sound from constant directivity can be tamed by a slope but IME, if you slope the on-axis, it sounds dull and not right. Instead, if the reduction in sound power is obtained through increasing directivity it sounds more natural—something you get more easily from cone and dome three ways.
The take away for diyers is to use strategically placed absorption (or a large room) to work with constant directivity designs but keep the on-axis flat. It is the combination of flat on-axis AND falling sound power that sounds right. Again, just my experience but others seem to agree.
No problem… I am glad to hear that I hear what others are hearing. We are way off topic anyway… but all this is fun.