[JohnR]

I had earlier been using a "naked" Neo3PDR. The on-axis FR always bothered me. I have since seen a similar effect in other measurements posted online. This is the measurement I took back then - as you can see, there is a fairly large dip around 8 kHz in the on-axis response (red line):




I finally got around to trying something else. It took a while of mucking about in Edge, but in the end I was able to get a fairly smooth response. Here is the measured response with the new baffle - purple is the tweeter with no EQ, green is with EQ:



There's still a dip at 8 kHz, but not much. The response extends lower and it's easy to EQ it flat. I haven't looked at the off-axis measurements yet. For all I know it could be a disaster out there.

Picture of the prototype "baffle" coming shortly.

[jonnoshore]

Q. could the tweeter be placed at an angle to the listener to get the blue curve and then EQ from 10kHz to get the curve flat?

Sound a bit loud off axis though I suppose, I was thinking more to smooth out the 8kHz dip.

[JohnR]

I'm not sure that would work so well as that is the 60 degree curve. The green one (in the first graph) is the 30-degree curve - that might be feasible... I'd never thought of that to be honest.

I'm not sure how audible the 8kHz dip is. But it's still annoying

[JohnR]

This is the prototype baffle. It seems that cutting off the "shoulders" of a baffle generally seems to be a good way to smooth out the response, on-axis anyway. It took me a long time to arrive at this shape - in retrospect it seems obvious 



Anyway, I've decided that this driver is going to be better in a baffle, of roughly this kind of shape. I recall that some here have used a Neo3/PDR, has anyone experimented with the shape of the baffle?

I'll try to get some off-axis measurements in the next few days, I'm more curious now that I've posted this.

[jonnoshore]

I would imagine that 8kHz relates to the distant to the edge of the square baffle when on it's own as all 4 side would have had this distance... I imagine the new shape will improve the dispersion ... would be good to see a measurement of it.

What is the wave length of an 8kHz wave then see if the edge is a 1/4 ratio of that length... Just a thought.

Like the pipe cleaners. The gap under the tweeter and above it are still the same dimension it seems... I'd look at changing one of them or sealing the one below... is it flush on that baffle? (difficult to see from that photo)

[lowtech]

John, I have noted similar irregularities with the same tweeter in my prototypes.  Note, however, that JohnK claims to achieve a supremely flat response in his NaO Note design due to the use of his custom "waveguides".  I'm currently in the midst of an equipment change but plan to take some measurements of my long work-in-process dipole project soon.  I'm basically locked into my baffle shape but will be fooling around with waveguide profiles.  Any remaining uneven response will be dealt with by shaping filters.

[Nate Hansen]

Ahh yes, the bane of the 'naked' Neo3's existence! I run them like this, and that 8khz dip bugs the hell out me. 

I also wonder how audible it is. I listen about 20 degrees off axis, where the response is pretty flat, but is that all I hear? IMO, we need to look at the whole picture, response on as well as off axis. For instance, here's a shot of the power average of my system a while back. Fairly flat through the treble, with actually a bit of rise at 8khz:



I've played with open cell foam and felt to try to tame that null on axis, and they only seem to help a little. These days I try not to worry about it   

As far as baffles go I guess that depends on your priorities. I'm a full-on dipole enthusiast, and I want dipole response as high in frequency as I can get........which means no baffle for me. That means more eq, and that damn null. Like lowtech pointed out I think the waveguided Neo in the Note is probably the best approach (besides nude) for a "pure" dipole.

Since I don't know anything about baffles I'll shut up now 

[Nate Hansen]

OK I'm not quite done. FWIW here's my measurement of a Neo3 PDR nude (nevermind the red overlays) 0-60 deg in 15 deg steps:



JohnR, I'd be interested in seeing a set of polar FR measurements of your Neo3 baffle if you've got 'em.

[Rudolf]

Needless to say that this dip is generated by edge diffraction. jonnoshore already explained it.
The frequency of that dip is a function of the "effective" baffle diameter. It represents the first dipole dip.
The depth of that dip is a function of the relation of baffle width to source width. It will be deepest for a point source on a large baffle. And it will be most shallow when the source is as wide as its baffle.

You can solve this "problem" if the source doesn't "see" the (distant) baffle edge at the dip frequency and above. This is what JohnK does by applying a waveguide to the Neo3 in the Nao Note.
You can lessen this problem by cutting off as much baffle area from the Neo3 as possible. (But please don't do that in the real world!)
You can smear this problem by making the baffle larger and giving it a less regular shape. Now it doesn't show up as prominently in your diagrams, but it will be acoustically present over a wider frequency range and over a wider angle in space.
You can live with this problem by listening outside the area of the dip (20° off axis). Recommended!
You can totally mess up things by correcting the frequency response on axis and making it worse for the remaining 95 % of space.

Isn't it as easy as that? 

Rudolf

[Rclark]

 I've been listening to this tweeter for the last week and a half, gotta say, I haven't yet hit the limit of what it can resolve. I'm shoving all kinds of material through this thing...

 Would love to hear the 10!

 Anyway, love this discussion, sorry to interrupt.

[JohnR]

Thanks guys, for the comments  I haven't been able to measure again but should in the next couple of days.

Rclark, the Neo10 is a midrange, much wider element. I have to admit I don't share your enthusiasm for this (Neo3) driver, but I'm trying to get the best out of what I have.

Rudolf, interesting but controversial (?) list. Is a non-circular baffle shape really "smearing" the response? I'm not sure I like the term, F and T are intimately related, as you know, and a "perfect" dipole cannot have a "perfect" impulse response, fundamentally...

Nate Hansen, thanks for those plots, perhaps yours is one that I saw posted that made me look into this again (it looks familiar). When you say "dipole response" you mean below the dipole peak? I have to ask why - ?

lowtech - thanks, I took another look at the Note page. JohnK is not giving anything away in terms of waveguide shape, fair enough.

[Rudolf]

Is a non-circular baffle shape really "smearing" the response? I'm not sure I like the term, F and T are intimately related, as you know, ...

Maybe "smearing" is put too bluntly.  I will try to explain it more detailed in a later post. I'm still collecting the proper wording ...

... and a "perfect" dipole cannot have a "perfect" impulse response, fundamentally...

Thanks for bringing this up. I hadn't made up my mind about this well enough, but maybe I got it clear now:
A perfect dipole has a perfect impulse response. If you compare with a "normal" driver in infinite baffle, you have to make the dipole separation infinite for a fair comparison. If the "normal" driver is boxed and the dipole is mounted on a baffle of the same size/shape as the box baffle, both will show the same diffraction pattern. Only difference: peaks and dips in the OB response have twice the SPL as for the boxed speaker.

The "imperfection" of a dipole impulse response in real life is a function of the bad/missing symmetry between front and back radiation. It is NOT fundamentally IMHO. Please compare these speakers:


It is the same driver mounted without any baffle and closed with a cup to kill the back radiation. This are the impulse responses for both speakers. They are at the resolution limit of the measurement - I have no higher res. I have moved the responses slightly apart. Otherwise the green diagramm would be largely invisible:


The frequency response it not exactly the same, but similar enough for this comparison:


In the direct response we practically do NOT hear any separate delayed impulse from the rear of a dipole driver, but only the diffraction pattern of the baffle edges. I try to keep this diffraction pattern as uniform as possible 360° around the driver circumference - thereby keeping the resulting combined impulse response (of driver and baffle edge) as "original" as possible.

[lowtech]

I try to keep this diffraction pattern as uniform as possible 360° around the driver circumference - thereby keeping the resulting combined impulse response (of driver and baffle edge) as "original" as possible.

For the case of a "naked" NEO3PDR, the solution (to me) seems obvious.  Use their "waveguide" on both the front and rear of the baffle.  See my photo above. 

[Davey]

What frequency range are you fellas asking the Neo3 tweeter to cover in your projects?  Top two octaves only?

Cheers,

Dave.

[Rudolf]

What frequency range are you fellas asking the Neo3 tweeter to cover in your projects?

3 kHz at 18 dB/oct should be safe. Distortion is rising steeply just below 2 kHz.

[Nate Hansen]

What frequency range are you fellas asking the Neo3 tweeter to cover in your projects?  Top two octaves only?

Cheers,

Dave.

I'm crossing to a Neo8 @ 3700 lr4. I know some cross less than 2k but as Rudolf said distortion is an issue. I don't like how it sounds any lower than 2500.......

JohnR, when I said dipole response I mean ideally I'd like the off-axis response to keep to the dipole dispersion of -1db at 30, -3 at 45, and -6 at 60 all the way to 20khz. Obviously physics limits this, the Neo3 is (only?) good to that 8khz anomaly, better than back to back domes anyways!

As to why? I've come to prefer it over a single forward radiating tweeter, or even double-domes. Sounds much more natural IMO.........

[JohnR]

Here's my off-axis measurement of the new baffle.



(0, 30, 60, 90 degrees, no EQ)

With regard to "ideal" dipole behavior, before I was seeing that up to about 6k; with this one, it does that up to about 2.5k only. I'm not able to convince myself it's important... it still acts pretty much like a dipole but with a wider lobe....

[JohnR]

What frequency range are you fellas asking the Neo3 tweeter to cover in your projects?  Top two octaves only?

I've dropped it to 3.2k BW3 with this baffle, earlier I had it at 4k LR4 or LR8.

[JohnR]

The "imperfection" of a dipole impulse response in real life is a function of the bad/missing symmetry between front and back radiation. It is NOT fundamentally IMHO.

Hi Rudolf, what I meant was that a "perfect" dipole has two peaks in the impulse response. This is fundamental. Otherwise, it would not be a dipole.

[Rudolf]

Rudolf, interesting but controversial (?) list. Is a non-circular baffle shape really "smearing" the response? I'm not sure I like the term, F and T are intimately related, as you know, and a "perfect" dipole cannot have a "perfect" impulse response, fundamentally...

Hi Rudolf, what I meant was that a "perfect" dipole has two peaks in the impulse response. This is fundamental. Otherwise, it would not be a dipole.

Hi John,
it’s right to remember the first principles where dipoles start from. And your comments always come to the point. But in real life and for most of the frequency range dipoles are not two separated point sources, but a (point or expanded) source surrounded by a ring radiator – the baffle edge. JohnK has explained that in some depth in http://www.musicanddesign.com/Dipoles_and_open_baffles.html: In the 1/f-region (up to the first dipole peak) there is no second impulse for the equalized dipole (not even in an perfect dipole). If done right – making the baffle not too much larger than the radiating area of the driver – there will be no second impulse from the equalized dipole for higher frequencies either.

That said, in real life dipoles the prevention of the second peak in the impulse is not only desirable, but it is a prerequisite for a flat response. It is JohnK himself talking about “smearing” in this context:
“Third, while a larger baffle will increase the low frequency sensitivity the greater delay will translate a smearing of the impulse response over a longer period of time.”

I'm aware that above reads quite nit-picking. And it should not be looked at as "rules". But it is part of the understanding of a dipole. Whether the result (what result?) is worth the added effort and complication, is surely up to everybodies taste and demands.

Rudolf