Open baffle 4-ways under construction

[Jussi]

What about the Eton air motion tweeter? You might have to make a specialized mount for it to fit in the waveguide area, but otherwise should do really nicely.

I'm afraid it's too expensive, a bit pig in a bag and how well does it suit to <2Khz cross? That's the main problem with any ribbon tweeter.

Can you share measurement results for Esotars, with and without waveguide?

Here is one. There is on axis response with 20 degree and 45 degree off axis responses in the waveguide and on axis response with 45 degree off axis response at -10dB level without the waveguide. Both on axis response optimised to 2Khz 4th order Linkwitz-Riley highpass. I also have the raw measurements.

Primary problem is that notch around 9Khz. I guess it's created by the waveguide with such high profile dome. Some sound is radiated from the dome sides which isn't directed along the waveguide but bounce from it causing cancellation and stored energy.

Directivity without the guide isn't adequate. Such solution suffers from same problem as Orions. Radiation pattern spreads open at upper crossover frequency and with large baffle combined to large dome it beams back narrow towards top treble. Not good combination.

Jussi

 

[badman]

I'm sorry, a 'bit pig in the bag'? Could you reword the expression?

I brought it up specifically for the low crossover ability- Fs is 490Hz and it's quite hardy. 1k would be a stretch but 1.7k or 2k should be fine with a decent slope (2nd order or better), though you might have to notch the fundamental due to a rising response towards Fs.

Air Motion tweeters are tougher than their ribbon counterparts, just like planars are. The tweeter won't need dramatic power handling, max output is always a limitation with any reasonable open baffle anyway.

If I had a project like this, it'd be my choice. Too expensive... well.... seeing the excels and the baffle itself, I didn't think that was too tremendous an issue 🙂

 

[Timn8ter]

I'm sorry, a 'bit pig in the bag'? Could you reword the expression?

I think we would say "pig in a poke" or buying something sight unseen or unknown.

 

[Jussi]

I think we would say "pig in a poke" or buying something sight unseen or unknown.

Correct.

I brought it up specifically for the low crossover ability- Fs is 490Hz and it's quite hardy. 1k would be a stretch but 1.7k or 2k should be fine with a decent slope (2nd order or better), though you might have to notch the fundamental due to a rising response towards Fs.

Air Motion tweeters are tougher than their ribbon counterparts, just like planars are. The tweeter won't need dramatic power handling, max output is always a limitation with any reasonable open baffle anyway.

I have to consider them. Just like a bunch of other options. At this moment Accuton C2-12 and Seas 27mm aluminiums are the primary replacements. Both use different dome profile and therefore should work in waveguided environment.

If I had a project like this, it'd be my choice. Too expensive... well.... seeing the excels and the baffle itself, I didn't think that was too tremendous an issue 🙂 [/B]

I guess the "pig in a poke" issue is the problem here. I'd like to keep in known and tested top performers instead of trying new territories. This explains my midrange and lowmid selections. Unguided Esotar is also excellent unit but it just doesn't seem to cooperate with the guide so I have to consider other options.

Jussi

 

[Paul W]

Primary problem is that notch around 9Khz. I guess it's created by the waveguide with such high profile dome. Some sound is radiated from the dome sides which isn't directed along the waveguide but bounce from it causing cancellation and stored energy.

Yes, that is quite bad result. Which waveguide are you using? Just a guess at this point, but the compression driver waveguides everyone is trying (including me) may have the wrong throat profile for dome tweeters. A better throat for domes might be a conical flare beginning directly on the tweeter flange; with the largest practical roundover at the mouth of the waveguide.

The compression waveguides need to make a transition (curve) to the flare which isn't needed with a dome and, perhaps, actually works against the dome shape. Big caveat...YMMV, etc 😉

Maybe you can mold your own waveguide? Otherwise, the concave Accuton tweeters (with grill removed) might work well on a waveguide.

 

[Jussi]

Yes, that is quite bad result. Which waveguide are you using? Just a guess at this point, but the compression driver waveguides everyone is trying (including me) may have the wrong throat profile for dome tweeters. A better throat for domes might be a conical flare beginning directly on the tweeter flange; with the largest practical roundover at the mouth of the waveguide.

It's a manufactured guide for Finnish kit & speaker brand. Pretty much like this one:

http://www.hifitalo.fi/kuvat/set2.jpg

So it's a "conical flare" as you put it. Not a compression horn. Actually I don't think compression driver guides are that operational with normal tweeters. Especially soft domes. That throat structure causes tremendous stress to the diaphgram and even titanium domes in PA tweeters have problems handling it. Some 1" throat, 2" dome PA tweeters have several % THD at 100dB level which is pretty bad considering you can half that with regular waveguide and a good home tweeter. But the homesetup doesn't have 110dB efficiency and doesn't play +130dB levels.

These guides are made for 25mm Seas aluminiums. Matched. 27mm models fit with slight modification, not a problem. 27mm Seas or Accuton is propably my next step... Seas has usable grill which makes them safe for children. Pretty affordable as well.

Jussi

 

[Paul W]

We may be saying the same thing??? The "compression driver" waveguides I mentioned have constricted throats -like the large one shown here- but usually smaller versions are tried for Hi-Fi.

An externally hosted image should be here but it was not working when we last tested it.

Here is a link with the WG shape I suggested. Unfortunately, it also introduces a dip similar to yours, but higher in frequency.

Waveguide paper

 

[Jussi]

Here is a link with the WG shape I suggested. Unfortunately, it also introduces a dip similar to yours, but higher in frequency.

Looks reasonable. Constant directivity horn. That sounds familiar. Althought most recordings don't support all the way constant directivity. It's too bright.

Jussi

 

[Jussi]

So here is the problem. Problem that "bothers" almost all open baffle dynamic constructions. Orion, Beethoven, Phoenix and in this case my own.

Unwaveguided tweeter. I've wrote about the theoretical side of it and here is the measured result. Two on axis responses, at 0dB and -10dB levels. Two 60 degree off axis responses, "out" and "in" while my system has non-symmetrical tweeter positioning and therefore responses to both directions differ. Responses are measured in room so there are some reflections below 300-400Hz or so. Graphs are unsmoothed.

There is still some indications of dipole null around 1300-1500Hz which is expected. To 60 degree off axis null isn't perfect anymore so there is directivity left. In my case first reflection from sidewall to listening position arrives from around 80 degree angle if panels are pointed towards the listener. From that aspect null shouldn't be a problem.

But the tweeter. There just isn't that much directivity there. "Out" response runs to on axis response around 3Khz and "in" (the side the tweeter is) over wider range of 3-5Khz. Above that response drops pretty fast off axis and therefore runs power response down as well.

This is a complicated matter. First reflection from sidewalls isn't a good thing and with such uneven radiation pattern it's expected to have far more 2-6Khz sounds reflected towards listener than frequences for example below 1Khz. But that's just the 1st reflection from there. Below 1Khz this thing also radiates to rear and 1st reflection from front wall appears to the scene. Tweeters range doesn't have it.

In power response this is quite good solution up to 6Khz or so. While dipole rear radiation drops off at upper crossover frequency tweeters radiation pattern spreads wider and evens the 3dB theoretical gap in the power response. The problem is area above 6Khz. Dome size with panel size cause tweeter beam pretty fast and aggressive. This is also visible in the Orion measurements. Which leads to power response drop towards the top octave. Perhaps this makes the system sound calmer like the review of Orion once mensioned...

At least the deal with no waveguide messed up my speakercloth idea. I thought to use cloth just along the panel and that's one reason I sunked the other drivers deeper into the panel. But the Esotar, without its guide, as "bulged" dome which exceeds the panels level so the cloth would be contacted to the dome. Not good...

New ideas? Honest opinions?

Jussi

 

[Jussi]

A little addition. Using 2nd order crossover things smooth up a bit. Check the attachment. Dipolemids directivity follows the system a bit higher in frequency and while tweeter roll off towards the crossoverpoint from bit higher null directivity to 60 degree angles disappear. But how good idea is it to use 2nd order Linkwitz-Riley in MTM?

Jussi

P.S. Previous response uses 4th order 1800Hz cross.

 

[Paul W]

Jussi,
Since top-octave rolloff is determined by dome/ribbon size, it seems low end tweeter output is 1st problem to beat. Since WG didn't work with Esotar domes, maybe thick strips of wool felt (10-15mm?) and/or foam directly on either side of the dome to absorb side radiation might help.

If that works, you may be able to recess dome, plus control horizontal and vertical pattern, with carefully shaped opening in felt.
Paul

 

[Jussi]

Since top-octave rolloff is determined by dome/ribbon size, it seems low end tweeter output is 1st problem to beat. Since WG didn't work with Esotar domes, maybe thick strips of wool felt (10-15mm?) and/or foam directly on either side of the dome to absorb side radiation might help.

If that works, you may be able to recess dome, plus control horizontal and vertical pattern, with carefully shaped opening in felt.

I tried that earlier in one 2-way project. Didn't work that well. Phase shifts and diffraction problems. I think I'll have to experiment this with an actual crossover instead of simulation. If the response behaves like with that 2nd order 2Khz cross there shouldn't be that much problems.

Jussi

 

[Tenson]

I have only read the last two pages of this thread so forgive me if I am missing something, but why not try a phase-plug with the waveguide to fix the 9KHz dip? Since the dip in the response is not present at 45degrees off axis it seems this is a suitable solution. You may only need a small one directly in front of the dome at the throat of the WG.

 

[Jussi]

I have only read the last two pages of this thread so forgive me if I am missing something, but why not try a phase-plug with the waveguide to fix the 9KHz dip? Since the dip in the response is not present at 45degrees off axis it seems this is a suitable solution. You may only need a small one directly in front of the dome at the throat of the WG.

We (small local DIY group here in Finland) have a theory about this dip. Soft domes, like Esotar, have pretty high profile dome. Diaphgram is a bigger part of a "sphere" than aluminium and ceramic domes. For example Seas aluminiums (that work well in waveguide) are almost like a lense than a dome compared to soft domes. This causes the problem in the guide. Soft dome can radiate some sound from its sides which isn't directed along the waveguide but against its throat. And since it doesn't go along it reflects and causes some cancellations to some frequency. In this case around 9Khz.

So fixing diffraction caused (which also means stored energy problems in CSD) problems by adding a phaseplug which also is a potential source for stored energy isn't exactly what I had in mind.

But honestly, what's wrong with the responses without the guide? Sure it doesn't direct that much. Perhaps -3dB at 60 degrees but potentially enough to 80 degrees where 1st reflection from sidewall arrive. And while tweeter radiation isn't matched to dipoles front radiation there are less problems in the power response.

These with the assumption that the actual system works like the simulation. With 4th order cross it doesn't. Problems are similar with Orion. Compare if you like. I posted simulation with same responses using 1800Hz 4th order cross a bit earlier and here are the responses measured from Orion. Anechoic chamber, 0, 20 and 60 degrees. Power response in black. I assume gentler slope with Orion might fix some of that tweeter radiation spreading but it isn't possible with such large mid and low cross.

Jussi

 

[MBK]

Honest opinion - it depends... On one hand a truly even power response is hard to achieve with current technology. At some point all tweeters will beam, the question is, from what point on does it start to matter? In truth, most natural sound sources at HF will also beam, and/or have highly irregular directivity. Real listening of real instruments therefore likely also has a drooping power response. Close miking exaggeraterates the top end HF, and a downward sloping playback power response at HF somehow corrects for this, making for a result which in the end feels right. Ambient recording by contrast likely has much less HF content anyway, precisely because natural sound sources also have the same beaming and irregularities as tweeters, and because HF is easily, if irregularly, absorbed anyway in a concert venue.

What I really mean is: in the end, these problems may not really matter all that much. I acknowledge of course that one would always like to build the most perfect system! Compromise seems a dirty word but at some point one has to draw the line... In my mind, directivity is likely much more important as a factor in the 1-5k range, where our hearing is most acute. Here, the dipole-monopole transition creates enough headaches...

Also: in the Orion measurements, why does the power response dip between 2k and 5k when the off-axis radiation actually rises towards the 0 degree response (60 degree curve starts overlapping 20 and 0 degrees)? Is this due to decreased rear radiation (mid rolls off, tweeter doesn't "come around"?)

 

[Jussi]

On one hand a truly even power response is hard to achieve with current technology.

It's possible but I wouldn't recommend it. Just like many other things not that many recordings support such a frequency balance. With directly recorded material (no eq, no mixing, no mastering) such system works like charm. But such recordings are made in 50´s, 60´s and some in the 70´s. Present albums are balanced after recording to sound good in present reproduction equipment and common systems have roll off in their power response. This is just natural development. Normal speaker is a vented 2- or 3-way without anything that fancy on it. In radiation pattern way I mean. So it's a monopole at bass, turns into cardioid at some point depending on cabinet dimensions (creating bafflestep) and beam up to top treble. If the radiation patterns are matched so there isn't uneven behavior between different size and frequencyrange drivers this creates nicely rolling off power response. Poorer cases, without matched radiation patterns, roll off might have some notches and humps on it. This is also pretty common.

So constant directivity is a good idea. In theory. But practice doesn't support it. And what do you do with a ideal system in unideal world? Write an address to studios that they should stop messing up their albums? 🙂

Power response roll off towards treble is desirable. At what point should it start and how steep roll off it should be. That's much more difficult thing to consider. Linkwitz recommends it on his latest "design landmark" list (which is also understandable since his commercial kit has such balance) but also many long term DIY builders have run into the same conclusion.

Also: in the Orion measurements, why does the power response dip between 2k and 5k when the off-axis radiation actually rises towards the 0 degree response (60 degree curve starts overlapping 20 and 0 degrees)? Is this due to decreased rear radiation (mid rolls off, tweeter doesn't "come around"?) [/B]

That's correct. With gentler crossover loop that dip would propably be smaller or even invisible. In theory power response drops 3dB when rear radiation fades away. In Orions case tweeters low end range isn't matched in dipoles front radiation pattern so there isn't a drop, it actually rises a bit. If there is a waveguide to match the patterns there would be a 3dB drop in power response around the crossoverpoint. How steep depends on crossover order. In my case 4th order cross at 1800Hz would propably create pretty similar situation than in Orion. Dipole rear radiation suddenly drops off and tweeter takes over and while crossover order is high there is no MTM or dipole radiation pattern of the mids to assist the radiation pattern around 2-5Khz.

I have heard the Orions and they are one of my inspirations to start this project. I also know some of the weaknesses of the system and I'm trying to fix them. One is the wide tweeter radiation and at least the simulated crossover suggests gentler order crossover between mids and tweeter seem to fix it. Power response should be better since overlap between ranges is wider and overall radiation pattern doesn't change that dramatically.

Jussi

P.S. If you noticed above upper crossoverpoint Orions power response goes pretty much along with the 60 degree off axis plot. Here is a good hint while measuring normal box speakers: 60 degrees off axis response gives pretty good estimation how speakers power response looks like. Uneven behavior there and it's also uneven in power response. But this applies only for normal box speakers, dipole-monopole hybrids are more complicated.

 

[MBK]

Come to think of it - the drop in power response when the tweeter takes over should not quite be -3 dB (power).

The dipole has a directivity number of 3, and 10xlog(3) = -4.8 dB in power response (theory) w.r.t. a monopole. A tweeter on a flat baffle above bafflestep has directivity 2 (half space) and 10xlog(2) = -3 dB (power). So the difference is just 1.8 dB and a waveguide would have to restrict the tweeter's half space by an additional factor of 10^( -1.8/10)=0.66 to match a dipole mid, resulting in 2/3 of a half-space, corresponding to a 2x180/3=120 degrees symmetrical conical guide. Perfectly achievable.

(It is almost 2 am here so I may be hallucinating - please correct me if I'm wrong).

 

[Jussi]

Come to think of it - the drop in power response when the tweeter takes over should not quite be -3 dB (power).

The dipole has a directivity number of 3, and 10xlog(3) = -4.8 dB in power response (theory) w.r.t. a monopole. A tweeter on a flat baffle above bafflestep has directivity 2 (half space) and 10xlog(2) = -3 dB (power). So the difference is just 1.8 dB and a waveguide would have to restrict the tweeter's half space by an additional factor of 10^( -1.8/10)=0.66 to match a dipole mid, resulting in 2/3 of a half-space, corresponding to a 2x180/3=120 degrees symmetrical conical guide. Perfectly achievable.

(It is almost 2 am here so I may be hallucinating - please correct me if I'm wrong).

I didn't quite got the point. So mathematically dropping rear radiation off and play on with half space radiating unwaveguided tweeter drops power 1,8dB and the waveguide some more which results the 3dB. Achievable and pretty propable as well. Below tweeter+panel beaming area normal 1" dome doesn't direct the sound at all to 60 degree angle. Dipole is 6dB down. Using waveguide tweeters 60 degree off axis is easy to fix to that -6dB level. 45 degree -3dB and so on.

In my case it's a question which one is worse. Use waveguide, match front radiation patterns and cause 3dB drop to power response. Or not to use waveguide, let radiation pattern spread a bit and have less or even no drop to power response. At least the transition with 2Khz 2nd order cross is much smoother than it is with 4th order filter.

Jussi

 

[MBK]

Not quite.

If

- a monopole has a directivity of 1 and a power response of 10xlog(1)=0

then compared to these 0 dB:

- a dipole has a directivity of 3 and power response of 10xlog(3)=4.8 down (actually, 10xlog(1/3)=-4.8 dB)

- a half-space radiating tweeter has directivity of 2 and power response of 10xlog(1/2)=-3 dB.

So the tweeter has 1.8 dB *less* power response *drop*, compared to a monopole, than a dipole. Or in other words, to match the tweeter power response to the dipole's, the tweeter needs to drop an additional 1.8 dB in *power* to reach the -4.8 dB of the dipole. And this can be done by restricting the existing half-space radiation of the tweeter some more, to ca. 120 degrees (down from 180 degrees which it would have in pure half-space). And this conveniently also matches the radiation patterns between dipole and tweeter.

This is all true in theory only. When I measured my fairly wide U-frame dipoles, I found they were pretty close to omnidirectional above 300 Hz, only below this did they start to radiate in classic dipole pattern. And even the tweeter radiation managed to "get around" 180 degrees quite significantly. So, actual measurements can't be beat...

 

[Jussi]

So the tweeter has 1.8 dB *less* power response *drop*, compared to a monopole, than a dipole. Or in other words, to match the tweeter power response to the dipole's, the tweeter needs to drop an additional 1.8 dB in *power* to reach the -4.8 dB of the dipole.

Like you mensioned. Moving from full space to half space drops power 3dB. Situation is the same is it monopole to half space radiating cardioid (above bafflestep) or dipole to hypercardioid (dipoles front radiation only). In both cases total radiation drops 3dB when radiation pattern is halfed to front sector. In theory.

This is all true in theory only. When I measured my fairly wide U-frame dipoles, I found they were pretty close to omnidirectional above 300 Hz, only below this did they start to radiate in classic dipole pattern. And even the tweeter radiation managed to "get around" 180 degrees quite significantly. So, actual measurements can't be beat...

Naturally. Here is a measured response of my Esotars in one waveguide setup. Throat isn't treated and small measuring baffle causes some diffraction problems to around 7,5Khz. As you see radiation pattern up to 60 degrees off axis is very similar than ideal dipoles front radiation is. -3dB at 45 degrees, -6dB at 60 degrees. Directivity increases with the dome beaming towards top treble.

But that's pretty much it. To 90 degree off axis response is "only" 12-13dB down where dipole has it's null. Actually the null isn't perfect in any actual situation, something around 30-40dB at best. With any baffle used null is very narrow and limited by the dipole "D" caused null points. And front radiating waveguided tweeter doesn't have total null to 180 degree angle. It's also something like 30-40dB down depending on baffle size and so on.

So the 3dB drop at power response is theoretical. Just like dipoles 4,8dB theoretical directivity index. It occurs in the range where system really behaves as a dipole. In my case something like below 500Hz or so (where efficiency drops 6dB/oct). Above that system is just dipolar but still directive with the nulls. In my case null at 1400Hz or so. It's visible up to 45 degrees off axis, over that system still obtains some directivity. Fortunately MTM null at 1-2Khz area assists the power response while horizontal directivity drops for a while. Allthought this doesn't seem to be a problem even in standard MT systems.

In this system I expect pretty uniform power response without the waveguide using 2nd order filter. Front radiation isn't uniform so placement and sidewalls are more critical.

Directivity index thought from a dummy: Dipole has 4,8dB. Front radiating tweeter without any additional directivity 3dB. Orions power response goes about 10dB down from 3-4Khz to 15Khz. This goes along with the 60 degree off axis response. 60 degrees off axis goes pretty much on same lines with 20 degrees and 0 degrees responses between 2,5Khz and 5Khz. So if my total system has 60 degree response 2-3dB down at 60 degrees off axis directivity should be something like 5-6dB before dome beams. Below 500Hz system has it's 4,8dB directivity, above that maybe a bit less. Transition from dipole to tweeter with 2nd order filter is between 1000Hz and 4000Hz so power response change from dipolar radiation and 4(,8)dB index to front radiating tweeter and 5(-6)dB index should be pretty smooth.

But this is just theory. Tweeters off axis response depends on dipole mids radiation pattern. But I guess difference in power response would be bigger with the waveguide. Top end drops too fast in my opinion but I can't help it. That's just how 1" soft dome behaves.

I have heard one system that had complete constant directivity. Standard WMT configuration, rather futuristic design with all drivers in their own schoolbook perfect cabinets. Woofer used 12" SLS in resistance cabinet and hypercardioid radiation pattern. Pretty ideal against room modes. Midrange, covering 200-2000Hz range, in resistance cabinet having cardioid radiation pattern. At the top 25mm Seas aluminium in very special horn loading which lead it's response to constant directivity up to 20Khz. So tweeters off axis response follows the pattern all the way, it doesn't beam. -3dB at 60 degrees, -6dB at 90 degrees, following the cardioid radiation pattern of the midrange. But this setup doesn't have much recordings that support it. Most albums are made for rolling off power response. Such system is just too bright...

Jussi