Beyond the Ariel

One of the things that really bothers me about how the H1N1 swine flu situation is being handled is that little or no objective information is being disseminated about it. Instead, a set of drastic prescriptive measures is being implemented as if it were far worse than HIV. This puts me in mind of totalitarian North Korea's satellite/submarine launch, which they are still pretending to their captive people has been successful. This is very wrong, and the way the swine flu situation is being handled reeks of the same type of deceitfulness.
 
Thinking about alternatives to overcome some limitations in OB designs I have come to the conclusion that we might have to leave pure OB to proceed one step further.

Using a roughly 30cm / 12" wide baffle the first deep notch after the baffle peak is around 1,5 kHz.

No big problem for a fairly big mid - like my 8" Jantzen due to its smoothly increasing directivity – but already a severe limitation for the AMT tweeter with its very wide polar pattern at this frequency.

Net result is that there are three options left:

1. take a mid that has smooth response above 1.5kHz and increase XO to well above 1500Hz (my current 2.5kHz setup)
2. equalise the notch – which will result in peaks off axis as polar pattern is non-CD here
3. leave pure OB and go the way of wave guiding for that particular freqeuncy range.


First of all I had to jump my mental hurdle of not having ever considered (well – at least for a very long time) to use horns for home applications.

This thread really helped in this ;) - and after having done the hardest step and also got some start-up help - thanks Jean-Michel ! – I decided to do an experiment on dual hemisphere horns (or whatever the best name may be).

I've already built the first one – mouth roughly 30 x 15cm (12"x7" ) and 15cm (7") deep.

Two of them identically attached to the front and the rear of my AMT and you get the picture.

Being in the process of building a second "dual hemisphere horn" speaker for stereo and not having done any sophisticated equalisation but lowered the XO point to roughly 1,5kHz – my first audition is – well - very promising.

Has this been done previously? Anybody who has any experience with such an "dual hemisphere horns " approach?

Michael
 
skorpion said:
Here is a thread from Audiocircle 2007 discussing a similar subject: http://www.audiocircle.com/index.php?topic=36331.0 . JohninCR built front and rear waveguides for B200. I am afraid pictures are absent from AC at the moment so we cannot look at his efforts. But it was in the same direction that you are going now. Let us know the progress.

/Erling


Though physically similar, the result is *very* different between the two.

Presumably Michael has a high-pass filter that significantly attenuates the lower freq. response where the AMT driver (with waveguide) becomes increasingly "non"-directional. This method should substantially remove significant diffraction effects.

In contrast John's prototype will only add diffraction. (..it's a full-range driver operated lower in freq. than the waveguide's loading.)

*VERY* different results. ;)

BTW, the Neo 8 (std. version) is exceptionally well suited to the method Michael is experimenting with (..though with a much larger waveguide(s) and a lower operating freq.). Something similar, (though better loaded), to this could be done with the several Neo 8s:

http://www.audio-consulting.ch/?Products:Speaker:Rubanoide
 
As far as I remember the B200 was highpassed with an H-baffle housing a 15" bass below. But much detail was never given more than some photos. My remark was more in the spirit that 'nothing is new under the sky'. But I am certain, and also quite interested, that the approach can be a fruitful way of handling those bafflewidth - frequency - directivity issues addressed.

/Erling
 
mige0 said:
Thinking about alternatives to overcome some limitations in OB designs....


What limitations, specifically?


mige0 said:


1. take a mid that has smooth response above 1.5kHz and increase XO to well above 1500Hz (my current 2.5kHz setup)
2. equalise the notch – which will result in peaks off axis as polar pattern is non-CD here
3. leave pure OB and go the way of wave guiding for that particular freqeuncy range.


How about using multiple baffles, each one keeping its driver operating below the dipole peak?

Good luck with your dual horn. ;)
 
Took some time to do a decent measurement setup – Sorry for the delay in responding to you input.
I'm usually not so much interrested in FR and polar patterns - but have - at least slightly - changed my mind during the course.

Thanks for the pointer to the other thread - very much appreciated - and as already mentioned - yes my aim *is* different but may lead to comparable conclusions.
:)
I'm definitely not looking for constant directivity as a mayor goal – but from an equalisation point of view its a welcome gift if we get it "for free".

What I want to achieve is an improvement to the behaviour of OB above the first peak.

Depending on the baffle width with respect to the driver size there is a relatively deep notch following the peak. There are more peaks and notches to follow but usually they get less severe further up the frequency range.

As outlined, if I equalise for good on axis response that notch will show up as dip in the power response.


First pix is for the tweeter in dipole configuration.

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


This is a measurement – not a simulation !
I haven't done excessive equalisation, as its not needed to show the important thing – so there are some ripples left and above 18kHz there is no equalisation applied at all.

Next pix is for the same OB configuration – measured at angles of 0 10 20 30 40 deg

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


This traces ain't very specific for an AMT. Any 1" tweeter in an 30cm OB would perform roughly the same *if* it could be operated absolutely symmetrically (front to rear) as a dipole

If you look at the 1,5kHz behaviour you immediately see what I want to improve.
The peaks at 1500Hz are *not* due the baffle peak (this one is at slightly below 1kHz for that OB) . This peaks are an unavoidable "by product" if we want to get good on axis response for small speakers (tweeters) in a modest baffle.



With the next step we come to the solution I was looking for.
As the peaks and dips are caused by comb filtering – we successfully can try to avoid that in the frequency range of interest by adding some directivity by a wave guide / horn.

Next pix again is on axis measurement – but this time for the dual hemisphere horn

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


Again boring flat 4th order Butterworth HP ;) (though a completely different XO equalising of course)

Next pix is for the same double horn measurements at angles of 0 10 20 30 40 deg

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


Pretty well behaved considering the high zoom range – no?

What is most obvious is that we do no longer get the "stick out" 1500Hz in the power response – though it's still there – but at an attenuated level.

Telling form (fairly short term) auditing – I'm very happy with the results.
Overall presentation of this kind of "dual hemisphere horn" has a very smooth but still crisp top end and no roughness or "in the face " pattern at all. Nothing of the excellent resolution of the AMT got lost.
If I wouldn't see the horn in front of me – it could be a direct radiating speaker – beautiful soundstage as well as far as can tell for now.
What more could I ask for?

Michael


PS
don't get lost in speculations on the 10kHz irregularity.
I was in a hurry and didn't seal the cables.
 
mige0 said:

This is a measurement – not a simulation!
I haven't done excessive equalisation, as its not needed to show the important thing – so there are some ripples left and above 18 kHz there is no equalisation applied at all.

...

This traces ain't very specific for an AMT. Any 1" tweeter in an 30cm OB would perform roughly the same *if* it could be operated absolutely symmetrically (front to rear) as a dipole.

If you look at the 1,5 kHz behaviour you immediately see what I want to improve.
The peaks at 1500 Hz are *not* due the baffle peak (this one is at slightly below 1 kHz for that OB). This peaks are an unavoidable "by product" if we want to get good on-axis response for small speakers (tweeters) in a modest baffle.




Thanks for the measurements, very illuminating - especially as it applies to the general case of all open-baffle tweeters and midranges.

I've been wondering what would happen if the front and rear horns were intentionally mismatched in size and cutoff, so the ripples of each would be smoothed out. Maybe this would just make a big mess. The application I'm thinking of would have a single cone driver driving the front and rear horns (push-pull), so the two horns would "see" each other impedance-wise. I'm not sure if asymmetric or symmetric horns would be better in a push-pull application.

P.S. The "mystery" flu - H1N1 or not - is finally all gone. Still catching up on my sleep, but energy levels are recovering.
 
Lynn, no problem at all - my dual horn is exactly like that.

I know there is ongoing discussion about the impedance interaction of two horns at the same cone.

But for mine - I can look staight through the horn as the foil is pretty thin - beautiful yellow color when illuminated from the back... :)

So this experiment may also answer some of the basic questions floating around.
My guess is that it work at least as well as with one horn attached (though not tried)

Michael
 
The English expression "having your cake and eating it too" comes to mind for the dipole horn. You get better air-to-diaphragm impedance match at the lower frequencies, reducing diaphragm excursion and resulting IM distortion, as well as controlling bad behavior far off-axis in the same frequency range.

I'm thinking of a perverse twist on the Altec A-series short horn (as seen in the A2, A4, and A7 theater loudspeakers) with a dipole horn for the 414 12" driver. Since there's no need for extended LF response, I'm perfectly happy if the thing goes down to 100~150 Hz, and goes up to 800 Hz. The asymmetric loading between the front and rear horns might indeed smooth out the usual "horn bumps" that happen in the cutoff region of many horns - since this is a short horn, I'd like to operate it right through and below cutoff.

Quick question - did you use the larger or smaller horn in front of your AMT's?
 
mige0 said:
If you look at the 1,5 kHz behaviour you immediately see what I want to improve.
The peaks at 1500Hz are *not* due the baffle peak (this one is at slightly below 1kHz for that OB) . This peaks are an unavoidable "by product" if we want to get good on axis response for small speakers (tweeters) in a modest baffle.
The peaks at 1500Hz are due to the second baffle peak (the one above the first dipole dip). There is always the alternative approach to make the baffle not "modest", but reasonable for a dipole tweeter. Lets say 2" wide for the 1" driver. You will need to EQ 6-9 dB at 1.5 kHz to counter the baffle loss, but apart from that the response should be at least as smooth as with your double horn.
 
Thanks for your replies!

I finally found time to search for the post I was referencing to, it's this

http://www.diyaudio.com/forums/showthread.php?s=&threadid=144153

About 10 years ago I built an MTM using 4 of the well known Vifa P13WH. These speakers never got much use often going unplayed for a year at a time. Recently I decomissioned them to make some prototypes. After some stange results I measured them and the Fs was in the 150 to 160Hz range vs 65 to 70. I've run them overnight with a low frequency and only had maybe a 10 Hz drop.

That's a real pity as the Ariel is the only really interesting 'high-end' DIY-project I'm aware of.

Of course I greatly appreciate if you guys point out other reference projects that I might be missing.

Have fun, Hannes
 
To complete the course - here are the measurements for the AMT in the same baffle as for the dipole measurements – but with closed back volume – operating as "monopole":

first pix on axis – (ACOURATE) business as usual ;)

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



next pix AMT as monopole at 0 10 20 40 deg:

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




And last but not least
to get an idea of the benefits regarding (IM, thermal, etc) distortion for each setting of monopole, dipole and dual—hemisphere-horn:


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



- flat line is for reference taken from the AMT in monopole configuration
- with roughly 2 dB of gain we see the dipole configuration and
- with roughly 10dB of gain we see the dual horn configuration.

All measurements were taken at a diaphragm to mic distance of 1 Meter



Michael
 
Even from the mere measurements above we possibly could conclude that the dual horn approach is a pretty good idea to be blend over to a OB dipole mid.
:)

Possibly for this size of double horn it might be best to cross over around 1500Hz with active or even higher in passive – as Jean-Michel Le Cléac'h told us to not forget about the large increase of group delay towards cutoff.

--------------

Lynn Olson said:
The English expression "having your cake and eating it too" comes to mind for the dipole horn.

:snoopy:


Lynn Olson said:
Quick question - did you use the larger or smaller horn in front of your AMT's?

There are two identically horns attached to the front and to the back.- strange look ;)
AMT is as front / back symmetrically as are electrostats for example.



cuibono said:
Nice work Michael!! :eek:

Would you write a little about the horn - its dimensions, how you calculated them? A picture?

Incidentally, I was just wondering what the max SPL of your AMT is? Its a Beyma TPL-150?

Good work, and thanks!

Horn mouth is roughly 15cm x 30cm - if we measure the mouth at the 270deg position.

I asked Jean-Michel Le Cléac'h for help as I'm not really experienced with designing horn contours.
Many people did positive comments on his contour emphasising on aspects that I like to have too.
His horns also have the advantage of kind'a built in "edge treatment".

What Jean-Michel came up with was a "quasi-cylindrical wavefronts horn" - as he calls it.

The contour – for me – looks like his Le Cléac'h horns look like – so I guess it actually *is* a Le Cléac'h double hemisphere horn.

Pretty brute force OB edge treatment if we think about...
;)

As for the AMT – have a look at Stig-Eric's thread as he has experience with several different AMT's.
http://www.diyaudio.com/forums/showthread.php?postid=1797744#post1797744

Basically it does not matter too much which brand you take . The benefits of AMT are built into the design.
If you would like to go for cheep I'd even recommend to give the Aurum Cantus a try
http://www.diyaudio.com/forums/showthread.php?postid=1778053#post1778053

Max SPL ?
Possibly I'll do measurements when all neighbours are on vacation
;)

Rudolf said:

The peaks at 1500Hz are due to the second baffle peak (the one above the first dipole dip). There is always the alternative approach to make the baffle not "modest", but reasonable for a dipole tweeter. Lets say 2" wide for the 1" driver. You will need to EQ 6-9 dB at 1.5 kHz to counter the baffle loss, but apart from that the response should be at least as smooth as with your double horn.


Maybe – depends on how we look at it.
Simulations show different than my measurements,
If you look at the dipole gain in my previous posting – you will notice one valley at around 1500'Hz and the next valley at around 4kHz.
This should correspond to the notches / dips I mentioned – no?


Rudolf said:

There is always the alternative approach to make the baffle not "modest", but reasonable for a dipole tweeter. Lets say 2" wide for the 1" driver. You will need to EQ 6-9 dB at 1.5 kHz to counter the baffle loss, but apart from that the response should be at least as smooth as with your double horn.

Agree - but you heavily loose distortion free headroom if you make the tweeter OB that small - plus - we can't shift the issue out of pass band with AMT's of 80 – 100mm / 3"- 4" width.

Both no real options for my project that is intended for monitoring at high SPL levels as well.

By the way - I tried already – but double horn is by far the more advanced solution
;)

Michael