[HiFiNutNut]

David,

Thanks for your reply.

I don't think the Esotar is a disappointment but am curious of learning any better alternatives.

I thought of cutting the faceplate to shorten the diametre only after I built the cabinet - it was too late. But since I have damping in both the ceiling and the floor, off-axis response is less important, so the 140mm wide diametre of the Esotar still works for me.

I believe the "less solid" sound is due to the dip between 5-7kHz. I guess I could partially compensate this dip by a clever trick I worked out only yesterday.

Surely a compression driver (I have not heard one, actually) can deliver far more SPL driven by the same power. What if we have a high power amplifier to drive the Esotar? Would it still sound compressed comparing to a compression driver?

Regards,
Bill

[D OB G]

Hi Bill,

I'm interested of course to find out what the "clever trick" is.

With compression, I think it is a question of crossover point and slope.

Dynaudio used it at a bit above 2 kHz, I believe, at 6 dB per octave.  I've tried that, and I feel that if 6 dB per octave is to be used,  crossover should be no lower than 5 kHz (some distortion is apparent with 2 kHz crossover).  Alternatively of couse, a steeper slope.  Although SL uses the Millenium at 1.4 kHz (24 dB per octave), I'm inclined to think that to get best performance, reduced compression, the Esotar shouldn't be used below 3 kHz.  I don't know how SL does it, but there is only so much air a dome can move, which is much, much less than a waveguide loaded compression driver.

Dynaudio show a graph, with a 4.7 uF capacitor (which seems to be 3 dB down at 4 kHz) that shows NO compression up to 1000W peaks (10 ms).  So you're right about the high power amp.  (I think this lack of compression does set it apart from other domes).

David

[HiFiNutNut]

I can't hear any group deley of LR4 or below, so I use LR4 between 2kHz and 2.3kHz. That reduces the excursion substantially comparing to the way most people use it. For me it is important to keep a plenty of headroom for the tweeter, since this system was designed to have about 120dB SPL capability and 93dB sensitivity throughout most of the audio bandwidth. If it was not because I happened to have a pair of the Esotar T330D, I would possibly not have designed the speakers in that way.

With very little insertion loss (as I used active EQ for the woofer EQ) my 200W monoblocks should have more than enough power to drive my OB panels.

They seem to work fine (the speakers are still under development) so far. But I would be interested in comparing it to compression drivers since I have never heard them before.

To reduce the sonic signature of the 5-7kHz 1.5dB dip, an very low value inductor of 10uH-20uF, in parallel with a resistor of 0.22R (a fraction of a dB)  to 0.51R (about 0.5dB), connected in series with the tweeter in my very simple crossover (CL for the tweeter and LC for the woofer), would reduce the tweeter response starting from about 8-10kHz for 1/4dB to 1/2dB. You increase the inductor value to get a low corner frequency or decrease the value to get a higher frequency, and the resistor value determines the attenuation. I have both simulation and measurement to confirm the result of the above implementation.

Then increase the resistor value connected in series with the inductor which shunts to the ground would reduce the Q of the tweeter response, which can reduce the response below 5kHz.

In effect, the above reduce the response above 8kHz and below 5kHz therefore has an effect of boosting the response between 5kHz - 7kHz.

Although a couple of components are added, we know that resistors and inductors do not degrade the sound (while capacitors do) so the above method is OK for me to use.

Of course the whole circuit needs to be put into consideration. Different capacitor and inductor values of the main circuit result in different response of the tweeter.

Regards,
Bill

[Graham Maynard]

I always kind of thought he couldn't be quite right because everyone knows that a high Qts driver makes better bass in OB.

Smile.  More bass does not equal *better* bass !

[D OB G]

An update.

I began to get a sense that maybe the PHY wasn't as detailed as it might be.

I had a pair of Seas M15CH001 on hand.  Nominally 5".  Filled polypropylene.  Moving mass 5.8 g.  Neodymium magnets.  Very open basket as a result.  90 dB compared with 98 dB.

Result: quite an improvement!  A big surprise!  Definitely more detailed and, probably as a result, more dynamic.  An ideal OB driver.

The successor, same driver but with a magnesium cone, is used in the Jamo (nasty break-ups).

David

[D OB G]

An update to the update.

Still having doubts about the DE250s, I replaced the front waveguides with my Esotars and equalized with DEQX (so no little dips or shelves).

I can't deny it, as much as it would be convenient to do so, but the Esotar has more detail, and the soundstage is surprisingly more stable off axis.

So there is a mid/tweeter that hands over with a consistent polar response, but the tweeter obviously starts beaming i.e. there is a deficiency of power off axis.

So both waveguides are placed behind the baffle, in parallel, one pointing back, and one pointing upwards, on each side.  They are in series with a 3.5 uF capacitor.  So there is the opposite of a beaming response.  Output progressively increases with frequency, with wide dispersion, until about 15 kHz or so.

Because the waveguide loaded compression drivers are so efficient it is just a matter of padding them down and there is a flat on axis response, and a very good power response, approaching 360 degrees, becoming more dipolar as frequency decreases.

A big improvement (how many more improvements can there be ?!!).

David

[panomaniac]

Esotar?  Is that a dome tweeter?

[D OB G]

Yes, made by Dynaudio, and not available to DIYers for probably ten years.

They use the Esotar2 (neodymium magnets to reduce size) in their top of the line speakers.

[D OB G]

Since it is very easy to make sub-baffles to fit the main baffle I have now tried a large number of combinations.

Tweeters:

1.  DE250 in XT1086

2.  DE250 in XT1086 with rear firing DE250 in Parts Express 12" waveguide

3.  Esotar

4.  Esotar with rear firing DE250 in XT1086

5.  Esotar with rear firing DE250s in both XT1086 and PE waveguide (padding adjusted)

in all combinations with:

Midranges:

1.  PHY-HP  8"

2.  Seas M15CH001  5" (more like 4")

3.  Focal 7N303  6"

The final result?

PHY-HP with Esotar with rear firing DE250s in the waveguides (padded appropriately)

It's interesting how there is a synergy between certain drivers.  As I said earlier, I wasn't sure about the PHY-HP, but put it with the Esotar and the combination sings!

David

[D OB G]

I'm now an 18sound sorta guy.

I already had the XT1086 waveguide.  The 18sound NSD1095N compression driver, which is optimised for that waveguide, has proven to be better than both the DE250 and the Esotar.  It has a titanium nitride diaphragm- stiffer than other metals, but claimed to not break-up in the audio band (which measurements confirm).

With the midrange, I've added Fostex FE206E, modified by removing the whizzer cone, and Audio Nirvana Super 8 to the comparative list.  I've come away with the 18sound 6ND430.  This was rated very highly by Zaph, was in stock at a good price with the Australian distributors, AT Professional, because they are using so many of them successfully in pro applications, and, not that this has any weight, 18sound consider it the best driver they have made!

I'm crossing them over at 1.6kHz.

DEQX equalises, but to reduce the amount of work the DSP has to do, I run the mid through a 10 mH inductor, shunted by a 22 ohm resistor, to flatten the roll-off from the baffle to 90 Hz.  This is as low as the compensation can be achieved due to the rising impedance below that.

The 6ND430 has proven to be a good deal better than the other drivers, and I can highly recommend it.

David

[scorpion]

David,

I know you listen to classical music. How are your perception of the Symphony Orchestra 1st Violins playing at full strength affected by the 1.6 kHz crossover ? I suppose you can imagine the aim with my question.
I find the bows playing in high pitch the most difficult to reproduce correctly.

/Erling

[D OB G]

Hi Erling,

I've come to believe that the best sound comes from using drivers below their first break-up frequency.  I know that others believe there is a "critical" midrange to harmonic band, in which crossovers should be avoided.  I think that the break-up is the best thing to avoid.

With violins, I had a good tool in the form of the PHY-HP with respect to the wide range of crossover frequencies available.  There turned out to be no advantage, and indeed a disadvantage, to using higher crossover frequencies, and a lower frequency tends to avoid somewhat, I suppose, the Fletcher-Munson maximum auditory sensitivity region (if we say that it is centered on about 3 kHz).

Linkwitz and Geddes seem to try to get the crossover as low in frequency as possible (1.4 kHz for Linkwitz, and 800 Hz in Geddes "best" speaker).

I agree with you about the challenge of getting the violins right!

David

[scorpion]

Hi David,

I certainly appreciate your driver journey and opinon! I value especially your estimate of the 6ND430 which I have had on my short list for a long time. Not very many has ventured to test them. But it may very well be the unit to choose if going for the best. I also read 18Sounds own constructor comments about this speaker. It should be good. My MJAO, which is on a budget, I think is quite good but could be improved, as could almost all speakers. And the midrange is the weakest link I have concluded, althoug I specifically choose guite a bit above average both in terms of performance and price for this design. As Ric Schultz challenged me for time alignment http://www.audiocircle.com/index.php?topic=53821.100 of the MJAO I am also in the process of trying to better it althogether which is part of my current agenda. It can be a bit bettered by some new introductions. But it will always be something like 95 % perfect. And we all go for these last percentages.

Regarding 6ND430 there is no breakup until 4 kHz so why not test it up to 3 K. You seem to have all the means. I would be glad for an opinon. But going all that way up calls for a steep crossover and perhaps a notch for the 4K rise. Anyway I am impressed by your efforts to develop a speaker to live with long time. 

/Erling

[D OB G]

Hi Erling,

I've long admired your work with the Volks-OB and the MJAO.
Thanks for your contribution, which I believe has been a great influence, especially with your ingenuity and attention to detail.
You've probably noticed that above 1 kHZ the 6ND430 is flatter at 45 degrees than on axis (i.e. flat!). (i.e. cross speakers in front of listening position?).
Theoretically, if there has to be a crossover, I like to keep the acoustic centers as close together as possible. i.e. keep the distance much smaller than the wavelength at crossover. i.e lowest crossover frequency consistent with excursion limitations and onset of distortion.
In practice, I want the handover from mid to tweeter to have a smooth continuation of dispersion pattern.
I don't think there would be any problem with a notch filter, and I'm not worried by a steep filter (I tend to think that 6 dB per octave electrical, for example, can be made to work from bass to mid, but not so well from mid to tweeter- it would be interesting to find out how the MJAOs would sound with a mid/tweeter LR 24 dB per octave crossover!!), but there would be a discontinuity in dispersion crossing to the waveguide at 3 kHz (not much- but I think best to be avoided).

Regards,

David

[D OB G]

Thought I'd share some results of a bit of mucking around I've been doing.

All about baffle edge diffraction.

The baffle is 60mm thick and I've got a quarter circle round over all around (with the intention of reducing diffraction).

I wondered what the effect would be of adding short sides, to make a very shallow U baffle, which would end up without any sharp edges on the front of the baffle.

If the sides are short there should be minimal cavity resonance. 

At lower freqs (the baffle operates from 120 Hz up) it should still act as a dipole.  At higher freqs, around where the wavelength is shorter than half the width of the baffle, it will no longer be a symmetrical dipole, but there should be a reduction in diffraction.

I experimented with pieces of cork mounted around the sides of the baffle, from 20mm depth of U, to 200mm. 

The 200mm sides muddied the sound.

At 75mm deep (i.e. sides  of 135mm, with roundover) there was no negative alteration to the clarity,  but there was an improvement in imaging and sense of space.

So now I have to make some permanent arrangements!

David

[D OB G]

The Audiosyncrasy has new clothes!



The baffle is now 1200mm X 450mm X 25mm acrylic.

The bass baffle stays the same, but the mid has been reversed for the tweeter.

There is a cork covered piece of wood wedged firmly between the acrylic and the cork covered steel frame of the woofer baffle, which stabilizes the acrylic very nicely.

The sound is very clear  aa.

David

[scorpion]

David,

Real nice, indeed ! No 'Emperor's New Clothes' I am sure even if there might be some resemblance ! 

It would be nice to know crossover frequencies and the amount of EQ applied.

/Erling

[D OB G]

Thanks Erling.

Great turn of phrase!

I'm crossing over at 170Hz now, at 24 dB per octave LR.

The baffle starts rolling off at about 200Hz, so the DEQX is removing the hump as much as anything.

The mid crosses over to the tweeter at 1.6kHz (at an FIR equivalent of 260dB per octave!!!).

The woofers cross over to a sealed sub for each channel at 40Hz, 24dB per octave.

Work on the subs should finally bring the speakers to a conclusion.  There will be 4 X Peerless 10" XXLS drivers, back to back, per box.  Distortion should be minimal. 

[D OB G]

Many post ago I said that I hoped the speakers as they were then would be my "definitive"  speakers.

How crazy could I be?

The current incarnation continues the acrylic theme.





The baffle is narrower, but not as narrow as Rudolf and johnK make their case for, with respect to polar response.
I had to decide whether to step it in under the waveguide.
Mid is the tried and true 6ND430.
Tweeter is the DE250 on the Dayton 12" waveguide.  This requires an adaptor, which I have done my best to merge "seemlessly" (filling and grinding).
The waveguide is made of plastic, so I have covered the outside with a thick layer of plasticine (modelling clay).
It is filled with laminations of 30 pores per inch open-cell foam (a la Geddes).
There is a final disc of foam over the  front to complete the aesthetics.
The crossover is 48 dB/octave linear phase (DEQX) at 2.4 kHz.
I've used panomaniac's sonogram CSD plot as a guide to this frequency.




This has been checked against the polar response, and listening tests.

The woofer (the same EMI 1550) now runs from 80 Hz (24 dB/octave to the subwoofer- one each channel- soon to be two per channel) to 200 Hz at 24 dB/octave.  Only one woofer needed per side with that crossover. 
I've raised the sub crossover frequency.  Interesting that Linkwitz has now raised his from 40 Hz to 50 Hz.





It is mounted on a new sub-baffle.  The baffle doesn't touch the acrylic baffle, so there is space between, and the baffle is open on the sides.  It is spiked, with rubber cups where it sits on the main baffle base.  It is angled with respect to the acrylic, and the result is an absense of resonances.  The woofer is delayed with respect to the acrylic baffle.

The base is now far more compact.  Better (I wouldn't go so far as to say good  ) WAF.

Regards,

David

[scorpion]

David, clever mounting of the bass and separate but nonobstructing baffles. Real nice. 

I am a bit curious about the DE250 and the Dayton Waveguide. Do you prefer this combo over the 18Sound units ? Does the foam make a difference ?
I have used the same units and silenced the Daytons completly for own noise with bitumen but never worked it through inside the horn like you have done.
And I always have had the feeling of a hard sharp edge to the sound even off axis. More work pays off ?

You inspired me to try 6ND430 which I don't regret. Still in the MJAO baffles they play active with the standard Neo3 without backcup in dipole mode, thus, and the Alpha15 pair. I really like these units together and the sound they make. Crossover frequensis are 200 Hz 24 dB/oct and 3 kHz 48 dB/oct L-R. To me the weakest point now are the Alpha15s.  There might be call for change there somehow.

/Erling