[Oublie]

Hi Folks i'm new here and it's time for my next project.

I recently aquired 64 wharfedale drivers from ebay which i plan to run open baffle in a line array configuration alongside a full height electrostatic tweeter.

heres a pic of my idea.



the T/S params for the drivers are as follows

QES 0.63
QMS 2.84
QTS 0.516
FS 89.98
VAS 1.62L
MMS 5.3
RE 3.8
BL 4.24
SD 43.9

i notice willbewill did a similar project although he was only using a single tweeter from what i recall.

the unusual  thing about these drivers is that part of the magnet along with magnetic shielding is removable so first question is what effect would this have on the t/s params.

secondly can anyone give me any guidelines regarding baffle width and placement for the drivers.  I know that most of the ob's ive seen are dealing with single large drivers so this is a little different.  i more concerned about the effect of having such a tall baffle.  I'd like to use a fairly narrow but curved baffle (semi u baffle?) but am worried about bass rolloff.

my plan is to cross over to a 15-20 cm wide by 2.5m high curved esl panel from about 2.5- 3 khz up

i also plan to build a pair of inverted tapped horns for bass duties.

can a linkwitz transform be used in an open baffle setup to lower fs at the expense of excursion - no a lot but basically to get a flatter response down to around 100hz prior to crossover?

I'm not too worried about using up all my excursion or the onset of distortion as with 64 drive i.e. 32 per side i don't plan on getting anywhere near their xmax even when playing very loud.

thanks folks and any help is greatly appreciated.

[D OB G]

Hi Oublie,

Sounds VERY interesting.

Here are a few thoughts.

With the baffle, I'm coming to wonder if a curved baffle might have some advantages.  The dipole radiation pattern will suffer, but diffraction may be improved (it is from my experience).  I have just recently noticed that Visaton's NoBox uses a shallow "U" at the height of the midrange (mostly for structural reasons).  Supravox use a "U" with heavy chamferring i.e. approximating a curve.

Rear radiation is never symmetrical with front radiation at higher frequencies of a cone driver, due to obscuring by the magnet and chassis.  At low frequencies even a squared "U" baffle works. 

I'm coming to believe that, psychoacoustically, the "coherence" of the rear radiation is diminished in importance, not least beacause of the incoherent rear reflections, which do not arrive with a flat frequency response, due to variable absorption, or with coherent time versus frequency, due to the geometry of placement.  It would seem to be important that the woofer and tweeter towers are time aligned with respect to front radiation, but because of the precedence effect, it is the power response of the rear radiation that seems to me to be most important.

All this adds up to me supporting your configuration .

Magnet sounds unique?

I presume that removing parts of the magnet would raise Qt i.e. may be of benefit with an open baffle.

For width, and the effect of the magnet changes, I think you have two options.  Some measurement gear, or simulations. 

There is a free program called The Edge, by Tolvan, that allows you to try different baffle sizes, configurations, and driver layout, to produce an expected baffle frequency response.

His main program allows you to input TS parameters, to give a much more accurate result, which you would need if you are going to use a Linkwitz transform.

I believe MJK's program may do so also.

Depending on the diameter of the drivers (what are they?), I would think that 100 Hz should be achievable.

Hope this is of some help.

Regards,

David

[Oublie]

David,

Thanks for the info, the drivers in question are 4" so as you say they should be good for 100hz.

I think ill test the t/s params with and without the magnets and run a few sims with winisd to see the difference in response. I've run a few tests with win isd with the drivers speced in a closed box with an internal volume equal to the room the speaker will be going in (3500 cubic feet) In order to get and initial indication of the practicalities of running open baffle.

I've been using edge but without the proper parameter input it really only gives me a 'feel' for the response.

I think my next step will be to build a pair of test baffles, mount the drivers and see what i get.

I'm very glad to hear that the curved u baffle should work sufficiently well as i've been reading everything i can find regarding both OB and line array design including the work of Siegfried Linkwitz and Jim Griffin.

The reason i've decided to go with Electrostatic over ribbon which most people seem to opt for is three fold. I run a pair of stax Omega 2a headphones and absolutely love the sound but don't really like the idea of running huge esl's for midrange and high range duties plus i wont get the spl i really require at the lower end of the mid frequency range.  Secondly Comb filtering just wont exist as it does with standard tweeters even the little half inch ones.  And finally its cheaper for me to build a pair of esl tweeters than it is to buy a large box of tweeters (i Live in northern ireland so nothing comes cheap ) also i have both the step up transformer and most of the components for the hv dc bias supply already.


I have a personal theory regarding comb filtering that due to the nature of open baffles and particularly when used in a line array alignment that the rear wave will go quite some way toward mitigating any effects of comb filtering that could be experienced.  The fact is that although the rear wave does have less of an effect due to the basket etc the reflections of the rear wave and their uneven dispersion over the full length of the array should go some way toward averaging out any peak or dips in response.

Steve.

[Oublie]

I've added some plots and preliminary information to my photo album

here the link

http://www.audiocircle.com/index.php?action=gallery;area=browse;album=2294

[zobsky]

1. removing a shielding magenet should LOWER the Qts.
2. don't mount the driver line EQUIDISTANT from both the left and right edges

[lowtech]

If your goal is to build a dipole the baffle should not be curved.  (Edge) diffraction does not exist, in the conventional sense, with a dipole design.  I suggest you read all you can at the Linkwitz site - http://linkwitzlab.com

[Rudolf]

Oublie,
isn't there a major design fault in your project?
For a cross over at 2.5-3 kHz the centers of the drivers involved should be less apart than 1/2 wavelength of the xover frequency: about 6 cm. This applies for the horizontal distance even more than for the vertical. IMHO you need to place the tweeter line almost immediately next to the wharfedales. Placing two wharfedale lines as wide apart as shown in your EDGE simulation also calls for disaster. Try to simulate for 45deg to get the picture.

[Oublie]

Oublie,
isn't there a major design fault in your project?
For a cross over at 2.5-3 kHz the centers of the drivers involved should be less apart than 1/2 wavelength of the xover frequency: about 6 cm. This applies for the horizontal distance even more than for the vertical. IMHO you need to place the tweeter line almost immediately next to the wharfedales. Placing two wharfedale lines as wide apart as shown in your EDGE simulation also calls for disaster. Try to simulate for 45deg to get the picture.

True,  Comb Filtering could be a big issue (if i can hear it) but with the drivers i have I would have to crossover around 1700hz or thereabouts in order to stop this.  If it comes to it i could adjust the crossover accordingly.

I was thinking of an MTM configuration with the electrostat tweeter running down the middle of the panel thats why the spacing was there. Thanks for the advice i'll try a bit of remodelling and see how i get on maybe MMT is a better option.

Lowtech,

The curve is not really exact but was to give me a means of visualising the start of an idea.  The chances are i will end up with either a U baffle or B Baffle (my own little idea). From what i've read my baffle width will be as narrow as possible and there may even be a small open air gap running the vertical height of the panel between the esl and the mid.

I suppose the sooner i get my test baffles built and do some listening the better.  Have to source some 50% open area metal for the stators but i can get the mids up and running on test hopefully this coming weekend.

[Rudolf]

True,  Comb Filtering could be a big issue ... i'll try a bit of remodelling and see how i get on maybe MMT is a better option.

MMT is a very good idea IMHO . You could roll off the outer M column gently above 200-300 Hz (6dB). This would partly counter the inherent "bass" rolloff of the small baffle and avoid lobing at the same time.

Rudolf

[D OB G]

Hi Oublie,

Sounds like any advice I might have is clearly superfluous (or wrong!).

Hi lowtech,

As I alluded to in my little contribution, a curved baffle would not be a dipole at middle to higher frequencies, but is that the only reason for using open baffle?

Reading Linkwitz, is not a W baffle a dipole in the pass band involved? i.e. a dipole without a flat baffle over a certain range of frequencies.  In addition to the Supravox design, I notice that the PHY design approximates a curve. i.e. open baffle, but a dipole only over a certain range of frequencies.

As I understand it, there will be a "hole" developing in true dipole behaviour of even a flat baffle above, say, 1 kHz (Linkwitz obviously gets to 1.4 kHz with the driver he uses) when crossed over to a tweeter, due to the non-symmetrical output (depends on the radiation pattern of the tweeter(s)- whether front, front and back, or "dipole" tweeter too).

Kreskovsky gives the option of rear tweeter on or off (although his design is optimised for both tweeters).  Linkwitz originally didn't use a rear tweeter, and now that he does, I am under the impression that it is 3 dB down. i.e. a nominally dipole speaker does not have to be a dipole at higher frequencies. 

To clarify a principle for me, is a cone driver operating to, say, 20 kHz (no tweeter), say a B200, a dipole at higher frequencies? i.e. is there not overwhelmingly front radiation?

If it is not a dipole, isn't there going to be edge diffraction at higher frequencies?  Isn't true dipole cancellation the reason that conventional box edge diffraction does not occur?

By the way, can you help me here?  What role ACOUSTICALLY (or not!) do the side panels of his Orion have?  They obviously strengthen the baffle, and provide somewhere to hang a grill!, but since he uses magnet mounting for the midrange, the baffle is only supporting the tweeter(s). 

Regards,

David

[Rudolf]

If your goal is to build a dipole the baffle should not be curved.

I would not say that with your exclusiveness. It very much depends on the ratio wavelength : frequency how a radius will influence the response

(Edge) diffraction does not exist, in the conventional sense, with a dipole design.

Uhh? Did you read http://www.linkwitzlab.com/faq.htm#Q8 and http://www.linkwitzlab.com/diffraction.htm? Doesn't it show that diffraction in dipoles is twice as severe as in boxes? Look in EDGE, what happens, if you check off the "Open baffle" check box.

I suggest you read all you can at the Linkwitz site - http://linkwitzlab.com

Better read it twice ...

[Oublie]

Hi Oublie,

Sounds like any advice I might have is clearly superfluous (or wrong!).

David

Far From it David,

I Really appreciate your input on this.  A lot of the Open baffles i have seen used U baffle so it is an alignment which obviously works.

As you indicated in your last post if a curved baffle is not a dipole at higher frequencies i wont worry as i will be crossing over to the electrostatic panel which of course is a dipole tweeter and should work in a similar manner to Linkwitz's Orion when the rear facing tweeter was added. 

Is the change from dipole related to the width of the panel or the curve or is it a mix of the two?

My understanding is that a wider baffle allows higher frequencies to roll off due to the sound travelling along the width of the baffle but will augment the lower frequencies am i correct?

I can't wait to get my test panels built particularly because of this whole issue regarding comb filtering to see if i can hear it.  A number of both pro and amateur designers have said the issue doesn't exist for them due to the amount of room interaction and the fact that all speakers have a certain amount of combing to a greater or lesser extent.

Upon reading all that has been suggested the initial baffles will be built in the following way.

1. full room height
2. Minimal centre to centre spacing between drivers
3. Initially at least minimal baffle width
4. when tweeter is added a small air gap between the mid and tweeter panel to stop the mid seeing the tweeter as part of the panel
5. ill cut a load of different add on panel to 'tune' panel width and depth.
6. ill have to try not only panel widths but speaker to rear and side wall placement  to see what affect this has on the rear wave from the speaker and.
7.  Tweeter panel and mid with and wihout felt backing to drop spl of the rear wave.

can anyone think of anything else i should try?

[D OB G]

It will be interesting to see what you think of the felt. 

It is widely believed that the Quad ESL57s are dipole electrostatics, but in fact the back is heavily covered with felt, right up next to the membrane, and in their original form are designed not to radiate to the rear. 

When I had a pair I was single, so I could remove the protective mesh (kVoltage on the dustcovers!!)- sounded much better not surprisingly.  I also removed the felt from the rear- another improvement.  It seemed to improve the dynamics as much as anything.

So the speaker became a dipole, and IMHO sounded better for the change (although there was a reduction in the proportion of bass- also not surprisingly).  But what was the reason?  Dipole, or increased dynamic range- no compression (or both)?  i.e. dipole or open baffle?

Maybe you don't need a gap between the mid and tweeter panels.  Having the center of mid to edge dimension different on each side may give a smoother response as mentioned earlier.
(Apogee- trapezoidal! and Magnepans come to mind).

Another comment on the curved baffle.  Geddes maintains that the ideal shape for a dipole speaker would be a torus i.e. doughnut shaped (to minimize (or eliminate?) diffraction).  So your B baffle (maybe back to back Bs) may be on the way there.

Regards,

David

[Oublie]

Bit of an update,

I've started cutting holes but SWMBO got in the way and I had to replace a set of bath taps which some idiot had brazed to the pipe rather than attached properly.  this took the best part of Sunday as I had to cut and bend new pipes and try to  remove the old connections etc so I didn't get all my holes cut. 

I'm going to try and do a bit each night and see how i get on.

I have decided to rear mount the drivers and since the magnet has a connector  hole (M5 bolt i think) ive decide that this may be the best way to go.

I have now aquired about 20 8' lenghs of 10cm diameter cardboard tubing (inners from rolls of carpet) which will be used as part of the baffle.

I'm planning to try a flat baffle, B Baffle, Figure of 8 baffle and teardrop baffle with a set of tubes flattened to mimic a teardrop shape.

I'll try and post an updated sketchup later to give an idea of the plan.

[Oublie]

The teardrop is not to scale in terms of width but panel height and depth gives an idea of scale.

[Kludden]

Hi!
It looks like something i have done long time ago.
If i remember right it sounds very good.
Regards Kludden

[Oublie]

Woo spooky and there was me thinking i'd never seen the idea before.

did you you any tuning regarding the front to rear wave ratio i.e. more curve to the front or back?

are they open baffle? - they look closed/ported

i noticed you are running wizzer cones - any comb effects?

did you do anything regarding the materials used? - I'm thinking if i like the prototypes ill make the sides out of 50% open tubular metal mesh as my esl panel will be incorporated into one side and stuff the unused side.

[Kludden]

Hi!
The black part is a box, open in the back. It sounded better when i put in the polyester foam you see in the picture.
I have not done any meassurment, Just listen with my ear. The last I do before I made my MJAO clones was to take away the tubes and just put some wood instead. 95 mm on both sides of the baffel and in flush with it. This sounded better. I think maybe it was the distans or the canyon between the baffel and tubes that do something with the sound.
Maybe I should have tested two different sized of the wood, so the speakers not was in center. It could have sounded better, as some have mentioned before.
The tubes was carpenter tubes as yours, but i have painted them polyester and hardener ( the same you use for repering boots) They be very stiff, nearly like metall
The loudspeaker are still alive but today as my front speakers to my movie.
Keep on building.
Kludden

[Oublie]

Thanks Kludden,

I'm intending to have a good play around with baffle spacing / angles to see what can and can't be achieved.

Rear wave alignment will also be addressed but one step at a time.

Off centre spacing along with standing waves etc all need to be looked at.

Does anyone know what issues i would have if i had 2 curved esl tweeter either side in essence for my figure of 8 configuration have a curved tweeter baffle for each side will this give me problems a high frequency? Im thinking of ways to get tweeter esl surface area up.  Obviously this would mean the rear of the tube would also need to be open - probably 50% like the stator metal.

[D OB G]

Lynn Olson speculated about a perforated baffle, with hardly any holes near the driver, progressively increasing in density to the edge. i.e. more open towards the edge, to reduce diffraction.  But I think the perforation you describe MIGHT produce too much cancellation. i.e. if the tweeter is effectively open at the back, then the baffle more or less doesn't exist.

You can't have tweeters both sides of the mid-range tower and, as you describe it, I believe the time alignment discrepancy might be significant.

Regards,

David