[JohninCR]

I try to stay active in some DIY forums, but it tries my patients at time.

I thought your patients were the DIY folks. Hmm

Davey, we're still waiting for the measurements you promised months ago

John, I still don't see a single solitary measurement from you. The OB/RLH/worlds only,etc,etc? That tapered H baffle mid (I think you called it a waveguide mid)? What ever happened to proving all the measurement geeks wrong? What about your promise to me?

I much prefer when I get results that agree with the theoretical

I still have no idea how you do that without measuring

C'mon John, point a mic at one of those contraptions of yours so we can see what's really going on 

cheers,

AJ

AJ,

I actually do use measurements, just very crude one's that I don't record.  BTW, I'm listening to my "tapered H baffles" right now....good one, no H ever dreamed of what these do.  Adding something to address edge diffraction on both pairs of OB-RLH's is one of the things I want to finish before the measuring extravaganza begins.  I admit that I've been dragging my feet and putting new ideas first, but measuring and documenting doesn't sound like nearly as much fun as building new ideas.  At least I share as I go.  Why don't you take a vacation to CR and help me measure all this stuff I have to do?  Then you can make sure I do it right.

[FlorianO]

Sorry for dropping the ball on this one, I first wanted to cover the basics before posting more, but work got in the way of going through the material

Speaking of which, John's post on DIYaudio forum some years ago (namely this:         http://www.diyaudio.com/forums/showthread.php?postid=482480#post482480 ) was really really helpful.

Here's my advice for what it's worth, since the number crunchers don't think I know what I'm talking about:
[...]

2.  Set your goals in terms of bass extension and SPL, along with a baffle size you can live with.  Your listening habits and musical tastes are key in setting your goals.  Then you can plug that info and driver Sd into DipoleSPLmax, and you can quickly back into what you need in the way of woofers.

3.  Woofer selection is going to determine what kind of fullranger you need.  If you need a lot of deep bass, then your woofers will be of a type that can't play very high, so you need a higher Q main driver which will reach down to the 100hz range.  This is where Martin's initial OB attempt broke down despite his "afternoon" optimizing simulations.  Those Dayton subwoofer type drivers simply weren't up to the task of blending with his Lowthers that start rolling off much higher than a typically recommended OB driver.  If you're set on using low Q fullrangers, then you have 2 choices, pro type woofers or multiple small woofers, both of which can be used higher in frequency, but you can forget about deep bass because the Fs is too high.

Many thanks for the advice John. Highly appreciated.

Wrt to simulations, we both agree, they must be reality-checked (I alluded to that in my post above) and I do find myself longing for some freq. measurements that confirm the simulation results.  Like Dick Olsher discussion around the Visaton NoBox project that has freq. response measurements to back it up (well, for what they're worth, of course).

Back on topic, smth which I found an extremely interesting read was  the discussion about low vs high Qts (here:        http://www.geocities.com/kreskovs/Box-Q.html). To sum it up, a large Qt (i.e. Qt=1.0) is more accurate in response but overshoots in transients, whereas a Qt=0.5 is overdamped but does not overshoot. But more importantly what makes it a "boomy" bass with large Qt drivers is the room response with 12 dB/octave increase in SPL towards low frequencies (bass traps, anyone ?).


Also, based on the front wave / back wave interaction I do understand (well, I _think_ I do...) why a thick baffle can compensate for its relative smaller width (ala Quasar), namely that a depth unit accounts for two width units due to the extra length added to the backwave path. Moreover, the added baffle thickness accounts for added stiffness and can only improve things. Of course, nothing is for three and the added thickness lessens the side canceling effects which -- as I understand it -- is the main benefit of flat OBs.  A valid question in that respect is how much that lessened canceling is audible (guess it's a factor of the wavelength since at longer wavelengths the thickness becomes negligible). (Again, this was covered at your own post above, just adding it here for referencing). All in all it makes Quasar quite an interesting design.


One last thing that still bothers me: We discussed only the baffle width (as the smaller dimension I guess) in determining the baffle diffraction loss point. Still, apparently baffle _surface_area_ is more relevant than the width, but I failed to grasp exactly why.

Moreover, in Dick Olsher discussion about the NoBox project he quotes that baffle dimensions (0.5m x 1.21m) as "approximating in extent the critical 100Hz half-wavelength, it's just right".  But he's not addressing baffle width at all. And the effect of the folded back wings (0.3m deep, but sloped).

IOW he's addressing only baffle length, but ommiting to discuss the width and the effect of the back-folded wings....

Am I missing smth ? Anyone care to comment on that ?

Thanks again,

Florian

P.S. Wrt to the NoBox design, it looks also quite intresting: back folded wings to add to the baffle width, but sloped (less back folds around the FR) _and_ supporting the front baffle for added stiffness....

[FlorianO]

Come to think of it, there are other issues with the Quasar design I haven't thought about:


First, the thickness of the baffle should be counted twice than above: Once from the back of the cone to the back of the baffle and then again to the front. Interesting to note that this doubling was not noted on Linkwitz site when he makes the equivalence from model f to model g open baffles (here: http://www.linkwitzlab.com/models.htm).

Second, and more importantly, if the drivers are mounted flush with the front baffle and if the cut-out through the baffle is cylindrical, then the thickness of the baffle will form a cavity that would resonate and modulate the back wave accordingly.

IOW all the drivers would be placed into U-frames (albeit pretty shallow ones). The question is:

Given that the tube cavity is shallow enough (156 mm in the Quasar case) compared to the diameter (242 mm) can the cavity effect can be ignored ? Note that U frames models I've seen (on Linkwitz site and on the musicanddesign.com) are under the assumption that the cylinder diameter is much smaller than the tube length, which is not the case here. Also, the 156 mm "tube" depth  here is the quarter wave for 550 Hz...

 I can see two possible variatons if that is a problem (will now go and re-check the models):

- flare the drivers cut-out towards the back e.g. liniarily, whatever.
- Leave them as they are and count for the U-cavity waveguide for the woofers. OTOH leave the FR on a "thin" baffle on top and avoid that alltogether.


Any thoughts ?

Florian

[johnk...]

Well, first when I said I don't post much any more actually I wasn't referring to this forum. There ia a lot of common interest here. Many of the other forums off little interest.

For a typical midrange driver on a 1" thick baffle there isn't any real concern with the depth of the cut out. Even rounding over the cut out from the rear doesn't change much in the dipole range. Its less obtrusive than the magnet structure and basket. But a 156 mm length tube even with a 242mm diameter (6.14" long x 9.52" dia) is certainly going to cause a resonance.



The low ratio of L/D = 0.645 makes the resonance lower in frequency and lower in amplitude too. Reducing D while keeping L constant makes the U-frame behave more like a 1 dimensional open ended tube and the resonance, while stronger, will move higher in frequency, closer to the theoritical 1/4 wave frequency.

[FlorianO]

Well, I appologize for that but I quoted wrong diam for the cut-out. It is actually 350 mm for the BD15 woofer.

Sorry for that

[Rudolf]

I can see two possible variatons if that is a problem ...:
- flare the drivers cut-out towards the back e.g. liniarily, whatever.
- Leave them as they are and count for the U-cavity waveguide for the woofers. OTOH leave the FR on a "thin" baffle on top and avoid that alltogether.

I remember a remark of Bert Doppenberg that the tube cavity in the Quasar is essential for raising the low end SPL of the FR. He explicitly voted against widening it.

[FlorianO]

I remember a remark of Bert Doppenberg that the tube cavity in the Quasar is essential for raising the low end SPL of the FR. He explicitly voted against widening it.

Well, nice to know, but I would like to know the reasoning behind that, thank you

Since we are talking about 8 in FR that works in conjunction with two 15 in woofers -- in a  bi-amped setup, nonetheless -- I would take that with a 1 kg grain of salt....

[johnk...]

Well, I appologize for that but I quoted wrong diam for the cut-out. It is actually 350 mm for the BD15 woofer.

Sorry for that

That would just push the first resonance down to about 280 Hz, according to the simulation.

Regarding Rudolf's comment. I would not recomment trying to use a cavity resonance of this type to boost the low end. This isn't a TL and the resonance it far too high. 

[FlorianO]

John,

First of all many, many thanks, not only for addressing this, but also for the plot above. I should have posted the right diameter to begin with . Again, sorry for that slip.

Regarding Rudolf's comment. I would not recomment trying to use a cavity resonance of this type to boost the low end. This isn't a TL and the resonance it far too high. 

My thoughts exactly (see my comment above. this gets a bit weird since we're both posting quasi simultaneously ). I would contend that a cavity resonating in the midrange wouldn't exactly help the FE206 . That's exactly why I mentioned above leaving the FR on a "thin" baffle: I would assume that the baffle stiffness is not so much of an issue for the 8 in FRer  (well, barred that the baffle receives also the vibrations from the woofers below) and make the cavity issue moot.

Thanks again John.

[Rudolf]

Just in case someone would take my comment regarding the Quasar as something like a recommendation:
I only tried to forward what Bert wrote. I haven´t listened to the Fostex version of the Quasar and don´t know whether that cavity changes the response to good or bad.

[FlorianO]

Rudolf,

I perfectly understood that, and my reply was all in good faith. While Bert is an expert designer, I have a very old habit of being skeptic of claims without proofs -- wherever those come from. The fact my (albeit beginner) first impression says otherwise doesn't really help.

Still, thanks for passing the info

Take care,

Florian

[scorpion]

FlorianO,

Regarding this last discussion you might like to take a look at the Jamo R90 design, http://www.jamo.com/Default.aspx?ID=5966&M=Shop&PID=17139&ProductID=17649 , and Nick Whetston's design for his Silver Iris review on TNT, http://myweb.tiscali.co.uk/nuukspot/decdun/poly_baffle.html . Thick baffles are just to avoid any baffle resonances with bad effects for the bass and midrange sound.

Imagine what Jamo could have done if they had used their aluminium also for magnet-mount instead af baffle mount. Nick's design is a simple
yet I think effective solution to the problem of avoiding baffle resonances. The Jamo is impressive in the way it is build and looks but also that
it is all passive and yet claims a frequency response of 25 - 30000 Hz from that baffle design.

/Erling

[MJK]

Come to think of it, there are other issues with the Quasar design I haven't thought about:


First, the thickness of the baffle should be counted twice than above: Once from the back of the cone to the back of the baffle and then again to the front. Interesting to note that this doubling was not noted on Linkwitz site when he makes the equivalence from model f to model g open baffles (here: http://www.linkwitzlab.com/models.htm).

Second, and more importantly, if the drivers are mounted flush with the front baffle and if the cut-out through the baffle is cylindrical, then the thickness of the baffle will form a cavity that would resonate and modulate the back wave accordingly.

IOW all the drivers would be placed into U-frames (albeit pretty shallow ones). The question is:

Given that the tube cavity is shallow enough (156 mm in the Quasar case) compared to the diameter (242 mm) can the cavity effect can be ignored ? Note that U frames models I've seen (on Linkwitz site and on the musicanddesign.com) are under the assumption that the cylinder diameter is much smaller than the tube length, which is not the case here. Also, the 156 mm "tube" depth  here is the quarter wave for 550 Hz...

 I can see two possible variatons if that is a problem (will now go and re-check the models):

- flare the drivers cut-out towards the back e.g. liniarily, whatever.
- Leave them as they are and count for the U-cavity waveguide for the woofers. OTOH leave the FR on a "thin" baffle on top and avoid that alltogether.


Any thoughts ?

I have been thinking about this issue for a few days and wondering if the baffle thickness really does play a part in the OB design and if the Quasar concept would indeed behave differently compared to a flat OB made from 3/4 inch plywood. So I added the thickness of the baffle as a variable in one of my OB worksheets and ran a few simulations.

Baffle size was set to 24 inches wide by 48 inches tall (similar to the Quasar design).
Eminence Alpha 15A driver mounted centered and 10 inches off of the floor.
Baffle thickness was set to 0.75", 3", and 6" with a tube like cavity extending behind the woofer.

I plotted the on axis SPL response and the polar SPL response for each baffle thickness. Bottom line is that the on axis SPL responses in front of the driver were virtually identical below 400 Hz, crossing the woofer(s) over to a full range driver in the 200 Hz range would lead to identical responses. Between 400 Hz and 1 kHz the differences in the magnitudes of the small wiggle were less than 1 dB. I then plotted the polar SPL response at 100 Hz and all three configurations produced a dipole figure "8" radiation pattern. The only difference was the rear lobe became a little more efficient as the baffle increased in thickness. The front lobe was essentially the same for all three thicknesses.

People can decide for themselves if they believe that simulations are accurate. But my conclusion based on these results is that the primary benefit of the thick Quasar baffle is to increase the stiffness of a tall relatively narrow baffle. The impact on the acoustic performance appears to be negligible.

Martin

[FlorianO]

I have been thinking about this issue for a few days and wondering if the baffle thickness really does play a part in the OB design and if the Quasar concept would indeed behave differently compared to a flat OB made from 3/4 inch plywood. So I added the thickness of the baffle as a variable in one of my OB worksheets and ran a few simulations.

Many thanks for the input Martin, highly appreciated.

With the risk of asking the obvious:

When you say you accounted for the thickness of the baffle and compared it with a flat OB, how exactly did you account for that  ? Double the thickness added to each side of the baffle to get the equivalent width  of a flat OB and then you compared it with that one ? IOW  took an OB with width W and thickness D and compared it with a flat OB with total width of W + 4D ?

In this case if we ignore any resonances in the tube cavities (and that's quite an assumption) the baffle of thickness D would be equivalent with a flat baffle with back wings of same depth (and ignoring any resonances in the latter too ).

Regards,

Florian

[MJK]

When you say you accounted for the thickness of the baffle and compared it with a flat OB, how exactly did you account for that?

I modeled a baffle that was 24" wide by 48" tall with the front of the driver mounted flush with the front face of the baffle. The back of the driver radiated into a short transmission line with a length equal to the baffle thickness. The terminus of the transmission line was flush with the back face of the baffle. The edge sources for the front and rear were separated by the thickness of the baffle. All dimensions were input as the real physical values without any "fudged" dimensions.

In this case if we ignore any resonances in the tube cavities (and that's quite an assumption) the baffle of thickness D would be equivalent with a flat baffle with back wings of same depth (and ignoring any resonances in the latter too ).

For a baffle less than 6 inches thick, there will not be any cavity resonances for the bass frequencies. Where you might see resonances is in the mid range and high frequencies. If you mount a full range or mid range driver in a geometry like this you need to be very careful balancing the depth of the baffle and the size of the short rear transmission line's cross-sectional area. Once the cross-sectional area is big enough with respect to the frequency being radiated, the acoustic impedance is almost purely resistive so you will not see any resonances just like in a well designed horn. If the cross-sectional area is not large enough with respect to the frequency being radiated, the acoustic impedence will have a mass and a resistive component so you will experience standing wave resonances. I really doubt that standing wave resonances behind the full range or mid range driver would be a desirable feature of the design's acoustic performance.

Martin

[JohninCR]

Martin,

Does the newest version of your spreadsheets handle the higher frequencies too?  eg  The problem I expect with the Quasar is that short pipe cavity behind the fullranger as I mentioned back on page 1.  For the woofer section, I think that baffle will just diffuse the region of greatest null and shift it slightly rearward from the baffle plane.  At least that's what's happened with similar shapes I've used for woofers.

[MJK]

The new worksheets calculate and plot the response up to 1 kHz. I have versions that go as high as 10 kHz but the run times become too long for iterating a design.

[JohninCR]

The new worksheets calculate and plot the response up to 1 kHz. I have versions that go as high as 10 kHz but the run times become too long for iterating a design.

I'd hate to do those calculations by hand.