Poor verticle directivity response - also pinging Svante

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have just played with the Edge and Xdir programs by Svante and they are absolutely excellent. They are the tools handy for anyone designing speakers.

But preliminary results raised my questions about how many speakers sound good at all.

Consider an example of a speaker with a tweeter to work with one or two 6.5" midwoofers, a very common choice of a two way design. The tweeter has a faceplate with a diameter of 140mm. A 6.5" midwoofer has a diameter of 180mm. Let's place them next to each other without any gap to achieve the best directivity response (findings from Xdir). So the distance would be (140+180)/2 = 160mm at the very minimum. So the highest frequency for the XO would be 344m / 0.160m = 2150Hz. Let's cross at 2150Hz because most tweeters do not like crossing at too low, even if LR4 is used. Because of the relative large sizes of the drivers and there is no gap, physical time alignment is not possible (neither is it practical for most DIYers). The driver physical offset plus the drivers' frequency responses would probably introduce a phase shift of around 70 degrees at 2150Hz, which is typical. All these numbers are the most typical, if not already optimized numbers chosen.

The following is the modelled response from Xdir:


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



The directivity does not look very good.

Most speakers may not have drivers placed so close to each other. Most speakers may not have the XO point to be this low. In other words, most speakers may have a directivity response worse than the above graph.

That is in thoery only. We know there are many good sounding speakers out there.

Any thoughts on this? Do speakers need strong directivity (like dipole) to sound good?


Svante,

Thank you very much for your programs. The Edge program is extraordinarily helpful. Just a question, have you done any measurements / experiments to prove the Edge graphs conforms to the real measurements? Also, with "Speaker source density", when do you use 1 or 2? Do you use higher density for woofers or tweeters? please explain how it relates to directivity. Thanks again.

Regards,
Bill
 
The directivity does not look very good.

Agree.

Most speakers may not have drivers placed so close to each other. Most speakers may not have the XO point to be this low. In other words, most speakers may have a directivity response worse than the above graph.

Agree.

That is in thoery only. We know there are many good sounding speakers out there.

Agree.

Any thoughts on this? Do speakers need strong directivity (like dipole) to sound good?

No.


b

1(1)
 

Attachments

  • odd-even.gif
    odd-even.gif
    84.6 KB · Views: 277
It would seem to me that dropping a crossover's frequency down right in the middle of where the human ear is most sensitive just to reach some theoretical ideal is throwing the baby out with the bath water.

Full-rangers may sound better in some rooms because of beaming, but you will never be able to rule out the fact that they don't have x-overs mucking things up throughout the midrange.
 
bjorno,

Unfortunately, no many tweeters can happily do 1671Hz. High excursion of tweeter never sounds good. The Esotar T330D I have has a faceplate of 140mm in diameter. The above graph is the best of what can be done with the Esotar T330D and the SS18W8545 I am planning to use.

Regards,
Bill
 
Complicated mather.

Basically I think constant or othervice well behaved directivity works best. But this doesn't mean high directivity is necessary. Full monopole is one type of constant directivity system.

There are good sounding MTM 2-ways out there but overally configuration isn't that perfect. Mids are too close to one another that construction doesn't direct at low frequences (depending on driver distances), then it beams (again depending on driver distances) and then spreads back open again when tweeter comes along. With non-waveguided tweeter difference is quite significant. Angles where direct floor reflection arrives is around 40-45 degrees vertical. To those angles non-waveguided tweeter doesn't have any directivity up to around 6-7Khz or so. Meanwhile MTM with +-160mm driver distance has 4dB directivity at 1300Hz, pretty much perfect null (in theory) at around 750Hz and partly formed null at 1850Hz (see the attachment).

Situation is fixed by using a 2,5-way crossover or using WMTMW configuration to add more directivity to bottom of the range. With gentle slope transition in vertical level is also quite smooth. I'd also recommend directivity improvements for the tweeter using waveguide, felt absorbers or similar.

Do speakers need strong directivity (like dipole) to sound good?

Depends what you consider being "good". It's all compromisses. Less directive systems have some strong points, directive systems some others.

Jussi
 

Attachments

  • mtm_+-160mm_4th_2150hz.jpg
    mtm_+-160mm_4th_2150hz.jpg
    80.2 KB · Views: 175
Jussi,

Thanks for your comments and your comments have always been useful to me. It reminds me the floor bouncing issue that I had not considered previously. I guess I may add on another thick rug on the floor if necessary.

Anyway, I am designing a new pair of speakers using the drivers I have on hands (Esotar T330D, 2 x SS18W8545 and 2 x SS26W8861 that should give a 93dB sensitivity). Although WMTMW seems attractive, I am thinking about separating the woofer box with the MTM so if the MTM does not work out well I can scrap the MTM but keep the woofer box to mate with a new MTM or something. I can't do so with a WMTMW. Another problem with WMTMW with 10" drivers is the wide baffle for the MTM. Yet another factor is that my small kids often run around and I am always worried that they may run into my 1.8m high 100kg weight? WWMTMWW box and imagine it fell over (it can fall if you push it)...

I am thinking about MTMWW. This should improve over the MTM as far as vertical directivity is concerned, probably not as good as WMTMW in thoery but should still give a large improvement. Interestingly, Using Xdir with a lowish XO point of 100Hz (or even 200Hz) the directivity does not seem to be much different between MTMWW and WMTMW. They all look pretty fine.

The SS18W8545 is pretty rough around 2-6k and the Esotar T330D has a 2dB dip before 6k therefore the passive crossover (LR4 at 2k) is quite difficult to design. After 8 hours playing with the Edge program, I have found some driver positions that allow the baffle response (up to 2dB bumps and dips) to nearly perfectly compensate the bumps and dips of the XO response. The combined response (after the default 6dB baffle step compensation from Edge) is flat from 200Hz to 1k, then slowing going down, up to -3dB between 3k to 5k, then moving back up 3dB. Does this correspond to a BBC dip? Did SL intentionally introduce a 3dB dip around this region in his Orion? This was not my original intention (I would prefer to have a 1.5dB dip and at most 2dB) but I had no choice - the driver physical offset (32-35mm) is not the easiest to deal with and I most certainly want to keep the number of components as small as possible - currently the passive LR4 XO is very simple with only a single LC to go with the two midwoofers, and a RCL with the tweeter, that is all. They are modelled with SpeakerWorkshop. The FRD and ZMA files rely on the manufacturer's datasheets. I don't have measurement equipment.

Of couse, I am placing all my trust into the Edge program and if it does not correspond to real life measurements, I would be in deep trouble. If it does, I will buy as many beers as Svante can drink if he comes to OZ.

I expect that the speakers may not to have the absolute pin-point accuracy like the Wilson's but may be only slightly less, while the speakers may give a large image and large depth and a strong pressence.

The only thing now worries me is your comment that when the M is too close to the other M in a MTM it does not direct at low frequencies. To keep the directivity right I have to place the tweeter right next to the midwoofer. Since the tweeter has a bit of vertical offset, the distance between the centres of the two Ms would be around 307mm, which is not far apart. I hope the WW beneath the MTM fix the problem.

Regards,
Bill
 
HiFiNutNut said:
Svante,

Thank you very much for your programs. The Edge program is extraordinarily helpful. Just a question, have you done any measurements / experiments to prove the Edge graphs conforms to the real measurements? Also, with "Speaker source density", when do you use 1 or 2? Do you use higher density for woofers or tweeters? please explain how it relates to directivity. Thanks again.

Regards,
Bill

Thank you.

Yes I have done measurements and compared them mainly with Basta! (which has the same engine for baffle diffraction, but also models the drivers). Even though The Edge/Basta! makes some simplifications of the theory, results are surprisingly good. It is rather informative to compare measurements and simulations, pretty often it opens the eye to little whiggles in the measured response curve and explains them.

There are of course things that are left unexplained, since neither of The Edge/Basta! simulates cone break-up, and often differences due to imperfect equipment shows up, at least if you are a poor DIYer :D

I had a look in my "archives", but it turns out that I am terribly poor at saving my measurements. I did, however find this "very serious" design, a box made of cardboard and tape, and a simulation of the box mounted on a wall in Basta! The measurements were made with the box lying on the floor like this:

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


For the tweeter the measured/simulated responses looked like this. The Behringer ECM8000 mic has an increase at high frequencies.

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


For the midwoofer, the measured/simulated responses looked like this. At low frequencies the match is rather good if I may say so, toward higher frequencies the driver has break-ups and these are not in the model.

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

WMTMW would indeed be better since it's symmetrical construction to both directions. But I understand your compromisse, personally I'm doing the same thing in my own system.

I'd say cross needs to be higher that 100-200Hz if you really want to punish that floorbounce. Or perhaps 200Hz is enough if you use gentle crossover slope, 1st order if possible. Broad overlap region is the key. More sound coming from large vertical space compared to smaller sources using steeper crossover.

I'd place the mids as close to the tweeter as possible and even then cross as low as possible. MTM lobing goes pretty wild if you run them high with large physical distances. You can patch the low frequency vertical directivity much better with the WW department.

Orion doesn't have "psycho acoustic dip" if I remember correctly. Phoenix had it. 3dB at 3Khz or so. But the overall responses aren't that great. Lots of diffractions coming from those sidepieces. I got my MTM with waveguided Esotar within +-0,75dB pretty easily. With unsmoothed responses. On axis off cource. Orion diffractions are propably the reason why it's CSD isn't that great. It has quite much midrange ripple. How audible, I don't know. But it could be better.

Jussi
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.