This is the way I understand it. The amount of toe in depends on radiation pattern. Generally as you gradually toe them in, you will hear the focus becomming clearer. When the axes cross in front of the listener, the transients gradually become muffled. This does not mean once you have the right toe in, that your going to be able to move laterally and have the image at the same location, but it does reduce shifting as you mentioned.Russell Dawkins said:
MBK said:
MBK said:
Both factors are true and of equal importance (and discussed that way in my papers) and the designs were done to accomodate both of them without compromise. If you look at the radiation pattern of the ESP line you will see a gradual reduction in level as one moves off axis. This makes the "sweet spot" much wider as I think that you can understand. If one move back slightly as one moves to the side the effect is even more pronouced becaues the time of travel is also being compensated.
Only a highly controlled pattern with this precise falloff will have this effect. You can't just toe-in any old speaker and expect it to work the same as Duke notes.
There are so many factors in a loudspeaker design, which is why one must get a handle on what are the important ones. Directivity is very high if not on the top and yet is left relatively uncontrolled in almost all designs. Big mistake IMHO.
gedlee said:
Oh I don't doubt that, it is clear from the concept, I just didn't know it worked so well for the phantom images when done properly. As I noted my own model has the drawback of the flat baffle tweeter here and so is not really suitable for this solution - something I had wanted to fix for a while with a waveguide but haven't gotten around to yet. But I still like to experiment anyway. That's how you learn.
I find it strange that few if any people ever discuss this issue. People talk about the sweet spot issue - in itself not so disturbing, if inconvenient - but no one points out the far nastier psychoacoustic (dis-)illusion of the phantom image moving in the wrong direction when the listener moves (very disturbing, to me).
Here's another try at widening the sweet spot using a Defined Coverage horn. From people who have heard them they work quite well. Link to the original Project Everest developed in the 80's. Similar horns were used in the 2600 and 3100 speakers from the late 90's
http://www.audioheritage.org/vbulletin/showthread.php?t=5671
http://www.audioheritage.org/vbulletin/showthread.php?t=5671
MBK said:
I would agree "that few" people discuss this issue, but not "if any". I not only discuss it but I solve it.
Robh3606 said:
I have to admit to being strongly influenced by this paper and that project. John Eargle was a good friend and I have read virtually everythng that he ever wrote. His AES papers on Constant Directivity were foundational. I consider what I am doing as the next evolution of what John started. He had the right ideas all along, he just lacked some of the more advanced concepts in the waveguide development. John was more of a recording engineer than a scientist - especially in his later years.
but when you calculate the needed directivity to cancel out the Haas effect (amplitude to time compensation), you get a directivitiy factor that is much bigger than what you can find in most direct radiators or horns ! Plus, you need the directivity to stay relatively constant to quite low frequencies : not so easy.
The idea is old, very interesting but difficult to realize.
I did this work in 1981, I have to scan those old documents !
Based on DIT and DII models from Kates, Haas compensation curves from Meyer and Shodder and models of hearing localisation from Franssen (time has passed, I agree), I calculated "optimal" directivity curves for stereophonic setup (2 speakers). Calculations gave me directivities with about -10dB at 30° (and quite constant from 300 and above). And it means speakers really "toed in". But what I could never really do is to build speakers with those characteristics and see if theory is wrong or right !!!
Because even at 500Hz, I think that it's not so easy to achieve those figures without having a really big horn or waveguide (big for domestic use, I mean). Another solution is with cardioid directivities and maybe this is feasible even at low frequencies (ie Geithain RL901)
Earl, I haven't seen your directivity figures but would be interesting...
I think that it is quite easy to make speakers that are really near perfection in mono in an anechoic room. What is missing is to be perfect for binaural hearing into a real room...
jlo said:
This completely contradicts your first statement as this would imply a directivity of about 45 dgrees to the -6 dB points - which is EXTREMELY NARROW not wide as you stated. I would completely concur with this number as it is exactly what I design to.
I would also prefer to go down to 300 Hz with this narrow coverage, but this, as you say, is almost unfeasible. It would require a waveguide of about 50" in diameter. My results indicate that going down to about 500 Hz is a very very good compromise.
My speakers are toed in at about 45 degrees.
JLO
Further - sorry, I may have misread your first post. I took the ambiguity to mean that you were implying that a wide directivity is more desirable, but I see that you could have meant that it needs to be very narrow. This I concur with - far narrower than any piston could achieve, but eminately possible with a good waveguide.
My appology if I misundertood you.
My design achives -6 dB at 45 degrees and holds this virtually flat from about 800 Hz and up. Toed in to about 45 degrees, the lsitener - on center about 10 feet back - is off axis about 22.5 degrees. As they move from side to side the falloff in level from one speaker is large while the increase from the other is not as great but does exist. This is very effective at controlling the "sweat-spot" even if it is not theoretically optimal (the later being impractical IMO).
Further - sorry, I may have misread your first post. I took the ambiguity to mean that you were implying that a wide directivity is more desirable, but I see that you could have meant that it needs to be very narrow. This I concur with - far narrower than any piston could achieve, but eminately possible with a good waveguide.
My appology if I misundertood you.
My design achives -6 dB at 45 degrees and holds this virtually flat from about 800 Hz and up. Toed in to about 45 degrees, the lsitener - on center about 10 feet back - is off axis about 22.5 degrees. As they move from side to side the falloff in level from one speaker is large while the increase from the other is not as great but does exist. This is very effective at controlling the "sweat-spot" even if it is not theoretically optimal (the later being impractical IMO).
jlo said:
From this sentence, I understand that jlo means you need a high directivity factor, which means narrow directivity. Or ?
Anyway, very interesting topic. Because I couldn't read everything yet: gedlee, is there a chance to benefit from your results with other sources like dipoles/cardioides or is their directivity just to wide ? I mean, of course not to the same extent as with waveguides, but still?
gedlee said:
Do you think that a pair of 12 or 15 inchers per speaker mounted side by side could help to increase horizontal directivity in the 300 Hz region ?
I'm thinking of something like TAD TSM2401 :
Another design I'm thinking of is the compact, full horn, and EQed Onkyo GS-1 :
An externally hosted image should be here but it was not working when we last tested it.
( http://page.freett.com/knisi/gs-1.htm )
Obviously a TSM2401 speaker is way bigger than a Summa and a GS-1 is more difficult to build and requires (probably active) EQ.
IIRC in your Summa paper you say that the room must provide some damping in the low range to compensate for the speaker being omni in this range. Do you think that the TAD and Onkyo designs could be of some help here, reducing the need for room damping to some extent ?
bzfcocon said:
The problem is that these types of sources are not constant in directivity. A dipole is a dipole only up to a point and then it narrows more like a piston. Cardoids will have the same problems. Unless you use the dipole only over a very narrow range where it is a true dipole, its directivity is not being constant necessitates the use of a lot of sources since the bandwidth of CD is pretty small. This leads to directivity problems from the multiplicity of crossovers, and before you know it the design has gotten complex and unworkable.
I looked at both dipole and cardiod designs (for the midrannge)and found the above to be the case. These types of sources do have the potential to lower the controlled directivity (CD) down about an octave below what a waveguide can feasibly do, but the cost in terms of crossover design, enclosure size and overall complexity is very high. And in the end there does not seem to be a great deal of benefit, because, as I said, it does not appear that one really needs to go all that low in frequency to correct for the "sweet spot" effect. Our hearing is simply not that accute at those frequencies as regards localization.
I think if we know what the dispersion angle is at different frequencies, it is quite easy to plot an angle of interest and toe in the speakers such that the relfection route hits at least two walls before reaching the listener, then you have a good toe in angle. the thing is that now since the center spot might be so much off axis that you might experience roll-off at the high's, so it's necessary to select speakers that have a slight rise in on-axis response and somewhat flatter off-axis response, then the match is good.
Giarsun said:
Yes, multiple woofers is an option that I have considered and even have designs for. But as you say, the practicality of going this route is questionable. And one would ideally like to narrow the directivity in both the vertcal and horizontal planes. Four LF sources can provide that and placing the waveguide in the center makes for an attractive coincidence of midrange and LF sources. But this is getting even more impractical.
I deal in the practical world of what can be made and sold resonably. 99% of the customers don't like unreasonable design solutions.
I think that you misread the Summa paper. The LF damping is primarily needed to smooth out the LF modal overlap. The higher power response of the monopole source in these kinds of rooms tends to offset the higher absorption in this region yielding a smooth and nuetral perceptual response.
FYI to the Chinese readers
http://www.almainternational.org/content/ALMA_China_2007_Meeting_Agenda_Chinese.pdf
http://www.almainternational.org/content/ALMA_China_2007_Meeting_Agenda_Chinese.pdf
gedlee said:
I think this is what Rey Audio tries to achieve with their Warp series :
( http://reyaudio.com/warp-e.html )
The "smaller" model using four 8 inchers is 2.6 ' wide, but I think it could me made narrower.
Thanks for that clarification.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.