Thanks.
Really curious about the waterfall plots, that haven't been published yet.
There is a reason why simulation tools exist so we don't guess and oversimplify things and make us look uneducated at the same time.
S1 line would be an 8 ohm resistive speaker, other curves are self explanatory.
Good heavens! You think I meant by all of this that you are uneducated? Have you ever tried to build such a circuit with the simulated values, and it performed exactly as you expected? Then you must be very lucky, because it never worked for me, at least not with the calculated values.
Believe me, there is MUCH more involved in speaker design than science, theories and simulations. Intuitions, a good inner hearing, sometimes sheer luck can play just as much here as the factual part. I also understand that most people today NEED solid facts, formulas, diagrams that can be proven by science, to understand and accept how something works.
I am not going into this any further, maybe after all I am really ignorant and uneducated, all I was trying to tell is that -according to the ancient wisdom- there is more than one path that leads you to the river. And my way is different, but sometimes even I can get there...... Good luck!
Last edited:
Never seen TG using so many notches in an xover!😱
Guess he didn't really enjoy the homework...😀
Btw, it always seems that TG cloning his previous designs when trying a new one. Always the same curves, though this time he had to use more caps and coils to achieve the TG sound...🙄
Btw too, like the LR2 target: a wide band midwoofer helps!
Last edited:
dragon, I was merely pointing out that I prefer to look at the facts about dissipation and the stuff that involves Ohms law and so should everyone else interested in electronics.
@GDO: TG is consistent with his view of good sound and I think that's fair enough.
@GDO: TG is consistent with his view of good sound and I think that's fair enough.
Last edited:
dragon, I was merely pointing out that I prefer to look at the facts about dissipation and the stuff that involves Ohms law and so should everyone else interested in electronics.
I understand. I -on the other hand- was trying to point out, that in the case of loudspeakers- involving rapidly changing AC signals in a rather complex electromechanical system - there is much more involved beyond Ohm's law.
In fact I am not aware of any simulation software or anything else that gives satisfactory -"in vivo"- explanation about all aspects of such a system.
There are countless static models exist which tell you very little about the numerous interactions when feeding the circuit with actual musical signal.
Last edited:
Of course all of us have such a "built in" analyzer in and on our head, but the interpretation of the results is strongly subjective, biased, and lacks any scientifically demonstrable factuality... 🙂
Last edited:
Well, I haven't tried to do such measurements yet, but I guess it can be done by feeding the X-over-speaker complex with the resonant frequeny of the RLC circuit, and observe (record) the waveform at the speaker terminals with a suitable device (storage scope for example). But then again this will show you (if it will show anything at all) only the static behaviour of the circuit at that frequency, with complex signals -such as music- I have no idea how it could be done.
One thing I know for certain -based on countless listening tests- that the inclusion of such a circuit in te XO somehow robs the life, the essence of the music, my finding was the lack of attack and fine detail. It is likely that the sudden sharp changes in phase and impedance response what are responsible for this, but I cannot really explain it any further.
I can relate to this. But then again, if you have to use high Q deep notches, then the drivers or the design might be bad. It should be done that way only out of passband - killing the resonances of cone break up.
That doesn't sound right - passive or DSP.
cheers
Last edited:
When break up modes show mostly on the axial response, putting a notch is simply a big shot to the power response.
Better chose a better design axis, or a different driver...
If you cross that LR4 at say 1600-1800Hz and add a notch or a tank filter at cone breakup (4.2KHz), i don't see the problem or how could that affect power response in a negative manner.
If you cross that LR4 at say 1600-1800Hz and add a notch or a tank filter at cone breakup (4.2KHz), i don't see the problem or how could that affect power response in a negative manner.
Yes i agree with your "out of the passband" comment...🙂
I insist that i find a terribly conventionnal and bad idea to choose the axial response as the design axis, because with wide band drivers it is always the worst possible design axis. Only good for the reverse-all-that-junk fir junkiards...😀
Last edited:
Problem exist because people are listening sweeps for distortion less than 1% (-38dB), but then leave cone resonance at say -25dB or even -20dB and that's "good enough". If you simply go off axis, the problem is simply hidden - not solved. You can not make that amount of energy magically disappear.
Cone resonance must be treated properly for a loudspeaker to sound right - either by steep xover or by notch filter or by a combination of both.
As for irregularities in the passband that require notch filters... Geddes had 2 notch filters on Summa 15 (predecessor of NS15). JBL 1400Array also. You can buy ruler flat drivers, or you can do it by implementing a good xover.
Cone resonance must be treated properly for a loudspeaker to sound right - either by steep xover or by notch filter or by a combination of both.
As for irregularities in the passband that require notch filters... Geddes had 2 notch filters on Summa 15 (predecessor of NS15). JBL 1400Array also. You can buy ruler flat drivers, or you can do it by implementing a good xover.
Last edited:
......
I insist that i find a terribly conventional and bad idea to choose the axial response as the design axis, because with wide band drivers it is always the worst possible design axis...
On axis is important as off axis is. Loudspeakers xover should be designed to work well on and off axis.
We should use term "listening window". Say +/-15deg vertical and +/-30deg (at least) horizontal.
I can relate to this. But then again, if you have to use high Q deep notches, then the drivers or the design might be bad. It should be done that way only out of passband - killing the resonances of cone break up.
cheers
Well this is exactly what my problem is... most of the newly developed 5-8 inch rigid cone drivers suffer from serious breakup modes somewhere between 4-10 kHz. (besides their usually rising response, as all of us know). I've seen examples with 20 dB on axis peak, but 10-15 dB is typical.
I admire their bass attack and precision, but if you don't filter out those -sometimes enormous - peaks, they ruin the overall sonics of a design, regardless of what type and slope XO being used. When I put in the necessary notch(es) to get rid of them, then most of the slap, the "speed" in the bass range suffers, and no, NOT because of the missing harmonic content in their upper range...
I am not trying to state that I am an enlightened or omniscient audio guru, but I've listened many-many hundred speakers in my experience, so this a real problem, not the "newbie is hearing things" category.
Also this does not mean necessarily that a given driver , or design "bad", just this is the way they were designed (you must sacrifice something to get something you want), se we have to live, work with them accepting these facts. After all you cannot always use 15-20 years old drivers in new designs....
...Well this is exactly what my problem is... most of the newly developed 5-8 inch rigid cone drivers suffer from serious breakup modes somewhere between 4-10 kHz. (besides their usually rising response, as all of us know). I've seen examples with 20 dB on axis peak, but 10-15 dB is typical...
Are we talking series or parallel notch filters (or both) that you hear differently - less vivid ?
Every Accuton, Seas Excel Mg, Seas Prestige AL, Wavecor AL or Illuminator AL suffers from resonance peak at cone break up. That's normal for hard cone drivers because of their pistonic behaviour.
I've had quite bad experience with soft cone speakers (plastic) - lack of detail and intelligibility, less vivid sound.
Last edited:
On axis is important as off axis is. Loudspeakers xover should be designed to work well on and off axis.
We should use term "listening window". Say +/-15deg vertical and +/-30deg (at least) horizontal.
Sure but you will not get flat at all directions, mainly in your listwning window .
So if you leave the axiaal response out of it , you dont have to worry about havving a rising axial response , you simply need to take care that it is smooth enough.
Are we talking series or parallel notch filters (or both) that you hear differently - less vivid ?
Every Accuton, Seas Excel Mg, Seas Prestige AL, Wavecor AL or Illuminator AL suffers from resonance peak at cone break up. That's normal for hard cone drivers because of their pistonic behaviour.
I've had quite bad experience with soft cone speakers (plastic) - lack of detail and intelligibility, less vivid sound.
Gosh, I got confused in my deleted post (language problem...😱)
I don't live in english speaking territory for 17 years, so my skills start to rust.😎
I usually find series notches more harmful, than parallel ones. I use parallel type (only when really necessary) to reduce edge reflection "bumps" at 1-2 kHz, but they seem to hurt less than the other type.
Agreed, cone breakup mode is (almost) unavoidable with pistonic mode drivers -at least with the recently used materials- but with smart combination it can be reduced to minimal. Good example is the not so new SB 17 CRC midwoofer with its Rohacell-Carbon fibre sandwich cone, where the peak is a "mere" 10dB, and can be tamed wery well, while maintaning the advantages of a rigid cone driver. This is one case where I was able to get away without using any kind of extra filtering (crossed at 1.8k to an SB26ADC)
Same opinion about plastic cones, they don't play any more in my party.....
Attachments
Last edited:
The LRC filters neglect the dc resistance of the coil. So a series resistance should be added to the coil at least for a series LCR. That coil resistance may ofcourse affect the sound.
An elliptical shape e.g. in a tweeter has (in vertically mounted position) some potential advantages:
This is one of the most interesting (and btw. self-evident) interpretations of the "dome tweeter" in the last decades IMO: One should simply appreciate that attempt ...
http://www.scan-speak.dk/?page_id=638&single_prod_id=241
However it seems to me, that further attempts could be made, to have a faceplate or waveguide even further matching/supporting/reinforcing that particular radiation characteristics of the transducer.
- dispersion in upper highs can be wider in the horizontal plane than in the vertical plane, which serves to lateral room reflections being less colored (less suffering from "dullness")
- also having the "primary listening window" for uncolored direct sound wider horizontally
- while having the (gradually) desired reduction in energy radiated towards upper highs by narrowing radiation in the vertical plane prefereably
- also breakup modes can be mitigated by deviation from a circular shape ( if design is accounting for that in detail also)
This is one of the most interesting (and btw. self-evident) interpretations of the "dome tweeter" in the last decades IMO: One should simply appreciate that attempt ...
http://www.scan-speak.dk/?page_id=638&single_prod_id=241
However it seems to me, that further attempts could be made, to have a faceplate or waveguide even further matching/supporting/reinforcing that particular radiation characteristics of the transducer.
Last edited:
Looking at TG measurments of his 2 ways i find these drivers a clear improvement on previous SS drivers.🙂
Btw, i don't see the need for any notch, nor for such a strict match to LR2 target which does not necesarily means closer to perfection, as this optimization will be valid at one point/axis only.
An externally hosted image should be here but it was not working when we last tested it.
Btw, i don't see the need for any notch, nor for such a strict match to LR2 target which does not necesarily means closer to perfection, as this optimization will be valid at one point/axis only.
Last edited:
- Home
- Loudspeakers
- Multi-Way
- Scan-Speak has a new line: The Ellipticor