If it works, it works. Just like crossovers. They don't have to be textbook schematics and values. In fact, it's usually better if they are not.
I've heard speakers that use 8, 10 and even 12" midrange drivers that work very well indeed. With the right driver and right crossover, all sorts of things are possible.
That is certainly a respectable position. It is my position as well.
But what do you do when the measurements say that a large woofer works just fine, but your ears tell you otherwise? You have already said that in that case you don't trust the measurements. This is how we got here right? I trust the measurements and you don't.
When the measurements and your ears don't agree then you have a problem. How you resolve this problem is the key to moving forward. I have been at this junction many times. I and virtually everyone that I know in this field have always found that the measurements are right and the ears are wrong. Just read Sean Olive's work on blind testing. The measurements agree with the blind tests almost perfectly, but sighted things change. No one is able to over come this problem.
When I was at Ford I had a gauge capability study done of the "Expert Listening Panel"- ten self proclaim audiophiles - all objective. The study showed that of the ten only two were reliable, the other eight judgments were all over the place. So yes, there were two who were consistent and reliable, but that is a 4:1 bet that someone chosen at random from this panel was not.
Last edited:
Hello Dr Geddes,
Possibly a trivial question, but are you in a way saying that HD is irrelevant because its inaudible ( esp when music is played ) below a level which most current good drivers (esp the B&C you use) have achieved. And that level is is quite high when real music is involved instead of test signal.
Thanks...
Possibly a trivial question, but are you in a way saying that HD is irrelevant because its inaudible ( esp when music is played ) below a level which most current good drivers (esp the B&C you use) have achieved. And that level is is quite high when real music is involved instead of test signal.
Thanks...
Hi HiFiSound
Not a trivial question at all, its right on the issue.
Let me make a few key points, some of which you have already made:
1) there is a threshold of nonlinearity below which we cannot detect it because of our ears inherent masking of such things.
2) this threshold can be vastly different for steady state signals versus transient ones. Steady state is far easier to detect.
3) (and here is the first key point) the numbers produced by THD or IMD do not correlate with the detectability of the nonlinearity in the underlying system. One form of nonlinearity, crossover distortion, is detectable at .1% THD while another, pure second order nonlinearity is not detectable even at 10%. Hence THD and IMD are useless at sorting out the "good" from the "bad". This is true for any signal steady state or transient.
4) it is the higher orders of nonlinearity that are audible, not the lower orders. Masking tells us why this is true.
5) the types of nonlinearity found in a loudspeaker are normally of the most benign type, low order. Electronics, the most insidious type, high order.
6) good loudspeakers are well below the thresholds of audibility for most signals, but certainly music, except when pushed beyond their design limits. A 15" woofer and a compression driver are never going to be pushed beyond there limits in a home situation.
There are metrics that do correlate with subjective assessment, but no one seems to want to adopt those - better to use pointless numbers than risk being shown to be at fault - being ignorant of the truth is better than being wrong.
An odd thing about the above is that it is very possible to make a loudspeaker with higher levels of THD that will actually sound better because the overall amount of higher order nonlinearities have been minimized. B&C actually does this, but they never even mention it. It is only apparent when one looks at the Klippel data. Their BL curves are deliberately NOT flat!! How's that for changing the paradigm of loudspeaker design!
I used to believe that nonlinearity was a key aspect of transducer design - my ears told me that. But the numbers don't lie and now I know that I was wrong all those years and wasted so much time chasing something that was unimportant.
I do trust your measurements and agree completelly but i think that there are some things that aren't measurable yet... or they are measurable but they are neglected for some reason. We are witnesses that THD and IMD (as you said earlier in this thread) are not being considered as an important factor any more. If there is the case where something was considered very important and now is neglected i think that the other way arround is possible too. The problem of me not being an expert in acoustics is because i can't explain what it is i hear through measurements because i wouldn't know where to begin. If it was just me, i would rule it off as a brain manipulation as you stated earlier but number of people heard the difference in AB test. Actually all 4 of them.
My knowledge in acoustics is substantial for an average music lover (hifi enthusiast, audiophile) but does not reach that far into the exotic measurements that may reveal and explain what it is we all hear - if we rule out mass hypnosis
Last edited:
Zvu
I do not doubt that you heard a difference, that is not only understandable, but highly likely. My objection is jumping to conclusions about the cause of this difference without having any objective data to support it. Also, "different" is not always "better". Without some objective data you don't know which is more "accurate", you just know that they are different. And if you are saying that "accurate" is not the goal, just "preference", then we do have different objectives.
You want to know an area that is completely unexplored: low level signals. Does anyone do frequency response at different levels, especially very low ones? I don't and I don't know any who have. I've done it at high levels and there are lots of changes. But very low levels are very difficult to get good data - especially for me in my noisy environment.
I loved the plot of the coaxial KEF speaker - that was simply brilliant. The next time I do a measurement setup I will do this exact same test with my compression driver and woofer. If that shows some results the there would be some objective data that larger woofers sound "different" - they have "different" diffraction. Which is better is a much harder question.
I do not doubt that you heard a difference, that is not only understandable, but highly likely. My objection is jumping to conclusions about the cause of this difference without having any objective data to support it. Also, "different" is not always "better". Without some objective data you don't know which is more "accurate", you just know that they are different. And if you are saying that "accurate" is not the goal, just "preference", then we do have different objectives.
You want to know an area that is completely unexplored: low level signals. Does anyone do frequency response at different levels, especially very low ones? I don't and I don't know any who have. I've done it at high levels and there are lots of changes. But very low levels are very difficult to get good data - especially for me in my noisy environment.
I loved the plot of the coaxial KEF speaker - that was simply brilliant. The next time I do a measurement setup I will do this exact same test with my compression driver and woofer. If that shows some results the there would be some objective data that larger woofers sound "different" - they have "different" diffraction. Which is better is a much harder question.
Last edited:
That's awesome to hear someone with your experience in this field digs the effort. I was surprised at the difference as well. Some have said it's an exaggeration of real world effects because (as most would expect) a high pass filter will typically be applied to drivers of this nature and there's some reasonable expectation of a crossover/SPL relationship here. In other words, it's not incredibly likely someone will driver a speaker to these kind of limits and moreso, it's only a steady state measurement. Still, I think it lends a lot of credence to the Doppler effect and absolutely shows, if nothing else, the potential for linear distortion causes simply by a nearby speaker (in this case, the concentric midwoofer). We typically jump straight to FR/HD/IMD; I think this example is another source of distortion that may be considered by others in the future. Now, how large of an impact it has is obviously going to vary wrt to excursion and how audible it is (I'm thinking of very fast transients) is subject. That said, I'd suppose if you were listening to music with a fast tempo beat, that verges from a dynamic state to a near steady state and increases the potential for perceived distortion.
Again, it's all hypothetical... just an off the cuff idea I had at the time before I closed up shop, so to speak, for this drive unit's testing. Makes me proud to see it be well received by you and some of the other folks on the forum.
- Erin
www.medleysmusings.com
The AM distortion I explained in post #48 in a full range low Xmax driver driven beyond Xmax (like Retsel's Lowther example in #37 that cleared up with a high pass filter) is a distinctly different issue than the frequency response deviation the DC offset Erin introduced in the KEF Q100 caused.
The cone is an extension of the HF driver horn, the DC offset creates a discontinuity at the midpoint of the horn, the 3mm wavelength of the discontinuity reflected in the peak and dip.
Many instruments have non symmetrical waveforms, so the HF frequency deviations could occur on transient peaks, while on average would not be audible with music not approaching the woofer's linear limits.
Performing the same DC offset on a full range driver, or a cabinet employing a separate woofer/ HF horn or cone or dome tweeter would result in different response problems.
All comes down to the SPL limits of small drivers, they can sound great at their designed levels adequate for near field use, but high SPL dynamics at distance require either a lot more small drivers, or larger units up to the task.
Art
Last edited:
You may well be on the right track with this comment. We did listen all of the speakers with few watts at the most because we were dealing with high efficient speakers (>98dB). I listened JBL 4355 at 1.5W and it was pretty loud at a distance of 3m, where i was sitting. It is very possible that with heavier Mms woofers just need a little more power to "wake up" and get the sound right.
And i don't remember that when i turned up my two ways with Beyma SM115/K and DE250 in large format WG, i had any issues with midrange. You might be right on that one and it sounds logical.
cheers
Thanks Art, as this likely describes why I observed the problem. I wonder if the problem starts to occur even before x-max is reached? Because our sensitivity for bass distortion is lower than higher frequencies, I wonder if we begin to hear distortion at higher frequencies before x-max is reached. I believe that x-max is measured to reflect a 10% increase in nonlineararity. But if distortion at higher frequencies is detected at lower nonlineararity levels, then the distortion at higher frequencies may begin to be detected at a lower level, say 2/3rds of x-max.
By the way, I am using 15 ohm Lowthers which have 3mm x-max, not the 1 mm x-max that the 8 ohm variety specs out at.
The piece I referenced earlier on speaks to phase distortion when the cone is moving in and out. This could also be a factor as well, although the authors of that study thought that the effect is not large.
I would like to comment on something that Art mentioned, that 15 inch drivers will not be driven into distortion at home listening levles. I think that this needs to be qualified since with the open baffle rage, 15 inch drivers are routinely being used up to and past their x-max when the drivers are being asked to counter the very fast rolloff due to backwave cancelling of the front wave.
Retsel
Actually, I never mentioned 15" drivers, I wrote:
"All comes down to the SPL limits of small drivers, they can sound great at their designed levels adequate for near field use, but high SPL dynamics at distance require either a lot more small drivers, or larger units up to the task."
Larger units run on open baffles can have the same amplitude modulation problems as smaller drivers in sealed or ported enclosures.
Although the distortion when Xmax is exceeded is certainly measurable, it's the low frequency modulation of upper tones that is more objectionable, and can be clearly heard simply by injecting a low frequency sine wave tone (say 60 Hz) of adequate amplitude while a vocal is playing through the same driver, the vocal will sound "gargly" before the 120 Hz harmonic becomes objectionable. The objectionable "gargling" sound is not Doppler (we don't hear a pitch change), or IM (per se) it is amplitude modulation of the upper frequencies by the lower frequencies.
Art
Thanks Dr Geddes!
So in general what should be the driver selection criteria ? With HD plots out of the way many drivers may now look quite pricey
You do need to insure that the X-max is not exceeded in practice, but if this is ensured then yes, you are correct, HD is unimportant.
I pay very close attention in a woofer to how well it handles its breakup. This is critical to me, but it will be for anyone with a fairly high crossover points. I pay no attention at all to TS parameters, I just use closed boxes and multiple subs.
I am just not in the camp that drivers are all important - I used to be. It is the architecture and the crossover that makes all the difference, not the drivers. (OK now, bring it on because that position will aggravate a lot of people. Just be forewarned that I have a lot of data to support me.)
I am not sure that there would be a pitch change for Doppler because it goes both up and down so the mean pitch remains unchanged.
At any rate, it is certainly true that one needs to keep below X-max at all times. In a home situation this should be a no-brainer, in pro, not so easy. There never seems to be enough headroom.
Opps, sorry Art, I mean Earl
In my post where I was trying to respond to Earl's comment that 15 inch drivers won't be driven past x-max in home situations, I accidently said that Art said it. Sorry Art...
Earl picked up on that and suggested that x-max should not be exceeded under all circumstances.
Retsel
In my post where I was trying to respond to Earl's comment that 15 inch drivers won't be driven past x-max in home situations, I accidently said that Art said it. Sorry Art...
Earl picked up on that and suggested that x-max should not be exceeded under all circumstances.
Retsel
Dr. Geddes,
Please un-confuse me. Drivers they aren't important as long as they do exactly what you want them to do? It sounds to me like drivers are very much important, in that you must carefully select one which meets your criteria.
I have the same confusion over compression drivers - you have said more than once that they are a commodity, and can be freely substituted (with appropriate crossover design), yet you have posted elsewhere that you have switched to using the more-expensive B&C DE500 over the DE250 due to its being "smoother". So does the change in the CD make a difference to the listener? Or does the "smoother" driver make crossover design easier?
Not trying to be sarcastic, just trying to get it straight in my head. It's kind of like the amplifier debates in the Lounge - all amps sound the same as long as they are good enough, begging the question of what is good enough.
Bill
It is all a matter of degree and that does not come across well. Drivers matter, but once you have one that works, it doesn't matter any more. I have serious doubts that the change from a DE250 to a DE500 made an audible difference (I don't actually know since I never listened to the difference.) You can see the measurement difference on my web. There simply is not that much difference. Drivers make about 20% of the difference and the architecture most of the rest. So yes, its all important, just not equally so.
I once made two identical architecture systems with different drivers sets - a 10:1 price difference. In a blind test of 16 listeners there was no statistical difference. But there was two other architecture loudspeaker systems (different) and the differences were all statistically different. That says a lot to me.
I once made two identical architecture systems with different drivers sets - a 10:1 price difference. In a blind test of 16 listeners there was no statistical difference. But there was two other architecture loudspeaker systems (different) and the differences were all statistically different. That says a lot to me.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.