I find those arrays can often be bad sounding, maybe that's why. But then there are also a lot of diffraction issues present.
I switched from the DE250 to the DE500 partly because the throat on the DE500 was shorter. I do not have the DE500 in my own system as yet, so I don't know if I would say that they sound any better. They measure better!
I have heard them several times and I disagree. They are fine at mid to loud levels, but quickly take off and become harsh without extreme SPLs. That can lead to pain. Maybe that's what you are hearing?
They are very clean at low levels, which makes you want to push them. The CD gets into trouble before it should, IMO. Might be interesting to test that verses another type of horn.
To me going above "loud levels" is "extreme SPL" and I have measured this at about 120 dB. To me that's extreme.
Pano,
You have pointed out one of the (many) dilemmas surrounding adoption of different horn profiles. The OS or conical expansions are "cleaner" regarding LD, but far less efficient at Fc than exponential or tractix expansions, which then drives up NLD at the higher drive levels they require down low. The HOM foam reduces HF level, again requiring far more more power than a horn which narrows in HF dispersion.
Since Earl states that he can't hear NLD, the trade off (at typical home levels) is OK for him. For those of us that can hear NLD, the choice is not as simple, especially at high SPL levels.
Art
Laurie Fincham gave a speech during dinner at ALMA this year. He mentioned (from memory, I hade wine and beer to dinner) that the devil is in the details.. believe your ears even if you initially can not measure something. If you hear something there are probably something causing it. The example he highlighted was nonlinearities from damping material inside a box. It loaded the driver in a nonlinear fashion causing audible distortion. Low level stuff, still audible.. acccording to Laurie Fincham.
Last edited:
Do you consider 120dB SPL being extreme "period" or extreme in some range, in some cases etc?
I have always measured the radiated NLD higher than driver current nonlinearity. Do you have different findings?
There's a guy and company named Klippel who have spent some time on this.. There's a test on their website anyone can take blind. Test contains the complex behaviour with many sources of nonlinear distortion.
The studies I've seen on this mirror my own. Time domain distortion aka phase shift is most audible in the low mid range, say around 100-500Hz give or take some. Above 1500Hz the ear gets worse and worse even though some phase detection remains. I'm actually working on that this week.
The fact that our amplitude sensitivity is highest in the range you mention does not mean that we are most sensitive to other phenomena in that range. We even have different sensitivity to sines vs narrow band noise.
Yes, HD is not very well masked with a sweep so the HD stands out clearly if above some level.
However, even if you do not hear distortion "jumping out" on you in playback does not mean it's not there making a worse sound. Only when you start to remove nonlinear distortion to lower levels you realize it was there. Of course, going to the extreme the sound will be so shitty you don't want to listen to it.
Also, as have been repeated for decades now, HD is one thing, but the key is the nonlinearities that casues the HD. The same will cause complex distortion with complex source material.
Last edited:
summation of what i've learned from this discussion:
distortion is irrelevant/doesn't exist/until levels are excessive.
hearing distortion at other than sublimely ridiculous levels is an anomaly which isn't as important as whether it's linear or non linear.
and that trying to understand horn design to further my understanding is an exercise in futility because it's readily apparent the far more knowledgeable people can't agree of what's important or what to do about it.
distortion is irrelevant/doesn't exist/until levels are excessive.
hearing distortion at other than sublimely ridiculous levels is an anomaly which isn't as important as whether it's linear or non linear.
and that trying to understand horn design to further my understanding is an exercise in futility because it's readily apparent the far more knowledgeable people can't agree of what's important or what to do about it.
Yes, for sure.
But I also wonder about the subjective perception of distortion in those ranges. For Example. I have built circuits that had a healthy dose of odd order HD in the bass range and people usually comment that it has "great" "powerful" or "massive" bass. Conversely I've built circuits that had a dose of high order harmonics starting above 2-3K. Comments where usually "harsh" "fatiguing" "digital".
Just curios about the perception, vs detection.
Yes, there is no general agreement on NLD audibility. The Klippel tests that Pan is mentioning are a good way how one can make his own conclusion, regardless suggestions of "authorities".
Listening Test
Though I appreciate Earl Geddes work on directivity, I do not agree with him regarding NLD importance, to me it is not unimportant and is worth investigating.
Listening Test
Though I appreciate Earl Geddes work on directivity, I do not agree with him regarding NLD importance, to me it is not unimportant and is worth investigating.
Pano's circuit distortion anecdote does resonate with me.
IMO from a perceptual standpoint low levels of distortion are dismissed as sonic qualities like fullness or harshness sonic caricatures/signatures if you will. increasing the percentage of distortion and maintaining the same apparent volume is hard,ask any guitar player. this is probably not contextually correct but i think that percentage of distortion combined with overall level is responsible for the problems guitar players encounter in developing their "distorted tone" the large envelope of distortion that sounds "full and lush" in isolation quickly suffers from masking effects in the presence of the other sounds and is interpreted by the ear/brain as dull and indistinct.(and usually results in a further increase in upper midrange boost in an attempt to cut through but only serves to exacerbate the situation.)
so for me the parallel is reducing distortion in a horn has the net perceptual effect of making all sounds coming from it better/cleaner more distinct.
IMO from a perceptual standpoint low levels of distortion are dismissed as sonic qualities like fullness or harshness sonic caricatures/signatures if you will. increasing the percentage of distortion and maintaining the same apparent volume is hard,ask any guitar player. this is probably not contextually correct but i think that percentage of distortion combined with overall level is responsible for the problems guitar players encounter in developing their "distorted tone" the large envelope of distortion that sounds "full and lush" in isolation quickly suffers from masking effects in the presence of the other sounds and is interpreted by the ear/brain as dull and indistinct.(and usually results in a further increase in upper midrange boost in an attempt to cut through but only serves to exacerbate the situation.)
so for me the parallel is reducing distortion in a horn has the net perceptual effect of making all sounds coming from it better/cleaner more distinct.
is it possible that if given distorted stimulus without either a reference or mix of clean signal simultaneously an average person's ear/brain simply concludes that all distorted stimuli is indistinct (no different)?
Of course, also when doing perception tests it's extremly imortant to use the right type of material/stimuli. Some nonlinearities is problem free with one signal (or one song/tune or whatever) but results in audible distortion with other signals.
Also room acoustics affects this. Some distortion is simply masked by room sound, goes for both linear and nonlinear distortion.
120 dB at the listening position, to me, is extreme by all standards. Even the least compressed recordings (with a handful of exceptions) would require an average level of 90dB+ to achieve 120 on peaks. 90 avg is very loud, and is starting to flirt with the low end of hearing damage though music is more complex than industrial machines.
Yes, where as I have never done THD measurements of my speakers - for obvious reasons - I have measured the sound levels at which they "loose it", as PANO has pointed out (all speakers will do this at some point.) In my room with the Summas it is about 120 dB (C) at the peaks. And yes this is way louder than I would ever listen normally.
- Status
- Not open for further replies.