Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
In Reply to: RE: Exemplary Review Standard posted by John Atkinson on January 26, 2017 at 04:41:13
Based on past experiences, I bet we're going to have to agree to disagree, but I have to put my 2 cents in on distortion measurements -- basically, they're really important to do.Total harmonic plus individual harmonic (we only do total due to system and time constraints, because we pay a lot to use a true chamber).
At the measurement-oriented companies I visit, distortion measurements are always done -- valuable working tool. Of course, not the only measurements done, but considered vital and part of the complete set. In our own experience, by measuring distortion, we have affected subsequent products for at least one dozen companies based on what they've seen when we've measured one of their current products. In a nutshell, we measure a speaker, they see the distortion, they fix it up for the next version or whatever comes next.
Furthermore, companies work to reduce distortion and really the only way to reveal what they've achieved is by measurement. When we measured the first Magico S5, it had the lowest THD of any speaker we'd measured and that information the company was happy to send out everywhere. They worked on certain aspect to ensure that distortion was down.
As a result, I wouldn't be so quick to sweep it under the carpet. As well, I think it's important to point out that most can't do proper distortion measurements because you need a really good environment to do so. This, too, is an advantage we have of using the NRC chamber -- built in a location and in such a way that has a quiet noise floor and we can measured to under 1%.
Again, you've written your paper, feel a certain way, and I'm happy again if we agree to disagree. But I do feel they're vital (and we wouldn't keep paying if they weren't), so people should know the other side.
IMO, Distortion measurements are normally done at too low a power level , usually 1watt , what level are you currently using .. ??
I agree with you about power level not being appropriate. We use output level -- single speaker at 90 and 95dB (the latter if the speaker can take it), anechoically at 2 meters. This means that it's the equivalent of 96 and 101dB at 1 meter (each halving of distance increases output by 6dB).
If you consider that two speakers playing adds 3dB, that's 99 and 104dB, but that's still not factoring in room gain. So pretty darn loud.
THD metrics are any more relevant with speakers than for electronics - where I find them utterly useless and often misleading.
Back as a teenager, I took my AR Integrated to one of the McIntosh clinics where Dave O'Brien himself measured my unit and presented me with a fancy chart. The results were consistently below the 0.5% spec using a sine wave and I thought that was pretty cool at the time.
And yet - it still sounded poor especially at low levels. The H-K Citation 11 / Crown D-150 combination that replaced it two years later (ca '74) was significantly better.
If we accept the notion that speaker distortion is usually higher than that of electronics, why wouldn't that be readily audible?
The importance of harmonic distortion originated in Radio engineering where broadcasting energy outside your assigned frequency was a big no-no. Here, what mattered was how much was outside the band and Total Harmonic Distortion did the trick.
When amplifiers were all pretty much the same technology, then too or at least at first, lower distortion measurements did correlate to less coloration.
The problem began when high amounts of closed loop feedback were used, instead of 10 or 15dB of loop, with SS amplifiers they were 60 even 90+ dB of feedback. While that made static numbers that were very low, the measurements often departed from what one heard. Because of perceptual masking in our auditory system, the farther away from the fundamental the harmonic is, the easier it is at the same level to hear AND when listening to music which is harmonically structured, odd harmonics sound bad but even harmonics are often perceived as "richness". With natural sounds or industrial sound (none of which are harmonic) then even and odd harmonics can change the timbre. The problem with added feedback is that it lowers the low order distortion components but it generates new ones much higher up. If one examines our hearing sensitivity curve one see's that where ones sensitivity is greatest (around 3-4Khz) one finds that when a loudspeakers distortion components fall in that range they are also most audible relative to the fundamental.
At risk of sounding like a broken record, there are SO MANY things in addition to distortion that a generation loss test WILL RAPIDLY reveal the sonic infidelity of any component especially loudspeakers which by far ARE the weakest link. It doesn't tell you what to fix but by hearing the warts, it often shows where to look and often too, you can hear them afterwards listening directly. This has been invaluable at work especially early on.
Danley Sound Labs
...AND when listening to music which is harmonically structured, odd harmonics sound bad but even harmonics are often perceived as "richness".
At risk of sounding like a broken record, there are SO MANY things in addition to distortion that a generation loss test WILL RAPIDLY reveal the sonic infidelity of any component especially loudspeakers which by far ARE the weakest link.
I would agree with the value of your holistic approach, but that is not what Doug is promoting using the NRC lab. I'm still not convinced that single generation sine wave testing for any audio component really tells you much about the results with music. So speaker A's value is 1.2% while speaker B is 0.5%. And?
There's information, but not knowledge.
I am not sure if you are asking me or John A.
I will just say that I never said they are more important than for electronics. I just said that distortion measurements are important.
to whom I responded if you notice the post placement. :)
I will just say that I never said they are more important than for electronics
If you return to my question, it was about being "relevant". I find THD metrics irrelevant to provide useful information about electronics. Why then are they relevant for speakers?
I just said that distortion measurements are important.
Which returns to my question of "why".
The audibility of distortions is certainly not clear cut -- lots of research going on -- but there are some areas where what you measure and what you hear are strongly correlated. I'll give you one very straightforward example: ribbon tweeters.
We've been measuring speakers with ribbon tweeters for well over a decade and, often, we've found that reviewers hear a harsh, aggressive sound from them when the music gets turned up. If you look at the distortion on ribbons, below about 3500Hz the distortion rises at an increasing rate until you get to about 2000Hz, which is where many companies choose to cross them over in two-ways to get a proper blend with the mid-woofer, where it wants to shoot through the roof. Oftentimes, it's over 10%, even higher.
What's happening? From what I can tell, most ribbons don't have the excursion to reproduce frequencies that low, so it's cranking out a distorted sound. Simple as that -- and it's easily measurable. Funny thing is, I was talking to a recording engineer about this and he said, "You know what, sax players love ribbon microphones, because they have bite." That bite, I bet, is that harsh, aggressiveness you can hear from a ribbon tweeter. In fact, nowadays when we get a ribbon-based speaker in, whenever I can, I'll often ask the designer, "How do you deal with the distortion?" When they know what they're doing, they'll tell you.
Anyway, that's one example where you can look at the distortion plot and correlate it very strongly with the sound.
I suspect you are talking about planer magnetic tweeters , Ribbons have far less issues and no harshness at 2K , at 2K they are still field saturated . As to ribbon mics, There is no harshness from Ribbon mics I'm aware of , most using ribbon mics like their tone , it sounds natural to them.
Btw , there is no speaker/audio system , I'm aware of capable of creating the percussive energy of live instruments , so "real" is relative ...
that example sounds more like overdriving to me.
The speakers I've heard using ribbons like Magneplanars, Scaenas and Nolas don't seem to make such sophomoric mistakes and employ them in at least three ways designs.
I would not characterize any of those speakers as sounding "harsh".
the maggie tymp iva certainly, at least to my ears, had distortion at the bottom of the tweeter's passband, and, if memory serves, HP heard the same phenomenon on their model 20, allegedly lessened in the 20.1. I've heard sumpn simlar from conventional tweeters in some speakers with 1st-order cross-overs.
Mind you, not everyone heard what I heard, or perhaps they heard but were not as bothered by it.
HP heard the same phenomenon on their model 20, allegedly lessened in the 20.1.
Allegedly? I cannot speak to the 20R having never heard it, but I can regarding the 20.1s I heard multiple times in multiple systems at Sea Cliff. Instead of conjecture, why not actually read what has been written by him and others?
Valin reviewed the 20R here and mentioned a number of things that HP did say about them at the time. More recently, I also heard the 3.7 at Sea Cliff during a visit by Hobson. As one who highly values coherency, my primary concern with the 20.1 was that it sounded like a multi-way speaker and lacked coherence. The 3.7 (and presumably the 20.7 which Warren Gehl at Audio Research uses for evaluation) are better in that respect. Read HP's commentary here.
There is mention of improved response from the current ribbon where HP thought perhaps the range was different. Diller indicated there were refinements, but not to the crossover point. You're welcome to disagree.
From personal experience, I differ with your assessment that "The speakers I've heard using ribbons like Magneplanars, Scaenas and Nolas don't seem to make such sophomoric mistakes . . . ." The Tympany IVa, to my ears, certainly did, but I acknowledge that what you've written allows for the possibility that you never heard those. It is possible that some might describe the phenomenon I heard as "discontinuity" or "lack of coherence" or "disconnectedness between mids and treble". I suspect that the sound I heard might be due to what Doug Schneider describes.
If I have mischaracterized, from old & vague memory, the views of HP or any of his cronies regarding subsequent Maggies, well excuuuuse me.
The Tympany IVa, to my ears, certainly did, but I acknowledge that what you've written allows for the possibility that you never heard those.
Have not. I've heard a range of Maggies over the past forty odd years including the Tympani 1U, Tympani IIIa, MG-II, MG-III, 1.6, 3.6, 20.1, 1.7 and 3.7, but my comments were primarily aimed at current production, aka ".7s".
It is possible that some might describe the phenomenon I heard as "discontinuity" or "lack of coherence" or "disconnectedness between mids and treble"
Perhaps. While Harry made that comment about the 20.1s fifteen years ago, that was not the case with the 3.7 even from six years ago. I'm guessing you didn't read the link I provided to that review.
" In saying it doesn't sound like its ancestors, I mean to suggest, before going into detail that the 3.7s do not sound at all discontinuous as they have in the past , but rather as if cut from a whole cloth. Before this (and perhaps the same with the 1.7, which I haven't heard), the perceptive listener could hear the differences between the ribbon tweeter, the midrange, and the separate bass planar elements, and these differences were audible not only as difference in rise time, but also as a kind of characteristic texture. "
And yet, they still use ribbons. I'm thinking that other speaker design considerations are responsible, not the inherent driver type. I'm not a ribbon apologist having never owned one, but I've heard wonderfully natural top end with great extension more than one loudspeaker employing them. I still find full range electrostats most coherent, but could easily live with 3.7s cum subs or 20.7s. I could not honestly say that about earlier models.
You're absolutely correct that I don't refer to 1985 models like the T-IV.
You asked for an example of where a distortion measurement matters -- and I gave it to you. And it's not a matter of simply overdriving, since it happens at fairly low output levels.
BTW, it's not a sophomore design issue -- it's a limitation of the driver type. BTW, even in three-ways the problem still surfaces if they can't cross the midrange to the tweeter high enough. But if you don't measure it, you might not know what's causing the issue.
BTW, it's not a sophomore design issue -- it's a limitation of the driver type.
What is sophomoric is employing a driver in a range not suited for its use.
BTW, even in three-ways the problem still surfaces if they can't cross the midrange to the tweeter high enough
Have you encountered any such?
To understand the complexity of the issue, please watch the video below. Monitor Audio acoustic designer Michael Hedges talks about the development of an AMT for their new Platinum II series, which provides greater surface area and lower distortion. When we created that video, I talked to Michael at length about the issue and he said that although he'd worked on getting ribbon distortion low enough, it wasn't satisfactory, which is why they went to an AMT for this range.
Back to the original point -- distortion is measurable and, often, correlates with what we hear. Dome-tweeter distortion at these frequencies is also problematic.
Love his guitar work and had the opportunity to experience him live on several occasions! Oops, wrong Michael. :)
Perhaps as JA seems to find, there is apparently difficulty associating an increase in distortion with perceived sound quality. One cannot find any such mention of such issues in this review of a two way Monitor Audio speaker. Excerpts from your conclusion:
" The ribbon tweeter and woofer are proprietary designs that are merged together splendidly...
Other strengths include a high level of transparency and detail along with super-extended highs that are effortless and clean. The sense of spaciousness these speakers can convey is awesome."
Despite those favorable comments, the THD curve is indeed much higher at the bottom of the tweeter's range.
Do you know of their current CP-IW460X suffers any issues with its three way design passing over to the ribbon at 3 kHz?
Admittedly, the speakers I've auditioned and found to be excellent run the ribbon tweeters above around 6 kHz.
I have no idea about the CP-IW460X.
Insofar as correlating goes, if one doesn't measure it, then it's impossible to correlate it.
That said, I wasn't looking at the distortion elements as closely as I should've at first and was attributed "hardness" at higher volumes to other things. But when I started looking closely and also measuring more, I song a strong correlation with ribbon quality, distortion levels, and what gets heard.
BTW, I want to reiterate that ribbons aren't the only problem. There are many other driver problems. I just want people to be aware -- distortion DOES matter to some degree.
Post a Followup:
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: