I think context here is important.
Because without a waveguide you have to have a little downwards slope.
Otherwise the tweeter will have to much energy and will sound to bright.
I have my doubts if distortion would cause any real significant problems. The ear is already a lot less sensitive to it at these higher frequencies.
But most importantly, a mismatch in directivity is far more audible than distortion.
Diffraction issues can also be at play here.
So I don't think you can't just give a general statement about this either.
I find it quite unlikely that distortion will be so significant here.
Typical 5-7" + 1" has directivity issue between ~ 2-6kHz, it radiates too wide. Downward slope of FR 1-2dB from 1-10kHz would solve this, but top end sparkle could suffer, overall sound could be little bit lifeless.
It is often the combination of wide radiation and diffractions, I usually solve this by lowering the problematic area by 1-2dB, it does not improve DI curve, it just partially compensates power response and the energy it radiates at those frequencies. It is not by the latest loudspeaker design mantra and Spinorama codex, but it works, and in normal rooms it makes listenable system.
Here is what I did for ClassIllu 2way, 18W/8545K00+6600. I cannot say it sounds worse or better then waveguided systems. It sounds just different, I can imagine in some rooms this could be auditioned and judged as too bright, but so far none of the listeners of those 2way complaint. Anyway, waveguided tweeters and speaker are less prone to sound bright and take less work to dial in the right balance.
It is often the combination of wide radiation and diffractions, I usually solve this by lowering the problematic area by 1-2dB, it does not improve DI curve, it just partially compensates power response and the energy it radiates at those frequencies. It is not by the latest loudspeaker design mantra and Spinorama codex, but it works, and in normal rooms it makes listenable system.
Here is what I did for ClassIllu 2way, 18W/8545K00+6600. I cannot say it sounds worse or better then waveguided systems. It sounds just different, I can imagine in some rooms this could be auditioned and judged as too bright, but so far none of the listeners of those 2way complaint. Anyway, waveguided tweeters and speaker are less prone to sound bright and take less work to dial in the right balance.
I always call that curve the DI (directivity index) curve. Is there a better name for it?
We all hear differently. John Krutke's white paper/blog entry makes it clear that he finds the distortion character of a driver to be very important... and he is a very experienced, very successful designer. My inclination is to assign a high level of credibility to his statement, even if I don't understand the rationale or the science behind it at this time. I was not familiar with this information, thank you @tktran303 .
Why do we all hear things differently? Why does one person like speaker A better than speaker B, while another likes speaker B more than A. Since no speaker is perfect, every speaker system has been designed with compromises and a balance of performance attributes. For my best listening enjoyment, I prefer attributes A, D, J, and Q to be prioritized higher... for someone else, they would prefer attributes B, C, N, R, and Y to be prioritized more highly...
I agree. When listening to my TXT system, which is non-waveguide, I have 3 DSP presets which adjust the slope from 100 - 10k. Preset 1 is flat on axis, preset 2 is -3 dB at 10k, and preset 3 is -5 dB at 10k. For most classical music, jazz, and naturally recorded acoustic music, I used preset 1 (flat). I also used this for all movies and TV shows. I used preset 2 for most pop music to tame the excessive brightness. Only a few recordings required preset 3.
With this waveguide system, I have found that I need just one preset for almost all recordings. The single preset seems to work well for both classical, jazz, opera, pop music, country, rock, etc. I have a preset which slopes down the response by -3 dB, but I find myself rarely using it.
So this is an interesting find. The TXT system is engaging the room more, it has less directivity, so the room has more influence on the tonal balance. I believe this makes the sound system more sensitive to small changes in treble energy.
I have experimented with various speaker/listener positions, but these alternate placements are only temporary. The primary listening spot is fixed, and the speakers are adjustable in terms of +/- 8 inches or so... no big changes in speaker position. Fortunately, my primary listening/speaker position is a very good spot in this room, particularly for smooth bass response.
We discussed this earlier in this thread.
https://www.diyaudio.com/community/threads/purifi-waveguide-project.394174/post-7404653
By moving closer to, or further from the speakers, I can mimic the effect of a higher or lower directivity.
j.
I know who he is, I have been following for 10-15 years
(and sometimes we cross paths).I only just don't quite agree with him on certain aspects, this is one of them.
With me quite some other "very experience and successful designers"
The ear is very sensitive around the 1kHz, but it very quickly goes down with higher frequencies (as well as lower, but that is not the point here).
To make a long story short, I think it's handy to read this blog post as a brief summary;
https://hifi-selbstbau.de/index.php/hsb-grundlagen/verschiedenes/klirrfaktor-wie-viel-ist-zu-viel
translated;
https://hifi--selbstbau-de.translat...l=auto&_x_tr_tl=en&_x_tr_hl=en&_x_tr_pto=wapp
Here is the K3 graph;
https://hifi-selbstbau.de/images/stories/grundlagen/klirrempfinden/PhonGF_HideK3a.png
This blog post is well in line with other literature about psycho-acoustics.
The masking effect is just enormous at higher frequencies.
@gedlee rightfully so can rant about that forever, and why distortion numbers just don't make sense on themselves.
But much more important, is that I would like to go back to what I said originally.
I never said that distortion isn't important.
I said that things like directivity, diffraction, frequency response etc are MORE important (aka more audible).
It's really hard to make two identical systems with only the distortion as the major difference.
That being said, I have done quite some testing with like distortion plugins and headphones.
With music, you can go quite high when it comes down to distortion.
I would also really highly ask to actually do these tests yourself!
Than still, it's nothing more than a nuance if we look at the order of magnitude of most good performing drivers nowadays.
This also seem to be the experience from Erin @bikinpunk
One of his best reviewed speakers, like the Dutch & Dutch 8C as just an example, don't perform that amazing on distortion.
Yet, I have never heard him complaining about distortion in general, only in cases when it was really obvious.
btw, can you provide a link to his white paper, so people don't have to go into the woods again?
Last edited:
It seems my naming "DI curve" caused confusion, at least that is how I read b_force comment. I call it DI curve as well, directivity curve is fine too.
It is all too common even among experienced people to decide the reason for something is down to what they have seen in a graph or measurement. I have no doubt that he was able to determine a set of rules that worked for him based on the measurements that he was taking, that is a useful thing for a designer to be able to do and for other's to be aware of. There are many other factors that can coincide with the reasons for why non linear distortion can occur.
It is problematic to throw away or disregard all the psychoacoustic research that has been done on non linear distortion perception. True cause and effect is very difficult to establish without a lot of rigorous testing. It doesn't alter the fact that the rule works, just clarifies why it works.
I don't know why, Jim and your use of the term was correct within the context of a CTA2034 graph, no need to change.
Well I certainly agree with that, especially within the context of good quality drivers which are low distortion to begin with.
I appreciate your thoughts, as always.
It might not be coincidental that more than several people call it a DI or Directivity Index curve... even Vituixcad hints to that
:Export a single image and it will even show that title in the legend.
I hope this isn't too off topic.
I am also a fan of switchable presets for responses that suit the source material. Maybe it is a deficiency in my design / speakers, but I have certain artists and masters that I like that need "taming" and others where a flat response is ideal and sounds great.
In anycase, I don't know if you have a dynamic / automated "switch" for your presets. If this concept interests you, I developed some simple Linux scripts to achieve this (assuming you go down a software DSP route - e.g. CamillaDSP). It also works for ALSAEqual (equaliser).
I also use this approach for double blind A / B crossover testing.
If you venture down the Linux path, I'm happy to share the above if of interest. (Might even package it up on Github)
I am also a fan of switchable presets for responses that suit the source material. Maybe it is a deficiency in my design / speakers, but I have certain artists and masters that I like that need "taming" and others where a flat response is ideal and sounds great.
In anycase, I don't know if you have a dynamic / automated "switch" for your presets. If this concept interests you, I developed some simple Linux scripts to achieve this (assuming you go down a software DSP route - e.g. CamillaDSP). It also works for ALSAEqual (equaliser).
I also use this approach for double blind A / B crossover testing.
If you venture down the Linux path, I'm happy to share the above if of interest. (Might even package it up on Github)
Thanks!
Well it raises an important aspect I think from why certain discussions (in general) seem to derail quickly.
That is that you won't fix these problems with speakers with a very low distortion.
It's not like your bad resonances, or directivity will totally disappear with better distortion.
Distortion, doesn't magically cancel out these problems
(in fact, because there is technically less masking, it might even sound worse).
So yes, of course we want the best performing drivers as possible.
But if those other issues are not being taken care of first, there is a point to be made how much of use that actually is.
It sounds as though you're currently in a(n understandable) broad state of indecision. You may be asking yourself (..that if typical amounts of driver distortion are audible,)...
1. Do they manifest as something simple like tonal balance, which can be adjusted, or
2. Are they something more tangible where you must toss out your regular drivers if you want a quality speaker..
I may not have been very clear. I am not in a state of indecision. When I wrote that
I was acknowledging that I take his thoughts and ideas seriously, and I will give them serious consideration. His perspective on distortion is interesting, and it is certainly different from the mainstream consensus on harmonic distortion. Being outside of the consensus can be indicative of innovation, but it can also be a sign of being wrong. Given John Krutke's long record of success, I am very reluctant to simply dismiss his observations and conclusions simply because it is outside mainstream thought. Perhaps his theory may explain why drivers with seemingly very similar measured performance, operating within their pistonic range, sound different. But it is also possible he is wrong.
At this point, I am not ready to change my personal design guidelines regarding harmonic distortion, which have guided me well.
j.
That's EXACTLY my experience when comparing non-waveguide monitor speakers (Adam) with wg speakers (Neumann) some time ago.
Speakers with WG are more "reliable". They sound more similar in different rooms and when you know/learn their sound you can judge the source material and work. With the Adams differences where bigger, I was not sure if I can trust what I hear.
I would for sure not need different filters for my Neumann KH120ii (esp. when measured and calibrated with MA-1) or my main speakers. I don't need to change the speaker, good source material will sound good and bad source material ... well, you need to work on that ;-)
That's interesting and some of these go well with my experience.
I also prefere flat response with my designs. I always thought that it's cause of my background of studio critical listening and I need enough hf level to easy and quick judge material. And I'm used to that sound. I also worked at AKG and like this headphone sound - no bloated bass.
Of course these designs are all low THD cause at the moment I only build high quality speakers cause I have to value my time. And getting something better as a Neumann KH120ii is not that easy ...
I value THD measurements during development a lot - it helps to define the linear working ranges of speakers. Make a frequency sweep with the tweeter over level and 2min later you know it's working range and how deep you can cross over. I'm also a fan of low tweeter crossovers to get a more stable off axis response!
THD in the endproduct speaker ... we are in a new situation now. We never had speaker chassis which can do <0,1% THD even at serious levels (like M74). With some tweeters you can't measure the 1% THD maximum level cause they die thermical before you reach that! So we can build speakers now with really low THD and I prefer to do so.
On the other side - I put a midrange in my cinema speakers which produce a H2 bump 600Hz - that's a sensitive range! But I prevered the sound of these and I could not hear a problem. In fact midrange was more detailed and "dry" as the 2nd chasis I compared (which measured a little better but sounded more "round") and that was the design goal - most details out of an 8" midrange.
So don't overestiamte the audibility of THD but it's still a very important measurement!
(and don't underestimate the annoyance of higher order harmonics! These an make the "nasty" sound. )
ProAC 2.5 clone ... that's one of these designs I say with one glimpse "can not work". And it doesn't ;-) - at least not well.
And still there seem to be many people who like it! I belive many are simply used to this 6-7"/1" sound and think speakers SHOULD sound like that. It's their reference point.