I tried out the aweful apt50 cd the other day, listened to it filtered but solo and the sound where dreadful with lots of hamonic distortions at all levels, what a piece of junk
You should download a free signalgenerator that plays sines, then open so many generators you would like to simulate, from fundamental to at least 5-7 harmonics and set levels for to try dist levels from 10% to 0.1%
You should download a free signalgenerator that plays sines, then open so many generators you would like to simulate, from fundamental to at least 5-7 harmonics and set levels for to try dist levels from 10% to 0.1%
Here are some sine tones with various amounts of THD included THD - Total Harmonic Distortion - Test Tones
I'm using H650 headphones now and can easily hear the 0.5% THD, 0.1% not so sure. There is a mention in the link that HD600 has 0.1% THD so 😀 Music is no single sine tone so distortions get masked. Anyway it is good to check this kind of things out to get some perspective 🙂
5% distortion = -26db
1% distortion = -40db
0.1% distortion = -60db
Wowza. Gotta target 1% or under if at all possible, if "distortion free" is the goal.
I'm using H650 headphones now and can easily hear the 0.5% THD, 0.1% not so sure. There is a mention in the link that HD600 has 0.1% THD so 😀 Music is no single sine tone so distortions get masked. Anyway it is good to check this kind of things out to get some perspective 🙂
5% distortion = -26db
1% distortion = -40db
0.1% distortion = -60db
Wowza. Gotta target 1% or under if at all possible, if "distortion free" is the goal.
Last edited:
Quite recently someone posted a link to a webpage with a demo. 1 % THD with uneven orders (i.e. several harmonics and not just 3rd) was definitely perceivable while I could definitley not detect 1 % THD with even orders only. At that one percent the uneven order distortion sounded bright but not shrill.
Regards
Charles
Regards
Charles
Here is Klippel listening test Listening Test at least. Might have been some other? Suggested my threshold with phones was around 5% distortion.
Quickie in youtube How much distortion can you hear? - Audio Distortion Test - YouTube , maybe under 1% with music as elevated brightness, around 5% clearly distorting. With tones around 1%. So, maybe the 1% is pretty much the target I'd say, at least for me. At least with headphones in an office environment 🙂
Quickie in youtube How much distortion can you hear? - Audio Distortion Test - YouTube , maybe under 1% with music as elevated brightness, around 5% clearly distorting. With tones around 1%. So, maybe the 1% is pretty much the target I'd say, at least for me. At least with headphones in an office environment 🙂
Last edited:
Sorry for the oftopic, just outputting something here for reference. Seems like very good speakers can achieve about 1% ~-40db THD+N from 100Hz and up, the worst being the low frequencies around 1% and a bit better in the mids. Often cited good speaker Revel Salon 2 measurements here SoundStage! Measurements - Revel Ultima Salon2 Loudspeakers (12/2009) THD+N is of the chart at the mid range, it looks to be less than -46db at 1k for example.
Perhaps, I thought about that but don't remember if I tried 5. It was so easy to go from 3 to 6, lazy me....
Last edited:
Let's settle on 1%. We could discuss some other aspects of cone breakup though. Like directivity irregularities. Or even the perceivability of the resonances themselves (well, that has been discussed extensively, Toole has some references IIRC).
Why? It should be pretty trivial to prove that for most people, 1% H2 is completely inaudible but 1% H5 (or god forbid H7) would be pretty obvious.
If I flip that question on you, that would mean that you believe that the different harmonic orders are always in roughly the same distribution, right? But to answer your question, no, no practical case that I know of. Except maybe, just maybe, when a website gives you a sound file with x% THD and asks you when you start to hear it... 😉
Last edited:
454Casull, yeah a THD listening test won't tell such detail but as a speaker builder better make sure none of the H products is higher than the threshold with THD, just in case. One could get away with some extra even order distortion if that is not so audible. Anyway, it is a bit hard to find data on speaker driver harmonic distortion distribution, most distortion measurements I find online list only the H2 and H3. Are distortion measurements doable at home environment?
edit: hah I've now realized the hearing system might catch on H5 or what ever and the THD reading could be anything above that. Well, anyhow I found it useful to get some perspective 🙂 I'd guess well designed speakers (that are similar system design) would have similar HD distribution between each other? Does cone resonance dominate the motor related distortion, does it dominate the speaker system distortion? What are the sources of distortion and how they relate?😀 What about phase distortion, is there more types of distortion? Many questions.
Answering to my self whilst trying to remember some distortion graphs that with excursion comes the motor distortion, mid range suffers the cone resonance. A motional feedback system or simply reducing the excursion could reduce the motor related distortion and cone resonances are subject on the crossover design. Talking to myself a little bit more, to overcome some of this build multiway speakers with huge bass drivers. Best would be multiple entry horn which suppresses some of the HD with bandpass features. Hmm, multiple entry horns seem to be the pinnacle of current speaker technology. They are a bit more complicated to design and build though, and rather large, but when top performance was easy? Never 😀
edit: hah I've now realized the hearing system might catch on H5 or what ever and the THD reading could be anything above that. Well, anyhow I found it useful to get some perspective 🙂 I'd guess well designed speakers (that are similar system design) would have similar HD distribution between each other? Does cone resonance dominate the motor related distortion, does it dominate the speaker system distortion? What are the sources of distortion and how they relate?😀 What about phase distortion, is there more types of distortion? Many questions.
Answering to my self whilst trying to remember some distortion graphs that with excursion comes the motor distortion, mid range suffers the cone resonance. A motional feedback system or simply reducing the excursion could reduce the motor related distortion and cone resonances are subject on the crossover design. Talking to myself a little bit more, to overcome some of this build multiway speakers with huge bass drivers. Best would be multiple entry horn which suppresses some of the HD with bandpass features. Hmm, multiple entry horns seem to be the pinnacle of current speaker technology. They are a bit more complicated to design and build though, and rather large, but when top performance was easy? Never 😀
Last edited:
I agree, I'd go for five or three; four would possibly make the bell mode worse.
Zaph has published all his measurements with H5, if within range. Check them out. And it not only is belief, it is physics. For an X order distortion product, the excitation of the oscillation goes down by X, the number of the order. Put otherwise, the energy transfer is the same, but the required energy to sustain the oscillation is higher. So H4 will be lower than H2 and H5 will be lower than H3.
Last edited:
Fortunately !
OTOH the audibility of harmonics rises with their order because they are not masked by the fundamental and lower-order harmonics.
Regards
Charles
Up to just under 8 V the H5 on this driver is greater than H4: Usher 8955A | HiFiCompass
From 400 Hz to 5 kHz H3 > H2 for this one, and H5 ~= H3 around 200 Hz: SEAS W22EX001 | HiFiCompass
Same here, H3 and H5 are higher than H2 and H4 respectively: Peerless 830875 | HiFiCompass
H2 ~= H3 for this one, H5 > H4: Vifa NE180W-04 | HiFiCompass
Zaph's measurement for the W17RC38-04 has the H5 nipping at the heels of H2. Peerless 830656 has H5 ~= H2 as does AC130F1.
Of course these are cherry-picked examples. Point is, there's no formula that predicts how the magnitude of the different orders.
Last edited:
Since the cause of even order harmonic distortion in electrodynamic drivers is different from that of the uneven orders, you cannot compare the patterns. I made the difference in my post. Also, be sure to shift the curves to the corresponding frequencies before comparing...
This is not really true. Where did you get this idea?
This isn't true either.
Seems to me that a lot of misconceptions are floating in this thread. I didn't read it all, but what I have read is very misleading.