No rocket science contained in this postFirst image is a raw Seas L15 aluminium cone driver with primary breakup ~ 8100Hz.
2nd image is with a steep order XO @ 2,300Hz with a notch filter on the breakup. You can see the 3rd order is somewhat attenuated, but there is little / no change on 5th order HD since the XO cuts above the 5th order distortion
I'll take FFT over rocket science, thanks
In several large blind subjective studies on the perception of nonlinear distortion, it was found that there was no correlation between the level of THD and its perception. It was also found that as much as 20% THD could be inaudible on a music source.
Correct. It's the higher orders that are the most audible. This has been shown several times. 2nd is almost inaudible at any level.
As stated, IMD and HD both derive from the same system nonlinearity. If I know one then I know the other. They are NOT different types of distortion, they are the same thing.
Hence, not surprisingly, IMD also does not correlate to perception. Basically the current methods for measuring distortion are meaningless.
Thanks Dr Geddes.
We often read measurments everywhere but non educated enthusiasts have pain to select informations. a simple illustration: you have some T&S measurments made by enthusiasts but unluckilly some are measuring speakers after one hour or two of a long and hard breaking in program that the driver will never see in real life musics materials.Some parameters like the Fs or Vas can slighty change, I readed, and in fact the driver should need more rest before measuring, if the Fs moves not so much to be noticed, this would not be the same story for the VAS that can change the sim of the volume box...
I don't know if a myth or not... just to illustrate than people want methodology, involve science, but are not always aware of what they really do - or at least for the educated one can be in a paradigm non precise enough waiting the better next one-. Not easy for the simple diy enthusiast to sort out all of that. If concistency of measurments are the only method to repeat quality and control, one need exactly to focus where it is needed. For instance again the added mass method for cone drivers would be far of the ideal if I understood half of what I readed...
So thanks again for your inputs (and btw also all the nice people that help here and elswhere)
We often read measurments everywhere but non educated enthusiasts have pain to select informations. a simple illustration: you have some T&S measurments made by enthusiasts but unluckilly some are measuring speakers after one hour or two of a long and hard breaking in program that the driver will never see in real life musics materials.Some parameters like the Fs or Vas can slighty change, I readed, and in fact the driver should need more rest before measuring, if the Fs moves not so much to be noticed, this would not be the same story for the VAS that can change the sim of the volume box...
I don't know if a myth or not... just to illustrate than people want methodology, involve science, but are not always aware of what they really do - or at least for the educated one can be in a paradigm non precise enough waiting the better next one-. Not easy for the simple diy enthusiast to sort out all of that. If concistency of measurments are the only method to repeat quality and control, one need exactly to focus where it is needed. For instance again the added mass method for cone drivers would be far of the ideal if I understood half of what I readed...
So thanks again for your inputs
(and btw also all the nice people that help here and elswhere)
Last edited:
No, I mean that H5 of 1kHz is 5kHz and you cannot compare that to the H3 of 1kHz, which is 3kHz. They are 2kHz apart.
I remember there are sources stating uneven orders are more audible than even orders. Don’t ask we which ones now.
I can’t compare them? Then how can we even discuss at all the relative levels of H2, H3, H4, H5... if the outputs are at different frequencies as you say?
I am sure your original statement was that H5 < H3 and H4 < H2, but now that is no longer quantifiable or provable?
I haven't done much with T-S parameters they only apply to LF tuning and I use closed boxes. Not much need for T-S with a closed box.
A loudspeaker will "break-in" in an hour or so with a permanent shift in compliance (everything else is stable.) This means a larger Vas and low Fs. But once this change happens nothing else does. But one should always be careful with T-S measurements to keep the signal level low and make sure that the driver is not overheated.
Mass added works fine. That's what I would use IF I did any T-S measurements.
I remember there are sources stating uneven orders are more audible than even orders. Don’t ask we which ones now.
I've heard that too, along with Trump won the election. Neither is true and who you listen to matters.
My original statement was that higher orders (almost) always are lower in level. What I meant was that the fundamental frequency excites higher orders less than lower orders, resulting in lower levels. So take the original metal cone with a resonance at 9k. H3 will manifest itself as a peak at 3kHz, while H5 comes up as a peak at 1,8kHz. You should in this case compare the peaks and not compare the levels of H3 and H5 on 1,8kHz or 3kHz respectively.
The caveat in this of course is that H5 could be more audible than H3, certainly if you would use a lowpass at 2,5kHz. I didn’t take the bandpass of the driver into account because it ran away from my original point.
Now, before someone goes on the loose again, I fully understand that the causes of distortion in a electrodynamic driver are rather complex and one should maybe not simplify. A lot of distortion control is trial and error, even with ‘big’ manufacturers. The extensive use of BEM and FEM analysis proves that more or less.
This -quite old- article on distortion still is quite valid and very readable. I might attend some readers at different causes of even and uneven orders of harmonic distortion which are described here... it was indeed that unsymmetrical behavior of the motor tends to produce even order harmonics, whereas symmetrical nonlinearities in the motor cause uneven order harmonics. Which is, in fact, perfectly logical.
Are you suggesting that a 9kHz component acting on a 9kHz resonance would sound different depending on its relationship to the fundamental that it came from?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.