I thought Michael showed in that other thread that the nonlinear distortion from thermal heating was very small indeed, it seems like Earl does not agree with that.
/Peter
/Peter
keyser said:
The major form of "peak clipping" in a loudspeaker is the BL product falloff with excursion. This is completely unrelated to the thermal properties of the VC. Well maybe not completely, but mostly. The thermal clipping does not happen at waveform speeds, but it does happen at music RMS envelope speeds. It's all about the time constants involved. Only rarely will the thermal time constants be short enough to overlap into the signal frequency spectrum, but will always be in the signal envelope spectrum for music. Cone motion depends on voltage and it reacts very fast. Voice coil temperature (and hence Re) on the other hand depends on the integrated signal power (volts^2), integrated over the thermal time constants. They are only rarely going to interact with each other and for all practical purposes you can assume that they don't.
I never read such disagreement in Earl's posts and the post above confirms that he also thinks that nonlinear distortion from thermal heating is small. It is too slow to cause substantial nonlinear distortion, but it is fast enough to follow the envelope of the music and therefore have an impact on the reproduction of the dynamics in the music.Pan said:
Mr. Geddes, it still isn't clear to me. It seems you did not fully understand what I was trying to say. I'm not sure how I could be much clearer, but I'll give it another try.
I am looking at peak limiting (exceeding xmax) and thermal compression separately. The first causes the stimulus to clip, which in turn leads to generation of harmonics. So, exceeding xmax leads to both power compression and harmonic distortion, which are directly interrelated and inseparable, as they are both results of the same issue.
Of the second, thermal compression, I always thought it to be a more or less steady state phenomenon. As you pointed out, it is actually dynamically related to the signal. You proposed an intricate method of measuring thermal compression. But now my point is: if thermal compression is dynamically related to the signal, then it alters the waveform; if it alters the waveform, it will invariably lead to harmonic distortion.
Exceeding xmax leads to clipping of the peak of the waveform, which leads to harmonics. Thermal compression would primarily occur when the cone moves through the zero-axis, and would therefore cause subharmonics.
Why not simply measure the spectrum and look for subharmonics to tell if power compression is occuring?
I am looking at peak limiting (exceeding xmax) and thermal compression separately. The first causes the stimulus to clip, which in turn leads to generation of harmonics. So, exceeding xmax leads to both power compression and harmonic distortion, which are directly interrelated and inseparable, as they are both results of the same issue.
Of the second, thermal compression, I always thought it to be a more or less steady state phenomenon. As you pointed out, it is actually dynamically related to the signal. You proposed an intricate method of measuring thermal compression. But now my point is: if thermal compression is dynamically related to the signal, then it alters the waveform; if it alters the waveform, it will invariably lead to harmonic distortion.
Exceeding xmax leads to clipping of the peak of the waveform, which leads to harmonics. Thermal compression would primarily occur when the cone moves through the zero-axis, and would therefore cause subharmonics.
Why not simply measure the spectrum and look for subharmonics to tell if power compression is occuring?
Keyser
I think that I understood you.
Lets just steer clear of the X-max discussion because one should not be using a driver near these limits, which are ill defined in any case. Lets just talk about "typical" loudspeaker distortion aspects.
"if thermal compression is dynamically related to the signal, then it alters the waveform; if it alters the waveform, it will invariably lead to harmonic distortion"
That depends on how fast it reacts. If it is slower than the waveform then its efect on the waveform is minimal. This is the case. Its slower than the waveform but NOT slower than the musical dynamics.
I think that I understood you.
Lets just steer clear of the X-max discussion because one should not be using a driver near these limits, which are ill defined in any case. Lets just talk about "typical" loudspeaker distortion aspects.
"if thermal compression is dynamically related to the signal, then it alters the waveform; if it alters the waveform, it will invariably lead to harmonic distortion"
That depends on how fast it reacts. If it is slower than the waveform then its efect on the waveform is minimal. This is the case. Its slower than the waveform but NOT slower than the musical dynamics.
clipping does the same thing to every identical/successive cycle. Hence the distortion is harmonically related - THD
thermal compression does different things to identical/successive cycles as the temperature "signal" is related to some RMS average of the waveform envelope. Therefore it modulates the waveform causing sidebands to appear - IMD.
If the time constants are slow enough then the sidebands are so close to the original signal that they are masked. It would be an interesting exercise to see what order of magnitude the time constant should be to keep the sidebands (on average) within the mask of the original signal.
...or we could just listen🙂
thermal compression does different things to identical/successive cycles as the temperature "signal" is related to some RMS average of the waveform envelope. Therefore it modulates the waveform causing sidebands to appear - IMD.
If the time constants are slow enough then the sidebands are so close to the original signal that they are masked. It would be an interesting exercise to see what order of magnitude the time constant should be to keep the sidebands (on average) within the mask of the original signal.
...or we could just listen🙂
Iain McNeill said:
Yes, thats exactly what I want to do someday.
Iain McNeill said:
Thats a sure road to misinformation!
FWIW
The pdf of the AES paper (7059):
"Heat Dissipation and Power Compression in Loudspeakers"
by Douglas J Button
is available on Harman's Scientific and White papers web pages
http://www.harman.com/about_harman/technology_leadership.aspx
The pdf of the AES paper (7059):
"Heat Dissipation and Power Compression in Loudspeakers"
by Douglas J Button
is available on Harman's Scientific and White papers web pages
http://www.harman.com/about_harman/technology_leadership.aspx
True. it is very true. I totally agree. I also agree that some measurements are not correct and may be measuring more than simply distortion, but then, so are our ears.
which leads me to:
how does Geddes and other distinguish the IMD and THD? or should i also say the WAY that IMD produced by CERTAIN generated harmonics masks them from audibilty in a way that OTHER generated harmonics are not. Since the IMD and THD are inseperable, im also unsure what 'listener perceptions' are telling us about the nature of the distortions. on a conscious level they may not be strictly audible (in the best case scenario), but to the subconscious mind they may in fact be recognised as unnatural distortions, as measurements could also recognise.
An easy experiment would be to mix two sines say fundamental and 2nd harm or fundamental and 3rd harm. at similar levels of harmonic content the 2nd would be tonally more acceptable and less noticeable, as it seems more natural and the brain ignores it.
repeat the experiment again, but shift the harmonics(detune them) by 5%, 10% etc. listen again. unless you are tone deaf you will hear the difference in the IMD 'beat' and also BOTH distortions will be more audible. surely this is the brain telling you its out of tune, and unnatural.
what id really like to know, is what is the 'average' detuning' of said harmonics that can be detected by ear, and then subsequently, how perfectly 'tuneful' are the 2nd and 3rd harmonics generated by noisy suspension and cone noises, (not to mention air leak, port noise). Id be fairly sure they arent picture perfect 2nd and 3rd harmonics. unless that is, im hearing doppler that is detuning the distortion from my 5" audax's, which i KNOW suffer audible 2nd and 3rd......
and more madness abounds.....
in tweeters, unfortunately, i have to say that THD thresholds may be high as the 'average' listener has untrained ears, and may just view the slightly spectral brightness as a 'nice hifi response'. after all, thousands of people still but cr@ppy stereos and swear theyre the best thing theyve ever heard. just as thousands still have, for all intensive purposes, disco speakers in the homes, with god awful phenolic compression driven horns and essentially, souped up, well finished PA drivers. No-one needs a kapton coated VC on grp former, on a speaker with a robust paper cone(made with hard resonant paper short fibre paper) and powered by a 5 watt SET.
5 WATTS WOULDNT EVEN TICKLE IT!
sorry for the caps but that level of thermal compression and VC heating does not exist outside of auditorium levels or a rave. lets not forget the increase in R with increase in T unit for unit, is tiny, and that the reduction in dynamic SPL caused by such a small dynamic and time lagged increase in VC R is miniscule fractions of a dB. Even if dramatic overheating and overdriving signlas were applied, the lag in VC heating wouldnt hamper transients directly, but change the tonal balance slowly until the heat is dissipated, particularly in non FR systems where the xo points would shift, changing tonality.
MAYBE, if you have a 5watt tweeter, powered by a small SET of 5 watts also, heavily distorting, THEN you may get some thermal effects. but i expect they would be more due to the excessive level of THD in the SET output at max output, and any 'lack of dynamic headroom' would likely be caused by the amp more than anything.
at reasonable levels, lack of dynamic headroom is more likely to be caused by overly resistive surrounds, overly heavy cones, and mechanical system that is the enclosure, than by VC heating.
Last edited:
i believe you tested this idea. what i dont get is how on earth you expect anyone to be able to perceive THD at 120dB, if thats what occurred. Or come to think of it...how you could expect any microphone to withstand 120dB and not generate its own THD in large amounts.
One thing i know. MY eardrums generate their own audible THD when i get anywhere even remotely close to 110dB
The playback was NOT at 120 dB. All signals were played back at the same level, probably 90 dB or so. The mics were large 1" B&K and are no problem at 120 dB.
ok well that explains alot. even though large studio 1" diaphragm mics are normally only good for max spl of 126 dB, and even then are far outside what is optimum.
still in the persuit of trying to keep an open mind about this, i will try it and see for myself. I may be surprised. however if i can hear white noise at -30db i am doubtful about the not hearing the THD at a similar level. and whilst using correct methods, a smaple of 25 or even 100 people does not give a large enough sample for any real correlation to be reliably made, about anything, let alone a very subjective phenomena.
But again i will have a go.
thats ok. i am in the process of reading it. i only question it as it seems a little too much of a leap of faith. whilst steady state signals are less able to hide THD, can the transient signal mask it so completely?
i have two wav's here. both normalised to the same level. one is pure sine at 1k and the other is the same sine, plus 1.3k harmonic only, at a level of -30dB relative to the fundamental. i can clearly hear the 'chord' the harmonic creates. i will work on a similar gated version, with gate times of perhaps 100ms, unless you have a better suggestion of gate times. whether the same level will be audible in the gated version i have yet to find out.
ok, have just tried to upload but im unable to. maybe the file size is too large (5MB for 30s)
nope file size isnt it. guess this forum doesnt like wav's 😀
i shortened the files to 50ms in length. the file with added harmonic IS quite easy to differentiate (at least to me) from the pure sine, although not to the same degree as a longer signal. this is not due to any loudness increase or other spurious cue. it does sound more harmonically rich, even with just a single harmonic. I can see why a negative correlation could be found, since it could be construed as a richer sound, but the pure sine signal still sounds like a purer sample to me. If i can find a way to post the wav on here i will, in case anyone thinks i have optimistic ears....🙄
i shortened the files to 50ms in length. the file with added harmonic IS quite easy to differentiate (at least to me) from the pure sine, although not to the same degree as a longer signal. this is not due to any loudness increase or other spurious cue. it does sound more harmonically rich, even with just a single harmonic. I can see why a negative correlation could be found, since it could be construed as a richer sound, but the pure sine signal still sounds like a purer sample to me. If i can find a way to post the wav on here i will, in case anyone thinks i have optimistic ears....🙄
These are the valid file extensions:
bmp doc gif jpe jpeg jpg pdf png psd txt zip
Says so when you hit the "Manage Attachments" button
bmp doc gif jpe jpeg jpg pdf png psd txt zip
Says so when you hit the "Manage Attachments" button
- Status
- Not open for further replies.