I found a graph on the attached from the old magazine. It is a measurement of harmonic distortion.
Does anyone know how to interpret it? How important does the measurement of harmonic distortion?
I see many sources nowadays don't have it. Also, I have loudspeaker building cookbooks. They didn't discuss how to measure harmonic distortion as well.
Anyone, please help to suggest to me how to measure it.
Thank you in advance
Does anyone know how to interpret it? How important does the measurement of harmonic distortion?
I see many sources nowadays don't have it. Also, I have loudspeaker building cookbooks. They didn't discuss how to measure harmonic distortion as well.
Anyone, please help to suggest to me how to measure it.
Thank you in advance
Attachments
Last edited:
Take it line by line. For instance, take the bottom line. That shows a 40Hz fundamental test signal signal, an 80Hz 2nd harmonic and just a trace of 3rd at 120Hz.
The fundamental is at about 50dB, the 2nd at about 10? 15dB? So the distortion is about -35dB, -40dB?
All other lines do this for different fundamental frequencies. The accuracy of the graph is very bad and it should not be used to determine accurate magnitudes but more to see how distortion evolves with higher and higher frequencies.
Harmonic distortion in speakers is almost never shown because a) it doesn't correlate well with what you hear and b) it is horrible compared to electronics thus makes for poor marketing ;-)
Jan
The fundamental is at about 50dB, the 2nd at about 10? 15dB? So the distortion is about -35dB, -40dB?
All other lines do this for different fundamental frequencies. The accuracy of the graph is very bad and it should not be used to determine accurate magnitudes but more to see how distortion evolves with higher and higher frequencies.
Harmonic distortion in speakers is almost never shown because a) it doesn't correlate well with what you hear and b) it is horrible compared to electronics thus makes for poor marketing ;-)
Jan
Last edited:
A larger concern is intermodulation distortion because it produces frequencies that aren't related harmonically. The same nonlinearities that produce harmonic distortion also produce IMD so it follows that even low order "euphonic/musical" second order HD is something to be avoided as much as possible.
You want to measure the distortion of a loudspeaker? Get a measurement mic (like the Dayton) and a USB audio card. You can even use your smart phone with certain apps and hardware.
On Windows seek out the REW software, or ARTA. Either will do. There is an active thread on measuring distortion with REW.
Harmonics are present in almost all natural sounds. An instrument plays a note at 240 Hz, and there are always harmonics. The second harmonic is at 480 Hz, the third is at 720 Hz, etc... It is the harmonics that make a trumpet at 240 Hz sound different than a clarinet at 240 Hz or an electric guitar at 240 Hz.
Harmonic distortion is the presense of harmonics which are not suppose to be there... they were added by the microphone, the electronics, or the speaker.
If you listen to music with high levels of harmonic distortion, it can be difficult to tell certain instruments apart, or it may be difficult to tell how many instruments are playing. It starts to sound like homogenized musical noise, rather than music. It actually takes quite a bit of harmonic distortion to make music or speach sound bad.
As ScottJoplin says above, in a loudspeaker the same mechanisms which create harmonic distortion also tend to create IM distortion. Harmonic distortion is somewhat tolerable, but IM is bad. So we look for low harmonic distortion because it is easy to measure, and it tends to indicate that other forms of distortion (which are hard to measure) will also be low.
Harmonic distortion is the presense of harmonics which are not suppose to be there... they were added by the microphone, the electronics, or the speaker.
If you listen to music with high levels of harmonic distortion, it can be difficult to tell certain instruments apart, or it may be difficult to tell how many instruments are playing. It starts to sound like homogenized musical noise, rather than music. It actually takes quite a bit of harmonic distortion to make music or speach sound bad.
As ScottJoplin says above, in a loudspeaker the same mechanisms which create harmonic distortion also tend to create IM distortion. Harmonic distortion is somewhat tolerable, but IM is bad. So we look for low harmonic distortion because it is easy to measure, and it tends to indicate that other forms of distortion (which are hard to measure) will also be low.
Second and third are often tolerated in small amounts. At the risk of getting subjective, some say that 2nd adds a smooth warmth and 3rd, a cold touch. Some notice that it can mask other system problems leading to more material being listenable.
It's not a substitute for the real thing (unless you're trying to recapture the sound of long ago). 4th and higher harmonics are particularly out of place imo when high enough on their own.
It's not a substitute for the real thing (unless you're trying to recapture the sound of long ago). 4th and higher harmonics are particularly out of place imo when high enough on their own.
Indeed. It is the harmonics that cause the difference in sound between, say, a violin and a trumpet playing the exact same tone.
Jan
You can't change a driver's harmonic distortion with the crossover but it's possible to change that in the completed (mainly multi-way) loudspeaker, with a clever driver and crossover choice.
Edit: you can change the single driver's (usually non-linear) harmonic distortion electronically (via crossover etc.) at the expense of more linear distortion (freqency response). But even that more linear distortion can be beneficial if we think in multi-way and not in a single driver.
Edit: you can change the single driver's (usually non-linear) harmonic distortion electronically (via crossover etc.) at the expense of more linear distortion (freqency response). But even that more linear distortion can be beneficial if we think in multi-way and not in a single driver.
Last edited:
For example if you put a high-pass filter to a tweeter to reduce the low-end frequency response, then it's possible that you reduced the harmonic distortion percentage at it's low-end range because the HD response is likely a non-linear type of distortions so it doesn't change constantly with the SPL. Usually more SPL = more HD percentage.
The job of the crossover is not only to seamlessly blend the drivers together to achieve a flat frequency response but also make each driver only play the frequency ranges where they don't produce a lot of distortion.
The trick to designing a high performing multi-way speaker is therefore to find a set of drivers where each driver plays it's allotted frequency range as cleanly as possible, and it doesn't really matter how badly behaved it is far above or below that frequency range since the crossover will attenuate those frequencies sufficiently that the driver won't be heard.
Say you have a woofer which produces lots of distortion above 4kHz but plays cleanly to 3kHz. You also have a tweeter which produces distortion below 1kHz, but plays cleanly above 2kHz. When you add a crossover at 2.5kHz, you use each driver only over the range of frequencies where it plays cleanly. At the frequencies where lots of distortion is produced, each driver is much attenuated by the crossover so the distortion that is produced is played so quietly that you can't hear it. The more overlap you have in the frequency ranges which the drivers play cleanly in a multi-way system, the easier it will be to design the crossover as you have more freedom to pick a crossover frequency which gives you flat frequency response and good phase alignment. Well behaved drivers will also be more forgiving in a simple crossover design without steep rolloffs (low order filters).
If you have a woofer that produces distortion above 1kHz and a tweeter that produces distortion below 3kHz then they are a poor match as neither driver wants to play cleanly between 1 and 3kHz. The only way to reduce the amount of distortion produced by the drivers between 1 and 3khz is to attenuate both drivers between 1 and 3kHz with the crossover, resulting in a large dip in the frequency response over that range.
I've also seen people add a gentle slope to the overall frequency response to compensate for non-linear distortion. E.g. if the woofer is the weak link in the design and produces comparatively higher distortion in the bass and midrange than the tweeter does in the treble, they will design the crossover so the overall frequency response tilts up a few dB towards the higher frequencies, the idea being that the distortion in the lower frequencies will make it seem louder than it really is. Still, it's no substitute for just using drivers which produce exceptionally low distortion across the entire frequency range.
The trick to designing a high performing multi-way speaker is therefore to find a set of drivers where each driver plays it's allotted frequency range as cleanly as possible, and it doesn't really matter how badly behaved it is far above or below that frequency range since the crossover will attenuate those frequencies sufficiently that the driver won't be heard.
Say you have a woofer which produces lots of distortion above 4kHz but plays cleanly to 3kHz. You also have a tweeter which produces distortion below 1kHz, but plays cleanly above 2kHz. When you add a crossover at 2.5kHz, you use each driver only over the range of frequencies where it plays cleanly. At the frequencies where lots of distortion is produced, each driver is much attenuated by the crossover so the distortion that is produced is played so quietly that you can't hear it. The more overlap you have in the frequency ranges which the drivers play cleanly in a multi-way system, the easier it will be to design the crossover as you have more freedom to pick a crossover frequency which gives you flat frequency response and good phase alignment. Well behaved drivers will also be more forgiving in a simple crossover design without steep rolloffs (low order filters).
If you have a woofer that produces distortion above 1kHz and a tweeter that produces distortion below 3kHz then they are a poor match as neither driver wants to play cleanly between 1 and 3kHz. The only way to reduce the amount of distortion produced by the drivers between 1 and 3khz is to attenuate both drivers between 1 and 3kHz with the crossover, resulting in a large dip in the frequency response over that range.
I've also seen people add a gentle slope to the overall frequency response to compensate for non-linear distortion. E.g. if the woofer is the weak link in the design and produces comparatively higher distortion in the bass and midrange than the tweeter does in the treble, they will design the crossover so the overall frequency response tilts up a few dB towards the higher frequencies, the idea being that the distortion in the lower frequencies will make it seem louder than it really is. Still, it's no substitute for just using drivers which produce exceptionally low distortion across the entire frequency range.
Last edited:
It's because the amplitude of the distortion rises non-linearly with the fundamental amplitude. e.g. a driver playing a fundamental at 100dB may produce a 3rd order harmonic at 60dB (-40dB). If you drop the amplitude of the fundamental to 80dB the 3rd order harmonic may drop to say 20dB (-60dB).
This is because operating over small portion of a non-linear transfer function approximates a (closer to) linear function. Same reason amplifiers produce lower distortion at lower power output levels.
This is because operating over small portion of a non-linear transfer function approximates a (closer to) linear function. Same reason amplifiers produce lower distortion at lower power output levels.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.