after getting flak for contesting the conventional classification of loudspeakers are "linear devices" i find this https://www.klippel.de/fileadmin/_m...linearities–Causes_Parameters_Symptoms_01.pdf
the first sentence in the introduction confirms something i've long felt was true and that response varies with volume.
the first sentence in the introduction confirms something i've long felt was true and that response varies with volume.
You only have too look at the data sheet frequency response to see loudspeakers are far from a flat frequency response.
Nice paper, thanks for posting it! I've found the klippel data very interesting in the Vance Dickason speaker test articles and this paper has made them even more interesting now that I have keys to really understand what the data is all about 😀
Klippel has developed some technology to address these nonlinearities in realtime! Don't know when an actual products are available... Very cool new tech though, Klippel Controlled Sound : Cancellation of Nonlinear Distortion Be sure to check them demo videos!🙂
Klippel has developed some technology to address these nonlinearities in realtime! Don't know when an actual products are available... Very cool new tech though, Klippel Controlled Sound : Cancellation of Nonlinear Distortion Be sure to check them demo videos!🙂
Last edited:
Klippel goes into some nice details and I appreciate his use of illustrations and graphs. I'll add another good read which is Geddes chapter on the subject from his book Audio Transducers. He discusses the measurment of nonlinearities via nonlinear transfer functions expanded into polynomial orders in a way that's eloquent and easy to follow. He also goes into the psychoacoustic perception of distortion which Klippel's paper does not cover.
http://www.gedlee.com/downloads/AT/Chapter_10.pdf
http://www.gedlee.com/downloads/AT/Chapter_10.pdf
Last edited:
This is semantics. Calling speakers linear devices is not a declaration that they are not non-linear, it classifies their primary behaviour for the purpose of analysis. This technical term has been misinterpreted.
Agreed.
This applies not just to lab analysis but to your living room as well. When properly designed and operated as intended, speakers behave as linear devices.
Last edited:
Even still (to be pedantic), even when they are showing non-linear behaviour, they are also linear devices.
The point is these are separate things. They can be talked about in isolation regardless of the other one.
As it turns out it is simpler to do linear analysis when non-linearity is lowest, but to read anyhting into this would be a red herring.
The point is these are separate things. They can be talked about in isolation regardless of the other one.
As it turns out it is simpler to do linear analysis when non-linearity is lowest, but to read anyhting into this would be a red herring.
You could talk about the performance of a car in a straight line, then introduce the fact that it also turns corners. You could point out that even for a given speed, the velocity changes when turning. This doesn't invalidate straight line performance, it merely complicates the analysis of it.
I assume you're working towards a point here so I'll speculate. HD at differing frequencies for a given level may vary, thereby diverting different amounts of energy away at other frequencies, so OK..
Measure the frequency response of any driver at at least three different drive levels 10dB apart from each other. The measured level differences between the three different curves will be exactly 10dB at frequencies where there are peaks in the impedance curve.
At frequencies where the driver has an impedance minimum, there will be less than 10dB between each of the curves. This is going to be due to thermal power compression from the increased heating from the increased current flow at this frequency.
At the very low end of the drivers response, there will be less than 10dB between each of the curves because the suspension will be limiting the excursion of the driver as level increases more and more.
You can also measure the driver's impedance curve at different drive levels. This makes it very easy to see some on the nonlinearities in the driver.
At frequencies where the driver has an impedance minimum, there will be less than 10dB between each of the curves. This is going to be due to thermal power compression from the increased heating from the increased current flow at this frequency.
At the very low end of the drivers response, there will be less than 10dB between each of the curves because the suspension will be limiting the excursion of the driver as level increases more and more.
You can also measure the driver's impedance curve at different drive levels. This makes it very easy to see some on the nonlinearities in the driver.
Yes, but only if you're hammering it.
I took a Faital 10HX230, a passive crossover, a mic that's happy up to 140dB, and a 1.2KW@8ohm amplifier.
I then ran a sweep with REW, increased the levels by 3dB, and ran the sweep again. And again. And again. And then some more times until the speaker started to show real signs of stress.
The final sweep recorded around 125dB, which is impressive for sine tones. Harmonic distortion got as high as 50% in places, and there was a few dB of compression above 10kHz.
Interestingly, the 3" port coped fine with the air flow, but there was a 100Hz highpass.
Driven with more bass, I'd expect the usual port compression to show up at the bottom of the range.
Chris
There was the gradual distortion rise with level, as you'd expect from flux modulation, suspension non-linearities etc.
However, the frequency response only changed at extreme levels. The 10" cone is rated for 250w long-term, and the HF driver more like 30w, so to dump a 1KW sine sweep in there is seriously abusive.
Chris
Over on HiFi Compass many of their driver measurements are done at increasing voltage levels. That might yield some insight.
https://hificompass.com/en/speakers/measurements
Nothing as extreme as Chris' torture test.
https://hificompass.com/en/speakers/measurements
Nothing as extreme as Chris' torture test.
Pano sorry not seeing any response v level comparisons (but i may not be looking in the right place)
does anyone know of any previous work or studies on this very subject?
i understand that the absence of previous work in no way constitutes a confirmation of response variability with volume, but for me it does raise the question is this something overlooked?
i wish i could go back in time to analyze the conditions of measurement i've done in the past as the response variations which i've seen are what led to the perhaps erroneous conclusion that frequency response varies with volume.
does anyone know of any previous work or studies on this very subject?
i understand that the absence of previous work in no way constitutes a confirmation of response variability with volume, but for me it does raise the question is this something overlooked?
i wish i could go back in time to analyze the conditions of measurement i've done in the past as the response variations which i've seen are what led to the perhaps erroneous conclusion that frequency response varies with volume.
Last edited:
Data bass does response vs level comparisons, obviously only for subwoofers but you'll get an idea of the general trends.
As Chris points out, the main differences tend to show up most at very high drive levels.
As Chris points out, the main differences tend to show up most at very high drive levels.