That depends. Distortion on its own can be from any source.
If, however, you had a peak in sensitivity at 3 kHz, then yes, 1 kHz would show a strong spike of 3HD, 600 Hz would show a strong spike of 5HD, etc.
Going through Zaph's HD plots of all his drivers will be most enlightening.
Last edited:
Yes, I thought about breakup peaks as can be seen on Zaphs plots; so if a breakup occurs at say 6kHz and we can see a HD3 peak at 2kHz on the graph, that would mean the 6kHz fr is distorded, not the fundamental 2kHz. Correct?
What Tom is describing and what this thread is about is something I've seen described as being called cone resonance amplification.
As an example lets take a theoretical motor system.
For the sake of this explanation lets assume that the motor has zero linear distortion - in other words its frequency response alone is perfectly flat.
Now lets say that the motor itself displays non linear behaviour, such that it will always produce 1% 2nd harmonic distortion, 0.5% 3rd harmonic distortion, 0.1% 4th order distortion and 0.05% 5th order distortion.
If we now attach a theoretically perfect cone to this motor, then the resulting frequency response will be perfectly flat and the individual harmonic distortion plot will also appear perfectly flat.
Now lets say we replace our perfect cone with a real world cone and what happens? The frequency response starts out as perfectly flat as the cone is acting as a perfect piston. However at some point the cone goes through breakup and as a result resonances occur. This causes the frequency response to no longer be perfectly flat.
As a decent example here is Zaph's measurement of a Jordan full range driver.
So what's happened here you could say is our perfect motor with a flat frequency response has now been modified by the cone to produce wiggles and bumps.
Lets not forget here also, that our perfectly flat motor also has perfectly flat harmonics that exist, as were detailed at the top of the thread. But what exactly does this mean?
Lets go back and look at our motor with the perfect cone attached to it again. Lets also say that we've got the loudspeaker producing 100dBs. In other words this means, that regardless of what frequency we call on the loudspeaker to reproduce it will always produce 100dBs, ie the frequency response is perfectly flat.
Now lets go back to the listed harmonics.
The first up is the 2nd harmonic at 1%. Lets do a bit of maths to that 1%.
20 * log (100/1) = 40.
In other words take the logarithm of 100% divided by 1% and multiply it by 20. In this case we end up with 40 or 40dB.
What this means is that if we're playing a 1khz sine wave (the stimulus) and the loudspeaker turns this into 100dB of sound pressure, then a second unwanted tone will be generated by the loudspeaker system at twice the frequency of the stimulus and at 40dB less then 100dB tone that the loudspeaker is reproducing.
Another name for the pure stimulus sine wave is called the fundamental. In this case we would say the loudspeaker will generate a second, unwanted tone, at twice the frequency of the fundamental and at 40dB less then the fundamental. Or we can simplify that even more to say, the loudspeakers second harmonic distortion is always @ -40dB.
For our perfectly flat driver which always produces 1% of second order harmonic distortion, we would expect the graph that plots the second harmonic to appear as perfectly flat and @ 40dB less then the fundamental.
It is exactly the same for the 3rd harmonic, in this case at 0.5%. Doing the same mathematics we can convert that 0.5% into a dB figure of 46dB.
This means that along with the perfectly flat 2nd harmonic @ 1% or -40dB, the driver will also produce a second tone that is unwanted but this time at three times the fundamental and now at 46dB below the fundamental.
Now lets say we remove the perfect cone and replace it with another cone that is identical except that it has a resonance at 1khz. The result is that when driven by the same 1khz stimulus as before, the driver now produces 105dB. At every other frequency the cone still produces 100dB. We could say that the cone is amplifying the signal at 1khz by 5dB.
Now lets go back and look at our second harmonic. We know that this is produced as a multiple of two of the fundamental frequency. We also know that the cone resonates at 1khz, so if we want to excite this 1khz with the second harmonic we will need to use a fundamental drive frequency of 500hz.
Now if we drive the loudspeaker with a 500hz fundamental it will reproduce this at 100dB. The motor is still exactly the same, so along with generating this 500hz fundamental it will also generate an unwanted second harmonic @ 1%, in this case 40dB less then that 100dB. However this time around the cone will resonate at 1khz and amplify the second harmonic by 5dB. Our 40dB less has now turned into 35dB less and as a result we've got a greater amount of distortion produced, around 1.8% now vs the 1% of before.
For the third harmonic it is exactly the same, only this time around a fundamental of 333.33333....hz is required. The motor will produce a third harmonic of 0.5% or 46dB less, but the cone will once again ring at 1khz and amplify this by 5dB, taking it up to 41dB less and increasing the distortion up to 0.9%.
This is what forms the essence of cone resonance amplification, ie the cones resonances amplify and modify the behaviour of the motor.
If we go back to the Jordan driver and look at the distortion plot we can see something very familiar.
Look at the upper end of the third, fourth and fifth harmonics. Notice anything? They practically mirror the upper frequency response of the cone, but at multiples of three, four and five times lower then where the response issues appear in the normal frequency response plot.
Why this doesn't occur in the second harmonic I don't know, if someone else more knowledgeable then me would like to explain this it would be appreciated.
As an example lets take a theoretical motor system.
For the sake of this explanation lets assume that the motor has zero linear distortion - in other words its frequency response alone is perfectly flat.
Now lets say that the motor itself displays non linear behaviour, such that it will always produce 1% 2nd harmonic distortion, 0.5% 3rd harmonic distortion, 0.1% 4th order distortion and 0.05% 5th order distortion.
If we now attach a theoretically perfect cone to this motor, then the resulting frequency response will be perfectly flat and the individual harmonic distortion plot will also appear perfectly flat.
Now lets say we replace our perfect cone with a real world cone and what happens? The frequency response starts out as perfectly flat as the cone is acting as a perfect piston. However at some point the cone goes through breakup and as a result resonances occur. This causes the frequency response to no longer be perfectly flat.
As a decent example here is Zaph's measurement of a Jordan full range driver.
So what's happened here you could say is our perfect motor with a flat frequency response has now been modified by the cone to produce wiggles and bumps.
Lets not forget here also, that our perfectly flat motor also has perfectly flat harmonics that exist, as were detailed at the top of the thread. But what exactly does this mean?
Lets go back and look at our motor with the perfect cone attached to it again. Lets also say that we've got the loudspeaker producing 100dBs. In other words this means, that regardless of what frequency we call on the loudspeaker to reproduce it will always produce 100dBs, ie the frequency response is perfectly flat.
Now lets go back to the listed harmonics.
The first up is the 2nd harmonic at 1%. Lets do a bit of maths to that 1%.
20 * log (100/1) = 40.
In other words take the logarithm of 100% divided by 1% and multiply it by 20. In this case we end up with 40 or 40dB.
What this means is that if we're playing a 1khz sine wave (the stimulus) and the loudspeaker turns this into 100dB of sound pressure, then a second unwanted tone will be generated by the loudspeaker system at twice the frequency of the stimulus and at 40dB less then 100dB tone that the loudspeaker is reproducing.
Another name for the pure stimulus sine wave is called the fundamental. In this case we would say the loudspeaker will generate a second, unwanted tone, at twice the frequency of the fundamental and at 40dB less then the fundamental. Or we can simplify that even more to say, the loudspeakers second harmonic distortion is always @ -40dB.
For our perfectly flat driver which always produces 1% of second order harmonic distortion, we would expect the graph that plots the second harmonic to appear as perfectly flat and @ 40dB less then the fundamental.
It is exactly the same for the 3rd harmonic, in this case at 0.5%. Doing the same mathematics we can convert that 0.5% into a dB figure of 46dB.
This means that along with the perfectly flat 2nd harmonic @ 1% or -40dB, the driver will also produce a second tone that is unwanted but this time at three times the fundamental and now at 46dB below the fundamental.
Now lets say we remove the perfect cone and replace it with another cone that is identical except that it has a resonance at 1khz. The result is that when driven by the same 1khz stimulus as before, the driver now produces 105dB. At every other frequency the cone still produces 100dB. We could say that the cone is amplifying the signal at 1khz by 5dB.
Now lets go back and look at our second harmonic. We know that this is produced as a multiple of two of the fundamental frequency. We also know that the cone resonates at 1khz, so if we want to excite this 1khz with the second harmonic we will need to use a fundamental drive frequency of 500hz.
Now if we drive the loudspeaker with a 500hz fundamental it will reproduce this at 100dB. The motor is still exactly the same, so along with generating this 500hz fundamental it will also generate an unwanted second harmonic @ 1%, in this case 40dB less then that 100dB. However this time around the cone will resonate at 1khz and amplify the second harmonic by 5dB. Our 40dB less has now turned into 35dB less and as a result we've got a greater amount of distortion produced, around 1.8% now vs the 1% of before.
For the third harmonic it is exactly the same, only this time around a fundamental of 333.33333....hz is required. The motor will produce a third harmonic of 0.5% or 46dB less, but the cone will once again ring at 1khz and amplify this by 5dB, taking it up to 41dB less and increasing the distortion up to 0.9%.
This is what forms the essence of cone resonance amplification, ie the cones resonances amplify and modify the behaviour of the motor.
If we go back to the Jordan driver and look at the distortion plot we can see something very familiar.
Look at the upper end of the third, fourth and fifth harmonics. Notice anything? They practically mirror the upper frequency response of the cone, but at multiples of three, four and five times lower then where the response issues appear in the normal frequency response plot.
Why this doesn't occur in the second harmonic I don't know, if someone else more knowledgeable then me would like to explain this it would be appreciated.
I'm sorry, I've read this topic from the beginning but it's not yet clear to me what to do in these cases, like with this Seas L18RNX woofer:
Zaph says on his all metal design:
Is this the way to go? or a 18dB or 24dB would be preferable to avoid the trouble area?
Thanks
Zaph says on his all metal design:
Is this the way to go? or a 18dB or 24dB would be preferable to avoid the trouble area?
Thanks
Hi,
Say you have a bass/mid that is flat except it has a +10dB peak around 6KHz before roll-off.
This will cause a 10dB peak in :
2nd harmonic distortion around 3KHz
3rd harmonic distortion around 2KHz
4th harmonic distortion around 1.5KHz
5th harmonic distortion around 1.2KHz
6th harmonic distortion around 1KHz
etc......
If you cross the driver over 4th order L/R acoustic @ 2KHz the driver
will be 6dB down at 2KHz, so you only get a mild peak for 3rd harmonic
and the 2nd harmonic peak is effectively suppressed.
You need low levels of the higher order harmonics.
Trap filters only affect the frequency response at in the above around 6KHz.
They have no effect on the distortion the mechanical peak produces.
So ideally for a high Q metal cone you push the frequency as high as possible,
though in some cases a lower frequency with better damped Q might work.
The lower the c/o frequency the better, but this is limited by the tweeter,
rendering many peaky larger units to be effectively bass units only.
rgds, sreten.
note how the F3, F4 and F5 peaks 1 to 2Khz area, drop in frequency
Say you have a bass/mid that is flat except it has a +10dB peak around 6KHz before roll-off.
This will cause a 10dB peak in :
2nd harmonic distortion around 3KHz
3rd harmonic distortion around 2KHz
4th harmonic distortion around 1.5KHz
5th harmonic distortion around 1.2KHz
6th harmonic distortion around 1KHz
etc......
If you cross the driver over 4th order L/R acoustic @ 2KHz the driver
will be 6dB down at 2KHz, so you only get a mild peak for 3rd harmonic
and the 2nd harmonic peak is effectively suppressed.
You need low levels of the higher order harmonics.
Trap filters only affect the frequency response at in the above around 6KHz.
They have no effect on the distortion the mechanical peak produces.
So ideally for a high Q metal cone you push the frequency as high as possible,
though in some cases a lower frequency with better damped Q might work.
The lower the c/o frequency the better, but this is limited by the tweeter,
rendering many peaky larger units to be effectively bass units only.
rgds, sreten.
note how the F3, F4 and F5 peaks 1 to 2Khz area, drop in frequency
Last edited:
- Status
- Not open for further replies.