Hi guys,
I have some (very basic) doubts about a well-known phenomenon and its implication in building a multi-way loudspeaker
Let's consider a rigid membrane mid-woofer with a breakup at 4khz. For physical reasons that I can't still fully understand, this hypothetical mid-woofer shows distortion spikes, K2 at 1/2 f-breakup, and K3 at 1/3 f-breakup (is it right?).
My question:
Are these spikes "triggered" by a signal of 3khz, or are they present in any case? In other words, the reason of these spikes is a crossover not low/steep enough? And what if I used (theoretically) a brickwall at 2.5khz? In this case no 3khz signal would reach the driver triggering the breakup; so would I still experience those nasty K2 and K3 distortion spikes?
Hope it's clear... sorry if this sounds too trivial...
I have some (very basic) doubts about a well-known phenomenon and its implication in building a multi-way loudspeaker
Let's consider a rigid membrane mid-woofer with a breakup at 4khz. For physical reasons that I can't still fully understand, this hypothetical mid-woofer shows distortion spikes, K2 at 1/2 f-breakup, and K3 at 1/3 f-breakup (is it right?).
My question:
Are these spikes "triggered" by a signal of 3khz, or are they present in any case? In other words, the reason of these spikes is a crossover not low/steep enough? And what if I used (theoretically) a brickwall at 2.5khz? In this case no 3khz signal would reach the driver triggering the breakup; so would I still experience those nasty K2 and K3 distortion spikes?
Hope it's clear... sorry if this sounds too trivial...
I'm pretty sure that this topic was handled here already, but in case it hasn't:
think of the driver as a coupled system of motor (magnet, voice coil) and membrane. Playing back a signal at 1/2, 1/3, 1/4... of breakup frequency will almost always generate non-linear distortion in the motor. This distortion is then coupled in the membrane, where it is linearly (!) distorted. Example: signal at 2 kHz, non-linear distortion at 2x this frequency (4 kHz) at -40 dB, breakup resonance +10 dB over normal level, resultant distortion at - 30 dB.
Only remedy: cross over far/steep enough below the distortion spikes. Usually, you can ignore harmonics higher than the 4th order, as long as the driver isn't crap they're not high enough.
think of the driver as a coupled system of motor (magnet, voice coil) and membrane. Playing back a signal at 1/2, 1/3, 1/4... of breakup frequency will almost always generate non-linear distortion in the motor. This distortion is then coupled in the membrane, where it is linearly (!) distorted. Example: signal at 2 kHz, non-linear distortion at 2x this frequency (4 kHz) at -40 dB, breakup resonance +10 dB over normal level, resultant distortion at - 30 dB.
Only remedy: cross over far/steep enough below the distortion spikes. Usually, you can ignore harmonics higher than the 4th order, as long as the driver isn't crap they're not high enough.
Baseballbat is correct. The non-linear distortion of the motor is shaped by the linear distortion of the cone. A 2.5kHz brickwall filter will not help, since the harmonics occuring at 4kHz are being introduced by the non-linearity of the motor at lower frequencies (1.33kHz for 3rd order) and then linearly distorted by the cone.
There are two mechanisms to fix it:
One is to shape the frequency response of the driver by mechanical means to reduce the amplitude of the breakup - for example placing the speaker in a bandpass enclosure, or applying a damping compound to the cone.
The second is to increase the efficiency of the driver at the fundamental by acoustically loading it with a ported/bandpass enclosure or horn/waveguide so that for a given SPL level the excursion is reduced. Therefore for the same SPL level it produces less non-linear distortion since the motor operates in a more linear region.
If those are not practical, the only option is to avoid creating any harmonics at 4kHz. 2nd and 3rd order will be the main problem - to avoid 3rd order you should cross below 1/3 of the breakup node, so maybe around 1kHz for a driver with a large breakup node at 4kHz. High 4th order distortion is not common however speakers with simple motors often produce audible amounts of 5th order, especially if the breakup node is large.
There are two mechanisms to fix it:
One is to shape the frequency response of the driver by mechanical means to reduce the amplitude of the breakup - for example placing the speaker in a bandpass enclosure, or applying a damping compound to the cone.
The second is to increase the efficiency of the driver at the fundamental by acoustically loading it with a ported/bandpass enclosure or horn/waveguide so that for a given SPL level the excursion is reduced. Therefore for the same SPL level it produces less non-linear distortion since the motor operates in a more linear region.
If those are not practical, the only option is to avoid creating any harmonics at 4kHz. 2nd and 3rd order will be the main problem - to avoid 3rd order you should cross below 1/3 of the breakup node, so maybe around 1kHz for a driver with a large breakup node at 4kHz. High 4th order distortion is not common however speakers with simple motors often produce audible amounts of 5th order, especially if the breakup node is large.
Last edited:
Hi,
Its quite simple. If a driver has a frequency peak at 4KHz
it will exacerbate 2nd harmonic distortion for 2KHz, 3rd
for 4/3KHz, 4th for 1KHz and 5th for 800Hz.
The x/o cannot stop the driver doing this.
rgds, sreten.
Nice writeup.
I would add a third remedy: reduce distortion using current drive amplification.
Here is an example with a small 4" driver (JBL C500G, used in the JBL LSR32 and LSR6332) with breakups occurring above 4kHz. It is a 4 ohm driver.
3th order HD is shown here, and two peaks can clearly be seen at 1.5kHz and 2kHz, reflecting breakups at 4.5kHz and 6kHz.
black curve is with a voltage source (close to 0 ohm output impedance)
green curve is with a 13 ohm series resistor
red curve is with a 26 ohm series resistor (as close as I could get to current source)
Level were adjusted to get the same SPL at 1kHz.
Attachments
Thank you guys for your detailed replies! Now let me switch to an actual case...
Actually I'm going to make a 2 way + sub with an Accuton midwoofer and a tweeter RAAL 70-10d.
The RAALs don't like low x-over points, whereas the Accutons shouldn't be crossed too high. Fortunately I will run them actively using steep FIR filters, basically the low and high pass for mid and tweeter will result in 100 db below the cross region which will be just one octave wide! It means that 1/2 octave after the x-over point there's no more signal, and 1/4 after the signal is already 25 db below!
In any case such a system can't be crossed below the midwoofer 2nd and 3rd distortion spikes.
Which midwoofer would you use among these 2?
accuton® Carefully selected loudspeaker drivers.
accuton® Carefully selected loudspeaker drivers.
The 6.25 inches is newer and has a better motor with pretty linear BL and inductance.
The 7 inches has a bit bigger cone area and the very effective cut-outs for dampening the resonance frequency.
For my tastes 85 db SPL is already a pretty high sound level...
There are even better Accuton drivers, but too expensive...
Actually I'm going to make a 2 way + sub with an Accuton midwoofer and a tweeter RAAL 70-10d.
The RAALs don't like low x-over points, whereas the Accutons shouldn't be crossed too high. Fortunately I will run them actively using steep FIR filters, basically the low and high pass for mid and tweeter will result in 100 db below the cross region which will be just one octave wide! It means that 1/2 octave after the x-over point there's no more signal, and 1/4 after the signal is already 25 db below!
In any case such a system can't be crossed below the midwoofer 2nd and 3rd distortion spikes.
Which midwoofer would you use among these 2?
accuton® Carefully selected loudspeaker drivers.
accuton® Carefully selected loudspeaker drivers.
The 6.25 inches is newer and has a better motor with pretty linear BL and inductance.
The 7 inches has a bit bigger cone area and the very effective cut-outs for dampening the resonance frequency.
For my tastes 85 db SPL is already a pretty high sound level...
There are even better Accuton drivers, but too expensive...
Last edited:
Special thanks to you and TMM!!
I read through your comments and now I got it! So if we had an ideal motor with no distortion at the harmonics of its breakup frequency, no distortion spikes were exhibited at 1/2 1/3...of the breakup f, we would only see the cone breakup spike, is it right?
Maybe there would be another remedy... Please look at the HD of this driver:
accuton® Carefully selected loudspeaker drivers.
It has a gorgeous motor with extremely linear inductance and BL. Unfortunately it's terribly expensive...
Last edited:
If you are talking about the 6.25 accuton:
accuton® Carefully selected loudspeaker drivers.
It has a 2nd harmonic spike at just less than 2kHz and just over 4KHz
When you play ~2kHz through the speaker it generates a [small] 2nd harmonic at ~4kHz. When you play ~4kHz it generates a 2nd harmonic at ~8kHz. I would say these harmonics will still be made even if you filter the electrical input to say 5kHz because they are mechanically made after the input voltage.
Hi bigbottom,
Actually I'd filter this midwoofer at 3khz and very steeply. So the driver would receive no electrical signal at 4khz, which seems to be the main breakup frequency. Of course, and as I learnt in this thread, the cone will still receive 4khz signals generated inside the motor by its distortion, but this is inevitable! This is a midwoofer and that is its distortion performance...
Actually I'd filter this midwoofer at 3khz and very steeply. So the driver would receive no electrical signal at 4khz, which seems to be the main breakup frequency. Of course, and as I learnt in this thread, the cone will still receive 4khz signals generated inside the motor by its distortion, but this is inevitable! This is a midwoofer and that is its distortion performance...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.