Hi fellow nerds!
The following is the results of some simulations I did on saturation in speaker motors.
I did some testing to see what the effect of saturation in speaker motors might be. Many years ago a driver developer told me that he would avoid saturation, not mentioning why, but it certainly made me feel that it could be a bad thing. It got me thinking, and I started to wonder if it could be wrong, and that saturation could be a good thing.
Some manufacturers have put bottle necks in their motors to achieve saturation points, so the idea is not unknown. Others seem to just develop drivers somehow by trial and error, and by listening in between, but not having an idea what they are actually doing. Anyway, I tried to put some numbers on this.
I made a traditional asymmetric motor structure. 10mm gap, 4mm coil (under hung), with 25mm pole piece diameter. I then made 3 versions:
1: A version with small magnet, working at around 1,4 Tesla at most.
2: A version with very big magnet, working at around 2,4 Tesla at most, having a much higher force factor and heavy saturation both inside and outside the gap.
3: A version with a medium magnet, and some fine tuned saturation points to achieve the same force factor as number 1.
I then tried to simulate the force factor at different currents, and different force directions. I also simulated the nonlinearity of the force factor as it increases from 1A to 20A. Here are my findings:
1. Not saturated:
At 1A the asymmetry is at 4,1%. At 20A the asymmetry is at an incredible 135%. In positive direction at 20A the force factor is 39% less than espected, while in negative direction it is 37% more than expected.
2. Saturated, high force factor:
At 1A the asymmetri is at 0,24%. At 20A the asymmetry is at only 4,3%. In positive direction the force factor at 20A is 3,4% less than expected, while in negative direction it is 7,3% less than expected.
3. Saturated and matched force factor:
At 1A the asymmetry is at 0,0042%. At 20A the asymmetry is at 8,2%. In positive direction the force factor at 20A is 0,75% less than expected, while in negative direction it is 8,9% less than expected.
With the magnets replaced with steel parts, the force factor is just above 80N at 20A, regardless of polarity, just to give an idea of the magnitude of the DC-components. This field was strong enough to bring the pole piece close to saturation.
I think there are several interesting conclusions one can draw from this:
- A saturated motor has a significantly more ideal field.
- If the motor is not saturated the DC components does play a bigger role.
- If the motor is not saturated the PM contributes to the DC components when at heavy load.
- The huge modulation of flux density in the steel parts and PMs is likely to contribute to huge non harmonic time domain distortion.
All simulations are made in FEMM.
I also would like to point out that 20A is a very high current, but it is used to magnify the result.
The following is the results of some simulations I did on saturation in speaker motors.
I did some testing to see what the effect of saturation in speaker motors might be. Many years ago a driver developer told me that he would avoid saturation, not mentioning why, but it certainly made me feel that it could be a bad thing. It got me thinking, and I started to wonder if it could be wrong, and that saturation could be a good thing.
Some manufacturers have put bottle necks in their motors to achieve saturation points, so the idea is not unknown. Others seem to just develop drivers somehow by trial and error, and by listening in between, but not having an idea what they are actually doing. Anyway, I tried to put some numbers on this.
I made a traditional asymmetric motor structure. 10mm gap, 4mm coil (under hung), with 25mm pole piece diameter. I then made 3 versions:
1: A version with small magnet, working at around 1,4 Tesla at most.
2: A version with very big magnet, working at around 2,4 Tesla at most, having a much higher force factor and heavy saturation both inside and outside the gap.
3: A version with a medium magnet, and some fine tuned saturation points to achieve the same force factor as number 1.
I then tried to simulate the force factor at different currents, and different force directions. I also simulated the nonlinearity of the force factor as it increases from 1A to 20A. Here are my findings:
1. Not saturated:
At 1A the asymmetry is at 4,1%. At 20A the asymmetry is at an incredible 135%. In positive direction at 20A the force factor is 39% less than espected, while in negative direction it is 37% more than expected.
2. Saturated, high force factor:
At 1A the asymmetri is at 0,24%. At 20A the asymmetry is at only 4,3%. In positive direction the force factor at 20A is 3,4% less than expected, while in negative direction it is 7,3% less than expected.
3. Saturated and matched force factor:
At 1A the asymmetry is at 0,0042%. At 20A the asymmetry is at 8,2%. In positive direction the force factor at 20A is 0,75% less than expected, while in negative direction it is 8,9% less than expected.
With the magnets replaced with steel parts, the force factor is just above 80N at 20A, regardless of polarity, just to give an idea of the magnitude of the DC-components. This field was strong enough to bring the pole piece close to saturation.
I think there are several interesting conclusions one can draw from this:
- A saturated motor has a significantly more ideal field.
- If the motor is not saturated the DC components does play a bigger role.
- If the motor is not saturated the PM contributes to the DC components when at heavy load.
- The huge modulation of flux density in the steel parts and PMs is likely to contribute to huge non harmonic time domain distortion.
All simulations are made in FEMM.
I also would like to point out that 20A is a very high current, but it is used to magnify the result.
I am going to try to investigate the effect on Le, and see if vanadium permendur makes a significant difference.
Well done Snickers-is. I wish more could present deeper research in speaker-driver design. Keep it comming 🙂
Thanks Hylle, I am currently investigating the effects of vanadium permendur and aluminium nickel cobalt-magnets. The most interesting part, I think, is to see if one can design the circuit so well it is completely oblivious to the type of PM and paramagnetic material.
I am not sure how much time I will spend posting here if the respons is close to zero.
I am not sure how much time I will spend posting here if the respons is close to zero.
While saturation is very effective in preventing flux modulation from VC current it really is a waste of material and cost. Putting a copper or aluminum shorting ring (aka demodulation ring) below the gap is almost as good as saturated pole pieces.
I think your view is interesting, but my mind is not as much at material cost right now.
Shorting rings are very effective, but I think we see a clear transition from low to high frequencies. Shorting rings needs to be extremely large and well connected to the magnetic circuit in order to have any effect at low frequencies. I believe a combination of the two is needed to make the ultimate driver.
Shorting rings are very effective, but I think we see a clear transition from low to high frequencies. Shorting rings needs to be extremely large and well connected to the magnetic circuit in order to have any effect at low frequencies. I believe a combination of the two is needed to make the ultimate driver.
Ok, so then I have made some additional simulations using some more exotic materials:
1: The tuned circuit has been given Vanadium Permendur as its paramagnetic material.
Symmetric error 1A: 0,15%
Symmetric error 20A: 8,9%
Loss at 20A pos: 1,5%
Loss at 20A neg: 11,6%
2: The circuit above has received Alnico 5 magnets. Their dimensions have been tuned to match the same 1A force factor.
Symmetric error 1A: 0,14%
Symmetric error 20A: 14,8%
Loss at 20A pos: 1,5%
Loss at 20A neg: 16,2%
3: The circuit above is then equipped with larger Alnico 5-magnets.
Symmetric error 1A: 0,08%
Symmetric error 20A: 8,7%
Loss at 20A pos: 0,49%
Loss at 20A neg: 9,2%
4: The circuit above has ben changed to over dimensioned neo magnets.
Symmetric error 1A: 0,092%
Symmetric error 20A: 3,9%
Loss at 20A pos: 3,0%
Loss at 20A neg: 0,88%
5: A circuit like the one above but with the paramagnetic material swapped to steel.
Symmetric error 1A: 0,18%
Symmetric error 20A: 4,75%
Loss at 20A pos: 4,4%
Loss at 20A neg: 0,40%
One should be really careful making universal conclusions, especially because so many dimensional characteristics play an important role in the way a PM material behaves in real life. I do however think we can draw a few conclusions that are at least valid for this experimental setup as follows:
- Permendur, when saturated, gives a very subtile advantage over steel.
- Over saturated permendur gives about 7,5% more force factor than steel in this circuit.
- Over saturated circuits offer a huge advantage in performance.
- The shape of the paramagnetic parts has a huge impact on the symmetry (as we all know). But different saturation levels can give unexpected DC behavior that is signal dependent.
- When the circuit is perfected, there is no systematic advantage of using Alnico over Neo, while the advantage of using Permendur is subtile, and in some situations, questionable.
1: The tuned circuit has been given Vanadium Permendur as its paramagnetic material.
Symmetric error 1A: 0,15%
Symmetric error 20A: 8,9%
Loss at 20A pos: 1,5%
Loss at 20A neg: 11,6%
2: The circuit above has received Alnico 5 magnets. Their dimensions have been tuned to match the same 1A force factor.
Symmetric error 1A: 0,14%
Symmetric error 20A: 14,8%
Loss at 20A pos: 1,5%
Loss at 20A neg: 16,2%
3: The circuit above is then equipped with larger Alnico 5-magnets.
Symmetric error 1A: 0,08%
Symmetric error 20A: 8,7%
Loss at 20A pos: 0,49%
Loss at 20A neg: 9,2%
4: The circuit above has ben changed to over dimensioned neo magnets.
Symmetric error 1A: 0,092%
Symmetric error 20A: 3,9%
Loss at 20A pos: 3,0%
Loss at 20A neg: 0,88%
5: A circuit like the one above but with the paramagnetic material swapped to steel.
Symmetric error 1A: 0,18%
Symmetric error 20A: 4,75%
Loss at 20A pos: 4,4%
Loss at 20A neg: 0,40%
One should be really careful making universal conclusions, especially because so many dimensional characteristics play an important role in the way a PM material behaves in real life. I do however think we can draw a few conclusions that are at least valid for this experimental setup as follows:
- Permendur, when saturated, gives a very subtile advantage over steel.
- Over saturated permendur gives about 7,5% more force factor than steel in this circuit.
- Over saturated circuits offer a huge advantage in performance.
- The shape of the paramagnetic parts has a huge impact on the symmetry (as we all know). But different saturation levels can give unexpected DC behavior that is signal dependent.
- When the circuit is perfected, there is no systematic advantage of using Alnico over Neo, while the advantage of using Permendur is subtile, and in some situations, questionable.
Hi,
20 Amps rms into a typical 6R driver is 2.4KW,
hardly a realistic scenario for comparison.
rgds, sreten.
20 Amps rms into a typical 6R driver is 2.4KW,
hardly a realistic scenario for comparison.
rgds, sreten.
Hi, A result that is meaningless is no use compared to meaniful, rgds, sreten.
So what you are saying is that if this driver has a 3 ohm RDC coil, the peak power would be 1200W, and the RMS-power around 800, meaning a voltage swing of 60V P-P, and noone has such an amplifier?
I suspect the point is more that at these power levels, even serious PA drivers are right up against limits of excursion, power handling, cone breakup and suchlike.
I don't dismiss your thought train, but at extreme drive conditions drivers are struggling for survival. If you chose say 20 or 30 volts you would be more reasonable. Remember 30 volts is about 100 watts, which ishould for most unsociably loud.
I don't dismiss your thought train, but at extreme drive conditions drivers are struggling for survival. If you chose say 20 or 30 volts you would be more reasonable. Remember 30 volts is about 100 watts, which ishould for most unsociably loud.
But then you have to look past the fact that the experiment clearly state that these effects start somewhere below 1A, which then would be around 2WRMS... It only becomes so much bigger when the transients strike.
A midrange driver would easily take 1kW at a wide frequency band without having problems with excursion, and without having thermal problems given that the duration of the transient is short enough.
BTW, cone break up has nothing to do with power level.
A midrange driver would easily take 1kW at a wide frequency band without having problems with excursion, and without having thermal problems given that the duration of the transient is short enough.
BTW, cone break up has nothing to do with power level.
This document only shows a list of properties of materials commonly used in magnetic circuits. I do not see the relevance?
Simply put ...
... there is no discovery being made here. The properties of magnetic materials and how they react when their saturation levels are exceeded are well known. The loudspeaker market is highly competitive, and delivering maximum performance and reliability at minimum cost is were the engineering action is. WHG
... there is no discovery being made here. The properties of magnetic materials and how they react when their saturation levels are exceeded are well known. The loudspeaker market is highly competitive, and delivering maximum performance and reliability at minimum cost is were the engineering action is. WHG
You are welcome to discuss that in another thread. There is already a thread about it here:
http://www.diyaudio.com/forums/multi-way/132777-flux-modulation.html
This thread is about how to make better motors.
http://www.diyaudio.com/forums/multi-way/132777-flux-modulation.html
This thread is about how to make better motors.
Hi,
No it is not, it is discussing stupid power levels,
rather than realistic optimum motor performance.
rgds, sreten.
Last edited:
Isn't this what happens with too much "damping" in the cabinet? 😉
Sent from my pc1162 using Tapatalk
Sent from my pc1162 using Tapatalk
- Status
- Not open for further replies.