Hello!
I'm designing a crossover for a 200W 3-way loudspeaker (crossover frequencies 300Hz / 4kHz).
For the woofer inductor, I need 3mH. Considering price, size and "challenge", I chose to go with an E-I laminated iron core got from a 12V/1A transformer (360mm2 core section area).
I know 4 main drawbacks of using iron core.
Here's the way I worked around them for my case:
1. Saturation - The coil and core combination got saturated above 15A - I just need 10A peak (200W/4Ohm).
2. Non linear permeability - I used a gap in such a way that total permeability got limited to 15 - I built a 200uH air coil that when inserted in the iron core went to 3mH. Current x time curve looked quite linear.
3. Hysteresis - I measured the THD with the inductor in the circuit - got maximum 0.3% which is ok to me.
4. Core losses - Using it for a 300Hz low pass crossover frequency, no big losses I think.
Question: when I initially evaluated the THD, I found a strong 3rd harmonic at any power level, around -35dB below the fundamental which yields ~1.7% distortion).
After pulling my hairs for some hours, I decided to remove the core bracket leaving only the laminated core pieces. After that, the misterious 3rd harmonic fade out giving me what I wanted (THD <0.3%).
Anyone knows why the bracket creates the 3rd harmonic component?
I observed that when the bracket is mounted, the inductance goes up a bit - when I removed the bracket I readjusted the inductance by using a smaller gap.
I also observed the same with another ferrite rod core inductor that I had, when I get this braket closed to it the 3rd harmonic pops up too!
Here are the results and pictures.
Saturation Test
Frequency spectrum of core with the bracket
Frequency spectrum of core without the bracket (all harmonics below ~50dB)
The inductor and the bracket
Thank you!!
I'm designing a crossover for a 200W 3-way loudspeaker (crossover frequencies 300Hz / 4kHz).
For the woofer inductor, I need 3mH. Considering price, size and "challenge", I chose to go with an E-I laminated iron core got from a 12V/1A transformer (360mm2 core section area).
I know 4 main drawbacks of using iron core.
Here's the way I worked around them for my case:
1. Saturation - The coil and core combination got saturated above 15A - I just need 10A peak (200W/4Ohm).
2. Non linear permeability - I used a gap in such a way that total permeability got limited to 15 - I built a 200uH air coil that when inserted in the iron core went to 3mH. Current x time curve looked quite linear.
3. Hysteresis - I measured the THD with the inductor in the circuit - got maximum 0.3% which is ok to me.
4. Core losses - Using it for a 300Hz low pass crossover frequency, no big losses I think.
Question: when I initially evaluated the THD, I found a strong 3rd harmonic at any power level, around -35dB below the fundamental which yields ~1.7% distortion).
After pulling my hairs for some hours, I decided to remove the core bracket leaving only the laminated core pieces. After that, the misterious 3rd harmonic fade out giving me what I wanted (THD <0.3%).
Anyone knows why the bracket creates the 3rd harmonic component?
I observed that when the bracket is mounted, the inductance goes up a bit - when I removed the bracket I readjusted the inductance by using a smaller gap.
I also observed the same with another ferrite rod core inductor that I had, when I get this braket closed to it the 3rd harmonic pops up too!
Here are the results and pictures.
Saturation Test
Frequency spectrum of core with the bracket
Frequency spectrum of core without the bracket (all harmonics below ~50dB)
The inductor and the bracket
Thank you!!
We always used nonmagnetic brackets that we made from fiberglass or plastic for that reason.
https://www.foundryservice.com/product/glastic-utr-composite-angles-channels/
https://www.foundryservice.com/product/glastic-utr-composite-angles-channels/
Last edited:
The bracket shorts the airgap, and because of its tiny cross-section, it concentrates all the flux in a very small area, leading to high induction levels. Since the bracket is just ordinary mechanical-grade steel, it has horrendous magnetic properties.
Combine both and you create huge levels of distortion
12VA is very much less than the 300W it may be asked to choke-off. Is it big enough?
300Hz is 5X 60Hz. Which suggests 25 times 12VA, or 300W. Right on the nose? BUT power transformers are allowed to approach saturation. It stresses the power company, but who cares? (For small iron.) I'd want at least 3X multiplier. A 36VA core.
The strap distortion is another clue that audio chokes are more complicated than power iron. Who cares if a power transformer distorts? I wonder if a 36VA strap-mount would reduce the distortion? Obviously some, surely not enough.
Aside from iron distortion, in heavy saturation the choke stops choking. Because this is series to the woofer, the amplifier may not be harmed. But if on loud peaks the crossover jumps up an octave or three, the burst of midrange may be distressing.
Hi PRR,
Let me explain my calculations - please point out if I'm missing something.
I've based my choice on the 12V/1A from the B (mag flux density) value calculated.
I've used 2 formulas for checking the saturation (they are in fact just different views - but just double checking)
1. Checking the B
The coil was made of 89 turns on a square form (20mm x 20mm x 20mm).
This yields to a 0.2mH inductance if out of the iron core.
I adjusted, by trial and error, the gap to achieve 3mH, so the resulting total permeability (iron + air Gap) is 15 (3/0.2).
The field strentgh H is calculated as H=(I*N)/L (A.Turn/m)
I need to pass 10A.
H=(10*89)/0.02 = 44500 A.Turn/m
The mag flux density B is calculated as B=u*H, where u=ur*u0
u0=1.256x10^-6 (vaccum permeability constant)
ur=15 (resultant from iron core + air gap)
B=15*(1.256x10^-6)*44500
B=0.83T
Iron core saturates at around 1.5T, so I would be at half way of saturation.
2. Calculating B as function of L (inductance), I (current), N (turns) and A (area)
B=(L*I)/(N*A)
L=0.003mH
I=10A
N=89
Area=0.02*0.02=0.0004m2 (this is the core section area of the 12V/1A transformer)
B=(0.003*10)/(89*0.0004)
B=0.84T
The measurements I did with a pulse matches approximatelly what these calculations indicated.
The current versus time starts to change the slope at around 17A, where the B would be 1.4T indicating the start of saturation. See the picture I've taken from the osciloscope.
So, I chose the 12V/1A based on the core section area of 400mm2 (0.0004m2).
In a real crossover low pass filter application, I could test it only with 6.5A at this moment since I only had a 80W amp available. No distortion observed from 20Hz up to 500Hz (took some samples).See picture from the oscilloscope @55Hz).
Regards!
Let me explain my calculations - please point out if I'm missing something.
I've based my choice on the 12V/1A from the B (mag flux density) value calculated.
I've used 2 formulas for checking the saturation (they are in fact just different views - but just double checking)
1. Checking the B
The coil was made of 89 turns on a square form (20mm x 20mm x 20mm).
This yields to a 0.2mH inductance if out of the iron core.
I adjusted, by trial and error, the gap to achieve 3mH, so the resulting total permeability (iron + air Gap) is 15 (3/0.2).
The field strentgh H is calculated as H=(I*N)/L (A.Turn/m)
I need to pass 10A.
H=(10*89)/0.02 = 44500 A.Turn/m
The mag flux density B is calculated as B=u*H, where u=ur*u0
u0=1.256x10^-6 (vaccum permeability constant)
ur=15 (resultant from iron core + air gap)
B=15*(1.256x10^-6)*44500
B=0.83T
Iron core saturates at around 1.5T, so I would be at half way of saturation.
2. Calculating B as function of L (inductance), I (current), N (turns) and A (area)
B=(L*I)/(N*A)
L=0.003mH
I=10A
N=89
Area=0.02*0.02=0.0004m2 (this is the core section area of the 12V/1A transformer)
B=(0.003*10)/(89*0.0004)
B=0.84T
The measurements I did with a pulse matches approximatelly what these calculations indicated.
The current versus time starts to change the slope at around 17A, where the B would be 1.4T indicating the start of saturation. See the picture I've taken from the osciloscope.
So, I chose the 12V/1A based on the core section area of 400mm2 (0.0004m2).
In a real crossover low pass filter application, I could test it only with 6.5A at this moment since I only had a 80W amp available. No distortion observed from 20Hz up to 500Hz (took some samples).See picture from the oscilloscope @55Hz).
Regards!
Very interesting. My cheap Edcor output transformers have brackets like that. I wonder if I can measure a difference with and without them.
The bracket could make a difference if the amplifier is SE: it then has a gap.
For a PP type, the magnetic circuit is completely closed, and since the bracket is made of low-µ mechanical steel and has a small cross-section, its reluctance will be much higher than the transformer, and its influence will be negligible
For a PP type, the magnetic circuit is completely closed, and since the bracket is made of low-µ mechanical steel and has a small cross-section, its reluctance will be much higher than the transformer, and its influence will be negligible
Ferrite E-cores with gaps have the gap in the centre pole which keeps it away from any nearby metal. And since they use two E cores together you can have single or double gap by using one or two gapped pieces in the stack.