drum core polarity...

J

John Larkin

Guest
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

Most are 2-pin parts and without a polarity dot, we\'d have
to test and mark them.

Wurth has a 4-pin part with a dot.

https://www.mouser.com/ProductDetail/Wurth-Elektronik/7447211101?qs=HXx4m3XcTe0AfZ94%2FxVkYA%3D%3D
 
On Friday, August 4, 2023 at 6:46:43 PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.
Click to expand...

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.
 
whit3rd <whit3rd@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43 PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.
Click to expand...

Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.

I don’t understand the polarity issue though—with the same layout, flipping
the inductor doesn’t do anything to the field. It’s just the current flow
and the helicity of the winding.

For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.

The magnetic moment goes like the inductance, so if the two inductors
differ by 10%, the cancellation is uncertain at the -20dB level.

I expect that one cup core would work better than an array of drums in most
cases. (Joerg could tweak stuff to null out the coupling to a single
vulnerable area, but it’s hard to do globally. )

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics,
Electro-optics, Photonics, Analog Electronics
 
On Sat, 5 Aug 2023 09:06:45 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

whit3rd <whit3rd@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43?PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.


Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.
Click to expand...

I\'d want minimal spacing, for field cancellation and because
I don\'t have space. If they are phased for far-field
cancellation, wouldn\'t I get a little more inductance? More
L is good!

I don’t understand the polarity issue though—with the same layout, flipping
the inductor doesn’t do anything to the field. It’s just the current flow
and the helicity of the winding.

For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.

The magnetic moment goes like the inductance, so if the two inductors
differ by 10%, the cancellation is uncertain at the -20dB level.

I expect that one cup core would work better than an array of drums in most
cases. (Joerg could tweak stuff to null out the coupling to a single
vulnerable area, but it’s hard to do globally. )

Cheers

Phil Hobbs
Click to expand...

I\'ll have a lot of DC current. The unshielded drums win on
energy storage per pcb area, partly because they have good
cooling and partly because they steal a lot of energy
storage volume from the surrounding space.

Tall and skinny is good too, to save surface area. Buying
from Mouser stock for 80 cents is good too.

I need two inductors per channel, 100 uH each, 2 amps,
basically in series. They carry the same current so maybe
the mid-distance fields can sorta cancel. My boards are on
1.6\" centers and will have air flow.

This is for the 8-channel dummy load board. That\'s 16
inductors and a couple more for switchers. The customer
ordered a dozen or so boards and keeps changing the specs.
We specified 10 watts per channel and now they want 40 watts
per channel, total 160 per board maybe. I can\'t dump 40
watts with resistors.

We can use mosfets and I found a CPU cooler that would work.

https://www.dropbox.com/scl/fi/vy6f54r6dmajbxqk6khod/Cooler_9_fets.jpg?rlkey=upuhj2r5bva4j4mf2lh63domj&raw=1

We could use the 9th part as the temperature sensor. Might
even use the same fet, substrate diode. Nobody makes the
TO-220 LM35 any more.

Actually, it just dawned on me, the fields will be mostly DC
with a smaller AC component if their drive is PWM.
Neighboring boards wouldn\'t mind a DC field! Even 5:1
cancellation might be worth doing. Yikes!
 
John Larkin <jjlarkinsnipsnip2highlandtechnology.com@> wrote:
On Sat, 5 Aug 2023 09:06:45 -0000 (UTC), Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

whit3rd <whit3rd@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43?PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.


Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.

I\'d want minimal spacing, for field cancellation and because
I don\'t have space. If they are phased for far-field
cancellation, wouldn\'t I get a little more inductance? More
L is good!


I don’t understand the polarity issue though—with the same layout, flipping
the inductor doesn’t do anything to the field. It’s just the current flow
and the helicity of the winding.

For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.

The magnetic moment goes like the inductance, so if the two inductors
differ by 10%, the cancellation is uncertain at the -20dB level.

I expect that one cup core would work better than an array of drums in most
cases. (Joerg could tweak stuff to null out the coupling to a single
vulnerable area, but it’s hard to do globally. )

Cheers

Phil Hobbs

I\'ll have a lot of DC current. The unshielded drums win on
energy storage per pcb area, partly because they have good
cooling and partly because they steal a lot of energy
storage volume from the surrounding space.

Tall and skinny is good too, to save surface area. Buying
from Mouser stock for 80 cents is good too.

I need two inductors per channel, 100 uH each, 2 amps,
basically in series. They carry the same current so maybe
the mid-distance fields can sorta cancel. My boards are on
1.6\" centers and will have air flow.

This is for the 8-channel dummy load board. That\'s 16
inductors and a couple more for switchers. The customer
ordered a dozen or so boards and keeps changing the specs.
We specified 10 watts per channel and now they want 40 watts
per channel, total 160 per board maybe. I can\'t dump 40
watts with resistors.

We can use mosfets and I found a CPU cooler that would work.

https://www.dropbox.com/scl/fi/vy6f54r6dmajbxqk6khod/Cooler_9_fets.jpg?rlkey=upuhj2r5bva4j4mf2lh63domj&raw=1

We could use the 9th part as the temperature sensor. Might
even use the same fet, substrate diode. Nobody makes the
TO-220 LM35 any more.

Actually, it just dawned on me, the fields will be mostly DC
with a smaller AC component if their drive is PWM.
Neighboring boards wouldn\'t mind a DC field! Even 5:1
cancellation might be worth doing. Yikes!
Click to expand...

The coupling will reduce the inductance, I think, but not by much.

--
Dr Philip C D Hobbs Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics,
Electro-optics, Photonics, Analog Electronics
 
On Saturday, August 5, 2023 at 2:06:54 AM UTC-7, Phil Hobbs wrote:
whit3rd <whi...@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43 PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.

Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.

I don’t understand the polarity issue though—with the same layout, flipping
the inductor doesn’t do anything to the field. It’s just the current flow
and the helicity of the winding.
Click to expand...

Au contraire! The field outside the ferrite (in the very large gap)
is where important amounts of energy is stored, and that stored energy is the cause of
inductance.
For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.
Click to expand...

True, but that isn\'t the spacing that has the \'fields mostly cancel\' effect mentioned.

To keep exterior fields under control, the shielded drum core parts work well.

<https://www.we-online.com/components/products/datasheet/744029004.pdf>
 
On Sat, 5 Aug 2023 18:46:19 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

John Larkin <jjlarkinsnipsnip2highlandtechnology.com@> wrote:
On Sat, 5 Aug 2023 09:06:45 -0000 (UTC), Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

whit3rd <whit3rd@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43?PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.


Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.

I\'d want minimal spacing, for field cancellation and because
I don\'t have space. If they are phased for far-field
cancellation, wouldn\'t I get a little more inductance? More
L is good!


I don?t understand the polarity issue though?with the same layout, flipping
the inductor doesn?t do anything to the field. It?s just the current flow
and the helicity of the winding.

For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.

The magnetic moment goes like the inductance, so if the two inductors
differ by 10%, the cancellation is uncertain at the -20dB level.

I expect that one cup core would work better than an array of drums in most
cases. (Joerg could tweak stuff to null out the coupling to a single
vulnerable area, but it?s hard to do globally. )

Cheers

Phil Hobbs

I\'ll have a lot of DC current. The unshielded drums win on
energy storage per pcb area, partly because they have good
cooling and partly because they steal a lot of energy
storage volume from the surrounding space.

Tall and skinny is good too, to save surface area. Buying
from Mouser stock for 80 cents is good too.

I need two inductors per channel, 100 uH each, 2 amps,
basically in series. They carry the same current so maybe
the mid-distance fields can sorta cancel. My boards are on
1.6\" centers and will have air flow.

This is for the 8-channel dummy load board. That\'s 16
inductors and a couple more for switchers. The customer
ordered a dozen or so boards and keeps changing the specs.
We specified 10 watts per channel and now they want 40 watts
per channel, total 160 per board maybe. I can\'t dump 40
watts with resistors.

We can use mosfets and I found a CPU cooler that would work.

https://www.dropbox.com/scl/fi/vy6f54r6dmajbxqk6khod/Cooler_9_fets.jpg?rlkey=upuhj2r5bva4j4mf2lh63domj&raw=1

We could use the 9th part as the temperature sensor. Might
even use the same fet, substrate diode. Nobody makes the
TO-220 LM35 any more.

Actually, it just dawned on me, the fields will be mostly DC
with a smaller AC component if their drive is PWM.
Neighboring boards wouldn\'t mind a DC field! Even 5:1
cancellation might be worth doing. Yikes!



The coupling will reduce the inductance, I think, but not by much.
Click to expand...

Depends on the polarities. I\'d think that one north-up and
one north-down minimizes the far-field (simple
superposition) and increases total inductance.

Is that right? But yes, I\'d expect a small inductance change
from unshielded drum cores alongside one another. Most of
one\'s field misses the other.

I can test both assumptions.
 
On Sat, 5 Aug 2023 13:02:10 -0700 (PDT), whit3rd
<whit3rd@gmail.com> wrote:

On Saturday, August 5, 2023 at 2:06:54?AM UTC-7, Phil Hobbs wrote:
whit3rd <whi...@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43?PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.

Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.

I don’t understand the polarity issue though—with the same layout, flipping
the inductor doesn’t do anything to the field. It’s just the current flow
and the helicity of the winding.

Au contraire! The field outside the ferrite (in the very large gap)
is where important amounts of energy is stored, and that stored energy is the cause of
inductance.

For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.

True, but that isn\'t the spacing that has the \'fields mostly cancel\' effect mentioned.
Click to expand...

Closer is better for external field cancellation.,
specifically 1.6\" away on the next board in the crate. About
2x the drum core height.

To keep exterior fields under control, the shielded drum core parts work well.

https://www.we-online.com/components/products/datasheet/744029004.pdf
Click to expand...

That\'s 5 uH at 1 amp. I want 100 uH at 2 amps. About 80x the
LI^2 energy storage.

Unshielded drum cores seem to always win. They are tall, so
pack a lot of volume into available PCB footprint. They have
great cooling. Magnetically, they are a lot bigger than the
volume they occupy because their air gap is external, the
entire universe in fact.
 
whit3rd <whit3rd@gmail.com> wrote:
On Saturday, August 5, 2023 at 2:06:54 AM UTC-7, Phil Hobbs wrote:
whit3rd <whi...@gmail.com> wrote:
On Friday, August 4, 2023 at 6:46:43 PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

That\'s counterproductive, if you expect them to be uncoupled; that\'d put N pole
next to S pole, effectively making a loop with gaps. Same loop, without gaps,
and named \'transformer\' can be assembled with two coils and C cores.

To the extent that fields \'mostly cancel\', there\'s less inductance.

Nah, with ferrite cores the coupling is a nit unless the spacing is much
less than the diameters.

I don’t understand the polarity issue though—with the same layout, flipping
the inductor doesn’t do anything to the field. It’s just the current flow
and the helicity of the winding.

Au contraire! The field outside the ferrite (in the very large gap)
is where important amounts of energy is stored, and that stored energy is the cause of
inductance.
Click to expand...

You’re probably thinking of a flyback transformer. In that case, the core
is a ring, and the gap is narrow compared with the minor diameter. Energy
density goes as B dot H, and perpendicular B is continuous at a material
boundary, so the narrow gap has nearly the same B, and mu times higher H
than inside the material.

Thus the energy density in the air gap can be thousands of times higher
than in the same volume of iron, so that most of the energy is stored
there.

John’s case is quite different, because the cores are separated by much
larger gaps, so the field mostly reconnects in the annular region outside
the cores. That makes the unperturbed inductances much lower, and also
reduces the energy density in the air.


In a quasistatic system, the force on any object is minus the gradient of
the total energy with respect to its position.

Arranging two electromagnets to cancel the external fields basically means
putting them adjacent and antiparallel.

In that configuration, the cancellation improves as they get closer, as I’m
sure you’ll agree. They also attract each other magnetically, which means
that the total magnetic energy goes down as they move together. Clear so
far?

Inductance is the twice the ratio of the total magnetic energy to the
square of the current, so the inductance goes down as the magnetic
cancellation improves.

There are other configurations that can reduce the fringing fields, such as
arranging the magnets back-to-back. In that case they will repel each
other, at least at large distances, so that the inductance will increase a
little.

What happens at small separations is more complicated than in the
antiparallel case because of the effect of core saturation.

Cheers

Phil Hobbs


For spacing dx between inductors ant frequencies such that lambda >> dx,
the field falls off as 1/r**3 rather than 1/r**2. More symmetrical layouts
can do better, until they get too big or the variation between parts
becomes dominant.

True, but that isn\'t the spacing that has the \'fields mostly cancel\' effect mentioned.
Click to expand...
Do tell. If you have one inductor whose magnetic moment is 10% larger than
the other’s, at what separation does the mismatch become dominant?



--
Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC /
Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics
 
On a sunny day (Sat, 05 Aug 2023 13:16:31 -0700) it happened John Larkin
<jjlarkinsnipsnip2highlandtechnology.com@> wrote in
<v3btcipih8cua47lii51qbue4i8gcbpl54@4ax.com>:

Depends on the polarities. I\'d think that one north-up and
one north-down minimizes the far-field (simple
superposition) and increases total inductance.

Is that right? But yes, I\'d expect a small inductance change
from unshielded drum cores alongside one another. Most of
one\'s field misses the other.

I can test both assumptions.
Click to expand...

Fun to play with,
I have use those as transformer:
http://panteltje.nl/pub/1.2_to_5V_converter_coil_detail_img_2256.jpg
 
On Saturday, August 5, 2023 at 1:49:34 PM UTC-7, Phil Hobbs wrote:
whit3rd <whi...@gmail.com> wrote:

On Friday, August 4, 2023 at 6:46:43 PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

Arranging two electromagnets to cancel the external fields basically means
putting them adjacent and antiparallel.
Click to expand...

An easy way to accomplish that is to use axis-parallel-to-PCB
units, mounted end-to-end. The orientation wouldn\'t need any \'dot\'
markings at all, since the coils\' footprints have polarity according to
the connections\' positioning.
 
On Sun, 06 Aug 2023 05:55:43 GMT, Jan Panteltje
<alien@comet.invalid> wrote:

On a sunny day (Sat, 05 Aug 2023 13:16:31 -0700) it happened John Larkin
jjlarkinsnipsnip2highlandtechnology.com@> wrote in
v3btcipih8cua47lii51qbue4i8gcbpl54@4ax.com>:

Depends on the polarities. I\'d think that one north-up and
one north-down minimizes the far-field (simple
superposition) and increases total inductance.

Is that right? But yes, I\'d expect a small inductance change
from unshielded drum cores alongside one another. Most of
one\'s field misses the other.

I can test both assumptions.

Fun to play with,
I have use those as transformer:
http://panteltje.nl/pub/1.2_to_5V_converter_coil_detail_img_2256.jpg
Click to expand...

A cell phone charger pad would make a nice test field
generator, if it would turn itself on without handshaking
with a phone first.

But any AC mag field would work for comparing a drum core
phase with a reference part.

Drum cores are great, except for the field thing.
 
On Sat, 5 Aug 2023 23:57:27 -0700 (PDT), whit3rd
<whit3rd@gmail.com> wrote:

On Saturday, August 5, 2023 at 1:49:34?PM UTC-7, Phil Hobbs wrote:
whit3rd <whi...@gmail.com> wrote:

On Friday, August 4, 2023 at 6:46:43?PM UTC-7, John Larkin wrote:
Unshielded drum cores are cheap and have lots of L*I^2 for
their size. But they leak mag fields.

Maybe I can put two side-by-side and have their far fields
mostly cancel. I need two anyhow.

Arranging two electromagnets to cancel the external fields basically means
putting them adjacent and antiparallel.

An easy way to accomplish that is to use axis-parallel-to-PCB
units, mounted end-to-end. The orientation wouldn\'t need any \'dot\'
markings at all, since the coils\' footprints have polarity according to
the connections\' positioning.
Click to expand...

Standard surface-mount inductors, 2412s or something like
that, might be left-hand or right-hand threads, so their
fields are uncontrolled. I guess an identical pair, off the
same reel, would have far fields cancel if the currents run
in opposite directions.

But a vertical inductor uses board space better and has
better exposure to cooling air.

I don\'t know of any horizontal drum cores, but there may be
some. Still, they are area inefficient, and my board will be
very dense.
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top