Who makes small swinging chokes?

[Joerg]

Folks,

The subject line almost says it all. What I mean with small is a nice
flat inductor like this:

http://media.digikey.com/Renders/Bourns Renders/SRR6038_sml.jpg

Toroidal would work, too. Just not the usual ones where you need a
40-ton truck and pilot vehicles. Custom is not problem but I am
wondering if there is anything available off-the-shelf.

For the young folks: A swinging inductor is one where the inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers

 

[Joerg]

I wasnt aware of any application for them nowadays.

Switch mode converters, for example. When you must cover a wide range of
outputs then ripple can become critical at very low output levels.
That's where the converter chip can usually only work in pulse skipping
or burst mode because it cannot otherwise get below its minimum duty
cycle. This is where you'd want a really large inductance but you can't
afford it for cost and size reasons. A swinging inductor would fit like
a glove because it's small. The inductance largely collapses to the
usual small value at high current but that's ok.

 

[legg]

Folks,

The subject line almost says it all. What I mean with small is a nice
flat inductor like this:

http://media.digikey.com/Renders/Bourns Renders/SRR6038_sml.jpg

Toroidal would work, too. Just not the usual ones where you need a
40-ton truck and pilot vehicles. Custom is not problem but I am
wondering if there is anything available off-the-shelf.

For the young folks: A swinging inductor is one where the inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers

Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.

What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)

RL

 

[Joerg]

That's cool, Joerg. I wish I knew more about inductors/ magnetics.
If it's a large volume can't you ask someone to make it for you? X
many turns of Y daimeter wire on Z piece of material. I guess you'd
have to run the numbers on the coil. API Delevan, is just up the road
from where I live.

Yes, if we can't get it anywhere it'll have to be custom. But the
challenge will be Z. If that requires two toroids side be side we'll
receive lots of "decline to quote" replies.

"Can I have that with fries instead of mashed potatoes?" ... "No,
sorry". ;-)

 

[Joerg]

Thanks for explaining. Funny how obsolete tech comes back into use.

It's just that some engineers and even professors think it's obsolete.
In power industries swinging inductors are still in wide use. You can
order one today from ABB. But delivery will be on a flatbed trailer and
you'll have to provide a crane

Reminds me of a design review where a filter was flagged as inadequate
and they didn't have room or budget for anything better. I was just an
invited reviewer and then raised my hand. "Maybe we could use a
Q-multiplier there" ... "A what?"

 

[Joerg]

Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.

What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)

So far just calculations. I'll also simulate it but only if I can be
sure that we can get those things at reasonable cost. Else it'll be a
nice circuit but would remain a pie in the sky.

 

[Joerg]

Folks,

The subject line almost says it all. What I mean with small is a nice
flat inductor like this:
[snip]

For the young folks: A swinging inductor is one where the
inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers

Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.

What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)


How about paralleling two stock SMD inductors with different
saturation points?

(Joerg might wince at two parts and double real-estate, though.)

Um, on 2nd thought, series-connection's more reasonable, values-wise,
but the series resistance might be a nuisance.

Especially since they'd both have to be swinging chokes for it to work.

No, in a series connection they don't have to be. But the transfer will
be rather abrupt, cost roughly doubles and real estate needs are
doubled. For the high-inductance one you'd have to use a CM choke so its
wire can carry the current but those are large and usually saturate
whambam style.

Joerg's idea of stacking high- and low-mu cores and running the winding
through both of them sounds like the ticket.

But it'll be custom and can't be done with the usual SMT-friendly cores :-(

My dream would be if someone cooked up the needed material so it's one
core with a loooong saturation slope (meaning not totally flattened out
but almost).

 

[Joerg]

[email protected] wrote:
On Oct 14, 5:48 am, [email protected] wrote:
Folks,
The subject line almost says it all. What I mean with small is a nice
flat inductor like this:
[snip]
For the young folks: A swinging inductor is one where the inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers
Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.
What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)
How about paralleling two stock SMD inductors with different
saturation points?
(Joerg might wince at two parts and double real-estate, though.)
Um, on 2nd thought, series-connection's more reasonable, values-wise,
but the series resistance might be a nuisance.
--
Cheers,
James Arthur
Especially since they'd both have to be swinging chokes for it to work.

Joerg's idea of stacking high- and low-mu cores and running the winding
through both of them sounds like the ticket.

Cheers

Phil Hobbs

No, they don't have to both be swinging choke.


Here--put two in series, L1, a high-valued low-esr choke that
saturates easily, and L2, low-valued and high-saturation current.

======= =====
.-.-.-. .-.-.
_________| | | |_____| | |____
L1 L2


At low currents, the inductance is L1+L2.

Once L1 saturates, it's a resistor (low-valued, by design), and L2
takes over.

Yep. I now have my Spice model refined for a smoothly converging
saturating inductor, if anyone is interested.

<raising_hand>

Only if useful in LTSpice though. I have used saturable cores in LTSpice
but haven't used too elaborate a function. And I'll have to dig in
really old stuff to find it

The main issue with two inductors in series is that the mfg for L1 won't
release saturation data, at least not in writing. Because you aren't
s'posed to do that. But I hafta. You can measure it on the impedance
analyzer and use the found data in the design and everything is peachy.
Then one sunny day Murphy strikes, the day the mfg changes ferrite
vendors ...

 

[Joerg]

The saturation doesn't have to be all that abrupt--IME saturation
comes pretty gradually.

It often does but that might change when the mfg switch ferrite vendors.

I've used some inductors as swinging chokes all on their own, driving
them to ~2x their saturation rating ("saturation" defined as 20%
inductance reduction). The result is greatly reduced inductance at
high current--a swinging choke.

Those were custom-wound, with fat wire to keep I^2*R losses
reasonable. About a 3 to 5:1 swing, IIRC.

As for cost, you'll have to weigh the cost of spec'ing a custom,
boutique part, against buying two mass-produced units easily available
at DigiKey.

Actually, I've had cases whera a custom run in Asia (they didn't have
the value I needed) was cheaper than the catalog part here. At
quantities well below five digits which really surprised me. But they
won't do a swinging inductor, it has to be on their standard cores and
machines.

You can put those in series too, if it satisfies your swing ratio. A
super-wide ratio might be hard.

Ideally I'd like a super-wide ratio. Something like 50:1. The more the
better.

[...]

 

[Joerg]

Phil Hobbs wrote:

[...]

But it'll be custom and can't be done with the usual SMT-friendly
cores :-(

My dream would be if someone cooked up the needed material so it's one
core with a loooong saturation slope (meaning not totally flattened out
but almost).

You could run a thin saw cut most of the way through the core on one side.

Partial sawing would immediately raise the hackles with the engineers at
the inductor mfg

 

[Joerg]

On Oct 14, 10:16 am, Phil Hobbs
[email protected] wrote:
On Oct 14, 5:48 am, [email protected] wrote:
Folks,
The subject line almost says it all. What I mean with small is a nice
flat inductor like this:
[snip]
For the young folks: A swinging inductor is one where the inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers
Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.
What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)
How about paralleling two stock SMD inductors with different
saturation points?
(Joerg might wince at two parts and double real-estate, though.)
Um, on 2nd thought, series-connection's more reasonable, values-wise,
but the series resistance might be a nuisance.

Especially since they'd both have to be swinging chokes for it to work.
Joerg's idea of stacking high- and low-mu cores and running the winding
through both of them sounds like the ticket.

No, they don't have to both be swinging choke.
Here--put two in series, L1, a high-valued low-esr choke that
saturates easily, and L2, low-valued and high-saturation current.
======= =====
.-.-.-. .-.-.
_________| | | |_____| | |____
L1 L2
At low currents, the inductance is L1+L2.
Once L1 saturates, it's a resistor (low-valued, by design), and L2
takes over.

They both have to be swinging chokes if they're in _parallel_, because
otherwise all you get is the lower-valued one until the higher-valued
one saturates, at which point you get an air-core coil.

Well, (blush), it depends which way you swing, doesn't it?

(I understand about the series connection--I was just pulling your chain.)

Successfully pulled. BTW, who ever knew Joerg was such a swinger?

Well, you need the matching car:

http://www.bcautos.com/musclecars/72swinger.html

Note: No center console, one can easily scoot across the whole front
bench

 

[[email protected]]

I had a 1950 Nash... don't remember the model name ("Ambassador"
maybe), but it was the full-sized bath-tub shaped body. Front seats
folded flat toward the back, making into a bed... perfect teenager's
car ;-)

In the '70s, a friend bought his daughter a Mustang-II. He said *that* was a
perfect teenager's car. ;-)

 

[Joerg]

I had a 1950 Nash... don't remember the model name ("Ambassador"
maybe), but it was the full-sized bath-tub shaped body. Front seats
folded flat toward the back, making into a bed... perfect teenager's
car ;-)

Must have been a Statesman, Airflyte or Rambler, AFAIR the other large
ones didn't look much like bath tubs. The Nash Metropolitan was the
small bath tub.

 

[Joerg]

Jim Thompson wrote:
On Oct 14, 10:16 am, Phil Hobbs
[email protected] wrote:
On Oct 14, 5:48 am, [email protected] wrote:
Folks,
The subject line almost says it all. What I mean with small is a nice
flat inductor like this:
[snip]
For the young folks: A swinging inductor is one where the inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers
Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.
What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)
How about paralleling two stock SMD inductors with different
saturation points?
(Joerg might wince at two parts and double real-estate, though.)
Um, on 2nd thought, series-connection's more reasonable, values-wise,
but the series resistance might be a nuisance.
--
Cheers,
James Arthur
Especially since they'd both have to be swinging chokes for it to work.
Joerg's idea of stacking high- and low-mu cores and running the winding
through both of them sounds like the ticket.
Cheers
Phil Hobbs
No, they don't have to both be swinging choke.
Here--put two in series, L1, a high-valued low-esr choke that
saturates easily, and L2, low-valued and high-saturation current.
======= =====
.-.-.-. .-.-.
_________| | | |_____| | |____
L1 L2
At low currents, the inductance is L1+L2.
Once L1 saturates, it's a resistor (low-valued, by design), and L2
takes over.
Yep. I now have my Spice model refined for a smoothly converging
saturating inductor, if anyone is interested.
<raising_hand>
Only if useful in LTSpice though. I have used saturable cores in LTSpice
but haven't used too elaborate a function. And I'll have to dig in
really old stuff to find it
The main issue with two inductors in series is that the mfg for L1 won't
release saturation data, at least not in writing. Because you aren't
s'posed to do that. But I hafta. You can measure it on the impedance
analyzer and use the found data in the design and everything is peachy.
Then one sunny day Murphy strikes, the day the mfg changes ferrite
vendors ...
When you specify the data for the custom part, just include the needed
parameters for non-saturated currents and for saturated currents, that
fits your bill. If the go outside this tolerance band, the part is
discarded. That forces the supplier to deliver the same parts that you
did the initial testing on
Regards
Klaus

Subcircuit for Saturating Inductor...

******************************************************************
**** Saturating Inductor v2, JET, 08/01/2011 (While in NY) *****
.SUBCKT LSATv2 TOP BOT PARAMS: LMAX=305uH LMIN=60uH I1TH=7.75mA
+ I2TH=8.25mA
L_L1 N_2 N_3 {LMAX-LMIN}
V_VM N_4 BOT 0V
L_L2 N_3 N_4 {LMIN}
E_E1 N_2 TOP VALUE { V(N_2,
+ N_3)*(1+TANH(0.693147/(I2TH-I1TH)*(abs(I(V_VM))-(I1TH+I2TH)/2)))/2}
.ENDS LSATv2
******************************************************************

Should work with _any_ "Spice".

I1TH is current at which you are 1/3 of the way from unsaturated to
saturated. I2TH, 2/3 of the way. (Defining both breakpoint _and_
slope.)

...Jim Thompson
- Skjul tekst i anførselstegn -

- Vis tekst i anførselstegn -

General info on the design of swinging chokes from Magnetics corp:

http://www.google.dk/url?sa=t&sourc...ofiJBA&usg=AFQjCNEdDXyg8gB0Q3YnEoHyRToP-lOvcg

Thanks, Klaus. In the past I have received declines when requesting
quotes for anything with a "weird" gap. Probably because they'd have to
introduce a production step on the milling machine that isn't in their
usual repertoire or not ECO-released.

But this is a good hint, maybe Spang Magnetics would be willing to do
it. I'll ask them.

A simple step wouldn't work too well though, has to be a little more
elaborate. The reason is that the regulator loop can go berserk at the
transition.

 

[Joerg]

Jim Thompson wrote:
On Oct 14, 10:16 am, Phil Hobbs
[email protected] wrote:
On Oct 14, 5:48 am, [email protected] wrote:
Folks,
The subject line almost says it all. What I mean with small is a nice
flat inductor like this:
[snip]
For the young folks: A swinging inductor is one where the inductance is
really high at very low currents and then drops off significantly
towards the full rated current, due to partial core saturation.
"Partial" is the trick. So it has nothing to do with swingers
Not a commodity, because there's no concensus on what the slope should
look like. You can wind toroid sets on an smd platform (40705TC +
00M0804T200 gives a .310 OD x .25H 'compound' core). Alternately, you
can order irregularly ground gaps.
What are you using, so far, to prove that this is a cure to all of
your light-load problems? (See what I mean?)
How about paralleling two stock SMD inductors with different
saturation points?
(Joerg might wince at two parts and double real-estate, though.)
Um, on 2nd thought, series-connection's more reasonable, values-wise,
but the series resistance might be a nuisance.
--
Cheers,
James Arthur
Especially since they'd both have to be swinging chokes for it to work.
Joerg's idea of stacking high- and low-mu cores and running the winding
through both of them sounds like the ticket.
Cheers
Phil Hobbs
No, they don't have to both be swinging choke.
Here--put two in series, L1, a high-valued low-esr choke that
saturates easily, and L2, low-valued and high-saturation current.
======= =====
.-.-.-. .-.-.
_________| | | |_____| | |____
L1 L2
At low currents, the inductance is L1+L2.
Once L1 saturates, it's a resistor (low-valued, by design), and L2
takes over.
Yep. I now have my Spice model refined for a smoothly converging
saturating inductor, if anyone is interested.
<raising_hand>

Only if useful in LTSpice though. I have used saturable cores in LTSpice
but haven't used too elaborate a function. And I'll have to dig in
really old stuff to find it

The main issue with two inductors in series is that the mfg for L1 won't
release saturation data, at least not in writing. Because you aren't
s'posed to do that. But I hafta. You can measure it on the impedance
analyzer and use the found data in the design and everything is peachy.
Then one sunny day Murphy strikes, the day the mfg changes ferrite
vendors ...

When you specify the data for the custom part, just include the needed
parameters for non-saturated currents and for saturated currents, that
fits your bill. If the go outside this tolerance band, the part is
discarded. That forces the supplier to deliver the same parts that you
did the initial testing on

Regards

Klaus

Subcircuit for Saturating Inductor...

******************************************************************
**** Saturating Inductor v2, JET, 08/01/2011 (While in NY) *****
.SUBCKT LSATv2 TOP BOT PARAMS: LMAX=305uH LMIN=60uH I1TH=7.75mA
+ I2TH=8.25mA
L_L1 N_2 N_3 {LMAX-LMIN}
V_VM N_4 BOT 0V
L_L2 N_3 N_4 {LMIN}
E_E1 N_2 TOP VALUE { V(N_2,
+ N_3)*(1+TANH(0.693147/(I2TH-I1TH)*(abs(I(V_VM))-(I1TH+I2TH)/2)))/2}
.ENDS LSATv2
******************************************************************

Should work with _any_ "Spice".

I1TH is current at which you are 1/3 of the way from unsaturated to
saturated. I2TH, 2/3 of the way. (Defining both breakpoint _and_
slope.)

...Jim Thompson

Thanks, Jim. I have copied this post into my sim directory among the
"secret sauce recipes"

So far I have used simple tanh but your method looks more elaborate and
versatile.

 

[Fred Abse]

Must have been a Statesman, Airflyte or Rambler, AFAIR the other large
ones didn't look much like bath tubs. The Nash Metropolitan was the
small bath tub.

ISTR a saying:

"If you can't afford a car, and you're too proud to get a Volkswagen, get
a Rambler"

The Metropolitan was made in the UK, by Austin. Should have stayed there.

 

[Joerg]

ISTR a saying:

"If you can't afford a car, and you're too proud to get a Volkswagen, get
a Rambler"

The Metropolitan was made in the UK, by Austin. Should have stayed there.

It looked like it was supposed to become one of those odd three-wheelers
they had over there and then they decided to put on four wheels.

 

[Fred Abse]

It looked like it was supposed to become one of those odd three-wheelers
they had over there and then they decided to put on four wheels.

AIUI, it was designed to fill Nash's "perceived need" for a small car,
drawing on Austin's experience in that field. Wrong on both counts.

Meanwhile, VW were selling millions, worldwide.

 

[Joerg]

AIUI, it was designed to fill Nash's "perceived need" for a small car,
drawing on Austin's experience in that field. Wrong on both counts.

Meanwhile, VW were selling millions, worldwide.

They were almost indestructable, with the engine only revving 70% of
what the Nash and similar cars did at highways speeds. The downside was
that the Beetle took forever to get to 60mph but in that market this
wasn't important.

US companies had plenty of designs to draw from if there was a market
for small cars but back then there simply wasn't in the US. All they had
to do was to open a design cabinet drawer at their overseas
subsidiaries. We had the P4 model of this one, for example:

http://en.wikipedia.org/wiki/Ford_Taunus

 

[[email protected]]

They were almost indestructable, with the engine only revving 70% of
what the Nash and similar cars did at highways speeds. The downside was
that the Beetle took forever to get to 60mph but in that market this
wasn't important.

The other downside was Winter. My brother had a Beetle in Minneapolis.