Loosely coupled transformer windings

[Steve Carroll]

I would like to wind a transformer with one "closely-coupled" secondary and
one "loosely-coupled" secondary, for current-limiting. I understand that
increasing the air gap in the centre leg of an E-core will decrease the
coupling co-efficient, but how do I do it for only one winding? If I use an
E-core with a gap and wind two windings on top of each other on one side,
will they be more closely coupled than a winding on the opposite side of the
core and can I then vary the coupling of the secondary by varying the
air-gap, while maintaining closer coupling between the primary and the
feedback windings? (As you can tell, I'm a relative newbie to fluoro
inverter design, but I'm willing to learn.)
(I've posted a drawing on my website, (12.5K JPEG):-
http://www.dolevoiceaustralia.com.au/question.html )

This is for a complementary drive rapid-start fluoro inverter. I want a high
voltage, ~1kV, to get things started, then under load or reduced frequency
I want
the voltage to drop heaps to one or two hundred volts.

Any help would be appreciated.

.... Steve

 

[John Popelish]

I would like to wind a transformer with one "closely-coupled" secondary and
one "loosely-coupled" secondary, for current-limiting. I understand that
increasing the air gap in the centre leg of an E-core will decrease the
coupling co-efficient, but how do I do it for only one winding? If I use an
E-core with a gap and wind two windings on top of each other on one side,
will they be more closely coupled than a winding on the opposite side of the
core and can I then vary the coupling of the secondary by varying the
air-gap, while maintaining closer coupling between the primary and the
feedback windings? (As you can tell, I'm a relative newbie to fluoro
inverter design, but I'm willing to learn.)
(I've posted a drawing on my website, (12.5K JPEG):-
http://www.dolevoiceaustralia.com.au/question.html )

This is for a complementary drive rapid-start fluoro inverter. I want a high
voltage, ~1kV, to get things started, then under load or reduced frequency
I want
the voltage to drop heaps to one or two hundred volts.

Any help would be appreciated.

Steve, the way you show it is generally the way it is done,
except that the center leg of the core is often a separate
chunk of core with an air gap at each end, that is slid
between the primary and secondary sides of the outer core.
Almost any microwave oven that does not have inverter drive
will have one of these in a kilowatt size. Many high
pressure sodium and mercury lamp ballasts are made this way,
also. The difference for your case, I think is that the
core will be ferrite instead of laminated silicon steel.

You might be able to get by with a toroid core with a the
primary (and feedback) on one half, the secondary on the
other half, and a ferrite bar insert across the diameter
between them. But it would probably be more reproducible
with UU cores.

 

[Steve Carroll]

Steve, the way you show it is generally the way it is done,
except that the center leg of the core is often a separate
chunk of core with an air gap at each end, that is slid
between the primary and secondary sides of the outer core.
Almost any microwave oven that does not have inverter drive
will have one of these in a kilowatt size. Many high
pressure sodium and mercury lamp ballasts are made this way,
also. The difference for your case, I think is that the
core will be ferrite instead of laminated silicon steel.

You might be able to get by with a toroid core with a the
primary (and feedback) on one half, the secondary on the
other half, and a ferrite bar insert across the diameter
between them. But it would probably be more reproducible
with UU cores.

Excellent, thanks John.
I thought I had the concepts right, but just wanted to be sure. I've
wound my share of simple inductors in the past, but nothing like this. I
like your lateral thinking - a bar across a toroid.
In this application, the impedance of the secondary isn't too critical,
but I'll have a play around now that I know I'm on the right track and
increase it as much as possible.
A question - what do you mean by UU cores? (Two "U" cores?)

.... Steve

 

[John Popelish]

I thought I had the concepts right, but just wanted to be sure. I've
wound my share of simple inductors in the past, but nothing like this. I
like your lateral thinking - a bar across a toroid.

There are lots of other possibilities, also. Keep in mind
that the top and bottom bar in your drawing are essentially
equi-potential (from a magnetic field standpoint) nodes, so
the ordering of the parallel branches is not very important.
The flux shunt could be on either side of either the
primary or secondary branches. It doesn't need to be
between them. Think E cores with center or one leg ground
off to make a gap. There are E core shapes made that have
all 3 legs the same cross sectional area for this purpose.
And, of course, you could add an external inductor in series
with a tightly coupled secondary and get a very similar effect.

In this application, the impedance of the secondary isn't too critical,
but I'll have a play around now that I know I'm on the right track and
increase it as much as possible.
A question - what do you mean by UU cores? (Two "U" cores?)

Right. I don't have keyboard characters for a pair of Us
facing each other to form a loop. ;-)

 

[Steve Carroll]

Think E cores with center or one leg ground
off to make a gap.

My drawing probably wasn't too clear, but there was an air gap in the middle
leg. (It looks like a washer.) I figure that I can wind the 3 windings,
then increase the air gap to reduce the coupling to the secondary until,
under load, it drops from 1000V of "turns-ratio" voltage to about 100V-200V
of "loaded" voltage, while still keeping close coupling between primary and
feedback windings, (or two halves of a centre-tapped primary for push-pull
driving).

There are E core shapes made that have
all 3 legs the same cross sectional area for this purpose.

I don't quite follow what you mean here.

And, of course, you could add an external inductor in series
with a tightly coupled secondary and get a very similar effect.

Yep, I've also been thinking about that. It would make starting easier, too,
especially if I adopted a method of shorting across the second side of the
heater filaments to heat them, then open that circuit to dump the series
inductor's energy across the tube for ignition, similar to mains operated
fluoros.

Incidentally, in the end I'll also add heater windings, which should have
reasonably close coupling, but one thing at a time, I reckon.

.... Steve

 

[Steve Carroll]

Right. I don't have keyboard characters for a pair of Us
facing each other to form a loop. ;-)

Regards,
John Popelish

View in "Symbol" font:- ÌÉ

.... Steve

 

[Mark]

View in "Symbol" font:- ÌÉ

... Steve

Question for the OP
Why would you NEED a loosely coupled winding?
Why can't you use a tightly coupled winding?
Mark

 

[John Devereux]

There are lots of other possibilities, also. Keep in mind that the
top and bottom bar in your drawing are essentially equi-potential
(from a magnetic field standpoint) nodes, so the ordering of the
parallel branches is not very important. The flux shunt could be on
either side of either the primary or secondary branches. It doesn't
need to be between them. Think E cores with center or one leg ground
off to make a gap. There are E core shapes made that have all 3 legs
the same cross sectional area for this purpose. And, of course, you
could add an external inductor in series with a tightly coupled
secondary and get a very similar effect.

Or a capacitor - as used in CCFL inverters.

 

[John Popelish]

John Popelish wrote:

My drawing probably wasn't too clear, but there was an air gap in the middle
leg. (It looks like a washer.) I figure that I can wind the 3 windings,
then increase the air gap to reduce the coupling to the secondary until,
under load, it drops from 1000V of "turns-ratio" voltage to about 100V-200V
of "loaded" voltage, while still keeping close coupling between primary and
feedback windings, (or two halves of a centre-tapped primary for push-pull
driving).

I understand, but increasing the gap between the core halves
will also increase the magnetization current of the primary
because of the lowering if its inductance and also slightly
lower the coupling factor to the tightly coupled secondary.

I don't quite follow what you mean here.

Most E cores have a center leg with twice the cross
sectional area of the legs on each side, with the
assumption that all the coils will be placed on the center
leg, and the flux will split into two equal fractions and
half will pass through each of the outside legs. But I have
seen a few E cores that have a constant cross sectional area
in all parts. These allow for some creative winding designs
with various windings on different legs.

Yep, I've also been thinking about that. It would make starting easier, too,
especially if I adopted a method of shorting across the second side of the
heater filaments to heat them, then open that circuit to dump the series
inductor's energy across the tube for ignition, similar to mains operated
fluoros.

Incidentally, in the end I'll also add heater windings, which should have
reasonably close coupling, but one thing at a time, I reckon.

Sounds like a plan.

 

[Steve Carroll]

View in "Symbol" font:- ÌÉ

... Steve

Question for the OP
Why would you NEED a loosely coupled winding?
Why can't you use a tightly coupled winding?
Mark

I thought the purpose was clear. I want the secondary to have loose coupling
so that the output voltage is high under no load, then drops considerably
under load once the tube ignites.

.... Steve