Maker Pro
Maker Pro

Switch mode power supply secondaries queries

N

N Cook

Jan 1, 1970
0
If you want 2 secondaries of say + and - 20 volts with common zero.
Assuming you have the right primary and drive requirement, number of turns
and gauge of wire for the secondary voltages and loads , what arrangements
are permissible?
Obviously a continuous run of insulated magnet wire, tapped in the centre ,
but what about a doubled up run, bifilar,in the same sleeving, with commoned
at one end joined together for zero volts and the other end split off, one
to one polarity of rectifier and one to the other.
What about a bifilar run again but one end of each commoned,exterior, but
the "plus" line coming from one end of one wire and the "negative" line from
the other end of the other wire.

The first is out as too much bulk of sleeving and retain proper
primary/secondary isolation. I'm aware that for unipolar secondaries say
0,+5V,+10V then it is permissible to run more wires/thicker wire between 0
and 5 than between 5 and 10 to reduce sleeving bulk.

Is there any practical difference, at 40 to 50KHz operation, between say 10
turns of secondary wire with cross-section area A and 10 turns of quadrupled
thinner wire each of area A/4. ?
 
M

MooseFET

Jan 1, 1970
0
If you want 2 secondaries of say + and - 20 volts with common zero.
Assuming you have the right primary and drive requirement, number of turns
and gauge of wire for the secondary voltages and loads , what arrangements
are permissible?
Obviously a continuous run of insulated magnet wire, tapped in the centre ,
but what about a doubled up run, bifilar,in the same sleeving, with commoned
at one end joined together for zero volts and the other end split off, one
to one polarity of rectifier and one to the other.
What about a bifilar run again but one end of each commoned,exterior, but

Bifilar as a twisted pair works well to make the two secondaries have
the same voltage. A twisted quad with pairs hooked in parallel works
a little better.

[....]
Is there any practical difference, at 40 to 50KHz operation, between say 10
turns of secondary wire with cross-section area A and 10 turns of quadrupled
thinner wire each of area A/4. ?

A seven wire, wire rope works even better. The next stop is Litz
wire. Both lower the losses. At the harmonics caused by the sharp
edges, you have quite high frequencies. If you have very fast
switching in the semiconductors the circuit has lower losses in total
but a bigger loss happens in the transformer. Going to multiple
thinner wire helps to lower this.
 
If you want 2 secondaries of say + and - 20 volts with common zero.
Assuming you have the right primary and drive requirement, number of turns
and gauge of wire for the secondary voltages and loads , what arrangements
are permissible?
Obviously a continuous run of insulated magnet wire, tapped in the centre ,
but what about a doubled up run, bifilar,in the same sleeving, with commoned
at one end joined together  for zero volts and the other end split off, one
to one polarity of rectifier and one to the other.
What about a bifilar run again but one end of each commoned,exterior, but
the "plus" line coming from one end of one wire and the "negative" line from
the other end of the other wire.

The first is out as too much bulk of sleeving and retain proper
primary/secondary isolation. I'm aware that for unipolar secondaries say
0,+5V,+10V then it is permissible to run more wires/thicker wire between 0
and 5 than between 5 and 10 to reduce sleeving bulk.

Is there any practical difference, at 40 to 50KHz operation, between say 10
turns of secondary wire with cross-section area A and 10 turns of quadrupled
thinner wire each of area A/4. ?

Where does this "sleeving bulk" come from? Transformers are normally
wound with enamelled wire, and the enamel is a few microns thick
(depending on grade)

http://www.elektrisola.com/index.php?id=362&L=0

There is a practical difference between 10 turns of secondary wire
with a cross-sectional area of A and ten turns of quadifilar wire
where each wire has a cross-sectional area of A/4 for high frequency
operation.

It arises because of the frequency dependent "skin effect" which
forces the current flow towards the surface of the wire. Check out
Litz wire.

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

The Wikipedia article does propagate the myth that it is worth silver-
plating copper wires to minimise skin resistance - bulk silver does
have a slighty lower resistance than bulk copper (16 versus 1.7 nano-
ohms metre) but electroplated silver doesn't - you'd have to melt the
electoplated silver (perhaps with a pulsed laser) before you saw any
improvement.
 
N

N Cook

Jan 1, 1970
0
MooseFET said:
If you want 2 secondaries of say + and - 20 volts with common zero.
Assuming you have the right primary and drive requirement, number of turns
and gauge of wire for the secondary voltages and loads , what arrangements
are permissible?
Obviously a continuous run of insulated magnet wire, tapped in the centre ,
but what about a doubled up run, bifilar,in the same sleeving, with commoned
at one end joined together for zero volts and the other end split off, one
to one polarity of rectifier and one to the other.
What about a bifilar run again but one end of each commoned,exterior,
but

Bifilar as a twisted pair works well to make the two secondaries have
the same voltage. A twisted quad with pairs hooked in parallel works
a little better.

[....]
Is there any practical difference, at 40 to 50KHz operation, between say 10
turns of secondary wire with cross-section area A and 10 turns of quadrupled
thinner wire each of area A/4. ?

A seven wire, wire rope works even better. The next stop is Litz
wire. Both lower the losses. At the harmonics caused by the sharp
edges, you have quite high frequencies. If you have very fast
switching in the semiconductors the circuit has lower losses in total
but a bigger loss happens in the transformer. Going to multiple
thinner wire helps to lower this.

So current going in opposite directions in 2 closely spaced secondary wires
makes no difference, compared to separated wires otherwise generating the
same voltages/current ?

I would have thought there was some sort of mutual inductance effect.
 
W

whit3rd

Jan 1, 1970
0
So current going in opposite directions in 2 closely spaced secondary wires
makes no difference, compared to separated wires otherwise generating the
same voltages/current ?

I would have thought there was some  sort of mutual inductance effect.

The current in the primary goes clockwise, and in ALL the
secondaries the current goes counterclockwise. The 'split supply'
secondaries are different only in that end A of the (+) winding
is grounded, while end B spits out current, while end B of the (-)
winding is grounded and end A sucks current in.

Both suck current at end A, both spit it out at end B. The
current goes the same way.
 
N

N Cook

Jan 1, 1970
0
So current going in opposite directions in 2 closely spaced secondary wires
makes no difference, compared to separated wires otherwise generating the
same voltages/current ?

I would have thought there was some sort of mutual inductance effect.

The current in the primary goes clockwise, and in ALL the
secondaries the current goes counterclockwise. The 'split supply'
secondaries are different only in that end A of the (+) winding
is grounded, while end B spits out current, while end B of the (-)
winding is grounded and end A sucks current in.

Both suck current at end A, both spit it out at end B. The
current goes the same way.

Confirmed in a practical test today. The other configuration of both end 'A'
grounded produces a correct rectified voltage in one and approximately zero
in the other.

Could I recommend the use of interlayer material on the primaries.
I'm rebuilding a small SMPS transformer where there was a gap in the winding
(as originally wound or developed over time - don't know) about 1/10 way in
the first of 3 layers. No interlayer and about 9/10 along the next return
layer had dropped into the first layer , local hot spot and shorting at that
point, leaving about 1/3 of intended turns functional, but of course
knocking out the driver.

Rebuilding on a 50 KHz SMPS type toroid core rather than the original as it
is easier , hence the limitations of secondaries isolation sleeving rather
than primary / secondary divider isolation wrap.
 
Top