Maker Pro
Maker Pro

Another posting on transformer design

A

Adam Seychell

Jan 1, 1970
0
Oh dear, of all the hobbies to choose from why did I decide with
electronics. I'm getting headaches from constructing a simple SMPS
transformer. It looks to me there are as many variables in transformer
design as there are in jet engines.
From all the documentaion I have read on transformers I get the
impression most design variables are based on rules of thumb or from
the engineer's past experience, which I own none of. So a few
questions to the SMPS gurus (hello Genome ).
Requirements are; power 100W, primary 200V driven by half bridge,
secondary 6V and feed to synchronous current doubler rectifier, freq.
100 kHz, and should meet isolated standards. I chose ETD29 ferrite
core. This topology uses only one secondary winding and one primary
winding.

What can I expect the ratio of total copper winding cross section area
(both primary and secondary) to the bobbin winding window area ? Does
20% sound typical for an isolated transformer?

What would be conservative value for the total power dissipation for
ETD29 transformer when not relying on forced air cooling ?
Is a core loss power density of 100 mW/cm^3 a good figure to design
for ?

How important is interleaving ? I haven't measure or tried to do any
analysis on leakage inductance, but only disadvantage I can see is
added insulation which takes up precious winding area available for
copper.

What about a copper foil shield layer between adjacent primary and
secondary windings ? Again, this takes up more available winding area,
but does it simplify common mode EMI input filtering ?

Adam
 
C

CBarn24050

Jan 1, 1970
0
Hi, if memory serves me right you will struggle with that core. Try the next
size up. You will need to split your primary winding to reduce eddy current
losses. The faster you run the hotter it gets, so adjust your frequency to the
minimum that you can get away with.
 
R

R.Legg

Jan 1, 1970
0
Requirements are; power 100W, primary 200V driven by half bridge,
secondary 6V and feed to synchronous current doubler rectifier,
freq. 100 kHz,
and should meet isolated standards.
I chose ETD29 ferrite core.

You need Vin minimum, if this is a regulated output, to determine
turns ratio.

Using Half-Bridge pwm can offer problems getting synchronous
rectification, if less than continuous full duty is intended.

Assuming this is an unregulated DC-DC isolation stage:

[ ( Vin/2 - Vffet ) x Dmax ] / ( 2Vout - Vfrect) = Np / Ns

Effective input voltage is halved (Vin/2), because only half the
input voltage is available across the primary coupling capacitor.

Forward voltage drop of the conducting fet (Vffet) subtracts from
the voltage developed across the switched winding.

Maximum duty cycle is set by the modulator drive, switch delays
and leakage inductance. The latter term will have to be a rough
estimate, until you measure your first iteration. It is important
because you are asking the current to reverse from +imax to - imax
on each switch transition. This will take time. Assume 95% (0.95).

2Vout is the peak voltage required before the current doubler
rectifier, and the rectifier forward voltage drop (Vfrect)
subtracts from it.

{ (200/2 - 2) x 0.95 ] / ( 2x6 - 1) = 88.4/11 = 8

If this were a regulated topology, Vin would be replaced by Vinmin,
and turns ratio would reduce.

Actual turns is calculated from the output winding, as this
voltage is 'regulated'.

Nsmin = ( Vo x 2 + Vfrect ) x ( Dmax / 2f ) / ( Bmax x Ac )

(Vo x 2 + Vfrect) - is the secondary peak voltage required by
the current doubler circuit to develop the intended regulated
output voltage at dropout/full duty cycle (volts)
( Dmax / 2f ) - is the time period the secondary peak voltage is
present during a switching period, when developing the intended
output voltage and current at dropout/full duty cycle(seconds).
Bmax - is the total flux swing (ie 2x Bpk for a half bridge)
determined not to overheat or to saturate the core and core
material selected. (tesla) see note#1
Ac - is the published effective core cross-section of the core shape
selected (square meters).

note#1 - To get Bmax you have to determine total core loss
permissible in your situation (W), convert this into mW/cm^3
for the volume of the core shape selected, and then consult
the mfr data to relate the limiting power density to charted
Bpk vs f for your core material.

In this case .15 Bpk is assumed, allowing a Bmax of 0.3.
If this were a simple single-ended converter, Bpk would equal
Bmax; topologies developing a full flux swing, like the half
bridge, have the advantage here.

( 2 x 6 + 1 ) x ( .95 / 2E5) / (.3 x 76E-6) = 2.7 or 3Turns.

This topology uses only one secondary winding and one primary
winding.

Primary turns is

Mp/Ns x Ns, or 8x3 = 24
What can I expect the ratio of total copper winding cross section
area (both primary and secondary) to the bobbin winding window
area? Does 20% sound typical for an isolated transformer?

Typical commercially wound copper factor in an isolating
device is abysmal. Multiply whatever you can fit with care
into your hand-wound transformer by 0.7, and expect trouble
until the winders become familiar with th part.
20% sounds like what the winders will want to wind.

A full bobbin is not always an optimum design. An overfull
bobbin will never make it. In some cases, the use of multiple
insulated wire is permitted, which can ease the requirement
for internal creepage and margins. Not all safety agencies
recognize this as valid insulation, unless rises are restricted
to class A, and the requirements of a listed insulation system
do not apply.
What would be conservative value for the total power dissipation
for ETD29 transformer when not relying on forced air cooling ?

1.32 W total for a 30 degree surface temperature rise.
This assumes 1 degreeC rise for every milliwatt dissipated
per cm^2 of surface area of the completed part.
Is a core loss power density of 100 mW/cm^3 a good figure to
design for ?

Add it up youself.
The aim is total part loss, not a core loss density.
As secondary turns are incremental, in fairly large steps
( ie 2, 3, 4 ) at this low voltage, they create core loss
and copper loss situations that are characteristic for material,
operating frequency, topology and output voltage. You will
have to consider the effects of incrementing total turns upwards
or downwards in each instance, yourself.

How important is interleaving ? I haven't measure or tried to do
any analysis on leakage inductance, but only disadvantage I can
see is added insulation which takes up precious winding area
available for copper.

You'll find out. Leakage inductance will actually steal applied
voltseconds - these are primary volt-seconds applied (and contributing
to core loss) that are not transferred to the secondary - reducing
effective output duty cycle at higher load current. Leakage energy in
a half-bridge is recoverable, but detrimentally complicates a
regulated circuit, and contributes to generation of EMI.
What about a copper foil shield layer between adjacent primary and
secondary windings ? Again, this takes up more available winding
area, but does it simplify common mode EMI input filtering ?

It can simplify it, allowing levels of conducted EMI to be reduced to
levels that are physically impractical in it's absence.

http://www.dei.unipd.it/~pel/Pagine_Personali/Giorgio/Pubblicazioni/
 
R

Robert Baer

Jan 1, 1970
0
Adam said:
Oh dear, of all the hobbies to choose from why did I decide with
electronics. I'm getting headaches from constructing a simple SMPS
transformer. It looks to me there are as many variables in transformer
design as there are in jet engines.
From all the documentaion I have read on transformers I get the
impression most design variables are based on rules of thumb or from
the engineer's past experience, which I own none of. So a few
questions to the SMPS gurus (hello Genome ).
Requirements are; power 100W, primary 200V driven by half bridge,
secondary 6V and feed to synchronous current doubler rectifier, freq.
100 kHz, and should meet isolated standards. I chose ETD29 ferrite
core. This topology uses only one secondary winding and one primary
winding.

What can I expect the ratio of total copper winding cross section area
(both primary and secondary) to the bobbin winding window area ? Does
20% sound typical for an isolated transformer?

What would be conservative value for the total power dissipation for
ETD29 transformer when not relying on forced air cooling ?
Is a core loss power density of 100 mW/cm^3 a good figure to design
for ?

How important is interleaving ? I haven't measure or tried to do any
analysis on leakage inductance, but only disadvantage I can see is
added insulation which takes up precious winding area available for
copper.

What about a copper foil shield layer between adjacent primary and
secondary windings ? Again, this takes up more available winding area,
but does it simplify common mode EMI input filtering ?

Adam

I know not of the material and size you mentioned.
Start with the "frequency" - the inverse of the PRF without regard to
duty cycle.
Pick a core material with low loss at that frequency, and ideally at
the frequency related to the narrowest part of the driving waveform.
The specifications of that material will indicate the size interms of
cross-sectional area.
The rest falls into place.
 
Top