max flux density of 250mT for TDK PC40?

[reggie]

A rule of thumb I have heard about and want to check its validity is
that:

For a fan cooled forward converter transformer, employing feed forward
should have a flux density lower than 250mT as not to saturate for TDK
PC40.

1) Is this a reasonable estimate of max flux density at say 100 Deg C
100KHz.

I know that in PSU terms (not described accurately mathematically)

V=NdØ /dt

B=Ø/Ae

Thus:

V=N*Ae (dB/dt)

therefore following integration etc...

V *dt = N *Ae * dB

And thus

dB = (V*dt)/N*Ae---(1)

and by similar methods

dB = (L*dI)/N*Ae----(2)

(d indicated change or delta)

2) For a continuous forward converter, would this 250mT = (dB) limit
be as a result of

a) the ripple current peak to peak in the coil (from low value to high
value of the trapezoid) during the fet on time = (dT) or

b) the max current (from zero to the top of the trapezoid) during the
fet on time = (dT)

I think a)

Have I the wrong idea?

Reggie,

 

[Fred Bloggs]

A rule of thumb I have heard about and want to check its validity is
that:

For a fan cooled forward converter transformer, employing feed forward
should have a flux density lower than 250mT as not to saturate for TDK
PC40.

1) Is this a reasonable estimate of max flux density at say 100 Deg C
100KHz.

I know that in PSU terms (not described accurately mathematically)

V=NdØ /dt

B=Ø/Ae

Thus:

V=N*Ae (dB/dt)

therefore following integration etc...

V *dt = N *Ae * dB

And thus

dB = (V*dt)/N*Ae---(1)

and by similar methods

dB = (L*dI)/N*Ae----(2)

(d indicated change or delta)

2) For a continuous forward converter, would this 250mT = (dB) limit
be as a result of

a) the ripple current peak to peak in the coil (from low value to high
value of the trapezoid) during the fet on time = (dT) or

b) the max current (from zero to the top of the trapezoid) during the
fet on time = (dT)

I think a)

Have I the wrong idea?

Reggie,

The main limitation is power dissipation density (W/cm^3) for a maximum
temperature rise (usually 40oC) as a function of frequency, and 100KHz
is pushing it, almost always requiring a derating of Bmax,pkpk. You're
working with DC specs.

 

[reggie]

The main limitation is power dissipation density (W/cm^3) for a maximum
temperature rise (usually 40oC) as a function of frequency, and 100KHz
is pushing it, almost always requiring a derating of Bmax,pkpk. You're
working with DC specs.- Hide quoted text -

- Show quoted text -

Thanks Fred,

I take your point about power dissipation and temperature rise.
Are you saying that at 250mT Peak running at 100KHz the temperature
rise would be to great, ie above your 40 Deg C ?

I am not sure what you mean by DC specs, could you please explain.

In essence I am after what dI causes which dB in relation to the
current waveform and the resulting BH curve.

Reggie.

 

[Fred Bloggs]

Thanks Fred,

I take your point about power dissipation and temperature rise.
Are you saying that at 250mT Peak running at 100KHz the temperature
rise would be to great, ie above your 40 Deg C ?

I am not sure what you mean by DC specs, could you please explain.

In essence I am after what dI causes which dB in relation to the
current waveform and the resulting BH curve.

Reggie.

Those TDK datasheets should have a power density graph of dissipation
density versus frequency with flux density peak variation as a
parameter, you are interested in the ac-component of flux density. This
dissipation results from the energy required to force the material
through its hysteresis loop and is nonlinear with frequency or the rate
at which you flip the polarity of the magnetization. The 2500 Gauss
sounds more like a DC saturation flux density for the material, at
100KHz, you may have to back this off to 1000 Gauss or less. The DC flux
bias also sets a limitation on your ac-component of flux but does not
directly contribute to power loss and heating. And you're right, you
have to use the equations you mention relating flux to voltage,
frequency, and effective cross-sectional area, this is your starting
point for estimating the magnitudes you need to size the core.

 

[reggie]

Those TDK datasheets should have a power density graph of dissipation
density versus frequency with flux density peak variation as a
parameter, you are interested in the ac-component of flux density. This
dissipation results from the energy required to force the material
through its hysteresis loop and is nonlinear with frequency or the rate
at which you flip the polarity of the magnetization. The 2500 Gauss
sounds more like a DC saturation flux density for the material, at
100KHz, you may have to back this off to 1000 Gauss or less. The DC flux
bias also sets a limitation on your ac-component of flux but does not
directly contribute to power loss and heating. And you're right, you
have to use the equations you mention relating flux to voltage,
frequency, and effective cross-sectional area, this is your starting
point for estimating the magnitudes you need to size the core.- Hide quoted text -

- Show quoted text -

I have looked at the data below:

http://www.mhw-intl.com/assets/pdf/...0_PC47_PC90_PC95_Material_Characteristics.pdf

From page 7 onwards I get blank pages because of erros in installing a
new character set. But I see what you mean on other materials, Can I
get this data from anywhere else. I use to have it in a booklet from
TDK. Could I request this from someone? Anyway I see your point

Now, In a forward converter the flux swing = the working component =
ac component as you put it (I need a diagram!) on a B-H curve, moves
the working flux density up towards saturation. If there is too much
DC component the top end of the working flux density will start to
saturate and one gets that classic non linear spiky primary current
waveform. If you catch it before the fet blows up!

The 2500 Gauss I think is a rule of thumb not to exceed flux density,
when using feed forward and fan cooled, as the saturation flux density
of TDKs PC40 is 390 mT sorry 3900 Gauss at 100 deg C.

2500 Gauss must be the peak flux density, because if it were the
working (ac component) for PC 40, the circuit would have to be
discontinuous, ie the flux density would have to reset to zero.

I am assuming that the reset winding in a continuous forward converter
takes the flux density from the peak to the dc level.

Does this make sense?

The DC flux
bias also sets a limitation on your ac-component of flux but does not
directly contribute to power loss and heating

I understand, It's the "loop area" that contributes to heating.

Thanks again,

Reggie.

 

[legg]

Those TDK datasheets should have a power density graph of dissipation
density versus frequency with flux density peak variation as a
parameter, you are interested in the ac-component of flux density. This
dissipation results from the energy required to force the material
through its hysteresis loop and is nonlinear with frequency or the rate
at which you flip the polarity of the magnetization. The 2500 Gauss
sounds more like a DC saturation flux density for the material, at
100KHz, you may have to back this off to 1000 Gauss or less. The DC flux
bias also sets a limitation on your ac-component of flux but does not
directly contribute to power loss and heating. And you're right, you
have to use the equations you mention relating flux to voltage,
frequency, and effective cross-sectional area, this is your starting
point for estimating the magnitudes you need to size the core.

I think Reggie is working on a flyback with incomplete energy
transfer.

The 'peak' flux density could greatly exceed deltaB.

RL

 

[reggie]

I think Reggie is working on a flyback with incomplete energy
transfer.

The 'peak' flux density could greatly exceed deltaB.

RL

No Leg,

It is a forward with feedforward and a fan.

 

[Eeyore]

The 2500 Gauss sounds more like a DC saturation flux density for the
material, at
100KHz, you may have to back this off to 1000 Gauss or less.

The data gives a saturation flux @100C of *390* mT !
http://www.mhw-intl.com/assets/pdf/...0_PC47_PC90_PC95_Material_Characteristics.pdf

Why are you using Gauss btw ? That's 'old' cgs units.

Graham

 

[Eeyore]

the saturation flux density of TDKs PC40 is 390 mT sorry 3900 Gauss at 100 deg C.

You are CORRECT to use mT. Gauss is a deprecated cgs unit.

Can you use PC44 or PC47 ? These will have lower losses and the core calculations will hardly differ much.

Graham

 

[Terry Given]

No Leg,

It is a forward with feedforward and a fan.

in which case the magnetising current is what you are interested in, not
the reflected load current.

Cheers
Terry

 

[legg]

in which case the magnetising current is what you are interested in, not
the reflected load current.

......to determine peak flux density.

RL

 

[reggie]

.....to determine peak flux density.

RL- Hide quoted text -

- Show quoted text -

in which case the magnetising current is what you are interested in, not
the reflected load current.

.....to determine peak flux density.

I thaught so, I need to think...

Reggie.

 

[reggie]

I thaught so, I need to think...

Reggie.- Hide quoted text -

- Show quoted text -

I have been thinking, not alot (it hurts!)...

The questions I need answering are:

1) is 250mT peak, forward converter, with feed forward and a fan an ok
target?
2) how would I calculate the B peak, to make sure the transformer does
not saturate
3) is saturation B peak for TDK @100 deg C 390mT (or is it peak or
peak-peak)
4) can someone explain this with relation to ripple current and what
this looks like on a BH curve.

I want to be able to plot the BH curve for my forward converter using
instantaneous values of B, so to check that it isn't saturating.

I know that one uses different "B" formulas for bidirectional and
unidirectional driven chokes., since this is a forward converter in
need unidirectional, so need to take into account residual flux
density remaining in the core prior to the next cycle.

I think this is what I need..

Any ideas?

 

[legg]

1) is 250mT peak, forward converter, with feed forward and a fan an ok
target?

It won't saturate, but how hot will the transformer get? If you add a
fan, you give yourself some leeway, but you will need thermal
protection, because the safety isolation transformer is one of the
few items for which there are published acceptable maximum temperature
limits under single-fault ( ie jammed fan) conditions.

2) how would I calculate the B peak, to make sure the transformer does
not saturate

Bpk = v * t / ( N * A )

Best calculated at normal minimum operating voltage ~ dropout.

Bpk = peak flux in Teslas
V = applied primary voltage in Volts
t = maximum on time in seconds
N = primary turns count
A = core xsectional area im meters squared

3) is saturation B peak for TDK @100 deg C 390mT (or is it peak or
peak-peak)

Saturation is a flux level that can be achieved on purpose or
accidentally. In a forward converter, with flux effectively reset to
~0 during the off time, it would have to be achieved in one swiching
cycle. This is not impossible, but in a static/stable regulated
forward converter system it would not be expected.

4) can someone explain this with relation to ripple current and what
this looks like on a BH curve.

In a coupled transformer, flux levels are independant of load current,
due to transformer action. Secondary current flux cancels primary
current flux. Only magnetizing current is uncoupled and controlled by
Lmag.

Ripple current in the output choke has no effect on the transformer
magnetization.

I want to be able to plot the BH curve for my forward converter using
instantaneous values of B, so to check that it isn't saturating.

The Bpk formula is valid for any specific point in the swiching cycle.
If the voltage reverses in sign, the direction of deltaB reverses.


RL

 

[Eeyore]

It won't saturate, but how hot will the transformer get?

Oh, for heaven's sake, 250mT is *nothing* for PC40. It can be used WAY
beyond that.

The data gives loss figures anyway.
http://www.mhw-intl.com/assets/pdf/...0_PC47_PC90_PC95_Material_Characteristics.pdf

100kHz @ 200mT = 400mW / cm3 approx average. That will run COOL !

Use PC44 or 47 and it's more like 300mW / cm3


Graham