Tutorial Saturable Core
Tutorial Saturable Core
Tutorial Saturable Core
Tutorial on
How to Define the Saturable Core Element
Powersim Inc.
July 2006
-1- www.powersimtech.com
Tutorial on Saturable Core Element
The magnetic elements in PSIM (windings, leakage paths, air gaps, and magnetic cores) provide a
powerful way of modelling magnetic devices. The objective of this tutorial is to show how to
define the saturable core element.
The simulated B-H curve of the saturable core depends on the combined effect of all the
parameters above. The diagram below shows how the parameters are related to the B-H curve.
phi_sat / Ac
K2 & Kexp2
R K1 & Kexp1
AL
The inductance factor AL mainly affects the slope of the B-H curve in the linear region. A larger
value of AL will result in a steeper slope.
-2- www.powersimtech.com
Tutorial on Saturable Core Element
The resistance R determines the width of the hysteresis loop. The larger the resistance, the wider
the hysteresis loop.
The coefficient phi_sat is roughly equal to the core flux at the deep saturation. It can be
calculated as the flux density B at the deep saturation multiplied by the core cross section area Ac.
The coefficients K1 and Kexp1 determine when the B-H curve starts to saturate (the first knee
point), and how sharp the transition of the curve is. A good guess of K1 is the ratio of the flux
density at this point versus the flux density at the deep saturation. For example, if the flux density
at this point is 0.245, and the flux density at the deep saturation is 0.35 T, the ratio will be 0.7.
The initial guess of K1 will then be 0.7.
The coefficient Kexp1 determines how sharp the transition of the curve is around this point. The
larger the value of Kexp1, the sharper the transition. A normal range of Kexp1 is from 10 for low
permeability ferrite to 200 for metglas.
The coefficients K2 and Kexp2 are associated with the second knee point in the saturation region,
and are used mainly to better fit the curve in the saturation region. A good guess of K2 is again
the ratio of the flux density at this point versus the flux density at the deep saturation. The
coefficients K2 and Kexp2 are used in rare occasions such as for ferroresonant regulators. They are
normally set as K2 > 2 and Kexp2 > 20 to keep them from affecting the B-H curve.
To illustrate the effect of the coefficients K2 and Kexp2, two B-H curves without and with the
effect of K2 and Kexp2 are shown below. The flux density on the Y-axis is normalized so that the
saturation flux density is 1.
K1 = 0.7 K1 = 0.7
Kexp1 = 10 Kexp1 = 10
K2 = 3 K2 = 0.9
Kexp2 = 20 Kexp2 = 60
(a) (b)
Fig. 2: The B-H curve with: (a) the effect of only K1 and Kexp1; and (b) the effect of both K1, Kexp1, and K2, and Kexp2
Fig. 2(a) shows the B-H curve without the effect of K2 and Kexp2. Notice how the curve continues
to increase beyond the saturation flux density of 1. Fig. 2(b) shows the B-H curve with the effect
of K2 and Kexp2. In comparison, the curve in the saturation region is flatter than the previous case.
The following example is used to illustrate how to obtain the core parameters from the
manufacturer datasheet.
-3- www.powersimtech.com
Tutorial on Saturable Core Element
Inductance: 4.2 uH
Core: EFD-25
Core effective cross section area Ac = 58 mm2
Core effective length Length = 57 mm
Approximate air gap inductance factor = 0.1167 uH (or gap length
lg = 0.62 mm)
Core Material: 3F3 ferrite
Number of Turns: 6
Operating Conditions:
Maximum current = 25 A
Peak-to-peak current ripple = 4 A
Switching frequency = 200 kHz
The first step is to obtain the B-H curve of the core material from manufacturers. In this example,
the B-H curve from Ferroxcube (www.ferroxcube.com) for the 3F3 ferrite material is shown
below.
Fig. 3: The B-H curve of the 3F3 ferrite material from Ferroxcube
The next step is to create a test circuit in PSIM to simulate the core and match the simulated B-H
curve with the manufacturer’s B-H curve in Fig. 3. The circuit below is created to test and
measure the B-H curve of the saturable core element.
-4- www.powersimtech.com
Tutorial on Saturable Core Element
Fig. 4: Test circuit to measure the B-H curve of saturable core model
In the circuit, the values of the core parameters are from the inductor design. A square-wave
voltage source is used as the test source. The peak voltage and the dc offset of the source are
adjusted so that the simulated B-H curve gives the same range of the magnetizing force H as the
datasheet. In this example, the range of H is from –250 to 250 A/m. The source frequency is set
to be the same as the operating frequency of 200 kHz.
The resistor Rin is used to provide damping so that the circuit can quickly reach the steady state.
The extra node of the saturable core element gives the flux phi flowing through the core. By
dividing the flux by the core cross section area Ac, we have the flux density B.
By multiplying the current with the number of turns and then dividing by the core length, we
have the magnetizing force H.
After the circuit is simulated, we first show the flux density B in the Y axis in Simview. Then go
to Axis -> Choose X-Axis variables, and choose H as the variable to plot the B-H curve.
Determining the core eofficients is an iterative process. One would start with an initial guess.
Then run the simulation, and compare the two B-H curves. Then go back to changes the
coefficients and run the simulation again. It may take many iterations to come with a good match.
A good initial guess of the coefficients will help speed up the process. The sections below
describe how the initial values are set for this example.
Inductance Factor AL
From the EFD-25 core datasheet, we have the core area of 58 mm2, the core length of 57 mm, and
from the 3F3 ferrite datesheet, we have the relative permeability μ r of 4000 (at 100oC). Based on
the information, we can calculate the core inductance factor as:
-5- www.powersimtech.com
Tutorial on Saturable Core Element
μ o ⋅ μ r ⋅ Ac
AL =
Length
which gives a value of 5.1uH. This value is used as the initial value for AL. As it turns out, in
order to better fit the B-H curve, this value needs to be increased. The final value of AL used is
8.5 uH.
Resistance R
The resistance R determines the width of the hysteresis loop. We started with an initial value of 1
mOhm, but found that the loop area to be too small. The resistance was eventually increased to
0.06 Ohm.
Coefficient phi_sat
The coefficient phi_sat is relatively easy to determine. The flux density B at deep saturation is
around 0.35 Tesla. The initial value for phi_sat was set to 0.35*Ac (where Ac is the core cross
section area). After some iterations, the final value of phi_sat was set at 0.38*Ac.
The B-H curve from the datasheet, in Fig. 1, shows no clear transition from the linear region to
the saturation region. It means that the coefficient Kexp1 should be relatively small. By examining
the B-H curve, we started with the initial guess as: K1 = 0.3; Kexp1 = 10.; K2 = 2; and Kexp2 = 20.
In this example, adjusting K1 and Kexp1 alone will not result in a good fit, and K2 and Kexp2 also
need to be adjusted to obtain a better fit. After numerous iterations and trial-and-error, the set of
parameters below was obtained as: K1 = 0.82; Kexp1 = 6.; K2 = 0.95; and Kexp2 = 40.
-6- www.powersimtech.com
Tutorial on Saturable Core Element
Fig. 5: The simulated B-H curve from the saturable core model
By connecting the air gap element in series with the core, we have the model for the inductor.
The circuit below shows the 4.2-uH inductor in a buck converter.
The result of the circuit using the modelled inductor is found to be very close to the result using
the ideal inductor.
-7- www.powersimtech.com