Pcim 2012 MW
Pcim 2012 MW
Ron
N
2
+
(
(
s L
j
Ron
N
2
+
+ :=
[1]
Zoff
N j ,
s L
j
1 K ( )
s
2
K 1 K ( ) L
j
( )
2
s K L
j
Roff
N
2
+
(
(
s L
j
Roff
N
2
+
+ :=
[1bis]
Finally, we need to consider the effect of the secondary leakage inductance at the output of the
HV trafo; this will generate a voltage drop such that Vo (peak value) will be different from
Vcepk * N ( where N=N1 is the transformer turn ratio ). By keeping this into account, Vopk is
given by:
Vopkest
Vin K
Ron
N1
s L
j1
1 K
2
( )
Ron
N1
2
+
K
Roff
N1
Vcepk Vin ( )
s L
j1
1 K
2
( )
Roff
N1
2
+
+ :=
[2]
Where, again, different damping between Ton and Toff is allowed for. The set of equations
derived in the previous discussion are the basis for the proposed design procedure, as it will be
shown in the following. First of all, we need to decide which parameters are a priori given and
which need to be calculated. In our approach, only 5 parameters are given: Vo, Po, Vin(pk), Fsw
and Vcepk. Now, load resistance is simply derived by:
Rload
Vopk
2
Popk
:=
[3]
While, having specified Vcepk and Fsw, Toff and Ton follow:
Toff
Fsw
Vcepk
Vin
+
|
\
|
|
:=
[4]
Ton
1
Fsw
Toff :=
[4bis]
Toff equation is approximated; damping is not taken into account. On the other side, damping
could NOT be introduced yet because the HV trafo has still to be designed yet and reflection
ratio between its secondary and primary side is not known! Coupling factor for the HV trafo
needs to be selected. All the design equations are too much inter-dependent between each
other and some parameter needs to be fixed based on the designers experience. We have
chosen to impose a reasonable K value. At this point, L and N are the only two still unknown
parameters, and two independent constraints are necessary. The approach is the following:
drawing the 3D graph of Zon (Zoff) versus L and N, we get the picture of Fig. 4. Actually, what is
represented here is the ratio between the real and imaginary part of Zon(Zoff).
RatioOn
Figure 4: design optimization surface
The locus where the ratio is maximum, give the points L-N1 which maximize the power transfer.
Because Zon may differ from Zoff, there will be two solutions and an average has to be
chosen. Please remember this is a design guideline, not a complete and precise analytical
design tool. Once a solution has been chosen, L and N1 are known, and Cres can be calculated
as:
Cres
Toff
2
2
Lpoff
N1 j1 ,
|
\
|
Vcepk
Vcepk Vin
|
\
|
|
:=
[5]
Now all the design parameters are known. Vopk can be checked by means of Eq [2], while peak
power is given by:
Ploadpk
Vin
2
Fsw
Rpon
N1 j1 ,
Ton s 1.5 2 exp
Ton
s
|
\
|
|
+ 0.5 exp 2
Ton
s
\
|
|
\
|
|
(
(
Fsw Rpoff
N1 j1 ,
Toff
8
deltaI
2
+ :=
Where
deltaI
Vin
Rpon
N1 j1 ,
1 exp
Ton
s
|
\
|
|
\
|
|
:=
And
s
Lpon
N1 j1 ,
Rpon
N1 j1 ,
:=
In the above eq. N1 is the value of N chosen by Fig 4, and j1 is the index for L. Lpon and Lpoff,
as well as Rpon and Rpoff, are the L-R value reflected on the HV trafo primary side, during Ton
and Toff. The MWO converter has been thus simplified to a single ended parallel resonant
converter where the resonating components are Lpon ( and Lpoff ) and Cres, and the damping
is given by Rpon ( and Rpoff ).
Design Example, simulation and measurement verification
In the following example, the known parameters were:
RF power : 900W
Estimated magnetron efficiency : 55%
Switching frequency : 40kHz
Peak input voltage : 165V
Peak primary switch voltage : 600V
From the curve of Fig 4 (re-calculated), optimal turn ratio and leakage inductance are derived:
N1 = 15 ; Lleak = 30uH
Then, Cres = 390pF, Vopk = 4450V, Ploadpk = 2300W and Icepk = 73A
The same design is simulated by using spice. The ripple is almost equivalent, and the sum of
the peak voltages across C9 and C11 is 4400V, as predicted by the above described procedure
( see Fig. 5).
Figure 5: Pspice simulation results
Measurements were performed on a commercially available magnetron power supply, having
parameters very close to the ones calculated and simulated. Fig. 6 show the results. The
inverter was based on IRG7PH42UD1, 1200V trench IGBT by International Rectifier Crop. Good
agreement is found on both DC output voltage, primary switch peak voltage and peak current.
Figure 6 : experimental results
Conclusions
A simplified methodology for designing single ended resonant converter for microwave ovens
has been introduced. Despite the analytical treatment is not completely precise, such
methodology gives some guidelines to size the HV transformer, resonating components and
gives a reasonable approximation of the stress level across the primary switch and across the
voltage multiplier components.
References:
a) Analysis of MicroWave Ovens for 115V and 230V line; C.Bocchiola; internal IR Report
b) Whirlpool AWM839 user manual
c) EP1806954A1, Matshushita Electric Industrial Co.
d) The complete microwave oven service handbook; J.Carlton Gallawa; 1989-2009.