Sei sulla pagina 1di 5

26 IEEE Power Electronics Society NEWSLETTER Second Quarter 2011

Simulation Bits: Adding the


Reverse Recovery Feature to a Generic Diode
Natan Krihely and Sam Ben-Yaakov

Abstract – A simplified behavioral model of diode reverse recov- The proposed equivalent circuit model of Fig. 2 emulates the
ery, which is based on the lumped charge control concept, is physical behavior during the turn-off period of a typical power di-
proposed and tested experimentally. The model can be imple- ode by solving the equations of the simplified diode model. The first
mented in discrete transition simulators that do not apply full order differential equation (2) is emulated in the model by an RC
physical models of devices and it does not require heavy compu- network. The resistor–capacitor pair RtCt of the model reflects a time
tational resources. The parameters of the model can be easily constant t during 0 < t < ta (Fig. 1). The resistor–capacitor pair Rt_rrCt
extracted from a simple single measurement. The model is dem- reflects a time constant trr during the recovery period t>ta. E_id is a
onstrated in the PSIM platform and is verified experimentally for current controlled voltage source associated with the diode current
and is equal to t · iD(t). The voltage controlled current source G_qm
various turn-off conditions.
expresses the response qm(t)/Tm. The role of the switch Srr is to
select the correct time constant and is controlled by the state of the
I. Introduction ideal diode Dvf. The diode Dvf carries the sum of iD(t) 1 qm(t)/Tm.
As long as Dvf conducts with iDfv(t)>0, the switch Srr is ‘off’ and the
Power electronics simulators can be divided into two groups with time constant is t. According to reference [3], at time instant ta the
respect to the numerical algorithm used by them. One class are the diode current iD(ta) and qm(ta)/Tm have equal magnitudes with op-
SPICE [Berkeley SPICE2] based simulators such as Orcad [1]. SPICE posite sign. At this instant, Dvf stops conducting and the switch Srr
uses an adjustable step, Newton-Raphson algorithm to solve the turns on. From now on the current iD(t) will decay to zero with a
matrix of nodal equations when the circuit contains nonlinearity time constant trr as follows,
such as physical models of switches and diodes. The second class
includes the discrete-transition simulators such as PSIM [2] and oth-
ers. The ideal device models used by the discrete transition simula-
tors and the fact that they disregard the transition instances makes IF
them much faster than the SPICE based simulator and eliminate, to I(D)_PSIM
a large extent, convergence problems. On the down side, since
most discrete transition simulators do not apply full physical mod-
els of devices, they are incapable of showing parasitic processes. –ar
For example, diodes in PSIM are ideal and include only one param- –Irm
eter: the forward diode voltage drop. Consequently, simulation I(D)_Experiment
results are unrealistic (Fig. 1) when simulating some power elec-
tronics circuits such as snubbers that are affected by non ideal and I(D)_Pspice
parasitic properties.
0 ta
Here we propose a simple and portable modeling strategy for
simulating the diode reverse recovery process within discrete-tran-
sition simulators. Figure 1. Simulated and Experimental turn-off current waveforms of
power diode (MUR8100E, ON-Semi) with stray inductance.
II. Modeling procedure
and parameters extraction
The modeling method adopted in this study is based on the
lumped charge control concept [3]. This principle is described ana- Anode
lytically by the following equations:
Rτ – Rτ_rr Rτ_rr iD
qe 2 qm
iD 1 t 2 5 (1)
Tm Lp
τ · iD + +
dqm qm qe 2 qm
05 1 2 (2) Srr Cτ qm
dt t Tm E_iD – Vf
– qm/Tm
qe 5 ISt c expa b 2 1d
vD
(3) 1 Dvf
nVT
y 0 x G_qm
iDvf iDvf
where, iD (t) is the diode current, qe is the injected charge level at +
y = 1, x > 0
the junction, qm is the total forward bias injected charge, Tm is the y = 0, x ≤ 0 – Rs
drift region transit time, t is the life time, IS is the diode saturation E_Dvf
current, v D is the diode junction voltage, VT is the thermal voltage
and n is the emission coefficient. To minimize computational over- Cathode
heads as much as possible, the model was further simplified in this
study by disregarding the exponential dependency between the Figure 2. Proposed diode model with reverse recovery. The Lp
forward voltage v D and the charge qe of the power diode (3). inductor represents the parasitic lead inductance.
Second Quarter 2011 IEEE Power Electronics Society NEWSLETTER 27

I_D I_D_EXP I_D I_D_EXP


3.00
3.00
2.00 2.00
1.00
0.0 1.00
–1.00
–2.00 0.0
–3.00
0

38

63
.0

.2

.4

.6

.8

.0
.4

.5
12

12

12

12

12

00
98

01
,0

,0

,0

,0

,0

,0
,9

,0
14

14

14

14

14

14
13

14
Time (us) Time (us)
(a) (b)
I_D I_D_EXP I_D I_D_EXP
3.00 3.00

2.00
2.00
1.00
1.00
0.0
0.0
–1.00

–1.00 –2.00
0

0
0
75

88

13

25

.5

.0

.5

.0
.5

11

12

12

13
.8

.1

32

.8

.1
31

32

32

33

,0

,0

,0

,0
,6

14

14

14

14
,6

,6

14

,6

,6
14

14

14

14

Time (us) Time (us)


(c) (d)
I_D I_D_EXP I_D I_D_EXP
1.50
5.00

1.00 4.00

3.00
0.50
2.00

0.0 1.00

0.0
–0.50
–1.00

13,998.75 13,999.375 14,000.00 14,662.50 14,665.00 14,667.50 14,670.00


Time (us) Time (us)
(e) (f)

Figure 3. Reverse recovery model validation (PSIM simulation): (a) Comparison to the measurement data, (b) Lr 5 10mH, IF 5 2.5A, (c) Lr 5
6mH, IF 5 2.5A, (d) Lr 5 1.3mH, IF 5 2.5A, (e) Lr 5 6mH, IF 5 1.3A, (f) Lr 5 6mH, IF 5 4.3A.

t 2 ta
iD 1 t 2 5 2 Irm expa2
1 1 1
b (4) 5 1
trr t Tm
(6)
trr

where Irm is the peak reverse current (Fig. 1). where ar is the current fall slope (Fig. 1).
The model is calibrated by seven model parameters that need By substituting the experiment parameters Irm, ar, ta, trr into (5)
to be determined in the following order. The parameter trr can be one can extract t. Tm is calculated from t and trr using (6). Once
obtained by fitting (4) against measurement data over t>ta. In [3], it is the fundamental parameters are extracted, the calculation of the re-
shown that the following relationships hold, maining five model parameters may proceed as follows: The contact
resistance Rs 5 DvD/DiD (from diode forward static characteristic), Vf
Irm 5 ar 1 t 2 trr 2 c 1 2 expa 2 b d
ta (from diode forward static characteristic), Ct 5 Co (arbitrary num-
(5)
t ber), Rt 5 t/Co and Rt_rr 5 trr/Co.
28 IEEE Power Electronics Society NEWSLETTER Second Quarter 2011

Qs Lo
V_OUT

V
1.5m

Rs 40m

Daux1
Ls 1.35u 100k Rs1

50
Lr 2u V_Cr Ro
100u Co
V
Vin 4.55
V_DFW

10n Cr
V

Rs = 50m
V_F = 1 47K Rs2
Tm = 75.33n
A C

DFW tau = 144.8n Daux2


Rtau = 144.8
Ctau = 1n
Rtau_rr = 49.55

V_Trig V
Q K_AD
Q
C BLOCK
1/5
V_OnTime V dsPIC30F2020

Figure 4. Circuit diagram of simulation and experimental setup of dc-dc buck converter. Lr and Ls are the snubber inductor and stray
inductance, respectively.

III. Model verification depicted in Fig. 5. The simulation model replicated very well the
The diode tested in the experiments is a commercial fast recovery behavior of the experimental circuit (Fig. 5 (a)), except for some
power diode MUR8100E, (ON-Semi.) with a nominal reverse recov- negligible differences stemming from stray capacitance and induc-
ery time of 75 nsec. Test measurements were carried out with an tance and mismatch in forward voltage of the diodes between the
estimated reverse voltage of 50 V during the turn off period of the simulation circuit and the experimental. As expected, large devia-
diode. The model parameters were calculated from a single mea- tions are found between experiment and simulation when the origi-
surement data of a typical turn off waveform in the presence of nal PSIM diode is used (Fig. 5 (b)).
stray circuit inductance with the following experimental parame-
ters: initial forward current IF 5 2.5 A, peak reverse current Irm 5 IV. Discussion and conclusion
2A, current fall slop ar 5 4.5 A/122 nsec and the length of the turn The proposed diode reverse recovery model is suitable for very
off period ta 5 122 nsec. The estimated stray inductance L s was detailed study of device interactions with the rest of the circuit,
found to be 1.35 mH. Applying the proposed extraction method, the e.g., snubber design. The experiment results were found to be in
extracted model parameters were found to be: Rs 5 50 mV, Vf 5 1 V, good agreement with the simulation waveforms (Fig. 3). The model
Tm 5 75.33 nsec, t 5 144.8 nsec, Ct 5 1 nF, Rt 5 144.8 V and Rt_rr gave good results for forward currents below 2.5 A and exhibited a
5 49.55 V (trr was estimated by fitting (4) against measurement data reasonable accuracy for a forward current of IF 5 4.3 A (Fig. 3 (f)).
over t . ta). The comparison between the experimental and simu- The deviation can be explained by the fact that the parameters were
lated results of the proposed model for various turn off conditions fitted to give closest approximation to data in which the forward
is shown in Fig. 3. The figure shows a good fit of the simulation to current was IF 5 2.5 A. In addition, a slow decay of the tail to zero
the experimental data. was observed. This is probably because the single time constant of
The model was also tested in a dc-dc buck converter that in- the model is incapable of simulating accurately the decay of higher
cludes a passive lossless snubber network as shown in Fig. 4. A order waveforms. The practical usefulness of the model over the
comprehensive analysis of this turn-on lossless snubber for dc-dc standard ideal diode was demonstrated in a test bench dc-dc buck
Boost converter was presented in [4]. In the experimental circuit: converter. The diode model was found to reproduce faithfully the
Vin 5 50 V, Vo 5 12 V, Po 5 30 W, fs 5 24 KHz, Lr 5 2 mH, Cr 5 10 nF, experimental data (Fig. 5 (a)).
Daux1 and Daux2 were MUR415 (ON-Semi). The converter was digi- The strength of the model is in the ability to help designers to
tally controlled by dsPIC30F2020, which in the simulation is repre- examine physical phenomena such as reverse recovery and its in-
sented by a C block (Fig. 4). The simulation was carried out under teraction with other circuit components and at the same time to
the following conditions: (1) auxiliary diodes Daux1 and Daux2 were explore control needs for stability, soft switching and others on one
ideal with constant forward voltage, (2) the reverse recovery of platform. The model can easily be incorporated in other discrete-
the freewheeling diode was implemented by the proposed model. transition power electronics simulators by following the proposed
Simulation results and the experimental verification waveforms are model construction procedure.
Second Quarter 2011 IEEE Power Electronics Society NEWSLETTER 29

I(Daux1) I(Daux1)_EXP
3.00

2.00
Simulation
1.00

0.0

I(Lr) I(Lr)_EXP
3.00
Simulation
2.00
1.00
0.0
-1.00

V_Cr V_Cr_EXP
17.50
Simulation
8.75

0.0

14,840.00 14,850.00 14,860.00 14,870.00 14,880.00


Time (us)
(a)

I(Daux1) I(Daux1)_EXP
3.00
Simulation
2.00

1.00

0.0

I(Lr) I(Lr)_EXP
3.00
2.00
1.00
Simulation
0.0
-1.00

V_Cr V_Cr_EXP
15.00
10.00 Simulation
5.00
0.0
-5.00
14,840.00 14,850.00 14,860.00 14,870.00 14,880.00
Time (us)
(b)

Figure 5. Comparison between simulation (PSIM) and the experimental waveforms of the turn-on lossless snubber: (a) results when the
proposed diode model is included, (b) results when the original PSIM diode is used.
30 IEEE Power Electronics Society NEWSLETTER Second Quarter 2011

Acknowledgment and Ph.D. degrees in engineering from the


University of California, Los Angeles, CA, in
This research was supported by THE ISRAEL SCIENCE 1967 and 1970, respectively. He is currently an
FOUNDATION (grant No. 476/08) and by the Paul Ivanier Center Emeritus Professor with the Department of
for Robotics and Production management. Electrical and Computer Engineering, Ben-
Gurion University of the Negev, Beer-sheva,
Authors Israel, where he was the chairman during
1985-1989. He is presently the head of the
Natan Krihely was born in Beer-sheva, Israel, Power Electronics Group (http://www.ee.bgu.ac.il/~pel) there.
in 1975. He received the B.Sc. and M.Sc His current research interests include power electronics, circuits
degrees in electrical engineering from Ben- and systems, electronic instrumentation and engineering educa-
Gurion University of the Negev, Israel, in 2002 tion. He is a consultant to commercial companies on various sub-
and 2006, respectively. He is currently pursu- jects, including analog circuit design and power electronics.
ing the Ph.D. degree at the Ben-Gurion
University of the Negev, Israel. References
From 2006 to 2008, he was with Telkoor
Power Supplies Ltd., where he was involved [1] Cadence Design Systems, Inc..
in design of military power supplies. His current research interests [2] PSIM user manual, Powersim Inc., Version 8.0, 2009.
include dc-dc converters, rectifiers, modeling and simulation, soft [3] P. O. Lauritzen, and C. L. Ma, “A simple diode model with reverse
switching techniques and energy harvesting. recovery,” IEEE Trans. Power Electron., vol. 6, no. 2, pp. 188–191,
Apr. 1991.
Shmuel (Sam) Ben-Yaakov (M’87–SM’08–F’10) was born in Tel [4] H. Levy, I. Zafrany, G. Ivensky, and S. Ben-Yaakov, “Analysis and
Aviv, Israel, in 1939. He received the B.Sc. degree in electrical Evaluation of a Lossless Turn-On Snubber,” in Proc. IEEE Applied
engineering from the Technion, Haifa, Israel, in 1961, and the M.S. Power Electronics Conf., 1997, pp. 757–763.

Applied power system

Potrebbero piacerti anche