Power Electronics 490

15 15.1 The forward converter

The basic forward converter, sometimes called a buck converter, is shown in figure
15.2a. The input voltage Ei is chopped by transistor T. When T is on, because the input
voltage Ei is greater than the load voltage vo, energy is transferred from the dc supply Ei
to L, C, and the load R. When T is turned off, stored energy in L is transferred via diode
Switched-mode and Resonant D to C and the load R.

dc-to-dc Power Supplies

A switched-mode power supply (smps) or switching regulator, efficiently converts a dc

voltage level to another dc voltage level, usually at power levels below a few kilowatts.
Shunt and series linear regulator power supplies dissipate much of their energy
across the regulating transistor, which operates in the linear mode. An smps achieves
regulation by varying the on to off time duty cycle of the switching element. This
minimises losses, irrespective of load conditions.
Figure 15.1 illustrates the basic principle of the ac-fed smps in which the ac mains
input is rectified, capacitively smoothed, and supplied to a high-frequency transistor
ac mains voltage
chopper. The chopped dc voltage is transformed, rectified, and smoothed to give the feed-back for unity
required dc output voltage. A high-frequency transformer is used if an isolated output input power factor

is required. The output voltage is sensed by a control circuit that adjusts the duty cycle
of the switching transistor in order to maintain a constant output voltage with respect to
load and input voltage variation. Alternatively, the chopper can be configured and Figure 15.1. Functional block diagram of a switched-mode power supply.
controlled such that the input current tracks a scaled version of the input ac supply
voltage, therein producing unity (or controllable) power factor I-V input conditions.
The switching frequency can be made much higher than the 50/60Hz line If all the stored energy in L is transferred to C and the load before T is turned back
frequency; then the filtering and transformer elements used can be made small, on, operation is termed discontinuous, since the inductor current has reached zero. If T
lightweight, low in cost, and efficient. is turned on before the current in L reaches zero, that is, if continuous current flows in
Depending on the requirements of the application, the dc-to-dc converter can be one L, operation is termed continuous.
of four basic converter types, namely Parts b and c respectively of figure 15.2 illustrate forward converter circuit current
and voltage waveforms for continuous and discontinuous conduction of L.
• forward For analysis it is assumed that components are lossless and the output voltage vo is
• flyback maintained constant because of the large magnitude of the capacitor C across the
• balanced output. The input voltage Ei is also assumed constant, such that Ei ≥ vo.
• resonant.
491 Switched-mode and Resonant dc Power Supplies Power Electronics 492

iL = i o ∧ ∨ Ei − vo
∆ iL = i L − i L = × tΤ (15.1)
L
When T is switched off for the remainder of the switching period, τ-tT, the freewheel
diode D conducts and -v0 is impressed across L. Thus, assuming continuous conduction
v
∆iL = o × (τ − tΤ ) (15.2)
L
Equating equations (15.1) and (15.2) gives
( Ei - vo ) tT = vo (τ - tT ) (15.3)
This expression shows that the inductor average voltage is zero, and after rearranging:
vo I i t
= = T =δ 0 ≤ δ ≤1 (15.4)
Ei I o τ
This equation shows that for a given input voltage, the output voltage is determined by
the transistor conduction duty cycle and the output is always less than the input
voltage. This confirms and validates the original analysis assumption that Ei ≥ vo. The
voltage transfer function is independent of circuit inductance L and capacitance C.
The inductor rms ripple current (and capacitor ripple current in this case) is given by
∆i 1 vo 1 Ei
iL r = L = (1- δ )τ = (1- δ ) δτ (15.5)
2 3 2 3 L 2 3 L
while the inductor total rms current is
2
 ½ ∆iL  1  ∧2 ∧ ∨ ∨ 
2

iL rms = I L2 + iL2r = I L2 +   = 3 i + i × i + iL 
L L (15.6)
 3  
L

The switch and diode average and rms currents are given by
IT = Ii = δ Io I Trms = δ iL rms
(15.7)
I D = I o − I i = (1 − δ ) I o I Drms = 1 − δ iL rms
If the average inductor current, hence output current, is I L , then the maximum and
minimum inductor current levels are given by
v
i = I L + ½ ∆iL = I o + ½ o (1- δ )τ
∧
L
L
(15.8)
Figure 15.2. Non-isolated forward converter (buck converter) where v0 ≤ E1:  1 1− δ 
(a) circuit diagram; (b) waveforms for continuous output current; and (c) waveforms = vo  + 
for discontinuous output current. R 2f L
and
15.1.1 Continuous inductor current v
i = I L − ½ ∆iL = I o − ½ o (1- δ )τ
∨
L
L
The inductor current is analysed first when the switch is on, then when the switch is (15.9)
off. When transistor T is turned on for period tT, the difference between the supply  1 1− δ 
= vo  − 
voltage Ei and the output voltage v0 is impressed across L. From V=Ldi/dt=L∆i/∆t, the R 2 f L
current change through the inductor will be respectively, where ∆iL is given by equation (15.1) or (15.2). The average output
493 Switched-mode and Resonant dc Power Supplies Power Electronics 494

current is I L = ½  i + i  = I o = vo / R . The output power is therefore vo2 / R . Circuit

∧ ∨
that is
  conduction are shown in figure 15.2b.
L L

waveforms for continuous vo δ

= 0 ≤δ <1 (15.16)
Ei 1 − t x
Switch utilisation ratio τ
The switch utilisation ratio, SUR, is a measure of how fully a switching device’s power This voltage transfer function form may not be particularly useful since the dead time tx
handling capabilities are utilised in any switching application. The ratio is defined as is not expressed in term of circuit parameters. Accordingly, from equation (15.15)
P (E − v )
SUR = out (15.10) ∧
i L = i o tT (15.17)
p VT I T L
where p is the number of power switches in the circuit; p=1 for the forward converter. and from the input current waveform in figure 15.2c:
The switch maximum instantaneous voltage and current are VT and I T respectively. ∧ t
Ii = ½ i L × T (15.18)
As shown in figure 15.2b, the maximum switch voltage supported∧ in the off-state is Ei, τ
while the maximum current is the maximum inductor current i L which is given by ∧
Eliminating i L yields
equation (15.8). If the inductance L is large such that the ripple current is small, the
peak inductor current is approximated by the average inductor current I T ≈ I L = I o , 2Ii vo τδ Ei
= (1 − ) (15.19)
that is δ Ei L
vo I o v that is
SUR = = o =δ (15.11) vo
1× Ei × I o Ei 2 LI
= 1− 2 i (15.20)
which assumes continuous inductor current. This result shows that the higher the duty Ei δ τ Ei
cycle, that is the closer the output voltage vo is to the input voltage Ei, the better the Assuming power-in equals power-out, that is, Ei I i = vo I o = vo I L , the input average
switch I-V ratings are utilised. current can be eliminated, and after re-arranging yields:
vo 1 1
15.1.2 Discontinuous inductor current = = (15.21)
Ei
1 + 22LI o 1 + 22LI i
The onset of discontinuous inductor ∨operation occurs when the minimum inductor δ τ Ei δ τ vo
∨
current i L , reaches zero. That is, with i L = 0 in equation (15.9), the last equality At a low output current or high input voltage, there is a likelihood of discontinuous
1 (1 − δ ) inductor conduction. To avoid discontinuous conduction, larger inductance values are
− =0 (15.12) needed, which worsen transient response. Alternatively, with extremely low on-state
R 2f L
duty cycles, a voltage-matching transformer can be used to increase δ. Once using a
relates circuit component values (R and L) and operating ∨
conditions (f and δ) at the transformer, any smps technique can be used to achieve the desired output voltage.
verge of discontinuous inductor current. Also, with i = 0 in equation (15.9)
L
Figures 15.2b and c show that the input current is always discontinuous.
I L = I o = ½ ∆iL (15.13)
which, after substituting equation (15.1) or equation (15.2), yields
(E − v ) Ei 15.1.3 Load conditions for discontinuous inductor current
I L = I o = i o τδ or τδ (1 − δ ) (15.14)
2L 2L
If the transistor on-time tT is reduced (or the load current is reduced), the discontinuous As the load current decreases, the inductor average current also decreases, but the
condition dead time tx is introduced as indicated in figure 15.2c. From equations (15.1) inductor ripple current magnitude is unchanged. If the ∨load resistance is increased
∨
and (15.2), with i = 0 , the output voltage transfer function is now derived as follows sufficiently, the bottom of the triangular inductor current, i L , eventual reduces to zero.
L
Any further increase in resistance causes discontinuous inductor current and the linear
∧ ( Ei − vo ) v voltage transfer function given by equation (15.4) is no longer valid and equations
iL = tT = o (τ − tT − t x ) (15.15) (15.16) and (15.20) are applicable. The critical load resistance for continuous inductor
L L
current is specified by
495 Switched-mode and Resonant dc Power Supplies Power Electronics 496

vo v 15.1.4i - fixed on-time tT, variable switching frequency fvar

Rcrit ≤ = o (15.22) The operating frequency fvar is varied while the switch-on time tT is maintained
I o ½ ∆iL
constant such that the ripple current remains unchanged. Operation is specified by
Substitution for vo from equation (15.2) and using the fact that I o = I L , yields equating the input energy and the output energy, thus maintaining a constant capacitor
v ∆i L L charge, hence output voltage. That is, equating energies
Rcrit ≤ o = (15.23)
I o I L (τ − tΤ ) v2 1
½ ∆iL Ei tT = o (15.27)
Eliminating ∆iL by substituting the limiting condition given by equation (15.13) gives R f var
vo ∆iL L 2ILL 2L Isolating the variable switching frequency fvar gives
Rcrit ≤ = = = (15.24) vo2 1
I o I L (τ − tΤ ) I L (τ − tΤ ) (τ − tΤ ) f var =
½ ∆iL Ei tT R
Divide throughout by τ and substituting δ = tT / τ yields
v 2L 2L 1
Rcrit ≤ o = = (15.25) f var = fs Rcrit ×
I o (τ − tΤ ) τ (1 − δ ) R (15.28)
1
The critical resistance can be expressed in a number of forms. By substituting the f var α
switching frequency ( f s = 1/ τ ) or the fundamental inductor reactance ( X L = 2π f s L ) R
That is, once discontinuous inductor current occurs, if the switching frequency is varied
the following forms result.
inversely with load resistance and the switch on-state period is maintained constant,
v 2L 2 fsL XL
Rcrit ≤ o = = = (Ω ) (15.26) output voltage regulation can be maintained.
I o τ (1 − δ ) (1 − δ ) π (1 − δ )
If the load resistance increases beyond Rcrit, the output voltage can no longer be Load resistance R is not a directly or readily measurable parameter for feedback
maintained with duty cycle control according to the voltage transfer function in proposes. Alternatively, since vo = Io R substitution for R in equation (15.28) gives
equation (15.4). Notice that equation (15.26) is in fact equation (15.12), re-arranged. R
f var = f s crit × Io
vo (15.29)
15.1.4 Control methods for discontinuous inductor current
f var α Io
Once the load current has reduced to the critical level as specified by equation (15.26), That is, for I o < ½∆iL or Io < vo / Rcrit , if tT remains constant and fvar is varied
the input energy is in excess of the load requirement. Open loop load voltage regulation proportionally with load current, then the required output voltage vo will be maintained.
control is lost and the capacitor C tends to overcharge.
Hardware approaches can be used to solve this problem 15.1.4ii - fixed switching frequency fs, variable on-time tTvar
• increase L thereby decreasing the inductor current ripple p-p magnitude The operating frequency fs remains fixed while the switch-on time tTvar is reduced,
• step-down transformer impedance matching to effectively reduce the apparent resulting in the ripple current being reduced. Operation is specified by equating the
load impedance input energy and the output energy as in equation (15.27), thus maintaining a constant
Two control approaches to maintain output voltage regulation when R > Rcrit are capacitor charge, hence voltage. That is
• vary the switching frequency fs, maintaining the switch on-time tT constant so v2 1
that ∆iL is fixed or ½ ∆iL Ei tT var = o (15.30)
R fs
• reduce the switch on-time tT , but maintain a constant switching frequency fs,
Isolating the variable on-time tTvar yields
thereby reducing ∆iL.
If a fixed switching frequency is desired for all modes of operation, then reduced on- vo2 1
tT var =
time control, using output voltage feedback, is preferred. If a fixed on-time mode of ½ ∆iL Ei f s R
control is used, then the output voltage is control by varying inversely the frequency Substituting ∆iL from equation (15.2) gives
with output voltage.
497 Switched-mode and Resonant dc Power Supplies Power Electronics 498

1 In complying with output voltage ripple requirements, from this equation, the
tT var = tT Rcrit × switching frequency fs=1/τ must be much higher that the cut-off frequency given by the
R
(15.31) forward converter low-pass, second-order LC output filter, fc=1/2π√LC.
1 ESR: The equivalent series resistor voltage follows the ripple current, that is, it swings
tT var α
R linearly about
That is, once discontinuous inductor current commences, if the switch on-time is varied VESR = ±½ ∆i × RESR (15.36)
inversely to the square root of the load resistance, maintaining the switching frequency
constant, regulation of the output voltage can be maintained.
Again, load resistance R is not a directly or readily measurable parameter for feedback
proposes and substitution of vo / Io for R in equation (15.31) gives
Rcrit
tT var = tT × Io ∆i o iC
vo (15.32)
tT var α Io
That is, if fs is fixed and tT is reduced proportionally to Io , when I o < ½∆iL
or Io < vo / Rcrit , then the required output voltage magnitude vo will be maintained. τ∆i/8C o VC

15.1.5 Output ripple voltage

Three components contribute to the output voltage ripple o VESR vc = v o

∆i R
• Ripple charging of the ideal capacitor
• Capacitor equivalent series resistance, ESR
• Capacitor equivalent series inductance, ESL
L∆i/ton
The capacitor inductance and resistance parasitic series component values decrease as
the quality of the capacitor increases. The output ripple voltage is the vectorial o VESL
summation of the three components that are shown in figure 15.3 for the forward L∆i/toff
converter.
Ideal Capacitor: The ripple voltage for a capacitor is defined as
Figure 15.3. Forward converter, three output ripple components, showing:
∆vC = 1
C ∫ i dt left -voltage components; centre – waveforms; and right - capacitor model.

Figures 15.2 and 15.3 show that for continuous inductor current, the inductor current
ESL: The equivalent series inductor voltage is derived from v = Ldi / dt , that is when
which is the output current, swings by ∆i around the average output current, I o , thus
the switch is on
∆vC = 1
C ∫ i dt = ½ 1 ∆i τ
C 2 2
(15.33) V + = L∆i / ton = L∆i / δτ
ESL
(15.37)
Substituting for ∆iL from equation (15.2) When the switch is off
V − = − L∆i / toff = − L∆i / (1 − δ )τ (15.38)
∆vC = 1
C ∫ i dt = ½ C1 ∆2i τ2 = 1 1 vo
8 C L × (τ − tΤ )τ (15.34)
The total ripple voltage is
ESL

If ESR and ESL are ignored, after rearranging, equation (15.34) gives the percentage
voltage ripple (peak to peak) in the output voltage ∆vo = ∆vC + VESR + VESL (15.39)
∆vC ∆vo 1 1 ½ Forming a time domain solution for each component, then differentiating, gives a
= 8 LC × (1 − δ )τ 2 = ½π 2 (1 − δ )  c 
f
= (15.35) maximum ripple when
vo vo  f  t = 2CRESR (1 − δ )
s
(15.40)
499 Switched-mode and Resonant dc Power Supplies Power Electronics 500

This expression is independent of the equivalent series inductance, which is expected whence the transistor on-time is 25µs and the diode conducts for 75µs.
since it is constant during each state. If dominant, the inductor will affect the output v 48V
voltage ripple at the switch turn-on and turn-off instants. ii. The average load current is I o = o = = 48A = I L
R 1Ω
From power-in equals power-out, the average input current is
Example 15.1: Buck (step-down forward) converter I i = vo I o / Ei = 48V×48A/192V = 12A
The step-down converter in figure 15.2a operates at a switching frequency of 10 kHz.
The output voltage is to be fixed at 48 V dc across a 1 Ω resistive load. If the input iii. From equation (15.1) (or equation (15.2)) the inductor peak-to-peak ripple current
voltage Ei =192 V and the choke L = 200µH: is
E −v 192V-48V
∆iL = i o × tΤ = ×25µs = 18A
i. calculate the switch T on-time duty cycle δ and switch on-time tT. L 200µH
ii. calculate the average load current I o , hence average input current I i . From part ii, the average inductor current is the average output current, 48A. The
iii. draw accurate waveforms for required circuit voltage and current waveforms are shown in the following figure.
• the voltage across, and the current through L; vL and iL The circuit waveforms show that the maximum switch voltage and current are 192V
• the capacitor current, ic and 57A respectively. The switch utilising ratio is given by equation (15.11), that is
• the switch and diode voltage and current; vT, vD, iT, iD. v2 48V 2
P
o
Hence calculate the switch utilisation ratio as defined by equation (15.11).
SUR = out = R = 1Ω ≡ 21%
iv. calculate the mean and rms current ratings of diode D, switch T and L. Ei × i o Ei × i o 192V × 57A
v. calculate the capacitor average and rms current, iCrms and output ripple voltage If the ripple current were assume small, the resulting SUR value of δ = 33% gives a
if the capacitor has an internal equivalent series resistance of 20mΩ (C = ∞). misleading under-estimate indication.
vi. calculate the maximum load resistance Rcrit before discontinuous inductor
current. Calculate the output voltage and inductor non-conduction period , tx,
when the load resistance is triple the critical resistance Rcrit. E i -v o
vii. if the maximum load resistance is 1Ω, calculate
• the value the inductance L can be reduced to be on the verge of
discontinuous inductor current and for that L
• the peak-to-peak ripple and rms, inductor and capacitor currents.
viii. Specify two control strategies for controlling the forward converter in a
discontinuous inductor current mode.
ix. Output ripple voltage hence percentage output ripple voltage, for C=1000µF
and an equivalent series inductance of ESL=0.5uH, assuming ESR = 0Ω.

Solution
i. From equation (15.4) the duty cycle δ is (V )
v 48V
δ= o = = ¼ = 25% 192V
Ei 192V V Tran V Diode
Also, from equation (15.4), for a 10kHz switching frequency, the switching period τ is
100µs and the transistor on-time tT is given by I cap V Diode V Tran

vo tT 48V t 0 18 A
= = = T 25 125
Ei τ 192V 100µs Figure: Example 15.1
501 Switched-mode and Resonant dc Power Supplies Power Electronics 502

iv. Current iD through diode D is shown on the inductor current waveform. The vo rms = iCrms × RCesr
average diode current is
= 5.2A rms×20mΩ = 104mV rms
τ − tT
ID = × I L = (1 − δ ) × I L = (1 − ¼)×48A = 36A
τ
vi. Critical load resistance is given by equation (15.26), namely
The rms diode current is given by
v 2L
1 τ −t ∧ ∆iL 2 1 75µs 18A 2 Rcrit ≤ o =
τ ∫0 100µs ∫ 0
T
iDrms = (i L − t ) dt = (57A- t ) dt Io τ (1 − δ )
τ − tT
. .

75µs
2×200µH
= 41.8A = = 16/3Ω
100µs × (1-¼)
Current iT through the switch T is shown on the inductor current waveform. The
average switch current is = 5 1 3 Ω when I o = 9A
t Alternatively, the critical load current is 9A (½∆iL), thus from the equation
I T = T I L = δ I L = ¼×48A = 12A
τ immediately above, the load resistance must not be greater than vo / I o = 48V/9A=5⅓Ω,
Alternatively, from power-in equals power-out if the inductor current is to be continuous.
I T = I i = vo I o / Ei = 48V×48A/192V = 12A When the load resistance is tripled to 16Ω the output voltage is given by equation
(15.20), which is shown normalised in table 15.2. That is
The transistor rms current is given by
1 t ∨ ∆iL 2 1 18A 2  2 δ 2 Rτ ¼ 2 × 16Ω × 100µs 1
25µs
vo = Ei k  −1 + 1 +  where k = = = 8 thus
τ ∫ 0 ( i + tT t ) dt = 100µs ∫ 0 (39A+ 25µs t ) dt
T
iTrms = k 4L 4 × 200µH

.
L .

= 24.1A  2
vo = 192V × 18 ×  −1 + 1 +  = 75V
1
The mean inductor current is the mean output current, Io = I L = 48 A .  8

The inductor rms current is given by equation (15.6), that is The inductor current is zero for an interval of the 100µs switching period, and the time
2 2 is given by the appropriate normalised expression involving tx for the forward converter
½ ∆iL
I L rms = I L2 +   = 48A 2 +  ½ × 18A  = 48.3A
   in table 15.2 or by equation (15.16), which when re-arranged to isolate tx becomes
 3  3
  
 δ  ¼ 
t x = τ 1 − = 100µs ×  1 − = 36µs
v. The average capacitor current I C is zero and the rms ripple current is given by vo   75V 
   50V 
1 tT ∆i τ − tT ∆i   Ei 
τ  ∫ 0
iCrms = ( − 1 2 ∆iL + L t ) 2 dt + ∫ ( 1 2 ∆iL − L
t ) 2 dt 
tT τ − tT
.

0

vii. The critical resistance formula given in equation (15.26) is valid for finding
1  25µs 18A 2 75µs 18A 2  critical inductance when inductance is made the subject of the equation, that is,
= .

100µs  ∫ 0
(-9A+
25µs
t ) dt + ∫ 0
(9A-
75µs
t ) dt 
 rearranging equation (15.26) gives
= 5.2A (= ∆iL / 2 3) Lcrit = ½ × R × (1 − δ ) × τ (H)
The capacitor voltage ripple (hence the output voltage ripple), is determined by the = ½×1Ω×(1-¼)×100µs = 37½µH
capacitor ripple current which is equal to the inductor ripple current, 18A p-p, that is This means the inductance can be reduced from 200µH with a 48A mean and 18A p-p
vo ripple = ∆iL × RCesr ripple current, to 37½µH with the same 48A mean plus a superimposed 96A p-p ripple
current. The rms capacitor current is given by
= 18A×20mΩ = 360mV p-p
iCrms = ∆iL / 2 3
and the rms output voltage ripple is
=96A/2 3 = 27.2A rms
503 Switched-mode and Resonant dc Power Supplies Power Electronics 504

The inductor rms current requires the following integration +

V ESL = L∆i / δτ
1  tT ∨ ∆iL 2 τ −tT ∧ ∆iL 2 
iLrms = ∫
(i +
τ  0 L tT
. t ) dt + (i L −
τ − tT ∫
t ) dt  =0.5µH×18A/0.25×100µs = 360mV
 = − L∆i / (1 − δ )τ
0
−
V ESL
1  25µs 96A 2 75µs 96A 2 
= . ×
100µs  ∫ 0
(0 +
25µs
t ) dt + ∫ 0
(96A -
75µs
t ) dt 

= - 0.5µH×18A/(1 - 0.25)×100µs = -120mV
Time domain summation of the capacitor and ESL inductor voltages show that the
= 96/ 3 = 55.4 A rms peak to peak output voltage swing is determined by the ESL inductor, giving
or from equation (15.6) ∆vo = VESL
+
− VESL
−

= 360mV + 120mV = 480mV

iLrms = I L2 + iL2ripple
The percentage ripple in the output voltage is 480mV/48V = 1%.
= 48 + (96 / 2 3)
2 2

♣
= 55.4 A rms
15.1.6 Underlying mechanisms of the forward converter
viii. For R >16/3Ω, or I o < 9A , equations (15.29) or (15.32) can be used to develop a
suitable control strategy. The inductor current is pivotal to the analysis and understanding of any smps. For
(a) From equation (15.29), using a variable switching frequency of less than analysis, the smps internal and external electrical conditions are in steady-state on a
10kHz, cycle-by-cycle basis and the input power is equal to the output power.
R 5 13 Ω The first concept to appreciate is that the net capacitor charge change is zero over each
f var = f s crit Io = 10kHz Io
vo 48V switching cycle. That is, the average capacitor current is zero:
1 t +τ
f var
10
=
× Io kHz Ic = ∫ic ( t ) dt = 0
9 τ t

(b) From equation (15.32), maintaining a fixed switching frequency of 10kHz, the In so being, the output capacitor provides any load current deficit and stores any load
on-time duty cycle is reduced for I o < 9A according to current surplus associated with the inductor current within each complete cycle. Thus,
the capacitor is a temporary storage component where the capacitor voltage is fixed on
Rcrit 5 13 Ω a cycle-by-cycle basis, and because of its large capacitance does not vary significantly
tT var = tT Io = 25µs Io
vo 48V within a cycle.
25 The second concept involved is that the average inductor voltage is zero. Based on
tT var = × Io µs v = L di / dt the equal area criteria in chapter 11.1.1i
3
1 t +τ
∫
it +τ − it = vL ( t ) dt = 0 since it +τ = it in steady-state
ix. From equation (15.33) the output ripple voltage due the pure capacitor is given by L t
∆i τ Thus the average inductor voltage is zero:
∆vC = 1 t +τ
8C V L = ∫ vL ( t ) dt = 0
τ t
18A × 100µs
= = 225mV p-p The most enlightening way to appreciate the operating mechanisms is to consider how
8 × 1000µF the inductor current varies with load resistance R and inductance L. The figure 15.4
The voltage produced because of the equivalent series 0.5 µH inductance is shows the inductor current associated with the various parts of example 15.1.
For continuous inductor current operation, the two necessary and sufficient equations
are Io=vo /R and equation (15.2). Since the duty cycle and on-time are fixed for a given
output voltage requirement, equation (15.2) can be simplified to show that the ripple
 505 Switched-mode and Resonant dc Power Supplies Power Electronics 506

current is inversely proportional to inductance, as follows Equation (15.42) predicts that the average inductor current is inversely proportional to
v the load resistance, as shown in figure 15.4a. As the load is varied, the triangular
∆iL = o × (τ − tΤ )
L inductor current moves vertically, but importantly the peak-to-peak ripple current is
(15.41) constant, that is the ripple current is independent of load. As the load current is
1
∆iL α progressively decreased, by increasing R, the peak-to-peak current is unchanged; the
L
Since the average inductor current is equal to the load current, then the average inductor minimum current eventually reduces to zero, and discontinuous inductor
inductor current is inversely proportional to the load resistance, that is current operation occurs.
Equation (15.41) indicates that the inductor ripple current is inversely proportional to
I L = I o = vo / R
inductance, as shown in figure 15.4b. As the inductance is varied the ripple current
1 (15.42) varies inversely, but importantly the average current is constant, and specifically the
IL α
R average current value is not related to inductance L and is solely determined by the
load current, vo /R. As the inductance decreases the magnitude of the ripple current
IL IL Rload L IL ILp-p increases, the average is unchanged, and the minimum inductor current eventually
reaches zero and discontinuous inductor current operation results.

L
15.2 Flyback converters

asing
96A
72A Ω Flyback converters store energy in an inductor, termed ‘choke’, and transfer that
energy to the load storage capacitor such that output voltage magnitudes in excess of

decre
the input voltage are attained. Flyback converters are alternatively known as ringing
decreasing R LOAD

choke converters. Two versions of the flyback converter are possible

36A

• The step-up voltage flyback converter, called the boost converter, where no
48A 200µH 18A output voltage polarity inversion occurs.
1Ω
• The step-up/step-down voltage flyback converter, called the buck-boost
100µH
converter, where output voltage polarity inversion occurs.
37½µH
15.3 The boost converter

24A 2Ω
The boost converter transforms a dc voltage input to a dc voltage output that is greater
in magnitude but has the same polarity as the input. The basic circuit configuration is
shown in figure 15.5a. It will be seen that when the transistor is off, the output
verge of capacitor is charged to the input voltage Ei. Inherently, the output voltage vo can never
9A 16/3Ω discontinuous be less than the input voltage level.
inductor current
When the transistor is turned on, the supply voltage Ei is applied across the inductor
δ 1-δ δ 1-δ L and the diode D is reverse-biased by the output voltage vo. Energy is transferred from
0 25µs 100µs t 0 25µs 100µs t the supply to L and when the transistor is turned off this energy is transferred to the
load and output capacitor. While the inductor is transferring its stored energy into C,
(a) (b) energy is also being provided from the input source.
Figure 15.4. Forward converter (buck converter) operational mechanisms showing The output current is always discontinuous, but the input current can be either
that: (a) the average inductor current is inversely proportional to load resistance R continuous or discontinuous. For analysis, we assume vo>Ei and a constant input and
and (b) the inductor ripple current magnitude is inversely proportional to inductance L.
output voltage. Inductor currents are then linear and vary according to v = L di/dt.
507 Switched-mode and Resonant dc Power Supplies Power Electronics 508

15.3.1 Continuous inductor current that is, after rearranging, the voltage transfer function is given by
vo I i 1
The circuit voltage and current waveforms for continuous inductor conduction are = = (15.44)
Ei I o 1 − δ
shown in figure 15.5b. The inductor current excursion, which is the input current
excursion, during the switch on-time tT and switch off-time τ- tT , is given by where δ = tT /τ and tT is the transistor
∧
on-time. The maximum inductor current, which is
(v - E ) E the maximum input current, i L , using equation (15.43), is given by
∆iL = o i (τ − tT ) = i tT (15.43) ∧ Et
L L i = I L + ½ ∆iL = I i + ½ i T
L
L
(15.45)
ii = iL
Io vo  1 (1- δ ) δτ 
= + ½ (1- δ ) δτ = vo  + 
1− δ L  (1 − δ ) R 2 L 
∨
while the minimum inductor current, i is given by L

∨ Et
i == I L − ½ ∆iL = I i − ½ i T
L
L
(15.46)
I v  1 (1- δ ) δτ 
= o − ½ o (1- δ ) δτ = vo  − 
 (1 − δ ) R
1− δ L 2 L 
∨
For continuous conduction i L ≥ 0 , that is, from equation (15.46)
Et v (1 − δ )tT
IL ≥ ½ i T = ½ o (15.47)
L L
The inductor rms ripple current (and input ripple current in this case) is given by
∆i 1 vo
iLr = L = (1- δ ) δτ (15.48)
2 3 2 3 L
The harmonic components in the input current are
2 Eiτ sin nδπ 2 voτ sin nδπ
I in = = (15.49)
2π 2 n 2 (1 − δ ) L 2π 2 n 2 L
while the inductor total rms current is
2
 ½ ∆iL  1  ∧2 ∧ ∨ ∨ 
2

iLrms = I L2 + iL2r = I L2 +  = i L + iL × i L + iL  (15.50)

 3  3  
The switch and diode average and rms currents are given by
IT = Ii − Io = δ Ii = δ I L I Trms = δ iL rms
(15.51)
I D = (1 − δ ) I i = I o I Drms = 1 − δ iL rms

Switch utilisation ratio

The switch utilisation ratio, SUR, is a measure of how fully a switching device’s power
handling capabilities are utilised in any switching application. The ratio is defined as
Figure 15.5. Non-isolated, step-up, flyback converter (boost converter) where v0 ≥E1:
(a) circuit diagram; (b) waveforms for continuous input current; and
(c) waveforms for discontinuous input current.
509 Switched-mode and Resonant dc Power Supplies Power Electronics 510

Pout conduction are shown in figure 15.5c.

SUR = (15.52)
p VT I T The onset
∨
of discontinuous inductor ∨operation occurs when the minimum inductor
current i L , reaches zero. That is, with i L = 0 in equation (15.46), the last equality
where p is the number of power switches in the circuit; p=1 for the boost converter.
The switch maximum instantaneous voltage and current are VT and I T respectively. 1 (1- δ ) δτ
− =0 (15.56)
As shown in figure 15.5b, the maximum switch voltage supported∧ in the off-state is vo, (1 − δ ) R 2L
while the maximum current is the maximum inductor current i which is given by
L relates circuit component values (R and L) and operating conditions (f and δ) at the
equation (15.45). If the inductance L is large such that the ripple current is small, the verge ∨of discontinuous inductor current.
peak inductor current is approximated by the average inductor current such that With i = 0 , the output voltage is determined as follows
L
I T ≈ I L = I o /1 − δ , that is ∧ Et ( v − Ei )
i = iT = o (τ − tT − t x ) (15.57)
vo I o L
L L
SUR = = 1− δ (15.53)
vo × I o yielding
1−δ t
which assumes continuous inductor current. This result shows that the lower the duty 1 − τx
vo
cycle, that is the closer the step-up voltage vo is to the input voltage Ei, the better the = (15.58)
Ei 1 − t x − δ
switch I-V ratings are utilised. τ
15.3.2 Discontinuous capacitor charging current in the switch off-state Alternatively, using
∧ Ei tT
iL =
It is possible that the input current (inductor current) falls below the output (resistor) L
current during a part of the cycle when the switch is off and the inductor is transferring and
energy to the output circuit. Under such conditions, towards the end of the off period, I L − I o = ½δ i
∧

part of the load current requirement is provided by the capacitor even though this is the
L

period during which its charge is replenished by inductor energy. The circuit yields
independent transfer function in equation (15.44) remains valid. This discontinuous 2 Ei tT
( I L − Io ) =
charging condition occurs when the minimum inductor current and the output current δ L
are equal. That is Assuming power-in equals power-out
∨
2 v Et
IL− Io ≤ 0 I ( o − 1) = i T
δ o Ei L
I L − ½ ∆iL − I o ≤ 0 (15.54)
that is
Io E δτ vo Et δ2 v t δ2
−½ i − Io ≤ 0 = 1+ i T = 1+ o T (15.59)
1− δ L Ei 2 LI o 2 LI i
which yields
or
2L vo 1
δ ≤ 1− (15.55) = (15.60)
τR Ei Et δ2
1− i T
2 LI i
15.3.3 Discontinuous inductor current
On the verge of discontinuous conduction, these equations can be rearranged to give
If the inequality in equation (15.47) is not satisfied, the input current, which is also the E
I o = i τδ (1 − δ ) (15.61)
inductor current, reaches zero and discontinuous conduction occurs during the switch 2L
off period. Various circuit voltage and current waveforms for discontinuous inductor At a low output current or low input voltage, there is a likelihood of discontinuous
511 Switched-mode and Resonant dc Power Supplies Power Electronics 512

inductor conduction. To avoid discontinuous conduction, larger inductance values are Hardware approaches can be used to solve this problem
needed, which worsen transient response. Alternatively, with extremely high on-state • increase L thereby decreasing the inductor current ripple p-p magnitude
duty cycles, (because of a low input voltage Ei) a voltage-matching step-up transformer • step-down transformer impedance matching to effectively reduce the apparent
can be used to decrease δ. Figures 15.5b and c show that the output current is always load impedance
discontinuous. Two control approaches to maintain output voltage regulation when R > Rcrit are
• vary the switching frequency fs, maintaining the switch on-time tT constant so
15.3.4 Load conditions for discontinuous inductor current that ∆iL is fixed or
• reduce the switch on-time tT , but maintain a constant switching frequency fs,
As the load current decreases, the inductor average current also decreases, but the thereby reducing ∆iL.
inductor ripple current magnitude is unchanged. If the load ∨
resistance is increased If a fixed switching frequency is desired for all modes of operation, then reduced on-
sufficiently, the bottom of the triangular inductor current, i L , eventually reduces to time control, using output voltage feedback, is preferred. If a fixed on-time mode of
zero. Any further increase in load resistance causes discontinuous inductor current and control is used, then the output voltage is control by inversely varying the frequency
the voltage transfer function given by equation (15.44) is no longer valid and equations with output voltage.
(15.58) and (15.59) are applicable. The critical load resistance for continuous inductor
current is specified by 15.3.5i - fixed on-time tT, variable switching frequency fvar
v The operating frequency fvar is varied while the switch-on time tT is maintained
Rcrit ≤ o (15.62)
Io constant such that the ripple current remains unchanged. Operation is specified by
Eliminating the output current by using the fact that power-in equals power-out and equating the input energy and the output energy, thus maintaining a constant capacitor
charge, hence output voltage. That is, equating energies
I i = I L , yields
v2 1
vo v2 ½ ∆iL Eiτ = o (15.66)
Rcrit ≤ = o (15.63) R f var
I o Ei I L
Isolating the variable switching frequency fvar gives
Using I L = ½ ∆iL then substituting with the right hand equality of equation (15.43), vo2 1
halved, gives f var =
½ ∆iL Eiτ R
v v2 v2 2L 2L 1
Rcrit ≤ o = o = o 2 = (15.64)
f var = fs Rcrit ×
I o Ei I L Ei tT τδ (1 − δ ) 2
R (15.67)
The critical resistance can be expressed in a number of forms. By substituting the 1
switching frequency ( f s = 1/ τ ) or the fundamental inductor reactance ( X L = 2π f s L ) f var α
R
the following forms result. Load resistance R is not a directly or readily measurable parameter for feedback
v 2L 2 fsL XL proposes. Alternatively, since vo = Io R , substitution for R in equation (15.67) gives
Rcrit ≤ o = = = (Ω ) (15.65)
I o τδ (1 − δ ) 2 δ (1 − δ ) 2 πδ (1 − δ ) 2 R
f var = f s crit × Io
If the load resistance increases beyond Rcrit, the output voltage can no longer be vo (15.68)
maintained with duty cycle control according to the voltage transfer function in
f var α Io
equation (15.44). Equation (15.65) is equation (15.56), re-arranged.
That is, for discontinuous inductor current, namely I i < ½∆iL or Io < vo / Rcrit , if the
15.3.5 Control methods for discontinuous inductor current switch on-state period tT remains constant and fvar is either varied proportionally with
load current or varied inversely with load resistance, then the required output voltage vo
Once the load current has reduced to the critical level as specified by equation (15.65), will be maintained.
the input energy is in excess of the load requirement. Open loop load voltage regulation
control is lost and the capacitor C tends to overcharge, thereby increasing vo.
513 Switched-mode and Resonant dc Power Supplies Power Electronics 514

15.3.5ii - fixed switching frequency fs, variable on-time tTvar Rearranging gives the percentage voltage ripple (peak to peak) in the output voltage
The operating frequency fs remains fixed while the switch-on time tTvar is reduced such ∆vo δτ
that the ripple current can be reduced. Operation is specified by equating the input = (15.72)
vo RC
energy and the output energy as in equation (15.66), thus maintaining a constant
The capacitor equivalent series resistance and inductance can be account for, as with
capacitor charge, hence voltage. That is
the forward converter, 15.1.4. When the switch conducts, the output current is constant
v2 1 and is provided from the capacitor. No ESL voltage effects result during this constant
½ ∆iL Ei tT var = o (15.69)
R fs capacitor current portion of the switching cycle.
Isolating the variable on-time tTvar gives
vo2 1 Example 15.2: Boost (step-up flyback) converter
tT var =
½ ∆iL Ei f s R
The boost converter in figure 15.5 is to operate with a 50µs transistor fixed on-time in
Substituting ∆iL from equation (15.43) gives order to convert the 50 V input up to 75 V at the output. The inductor is 250µH and the
1 resistive load is 2.5Ω.
tT var = tT Rcrit ×
R
(15.70)
1 i. Calculate the switching frequency, hence transistor off-time, assuming continuous
tT var α
R inductor current.
Again, load resistance R is not a directly or readily measurable parameter for feedback ii. Calculate the mean input and output current.
proposes and substitution of vo / Io for R in equation (15.70) gives iii. Draw the inductor current, showing the minimum and maximum values.
iv. Calculate the capacitor rms ripple current.
Rcrit v. Derive general expressions relating the operating frequency to varying load
tT var = tT × Io
vo (15.71) resistance.
vi. At what load resistance does the instantaneous input current fall below the output
tT var α Io current.
That is, if the switching frequency fs is fixed and switch on-time tT is reduced
proportionally to Io or inversely to R , when discontinuous inductor current Solution
commences, namely I i < ½∆iL or Io < vo / Rcrit , then the required output voltage
magnitude vo will be maintained. i. From equation (15.44), which assumes continuous inductor current
vo 1 t
15.3.6 Output ripple voltage = where δ = T
Ei 1 − δ τ
The output ripple voltage is the capacitor ripple voltage. The ripple voltage for a that is
capacitor is defined as 75V 1 50µs 1
= where δ = = 3
50V 1 − δ τ
∆vo = 1
C ∫ i dt That is, τ = 150 µs or fs= 1/τ = 6.66 kHz, with a 100µs switch off-time.
Figure 15.5 shows that for continuous inductor current, the constant output current I o is
provided solely from the capacitor during the period ton when the switch is on, thus ii. The mean output current I o is given by
∆vo = 1
C ∫ i dt = 1
C
ton I o I o = vo / R = 75V/2.5Ω = 30A
From power transfer considerations
Substituting for I o = vo / R gives
I i = I L = vo I o / Ei = 75V×30A/50V = 45A
vo
∆vo = 1
C ∫ i dt = C1 t on Io = 1
C
ton
R iii. From v = L di/dt, the ripple current ∆iL = EitT /L = 50V x 50µs /250 µH = 10 A
515 Switched-mode and Resonant dc Power Supplies Power Electronics 516

For a load resistance of less than 22½ Ω, continuous inductor current flows and the
that is operating frequency is fixed at 6.66 kHz with δ = 1/3, that is
∧
i = I L + ½ ∆iL = 45A + ½×10A = 50A
L

∨
fs = 6.66 kHz for all R ≤ 22½ Ω
i = I L − ½ ∆iL = 45A - ½×10A = 40A
L

For load resistance greater than 22½ Ω, (< vo /Rcrit = 3⅓A), the energy input occurs in
Figure: Example 15.2a 150 µs burst whence from equation (15.66)
v2 1
½ ∆iL Ei × 150µs = o
R f var
that is
R 1 22½Ω 1
f var = crit =
τ R 150µs R
150
f var = kHz for R ≥ 22½Ω
R
iv. The capacitor current is derived by using Kirchhoff’s current law such that at any vi. The ±5A inductor ripple current is independent of the load, provided the critical
instant in time, the diode current, plus the capacitor current, plus the 30A constant load resistance is not exceeded. When the average inductor current (input current) is less
current into R, all sum to zero. than 5A more than the output current, the capacitor must provide load current not only
when the switch is on but also when the switch is off. The transition is given by
1 t 2 τ −t ∆iL ∧ 
τ  ∫ 0 I o dt + ∫ 0 (τ − tT t − i + I o ) dt 
T T
iCrms =.
L
2
equation (15.55), that is
  2L
1  50 µs 10A  δ ≤ 1−
100 µs
τR
150µs  ∫ 0
= .
 30A dt + ∫ (
2
t − 20A) 2 dt  = 21.3A
0 100µs  1 2×250µH
iC ≤1 -
3 150µs×R
This yields R ≥ 7½Ω and a load current of 10A. The average inductor current is 15A,
with a minimum value of 10A, the same as the load current. That is, for R < 7½Ω all
the load requirement is provided from the input inductor when the switch is off, with
excess energy charging the output capacitor. For R > 7½Ω insufficient energy is
available from the inductor to provide the load energy throughout the whole of the
period when the switch is off. The capacitor supplements the load requirement towards
the end of the off period. When R > 22½Ω (the critical resistance), discontinuous
Figure: Example 15.2b inductor current occurs, and the duty cycle dependent transfer function is no longer
v. The critical load resistance, Rcrit, produces an input current with ∆iL = 10 A ripple. valid.
Since the energy input equals the energy output ♣
½ ∆i × Ei × τ = vo × vo / Rcrit × τ
Example 15.3: Alternative boost (step-up flyback) converter
that is
2v 2 2×75V 2 The alternative boost converters (producing a dc supply either above Ei (left) or below
Rcrit = o = = 22½Ω
Ei ∆i 50V×10A 0V (right)) shown in the following figure are to operate under the same conditions as
Alternatively, equation (15.65) or equation (15.47) can be rearranged to give Rcrit. the boost converter in example 15.2, namely, with a 50µs transistor fixed on-time in
517 Switched-mode and Resonant dc Power Supplies Power Electronics 518

order to convert the 50 V input up to 75 V at the output. The energy transfer inductor is The critical load resistance for continuous inductor current is specified by Rcrit ≤ vo / I o .
250µH and the resistive load is 2.5Ω. By equating the capacitor net charge flow, the inductor current is related to the output
current by I L = I o /(1 − δ ) . At minimum inductor current, I L = ½ ∆iL and substituting
with ∆iL = Ei tT / L , gives
v vo vo vo 2L
io io Rcrit ≤ o = = = =
vC = vo /R = vo /R I o (1 − δ ) I L (1 − δ )½ ∆iL (1 − δ )½ Ei tT / L τδ (1 − δ ) 2
Thus for a given energy throughput, some energy is provided from the supply to the
load when providing the inductor energy, hence the discontinuous inductor current
R threshold occurs at the same load level for each boost converter.
L
ii. Since the boost circuits have the same ac equivalent circuit, the inductor and
vC capacitor, currents and voltages would be expected to be the same for each circuit, as
Figure:
shown by the waveforms in example 15.2. Consequently, the switch and diode voltages
Example 15.3
- circuits and currents are also the same for each boost converter.
The two principal differences are the supply current and the capacitor voltage rating.
i. Derive the voltage transfer ratio and critical resistance expression for the The capacitor voltage rating for the alternative boost converter is vo - Ei as opposed to
alternative boost converter, hence showing the control performance is identical to vo for the convention converter.
the boost converter shown in figure 15.5. The supply current for the alternative converter is discontinuous, as shown in the
ii. By considering circuit voltage and current waveforms, identify how the two boost following waveforms. This will negate the desirable continuous current feature
converters differ from the conventional boost circuit in figure 15.5. exploited in boost converters that are controlled so as to produce sinusoidal input
current.
iC
Solution

i. Assuming non-zero, continuous inductor current, the inductor current excursion,

which for this boost converter is not the input current excursion, during the switch on-
time tT and switch off-time τ- tT , is given by
L∆iL = Ei tT = vC (τ − tT )
but vC = vo − Ei , thus substitution for vC gives
Ei tT = ( vo − Ei )(τ − tT )
and after rearranging,
vo I i 1  δ 
= = = 1+
Ei I o 1 − δ  1 − δ 
where δ = tT /τ and tT is the transistor on-time. This is the same voltage transfer I s u p p ly
(A )
function as for the conventional boost converter, equation (15.44). This result would 80
70
be expected since both converters have the same ac equivalent circuit. Similarly, the
critical resistance would be expected to be the same for each boost converter variation. 30
I lo a d
Examination of the switch on and off states shows that during the switch on-state,
energy is transfer to the load from the input supply, independent of switching action. t (µ s )
This mechanism is analogous to autotransformer action where the output current is due
to both transformer action and the input current being directed to the load. Figure: Example 15.3 - waveforms
♣
519 Switched-mode and Resonant dc Power Supplies Power Electronics 520

15.4 The buck-boost converter 15.4.1 Continuous choke current

The basic buck-boost flyback converter circuit is shown in figure l5.5a. When transistor Various circuit voltage and current waveforms for the buck-boost flyback converter
T is on, energy is transferred to the inductor. When the transistor turns off, inductor operating in a continuous inductor conduction mode are shown in figure 15.6b.
current is forced through the diode. Energy stored in L is transferred to C and the load Assuming a constant input and output voltage, the change in inductor current is given
R. This transfer action results in an output voltage of opposite polarity to that of the by
input. Neither the input nor the output current is continuous, although the inductor E −v
current may be continuous or discontinuous. ∆iL = i tT = o (τ − tT ) (15.73)
L L
thus
vo I i δ
= =− (15.74)
Ei I o 1−δ
where δ = tT /τ. For δ<½ the output magnitude is less than the input voltage magnitude,
while for δ > ½ the output is greater in magnitude than the input.
The maximum and minimum inductor current is given by
I v  1 (1- δ )τ 
i = o + ½ o (1- δ )τ = vo 
∧
+ (15.75)
1− δ
L
L  (1 − δ ) R 2 L 

and
∨ I v  1 (1- δ )τ 
i = o − ½ o (1- δ )τ = vo  − (15.76)
1− δ
L
L  (1 − δ ) R 2 L 

The inductor rms ripple current (and input ripple current in this case) is given by
∆i 1 vo
iLr = L = (1- δ ) δτ (15.77)
2 3 2 3 L
while the inductor total rms current is
2
 ½ ∆iL  1  ∧2 ∧ ∨ ∨ 
2

iLrms = I L2 + iL2r = I L2 +  = i L + i L × i L + iL  (15.78)

 3  3  
The switch and diode average and rms currents are given by
IT = Ii = δ I L ITrms = δ iL rms
(15.79)
I D = (1 − δ ) I L = I o I Drms = 1 − δ iL rms

Switch utilisation ratio

The switch utilisation ratio, SUR, is a measure of how fully a switching device’s power
handling capabilities are utilised in any switching application. The ratio is defined as
P
SUR = out (15.80)
p VT I T

Figure 15.6. Non-isolated, step up/down flyback converter (buck-boost converter) where p is the number of power switches in the circuit; p=1 for the buck-boost
where vo ≤ 0: (a) circuit diagram; (b) waveforms for continuous inductor current; and converter. The switch maximum instantaneous voltage and current are VT and
(c) discontinuous inductor current waveforms.
521 Switched-mode and Resonant dc Power Supplies Power Electronics 522

I T respectively. As shown in figure 15.6b, the maximum switch voltage supported in The change from continuous to discontinuous inductor current conduction occurs when
the
∧
off-state is Ei + vo, while the maximum current is the maximum inductor current ∧
I L = ½ i L = ½ ∆iL (15.85)
i L which is given by equation (15.75). If the inductance L is large such that the ripple
current is small, the peak inductor current is approximated by the average inductor ∧

current which yields I T ≈ I L = I o /1 − δ , that is where from equation (15.73) i L = vo (τ − tT ) / L

vo I o
SUR = = δ (1 − δ ) (15.81) The circuit waveforms for discontinuous conduction are shown in figure 15.6c. The
( Ei + vo ) × I o /1 − δ output voltage for discontinuous conduction is evaluated from
which assumes continuous inductor current. This result shows that the closer the output ∧ E v
i = i t = − o (τ − tT − t x )
L (15.86)
voltage vo is in magnitude to the input voltage Ei, that is δ = ½, the better the switch I-V L L
ratings are utilised. which yields
vo δ
=− (15.87)
15.4.2 Discontinuous capacitor charging current in the switch off-state Ei t
1 − δ − τx
It is possible that the inductor current falls below the output (resistor) current during a Alternatively, using equation (15.86) and
part of the cycle when the switch is off and the inductor is transferring energy to the ∧

output circuit. Under such conditions, towards the end of the off period, some of the I L = ½δ i L (15.88)
load current requirement is provided by the capacitor even though this is the period yields
during which its charge is replenished by inductor energy. The circuit independent Eiτδ
IL = (15.89)
transfer function in equation (15.74) remains valid. This discontinuous charging 2L
condition occurs when the minimum inductor current and the output current are equal. The inductor current is neither the input current nor the output current, but is comprised
That is of components of each of these currents. Examination of figure 15.6b, reveals that
∨
IL− Io ≤ 0 these currents are a proportion of the inductor current dependant on the duty cycle, and
that on the verge of discontinuous conduction:
I L − ½ ∆iL − I o ≤ 0 (15.82) ∧ ∧ ∧
I i = ½δ i L = δ I L and I o = ½(1- δ ) i L =(1- δ ) I L where i L = ∆iL = ½ I L
Io I R
− ½ o (1- δ )τ − I o ≤ 0 Thus using I i = δ I L equation (15.89) becomes
1− δ L
which yields Eiτδ 2
Ii = (15.90)
2 2L
L  L 
δ ≤ 1+ − 1 +  −1 (15.83) Assuming power-in equals power-out, that is Ei I i = vo I o
τR  τR
vo Eiτδ 2 voτδ 2 τR
= = =δ (15.91)
15.4.3 Discontinuous choke current Ei 2 LI o 2 LI i 2L
On the verge of discontinuous conduction, these equations can be rearranged to give
The onset of discontinuous inductor ∨operation occurs when the minimum inductor E
∨
current i L , reaches zero. That is, with i = 0 in equation (15.76), the last equality I o = i τδ (1 − δ ) (15.92)
L
2L
1 (1- δ )τ At a low output current or low input voltage there is a likelihood of discontinuous
− =0 (15.84)
(1 − δ ) R 2 L conduction. To avoid this condition, a larger inductance value is needed, which worsen
relates circuit component values (R and L) and operating conditions (f and δ) at the transient response. Alternatively, with extremely low on-state duty cycles, a voltage-
verge of discontinuous inductor current. matching transformer can be used to increase δ. Once using a transformer, any smps
technique can be used to achieve the desired output voltage. Figures 15.6b and c show
that both the input and output current are always discontinuous.
523 Switched-mode and Resonant dc Power Supplies Power Electronics 524

15.4.4 Load conditions for discontinuous inductor current control is used, then the output voltage is control by inversely varying the frequency
with output voltage.
As the load current decreases, the inductor average current also decreases, but the
inductor ripple current magnitude is unchanged. If the load ∨
resistance is increased 15.4.5i - fixed on-time tT, variable switching frequency fvar
sufficiently, the bottom of the triangular inductor current, i L , eventually reduces to The operating frequency fvar is varied while the switch-on time tT is maintained
zero. Any further increase in load resistance causes discontinuous inductor current and constant such that the ripple current remains unchanged. Operation is specified by
the voltage transfer function given by equation (15.74) is no longer valid and equations equating the input energy and the output energy, thus maintaining a constant capacitor
(15.86) and (15.91) are applicable. The critical load resistance for continuous inductor charge, hence output voltage. That is, equating energies
current is specified by v2 1
v ½ ∆iL Ei tT = o (15.96)
Rcrit ≤ o (15.93) R f var
Io Isolating the variable switching frequency fvar gives
∧
Substituting for, the average input current in terms of i and vo in terms of ∆iL from vo2 1
L f var =
equation (15.73) , yields ½ ∆iL Ei tT R
v 2L 1
Rcrit ≤ o = (15.94) = fs Rcrit ×
I o τ (1 − δ ) 2 R
By substituting the switching frequency ( f s = 1/ τ ) or the fundamental inductor 1
f var α (15.97)
reactance ( X L = 2π f s L ) the following critical resistance forms result. R
v 2L 2 fsL XL Load resistance R is not a directly or readily measurable parameter for feedback
Rcrit ≤ o = = = (Ω ) (15.95)
I o τ (1 − δ )2 (1 − δ ) 2 π (1 − δ )2 proposes. Alternatively, since vo = Io R , substitution for R in equation (15.97) gives
If the load resistance increases beyond Rcrit, the output voltage can no longer be R
f var = f s crit × Io
maintained with duty cycle control according to the voltage transfer function in vo (15.98)
equation (15.74). Equation (15.95) is equation (15.84), re-arranged. f var α Io
15.4.5 Control methods for discontinuous inductor current That is, for discontinuous inductor current, namely I L < ½ ∆iL or Io < vo / Rcrit , if the
switch on-state period tT remains constant and fvar is either varied proportionally with
Once the load current has reduced to the critical level as specified by equation (15.95), load current or varied inversely with load resistance, then the required output voltage vo
the input energy is in excess of the load requirement. Open loop load voltage regulation will be maintained.
control is lost and the capacitor C tends to overcharge.
Hardware approaches can be used to solve this problem 15.4.5ii - fixed switching frequency fs, variable on-time tTvar
• increase L thereby decreasing the inductor current ripple p-p magnitude The operating frequency fs remains fixed while the switch-on time tTvar is reduced such
• step-down transformer impedance matching to effectively reduce the apparent that the ripple current can be reduced. Operation is specified by equating the input
load impedance energy and the output energy as in equation (15.96), thus maintaining a constant
Two control approaches to maintain output voltage regulation when R > Rcrit are capacitor charge, hence voltage. That is
• vary the switching frequency fs, maintaining the switch on-time tT constant so v2 1
½ ∆iL Ei tT var = o (15.99)
that ∆iL is fixed or R fs
• reduce the switch on-time tT , but maintain a constant switching frequency fs, Isolating the variable on-time tTvar gives
thereby reducing ∆iL. vo2 1
If a fixed switching frequency is desired for all modes of operation, then reduced on- tT var =
½ ∆iL Ei f s R
time control, using output voltage feedback, is preferred. If a fixed on-time mode of
Substituting ∆iL from equation (15.73) gives
525 Switched-mode and Resonant dc Power Supplies Power Electronics 526

1 frequency and varying the transistor on-time tT. However, the output voltage diminishes
tT var = tT Rcrit ×
R while the transistor is on and increases when the transistor is off. This characteristic
(15.100) makes the converter difficult to control on a fixed frequency basis.
1
tT var α A simple approach to control the flyback regulator in the discontinuous mode is to fix
R the peak inductor current, which specifies a fixed diode conduction time, tD. Frequency
Again, load resistance R is not a directly or readily measurable parameter for feedback then varies directly with output current and transistor on-time varies inversely with
proposes and substitution of vo / Io for R in equation (15.70) gives input voltage.
Rcrit With discontinuous inductor conduction, the worst-case condition exists when the input
tT var = tT × Io voltage is low while the output current is at a maximum. Then the frequency is a
vo (15.101)
maximum and the dead time tx is zero because the transistor turns on as soon as the
tT var α Io diode stops conducting.
That is, if the switching frequency fs is fixed and switch on-time tT is reduced Given Worst case
proportionally to Io or inversely to R , when discontinuous inductor current Ei (min) Ei = Ei (min)
commences, namely I L < ½∆iL or Io < vo / Rcrit , then the required output voltage
Vo Io = Io(max)
magnitude vo will be maintained.
Alternatively the output voltage is related to the duty cycle by vo = δ Ei Rτ / 2 L . Io (max) tx = 0
f (max) ∆eo
∧
15.4.6 Output ripple voltage Assuming a fixed value of peak inductor current i i and output voltage vo, the
following equations are valid
The output ripple voltage is the capacitor ripple voltage. Ripple voltage for a capacitor ∧

is defined as Ei (min)tT = vo tD = i × L i (15.103)

τ (min) = 1/ f (max)
∆vo = 1
C ∫ i dt (15.104)
Equation (15.103) yields
Figure 15.6 shows that the constant output current I o is provided solely from the
1
capacitor during the period ton when the switch conducting, thus tD = (15.105)
vo
+ 1)
∆vo = 1
C ∫ i dt = 1
C
ton I o f (max) (
Ei (min)
∧
Substituting for I o = vo / R gives Where the diode conduction time tD is constant since in equation (15.103), v0, i , and L i

vo are all constants. The average output capacitor current is given by

∆vo = 1
C ∫ i dt = 1
C
ton I o =
R
1
C
ton
Io = ½ i (1 − δ )
∧
i
Rearranging gives the percentage peak-to-peak voltage ripple in the output voltage
and substituting equation (15.105) yields
∆vo
= 1 ton = δ τ (15.102) ∧ 1
vo RC RC Io (max) = ½ i × f (max) ×
i
vo
The capacitor equivalent series resistance and inductance can be account for, as with f (max) ( + 1)
Ei (min)
the forward converter, 15.1.5. When the switch conducts, the output current is constant
and is provided from the capacitor. No ESL voltage effects result during this constant therefore
capacitor current portion of the switching cycle. ∧ vo
i = 2 × Io (max) × (
i + 1)
Ei (min)
15.4.7 Buck-boost, flyback converter design procedure and upon substitution into equation (15.103)

The output voltage of the buck-boost converter can be regulated by operating at a fixed
527 Switched-mode and Resonant dc Power Supplies Power Electronics 528

tD vo v. Determine
L= (15.106)
vo • the critical load resistance.
2 Io (max) ( + 1)
Ei (min) • the minimum inductance for continuous inductor conduction with 2.5 Ω load
vi. At what load resistance does the instantaneous inductor current fall below the
The minimum capacitance is specified by the maximum allowable ripple voltage, that
output current?
is
∧ vii What is the output voltage if the load resistance is increased to four times the
∆Q i tD critical resistance?
C(min) = = i

∆eo 2∆eo
that is Solution
Io (max)tD i. From equation (15.87), which assumes continuous inductor current
C(min) = (15.107) vo δ
v =− where δ = tT / τ
∆eo ( o + 1) Ei 1−δ
Ei (min)
that is
The ripple voltage is dropped across the capacitor equivalent series resistance, which is
75V δ
given by = thus δ = 3 5
∆e 50V 1 − δ
ESR(max) = ∧ o (15.108) That is, τ = 1/ fs = 100 µs with a 60µs switch on-time.
i i

The frequency varies as a function of load current. Equation (15.104) gives ii. The mean output current I o is given by
∧
Io i tT Io (max) I o = vo / R = 75V/2.5Ω = 30A
= = i

f 2 f (max) From power transfer considerations

therefore I i = I L = vo I o / Ei = 75V×30A/50V = 45A
Io
f = f (max) × (15.109) iL
(A )
Io (max)
80
and 75 10A
IL = 7 5 A
Io (min) 70
f (min) = f (max) × (15.110)
Io (max) IT ID IT ID
60µs

0 50 100 150 t (µ s )

Example 15.4: Buck-boost flyback converter iC

(A )
30 Io = 3 0 A
The 10kHz flyback converter in figure 15.6 is to operate from a 50V input and
produces an inverted non-isolated 75V output. The inductor is 300µH and the resistive
load is 2.5Ω. 0
i. Calculate the duty cycle, hence transistor off-time, assuming continuous 0 50 100 150 t (µ s )
inductor current.
ii. Calculate the mean input and output current.
iii. Draw the inductor current, showing the minimum and maximum values. 40
iv. Calculate the capacitor rms ripple current and output p-p ripple voltage if C =
50
10,000µF.
Figure: Example 15.4
529 Switched-mode and Resonant dc Power Supplies Power Electronics 530

iii. The average inductor current can be derived from 2

L  L 
I i = δ I L or I o =(1- δ ) I L δ ≤ 1+ − 1 +  −1
τR  τR
That is
Alternately, when
I L = I i /δ = I o /(1- δ )
I i − I o = 5A
= 45A/ 3 5 = 30A/ 2 5 = 75A
From v = L di/dt, the ripple current ∆iL = EitT/L = 50V x 60µs /300 µH = 10 A, that is Io
− I o = 5A
∧ 1−δ
i = I L + ½ ∆iL = 75A + ½×10A = 80A
L
For δ = ⅗, I o = 3⅓A. whence
∨
i = I L − ½ ∆iL = 75A - ½×10A = 70A vo 75V
R= = = 22½ Ω
L

I o 10 3 A
iv. The capacitor current is derived by using Kirchhoff’s current law such that at any
instant in time, the diode current, plus the capacitor current, plus the 30A constant load The average inductor current is 8⅓A, with a minimum value of 3⅓A, the same as the
current into R, all sum to zero. load current. That is, for R < 22½Ω all the load requirement is provided from the
1 t 2 τ −t ∆iL ∧  inductor when the switch is off, with excess energy charging the output capacitor. For
∫ ∫
T T
iCrms =
τ  0 I o dt + 0 (τ − tT t − i + I o ) dt 
2
L
R > 22½Ω insufficient energy is available from the inductor to provide the load energy
.

 
throughout the whole of the period when the switch is off. The capacitor supplements
1  60 µs 40 µs 10A 
= ∫
30A dt + ∫ t − 50A) 2 dt  the load requirement towards the end of the off period. When R > 37½Ω (the critical
2
.
 (
100µs  0 0 40µs  resistance), discontinuous inductor current occurs, and the purely duty cycle dependent
= 36.8A transfer function is no longer valid.
The output ripple voltage is given by equation (15.102), that is
vii. When the load resistance is increased to 125Ω, four times the critical
∆vo δτ 3
5 × 100µs
= = ≡ 0.24% resistance, the output voltage is given by equation (15.91):
vo CR 10, 000µF × 2½ Ω
τR 100µs × 125Ω
The output ripple voltage is therefore vo = Ei δ = 50V × 3 5 × = 137V
2L 2 × 300µH
∆vo = 0.24 × 10−2 × 75V = 180mV

v. The critical load resistance, Rcrit, produces an inductor current with ∆iL = 10 A ♣
ripple. From equation (15.95)
15.5 The output reversible converter
2L 2×300µH
Rcrit = = = 37½Ω
τ (1 − δ )2 100µs × (1- 35 ) 2 The basic reversible converter, sometimes called an asymmetrical half bridge
The minimum inductance for continuous inductor current operation, with a 2½Ω load, converter (see chapter 13.5), shown in figure 15.7a allows two-quadrant output voltage
can be found by rearranging the critical resistance formula, as follows: operation. Operation is characterised by both switches operating simultaneously, being
Lcrit = ½ Rτ (1 − δ ) 2 = ½×2.5Ω×100µs×(1- 3 5 ) 2 = 20µH either both on or both off.
The input voltage Ei is chopped by switches T1 and T2, and because the input
vi. The ±5A inductor ripple current is independent of the load, provided the critical voltage is greater than the load voltage vo, energy is transferred from the dc supply Ei to
resistance of 37½Ω is not exceeded. When the average inductor current is less than 5A L, C, and the load R. When the switches are turned off, energy stored in L is transferred
more than the output current, the capacitor must provide load current not only when the via the diodes D1 and D2 to C and the load R but in a path involving energy being
switch is on but also when the switch is off. The transition is given by equation (15.83), returned to the supply, Ei. This connection feature allows energy to be transferred from
that is the load back into Ei when used with an appropriate load and the correct duty cycle.
531 Switched-mode and Resonant dc Power Supplies Power Electronics 532

Parts b and c respectively of figure 15.7 illustrate reversible converter circuit 15.5.1 Continuous inductor current
current and voltage waveforms for continuous and discontinuous conduction of L, in a
forward converter mode, when δ > ½. When the switches are turned on for period tT, the difference between the supply
For analysis it is assumed that components are lossless and the output voltage vo is voltage Ei and the output voltage v0 is impressed across L. From V=Ldi/dt, the rising
maintained constant because of the large capacitance magnitude of the capacitor C current change through the inductor will be
across the output. The input voltage Ei is also assumed constant, such that Ei ≥ vo > 0, as ∧ ∨ E −v
∆iL = i L − i L = i o × tΤ (15.111)
shown in figure 15.7a. L
T1
When the two switches are turned off for the remainder of the switching period, τ-tT,
the two freewheel diodes conduct in series and Ei + vo is impressed across L. Thus,
assuming continuous inductor conduction the inductor current fall is given by
D1 +
E +v
∆iL = i o × (τ − tΤ ) (15.112)
D2 L
Equating equations (15.111) and (15.112) yields
T2
vo I i 2tT − τ
= = = 2δ − 1 0 ≤δ ≤1 (15.113)
Ei I o τ
(a)
T D T D
The voltage transfer function is independent of circuit inductance L and capacitance C.
ON
switch
Equation (15.113) shows that for a given input voltage, the output voltage is
tx determined by the transistor conduction duty cycle δ and the output voltage |vo| is
tT tT
period τ io τ always less than the input voltage. This confirms and validates the original analysis
iL assumption that Ei ≥ |vo|. The linear transfer function varies between -1 and 1 for
Io 0≤δ≤1, that is, the output can be varied between vo= - Ei, and vo= Ei. The significance
∨
iL
of the change in transfer function polarity at δ = ½ is that
t t • for δ > ½ the converter acts as a forward converter, but
ii
iL • for δ < ½, if the output is a negative source, the converter acts as a boost
∨
iL converter with energy transferred to the supply Ei, from the negative output
source.
t t Thus the transfer function can be expressed as follows
∨
-iL vo I i
= = 2δ − 1 = 2 (δ − ½) ½ ≤δ ≤1 (15.114)
iD Ei I o
iL
∨ and
iL
Ei I o 1 1
= = = 0≤δ ≤ ½ (15.115)
t
vD
t vo I i 2δ − 1 2 (δ − ½)
Ei
where equation (15.115) is in the boost converter transfer function form.

15.5.2 Discontinuous inductor current

t t
(b) (c) In the forward converter mode, δ ≥ ½, the onset of∨ discontinuous inductor current
Figure 15.7. Basic reversible converter with δ>½: operation occurs when the minimum inductor current i L , reaches zero. That is,
(a) circuit diagram; (b) waveforms for continuous inductor current; and (c)
discontinuous inductor current waveforms. I L = ½ ∆i L = I o (15.116)
533 Switched-mode and Resonant dc Power Supplies Power Electronics 534

If the transistor on-time tT is reduced or the load resistance increases, the discontinuous 15.5.4 Control methods for discontinuous inductor current
condition dead time t∨x appears as indicated in figure 15.7c. From equations (15.111)
and (15.112), with i L = 0 , the following output voltage transfer function can be Once the load current has reduced to the critical level as specified by equation (15.117)
derived the input energy is in excess of the load requirement. Open loop load voltage regulation
∧ E −v E +v control is lost and the capacitor C tends to overcharge.
∆iL = i L − 0 = i o × tΤ = i o × (τ − tΤ − t x ) (15.117) As with the other converters considered, hardware and control approaches can mitigate
L L
which after rearranging yields this overcharging problem. The specific control solutions for the forward converter in
section 15.3.4, are applicable to the reversible converter. The two time domain control
tx
vo 2δ − 1 − τ approaches offer the following operational modes.
= 0 ≤δ <1 (15.118)
Ei t
1 − τx 15.5.4i - fixed on-time tT, variable switching frequency fvar
The operating frequency fvar is varied while the switch-on time tT is maintained
constant such that the magnitude of the ripple current remains unchanged. Operation is
15.5.3 Load conditions for discontinuous inductor current
specified by equating the input energy and the output energy, thus maintaining a
constant capacitor charge, hence output voltage. That is, equating energies
In the forward converter mode, δ ≥ ½, as the load current decreases, the inductor
average current also decreases, but the inductor ripple current magnitude is unchanged. v2 1
½ ∆iL Ei tT = o (15.123)
If the load resistance is increased sufficiently, the bottom of the triangular inductor R f var
∨
current, i L , eventual reduces to zero. Any further increase in load resistance causes Isolating the variable switching frequency fvar and using vo = Io R to eliminate R yields
discontinuous inductor current and the linear voltage transfer function given by
1 R
equation (15.113) is no longer valid. Equation (15.118) is applicable. The critical load f var = fs Rcrit × = f s crit × Io
R vo
resistance for continuous inductor current is specified by (15.124)
v 1
Rcrit ≤ o (15.119) f var α or f var α Io
Io R
Substituting I o = I L and using equations (15.111) and (15.116), yields That is, once discontinuous inductor current occurs at I o < ½ ∆iL or Io < vo / Rcrit , a
constant output voltage vo can be maintained if the switch on-state period tT remains
v v 2vo L
Rcrit ≤ o = o = (15.120) constant and the switching frequency is varied
I o ½ ∆iL ( Ei − vo )tΤ • proportionally with load current, I o
Dividing throughout by Ei and substituting δ = tT / τ yields • inversely with the load resistance, Rcrit
v (2δ − 1) L • inversely with the output voltage, vo.
Rcrit ≤ o = (15.121)
Io (1 − δ )δτ
15.5.4ii - fixed switching frequency fs, variable on-time tTvar
By substituting the switching frequency ( f s = 1/ τ ) or the fundamental inductor The operating frequency fs remains fixed while the switch-on time tTvar is reduced,
reactance ( X L = 2π f s L ), critical resistance can be expressed in the following forms. resulting in the ripple current magnitude being reduced. Equating input energy and
v 2(δ − ½ ) L 2(δ − ½ ) fs L (δ − ½ ) XL output energy as in equation (15.27), thus maintaining a constant capacitor charge,
Rcrit ≤ o = = = (Ω) (15.122)
Io (1 − δ )δτ (1 − δ )δ π (1 − δ )δ hence voltage, gives
If the load resistance increases beyond Rcrit, the output voltage can no longer be v2 1
½ ∆iL Ei tT var = o (15.125)
maintained with duty cycle control according to the voltage transfer function in R fs
equation (15.113).
Isolating the variable on-time tTvar, substituting for ∆iL, and using vo = Io R to eliminate
R, gives
535 Switched-mode and Resonant dc Power Supplies Power Electronics 536

1 Rcrit
tT var = tT Rcrit × = tT × Io vo 48V
R vo ii. The average load current is I o = = = 48A = I L
(15.126) R 1Ω
1 From power-in equals power-out, the average input current is
tT var α or tT var α Io
R I i = vo I o / Ei = 48V×48A/192V = 12A
That is, once discontinuous inductor current commences, if the switching frequency fs
remains constant, regulation of the output voltage vo can be maintained if the switch
iii. The average output current is the average inductor current, 48A. The ripple current
on-state period tT is varied
is given by equation (15.113), that is
• proportionally with the square root of the load current, Io
∧ ∨ E −v
• inversely with the square root of the load resistance, √Rcrit ∆iL = i L − i L = i o × tΤ
• inversely with the square root of the output voltage, √vo. L
192V - 48V
= × 62.5µs = 45A p-p
VL 200µH
Example 15.5: Reversible forward converter (V )
144 equal
areas

The step-down reversible converter in figure 15.7a operates at a switching frequency of

10 kHz. The output voltage is to be fixed at 48 V dc across a 1 Ω resistive load. If the
0 25 50 75 100 1 2 5 t (µ s )
input voltage Ei =192 V and the choke L = 200µH:

i. calculate the switch T on-time duty cycle δ and switch on-time tT

-2 4 0
ii. calculate the average load current I o , hence average input current I i
iii. draw accurate waveforms for IL
• the voltage across, and the current through L; vL and iL (A )

• the capacitor current, ic 48

Iav e 45A
• the switch and diode voltage and current; vT, vD, iT, iD
iv. calculate
• the maximum load resistance Rcrit before discontinuous inductor 0
0 25 50 75 100 1 2 5 t (µ s )
current with L=200µH and
• the value to which the inductance L can be reduced before dis- (V )
(A )
continuous inductor current, if the maximum load resistance is 1Ω. 192V

V D io d e V T ra n
Solution Ic ap
V T ra n
V D io d e
i. The switch on-state duty cycle δ can be calculate from equation (15.113), that is
v 48V 0 25 50 62½ 75 100 1 2 5 t (µ s )
2δ − 1 = o = = ¼ ⇒ δ = 58
Ei 192V
Also, from equation (15.113), for a 10kHz switching frequency, the switching period τ Figure: Example 15.5
is 100µs and the transistor on-time tT is given by
t t
δ = T = T = 58 iv. Critical load resistance is given by equation (15.122), namely
τ 100µs
v (2δ − 1) L
whence the transistor on-time is 62½µs and the diode conducts for 37½µs. Rcrit ≤ o =
Io τδ (1 − δ )
537 Switched-mode and Resonant dc Power Supplies Power Electronics 538

(2 × 5 8 -1)×200µH 2(δ r − 1)
= = 32/15Ω I fly = × Ir
100µs × 5 8 × (1- 5 8 ) 2δ r − 1
(15.128)
= 2 2 15 Ω when I o = 22 ½ A 2(δ r − 1)
where = δ fly for 0 ≤ δ fly ≤ ½ and 0 ≤ δ r ≤ ½
Alternatively, the critical load current is 22½A (½∆iL), thus the load resistance must 2δ r − 1
not be greater than vo / I o = 48V/22.5A=32/15Ω, if the inductor current is to be Again the reversible converter always has the higher inductor ripple current.
continuous. Essentially the higher ripple current results in each mode because the inductor energy
The critical resistance formula given in equation (15.122) is valid for finding critical release phase involving the diode occurs back into the supply, which is effectively in
inductance when inductance is made the subject of the equation, that is, rearranging cumulative series with the output capacitor voltage.
equation (15.122) gives The reversible converter offers some functional flexibility, since it can operate as a
Lcrit = R × (1 − δ ) × δ × τ /(2δ − 1) (H) conventional forward converter, when only one of the two switches is turned off. (In
= 1Ω×(1- 5 8 )× 5 8 ×100µs/(2× 5 8 -1) fact, in this mode, switch turn-off is alternated between T1 and T2 so as to balance
switch and diode losses.
= 93¾µH
That is, the inductance can be decreased from 200µH to 93¾µH when the load is 1Ω 15.6 The Ćuk converter
and continuous inductor current will flow.
♣
The Ćuk converter in figure 15.8 performs an inverting boost converter function with
15.5.5 Comparison of the reversible converter with alternative converters inductance in the input and the output. As a result, both the input and output currents
can be continuous. A capacitor is used in the process of transferring energy from the
The reversible converter provides the full functional output range of the forward input to the output and ac couples the input boost converter stage (L1, T) to the output
converter when δ>½ and provides part of the voltage function of the buck-boost forward converter (D, L2). Specifically, the capacitor C1 ac couples the switch T in the
converter when δ<½ but with energy transferring in the opposite direction. boost converter stage into the output forward converter stage.
Comparison of example 15.1 and 15.4 shows that although the same output voltage
range can be achieved, the inductor ripple current is much larger for a given inductance
L. A similar result occurs when compared with the buck-boost converter. Thus in each L1 C1 L2
case, the reversible converter has a narrower output resistance range before
discontinuous inductor conduction occurs. It is therefore concluded that the reversible vo
converter should only be∼used if two quadrant operation is needed.
The ripple current I f given by equation (15.2) for the forward converter and
C2
equation (15.111) for the reversible converter when vo > 0, yield the following current
+
ripple relationship.
I f = (2 − 1/ δ r ) × I r
(15.127)
where 2δ r − 1 = δ f for 0 ≤ δ f ≤ 1 and ½ ≤ δ r ≤ 1
∼
This equation shows that∼
the ripple current of the forward converter I f is never greater
than the ripple current I r for the reversible converter, for the same output voltage. Figure 15.8. Basic Ćuk converter.
In the voltage inverting mode, from equations (15.73) and (15.111), the relationship
between the two corresponding ripple currents is given by 15.6.1 Continuous inductor current

When the switch T is on and the diode D is reversed biased

iC 1( on ) = − I L 2 = I o (15.129)
539 Switched-mode and Resonant dc Power Supplies Power Electronics 540

When the switch is turned off, inductor currents iL1 and iL2 are divert through the diode vo 2 L2 2 f s L2 XL2
and Rcrit ≤ = = = (15.141)
I o τ (1 − δ ) (1 − δ ) π (1 − δ )
iC 1(off) = I i (15.130)
This is the same expression as that obtained for the forward converter, equation (15.26)
Over one steady-state cycle the average capacitor charge is zero, that is which can be re-arranged to give the minimum inductance for continuous output
iC 1(on)δτ + iC 1(off) (1 − δ )τ = 0 (15.131) inductor current, namely
∨
which gives L 2 = ½ (1 − δ ) Rτ (15.142)
iC 1(on) δ Ii
= = (15.132)
iC 1(off) (1 − δ ) Io 15.6.3 Optimal inductor relationship
From power-in equals power-out
vo I I Optimal inductor conditions are that both inductors should both simultaneous reach the
= i = L1 (15.133) verge of discontinuous conduction. The relationship between inductance and ripple
Ei I o I L 2
current is given by equations (15.135) and (15.139).
Thus equation (15.132) becomes
δτ Ei δτ Ei
vo I I δ ∆iL1 = and ∆iL 2 =
= i = L1 = − (15.134) L1 L2
Ei I o I L 2 (1 − δ )
After diving these two equations
L2 ∆iL1
= (15.143)
15.6.2 Discontinuous inductor current L1 ∆iL 2
Critical inductance is given by equations (15.138) and (15.142), that is
The current rise in L1 occurs when the switch is on, that is (1 − δ ) Rτ
2
∨ ∨

δτ Ei L 2 = ½ (1 − δ ) Rτ and L1 =
∆iL1 = (15.135) 2δ
L1 After dividing
For continuous current in the input inductor L1, ∨

I i = I L1 ≥ ½ ∆iL1 (15.136) L2 δ
= (15.144)
L1 1 − δ
∨
which yields a maximum allowable load resistance, for continuous inductor current, of
v 2δ L1 2 f Lδ δ X L1 At the verge of simultaneous discontinuous inductor conduction
Rcrit ≤ o = = s 1 = (15.137)
I o τ (1 − δ ) 2 (1 − δ ) 2 π (1 − δ ) 2
∨
L2 δ ∆i v
This is the same expression as that obtained for the boost converter, equation (15.65), = = L1 = o (15.145)
L1 1 − δ ∆iL 2
∨
Ei
which can be re-arranged to give the minimum inductance for continuous input
inductor current, namely That is, the voltage transfer ratio uniquely specifies the ratio of the minimum
inductances and their ripple current.
(1 − δ ) Rτ
2
∨
L1 = (15.138)
2δ 15.6.3 Output voltage ripple
The current rise in L2 occurs when the switch is on and the inductor voltage is Ei, that is
δτ Ei The output stage (L2, C2 and R) is the forward converter output stage; hence the per unit
∆ iL 2 = (15.139)
L2 output voltage ripple on C2 is given by equation (15.35), that is
For continuous current in the output inductor L2, ∆vC 2 ∆vo 1 (1 − δ )τ 2
= = 8× (15.146)
I o = I L 2 ≥ ½ ∆i L 2 (15.140) vo vo L2 C2
which yields If the ripple current in L1 is assumed constant, the per unit voltage ripple on the ac
541 Switched-mode and Resonant dc Power Supplies Power Electronics 542

coupling capacitor C1 is approximated by iii The inductor ripple current for each inductor is given by the same expression,
∆vC1 δτ that is equations (15.135) and (15.139). Thus for the same ripple current of 1A pp
= (15.147)
vo R C1 δτ Ei δτ Ei
∆iL1 = = ∆iL 2 =
L1 L2
which gives
Example 15.6: Ćuk converter δτ Ei 35 × 100µs × 50V
L1 = L2 = = = 3mH
∆i 1A
The Ćuk converter in figure 15.8 is to operate at 10kHz from a 50V battery input and
produces an inverted non-isolated 75V output. The load power is 1.8kW. iv The capacitor ripple voltages are given by equations (15.147) and (15.146),
i Calculate the duty cycle hence switch on and off times, assuming continuous which after re-arranging gives
current in both inductors. v δ τ 100 35 ×100µs
C1 = o × = × = 1.92mF
ii Calculate the mean input and output, hence inductor, currents. ∆vC 1 R 1 25
8Ω
iii At the 1.8kW load level, calculate the inductances L1 and L2 such that the
vo (1 − δ )τ 2 100 1 (1 − 35 ) × 100µs 2
ripple current is 1A p-p in each. C2 = × 18 × = × 8× = 16.6µF
iv Specify the capacitance for C1 and C2 if the ripple voltage is to be a ∆vC 2 L2 1 3mH
maximum of 1% of the output voltage.
v Determine the critical load resistance for which the purely duty cycle v The critical load resistance for each inductor is given by equations (15.137)
dependant voltage transfer function becomes invalid. and (15.141). When both inductors are 3mH:
vi At the critical load resistance value, determine the inductance value to which 2δ L1 2 × 35 × 3mH
Rcrit ≤ = = 225Ω
the non-critically operating inductor can be reduced. τ (1 − δ ) 2 100µs × (1 − 35 ) 2
vii Determine the necessary conditions to ensure that both inductors operate 2 L2 2 × 3mH
simultaneously on the verge of discontinuous conduction, and the relative Rcrit ≤ = = 150Ω
τ (1 − δ ) 100µs × (1 − 35 )
ripple currents for that condition.
The limiting critical load resistance is 150Ω or for Io = vo /R = 75V/150Ω = ½A, when a
lower output current results in the current in L2 becoming discontinuous although the
current in L1 is still continuous.
Solution
vi From equation (15.137), rearranged
i. The voltage transfer function is given by equation (15.134), that is
τ R(1 − δ ) 2 100µs × 100Ω × (1 − 35 )2
vo δ 75V L1 crit ≥ = = 2mH
=− =− = −1½ 2δ 2 × 35
Ei (1 − δ ) 50V
That is, if L1 is reduced from 3mH to 2mH, then both L1 and L2 enter discontinuous
from which δ = 5 . For a 10kHz switching frequency the period is 100µs, thus the
3
conduction at the same load condition, 75V, ½A, and 150Ω.
switch on-time is 60µs and the off-time is 40µs.
vii For both converter inductors to be simultaneously on the verge of
ii. The mean output current is determined by the load and the mean input current discontinuous conduction, equation (15.145) gives
is related to the output current by assuming 100% efficiency, that is ∨
L2 δ ∆i v
I o = I L 2 = Po / vo = 1800W / 75V = 24A = = L1 = o
L1 1 − δ ∆iL 2
∨
Ei
I i = I L1 = Po / Ei = 1800W / 50V = 36A
3mH 3
1A 75V 3
= = = =
5
The load resistance is therefore R = vo /Io = 75V/24A = 3⅛Ω.
2mH 1 − 3 5 2 3 A 50V 2
♣
543 Switched-mode and Resonant dc Power Supplies Power Electronics 544

15.7 Comparison of basic converters

The converters considered employ an inductor to transfer energy from one dc voltage
level to another dc voltage level. The basic converters comprise a switch, diode,
inductor, and a capacitor. The reversible converter is a two-quadrant converter with
two switches and two diodes, while the Ćuk converter uses two inductors and two
capacitors.
Table 15.1 summarises the main electrical features and characteristics of each basic
converter. Figure 15.9 shows a plot of the voltage transformation ratios and the switch
utilisation ratios of the converters considered. With reference to figure 15.9, it should
be noted that the flyback step-up/step-down converter and the Ćuk converter both
invert the input polarity.
Every converter can operate in any one of three inductor current modes:
• discontinuous
• continuous
• both continuous and discontinuous
The main converter operational features of continuous conduction compared with
discontinuous inductor conduction are
• The voltage transformation ratio (transfer function) is independent of the load.
• Larger inductance but lower core hysteresis losses and saturation less likely.
• Higher converter costs with increased volume and weight.
• Worse transient response (L /R).
• Power delivered is inversely proportional to load resistance, P = Vo 2 / R . In the 1 step-down
discontinuous conduction mode, power delivery is inversely dependent on step-up 3
inductance. 1

¾
15.7.1 Critical load current
buck-boost
Po / PT /Cuk
Examination of Table 15.1 show little commonality between the various converters and 2 reversible
& ½ 4
their performance factors and parameters. One common feature is the relationship Cuk
between critical average output current Io and the input voltage Ei at the boundary of
continuous and discontinuous conduction. full
¼ bridge
Equations (15.14), (15.61), and (15.92) are identical, (for all smps), that is
Eτ
Io = i δ (1 − δ ) (A) (15.148)
2L
critical

This quadratic expression in δ shows that the critical mean output current reduces to 0 ¼ ½ ¾ 1
zero as the on-state duty cycle δ tends to zero or unity. The maximum critical load
δ = tT/τ
current condition, for a given input voltage Ei, is when δ=½ and
I o = Eiτ / 8 L (15.149)
c
Figure 15.9. Transformation voltage ratios and switch utilisation ratios for five
converters when operated in the continuous inductor conduction mode.
 545 Switched-mode and Resonant dc Power Supplies Power Electronics 546

Since power in equals power out, then from equation (15.148) the input average current The reversible converter, using the critical resistance equation (15.122) derived in
and output voltage at the boundary of continuous conduction for all smps are related by section 15.5.3, yields twice the critical average output current given by equation
vτ (15.148). This is because its duty cycle range is restricted to half that of the other
I icritical = o δ (1 − δ ) (A) (15.150) converters considered. Converter normalised equations for discontinuous conduction
2L
are shown in table 15.2.
The maximum output current at the boundary, for a given output voltage, vo, is
A detailed analysis summary of discontinuous inductor current operation is given in
I i = voτ / 8 L (15.151)
c appendix 15.9.

Table 15.1 Converter characteristics comparison with continuous inductor current Table 15.2 Comparison of characteristics when the inductor current is discontinuous

converter δ 2 Rτ converter
k= Forward Flyback Flyback
Forward Flyback Flyback 4L Step-down Step-up Step-up/down
Reversible
Step-down Step-up Step-up/down
2L 2L 2L
1 δ δ critical δ (1 − δ ) ≤
2
Output voltage
vo /Ei δ − 2δ − 1 δ ≤ 1− δ ≤ 1−
continuous I 1− δ 1−δ Rτ Rτ Rτ
2 LI i E δ 2t Eiδ 2τ vo  2
Output voltage
vo /Ei 1− 1+ i T k  −1 + 1 +  ½ 1 + 1 + 8k  2k
discontinuous I Eiδ 2τ 2L I o 2 LI o Ei  k
Output polarity with
Non-inverted Non-inverted inverted any tx  2 1  1
respect to input ½ 1 + 1 +  1 + 4k + 1 + 8k  1+
Current sampled δτ  k 4k  2k
discontinuous continuous discontinuous bi-directional
from the supply 
∧
δR 2
Load current continuous discontinuous discontinuous continuous I L× 4k 1 + k − k 1 +  4k 4k
Ei  k
Maximum transistor
V V Ei vo Ei + vo Ei
voltage
Maximum diode
voltage
V V Ei vo Ei + vo Ei 15.7.2 Isolation
In each converter, the output is not electrically isolated from the input and a
Ripple current ∆i A Eiδτ (1 − δ ) / L Eiδτ / L Eiδτ / L 2 Eiδτ (1 − δ ) / L
transformer can be used to provide isolation. Figure 15.10 shows isolated versions of
Maximum transistor iˆT voτ (1 − δ ) Eiτδ Eiτδ ( Ei − vo )τδ the three basic converters. The transformer turns ratio provides electrical isolation as
A Io + Ii + IL + Io +
current 2L 2L 2L 2L well as providing matching to obtain the required output voltage range.
switch utilisation Figure 15.10a illustrates an isolated version of the forward converter shown in
SUR δ 1-δ δ (1-δ)
ratio figure 15.2. When the transistor is turned on, diode D1 conducts and L in the
Transistor rms transformer secondary stores energy. When the transistor turns off, the diode D3
low high high low
current provides a current path for the release of the energy stored in L. However when the
Critical load 2L 2L 2L 2(δ − ½ ) L transistor turns off and D1 ceases to conduct, the stored transformer magnetising energy
Rcrit Ω
resistance τ (1 − δ ) τδ (1 − δ ) 2 τ (1 − δ ) 2 τδ (1 − δ ) must be released. The winding incorporating D2 provides a path to reset the core flux.
½ R (1 − δ )δτ A maximum possible duty cycle exists, depending on the turns ratio of the primary
Critical inductance Lcrit H ½ R (1 − δ )τ ½ Rτδ (1 − δ ) 2 ½ Rτ (1 − δ ) 2 winding and freewheel winding. If a 1:1 ratio (as shown) is employed, a 50 per cent
(δ − ½ )
duty cycle limit will ensure the required volts-second for core reset.
o/p ripple voltage τ 2 (1 − δ ) τδ τδ τδ The step-up flyback isolated converter in part b of figure 15.10 is little used. The
∆vo V vo vo vo vo
p-p
8LC RC RC RC two transistors must be driven by complementary signals. Core leakage and reset
functions are facilitated by a third winding and blocking diode D2.
547 Switched-mode and Resonant dc Power Supplies Power Electronics 548

The magnetic core in the buck-boost converter of part c of figure 15.10 performs a 15.7.2i - The isolated output, forward converter – figure 15.10a:
bifilar inductor function. When the transistor is turned on, energy is stored in the core. • vo = nT δ Ei or I i = nT δ I o
When the transistor is turned off, the core energy is released via the secondary winding • The magnetic element acts as a transformer, that is, because of the relative
into the capacitor. A core air gap is necessary to prevent magnetic saturation and an voltage polarities of the windings, energy is transferred from the input to the
optional clamping winding can be employed, which operates at zero load. output, and not stored in the core, when the switch is on.
The converters in parts a and c of figure 15.10 provide an opportunity to compare • The magnetising flux is reset by the current through the catch (feedback)
the main features and attributes of forward and flyback isolated converters. In the winding and D3, when the switch is off. The magnetising energy is returned
comparison it is assumed that the transformer turns ratio is 1:1:1. to the supply Ei.
• The necessary transformer Vµs balance requirement (core energy-in equals
core energy-out) means the maximum duty cycle is limited to
0 ≤ δ ≤ 1/ ( 1 + nf / b ) < 1 for 1:nf/b:nsec turns ratio. For example, the duty cycle
is limited to 50%, 0 ≤ δ ≤ ½, with a 1:1:1 turns ratio.
• Because of the demagnetising winding, the off-state switch supporting voltage
is Ei + vo.
• The blocking voltage requirement of diode D3 is Ei, vo for D1, and 2Ei for D2.
15.7.2ii - The isolated output, flyback converter – figure 15.10c:
• vo = nT Eiδ /(1 − δ ) or I i = nT I oδ /(1 − δ )
• The magnetic element acts as a storage inductor. Because of the relative
voltage polarities of the windings (dot convention), when the switch is on,
energy is stored in the core and no current flows in the secondary.
• The stored energy, which is due to the core magnetising flux is released (reset)
as current into the load and capacitor C when the switch is off. (Unlike the
forward converter, where magnetising energy is returned to Ei, not the output,
vo.) Therefore there is no flyback converter duty cycle restriction, 0 ≤ δ ≤ 1.
• The third winding turns ratio is configured such that energy is only returned to
the supply Ei under no load conditions.
• The switch supporting off-state voltage is Ei + vo.
• The diode blocking voltage requirements are Ei + vo for D1 and 2Ei for D2.
The operational characteristics of each converter change considerably when the
flexibility offered by tailoring the turns ratio is exploited. A multi-winding magnetic
element design procedure is outlined in section 9.1.1, where the transformer turns ratio
is not necessarily 1:1.
The basic approach to any transformer (coupled circuit) problem is to transfer, or refer,
all components and variables to either the transformer primary or secondary circuit,
Figure 15.10. Isolated output versions of the three basic converter configurations: whilst maintaining power and time invariance. Thus, maintaining power-in equals
(a) the forward converter; (b) step-up flyback converter; and (c) step up/down power-out, and assuming a secondary to primary turns ratio of nT is to one, gives
flyback converter. 2
vs ns ip ns Z s  ns 
= = nT = = nT =   = nT2 (15.152)
v p np is np Z p  n p 
Time, that is switching frequency, power, and per unit values (δ, ∆vo /vo), are invariant.
The circuit is then analysed. Subsequently, the appropriate parameters are referred back
to their original side of the magnetically coupled circuit.
549 Switched-mode and Resonant dc Power Supplies Power Electronics 550

If the coupled circuit is used as a transformer, magnetising current (flux) builds, which i. Calculate the switch duty cycle, hence transistor off-time, assuming continuous
must be reset to zero each cycle. Consider the transformer coupled forward converter in inductor current.
figure 15.10a.
∧
From Faraday’s equation, v = Ndφ / dt , and for maximum on-time duty ii. Calculate the mean input and output current.
cycle δ the conduction V-µs of the primary must equal the conduction V-µs of the iii. Draw the transformer currents, showing the minimum and maximum values.
feedback winding which is returning the magnetising energy to the supply Ei. iv. Calculate the capacitor rms ripple current and p-p voltage ripple if C = 1100µF.
E v. Determine
Ei ton = i toff
nf /b (15.153) • the critical load resistance
• the minimum inductance for continuous inductor conduction for a 22½ Ω load
ton + toff = τ
That is Solution
Ei δ =
∧ Ei
( )
nf /b
∧
1− δ
The feedback winding does not conduct during normal continuous inductor current
operation. This winding can therefore be ignored for analysis during normal operation.
∧ 1 Io = Io =
δ= (15.154) I i = 45A 10A 10A
1+ nf /b
1
0≤δ ≤
1+ nf /b
Ei =50V Cs vo =225V
From Faraday’s Law, the magnetizing current starts from zero and increases linearly to 300µH
∧ Rs =22½Ω
I M = Ei ton / LM (15.155)
where LM is the magnetizing inductance referred to the primary. During the switch off 1:3
period, this current falls linearly, as energy is returned to Ei. The current must reach
' '
zero before the switch is turned on again, whence the energy taken from Ei and stored Io = Io =
as magnetic energy in the core, has been returned to the supply. I i = 45A 30A 30A
Two examples illustrate the features of magnetically coupled circuit converters.
Example 15.7 illustrates how the coupled circuit in the flyback converter acts as an
inductor, storing energy from the primary source, and subsequently releasing that
energy in the secondary circuit. In example 15.8, the forward converter coupled circuit Ei =50V 9Cs vo' =75V
300µH
act as a transformer where energy is transferred through the core under transformer Rp =2½Ω
action, but in so doing, self-inductance (magnetising) energy is built up in the core,
which must be periodically released if saturation is to be avoided. Relative orientation
of the windings, according to the flux dot convention shown in figure 15.10, is thus Figure 15.11. Isolated output step up/down flyback converter and its equivalent
circuit when the output is referred to the primary.
important, not only the primary relative to the secondary, but also relative to the
feedback winding.
Figure 15.11 shows secondary parameters referred to the primary, specifically
vo = 225V vo' = vo / nT = 225V/3 = 75V
Example 15.7: Transformer coupled flyback converter Rs = 225Ω Rp = Rs / nT2 = 225Ω / 32 = 22½ Ω
Note that the output capacitance is transferred by a factor of nine, nT2 , since capacitive
The 10kHz flyback converter in figure 15.10c operates from a 50V input and produces reactance is inversely proportion to capacitance.
a 225V dc output from a 1:1:3 (1:nf/b:nsec) step-up transformer loaded with a 22½Ω It will be noticed that the equivalent circuit parameter values to be analysed, when
resistor. The transformer magnetising inductance is 300µH, referred to the primary: referred to the primary, are the same as in example 15.4. The circuit is analysed as in
551 Switched-mode and Resonant dc Power Supplies Power Electronics 552

I
example 15.4 and the essential results from example 15.4 are summarised in Table 15.3 transformer
and transferred to the secondary where appropriate. The waveform answers to part iii 80A
are shown in figure 15.12.
70A 70A
I primary I primary
Table 15.3 Transformer coupled flyback converter analysis

value transfer factor value 80/3A

parameter for nT = 3 for 70/3A 70/3A
primary analysis → secondary analysis I secondary I secondary
Io = Io
Ei V 50 3 150 10A

vo V 75 3 225 0µs 60 µs 100 µs t

2
RL Ω 2½ 3 22½
Figure 15.12. Currents for the transformer windings in example 15.7.
-2
Co µF 10,000 3 1100 ♣
Io(ave) A 30 ⅓ 10
Example 15.8: Transformer coupled forward converter
Po W 2250 invariant 2250
Ii(ave) A 45A ⅓ 15A The 10kHz forward converter in figure 15.10a operates from a 192V dc input and a
1:3:2 (1:nf/b:nsec) step-up transformer loaded with a 4Ω resistor. The transformer
δ p.u. ⅗ invariant ⅗
magnetising inductance is 1.2mH, referred to the primary. The secondary smps
τ µs 100 invariant 100 inductance is 800µH.
ton µs 60 invariant 60 i. Calculate the maximum switch duty cycle, hence transistor off-time, assuming
continuous inductor current.
toff µs 40 invariant 40
At the maximum duty cycle:
fs kHz 20 invariant 20 ii. Calculate the mean input and output current.
∆iL A 5 ⅓ 1⅔ iii. Draw the transformer currents, showing the minimum and maximum values.
∧ iv. Determine
IL A 80 ⅓ 80/3 • the critical load resistance
∨
A 70 ⅓ 70/3 • the minimum inductance for continuous inductor conduction for a 4 Ω load
IL
iCrms A rms 21.3 ⅓ 7.1 Solution
Rcrit Ω 37½ 3
2
337½ The maximum duty cycle is determined solely by the transformer turns ratio between
the primary and the feedback winding which resets the core flux. From equation
Lcrit µH 20 32 180
(15.154)
VDr V 125 3 375 ∧ 1
δ=
∆vo mV 180 3 540 1+ nf /b
∆vo /vo p.u. 0.24% invariant 0.24% 1
=¼=
1+ 3
The maximum conduction time is 25% of the 100µs period, namely 25µs. The
secondary output voltage is therefore
553 Switched-mode and Resonant dc Power Supplies Power Electronics 554

vsec = δ nT Ei ∧
I M = Ei ton / LM
= ¼×2×192 = 96V = 192V×25µs/1.2mH = 4A
The load current is therefore 96V/4Ω=24A, as shown in figure 15.13a.
This current increases the switch mean current to
Figure 15.13b shows secondary parameters referred to the primary, specifically
Rs = 4Ω Rp = Rs / nT2 = 4Ω / 22 = 1Ω I T = 12A + ½ × δ × 4A = 12½A
Figure 15.13c show the equivalent circuit when the switch is off. The output circuit
vo = 96V vo' = vo / nT = 96V/2 = 48V functions independently of the input circuit, which is returning stored core energy to
Lo =800µH L'o = Lo / nT2 = 800µH/22 = 200µH the supply Ei via the feedback winding and diode D2. Parameters have been referred to
Note that the output capacitance is transferred by a factor of four, nT2 , since capacitive the feedback winding which has three times the turns of the primary, nf/b =3. The 192V
reactance is inversely proportion to capacitance. input voltage remains the circuit reference. Equation (15.155), Faraday’s law, referred
Inspection of example 15.1 will show that the equivalent circuit in figure 15.13b is the to the feedback winding, must be satisfied during the switch off period, that is
∧
same as the circuit in example 15.1, except that a magnetising branch has been added. IM Ei toff
The various operating condition and values in example 15.1 are valid for example 15.8. =
nf /b n 2f / b LM
4 192V×75µs
800µH
= 2
24A 3 3 × 1.2mH
The diode D2 voltage rating is (nf/b+1)×Ei, 768V and its mean current is
RL=4Ω I 4A
I D 2 = ½ (1 − δ ) M = ½ × (1 - 0.25 ) × = ½A
Ei=192V nf /b 3
vo=96V
iii. The three winding currents for the transformer are shown in figure 15.14.
1:3:2
(a) I
transformer 61A
IM/3 IM =4A
48A 24A 57A
200µH 800µH 39A
RL=1Ω RL=4Ω I primary
10.8mH I primary
Ei=192V LM 4C
Ei=192V
1.2mH vo=48V 9LM vo=96V
IM 57/2A
I secondary 39/2A
I sec
IM 4A
(b) (c) 4/3A
Figure 15.13. Isolated output forward converter and its equivalent circuits when the
output is referred to the primary. 0µs 25 µs 100 µs t
Figure 15.14. Currents for the three transformer windings in example 15.8.
ii. The mean output current is the same for both circuits, 48A, or 24 A when referred
to the secondary circuit. The mean input current from Ei remains 12A, but the switch iv. The critical resistance and inductance, referred to the primary, from example 15.1
mean current is not 12A. Magnetising current is provided from the supply Ei through are 5⅓Ω and 37½µH. Transforming into secondary quantities, by multiplying by 22
the switch, but returned to the supply Ei through diode D2, which bypasses the switch. give critical values of RL = 21⅓Ω and L = 150µH.
The net magnetising energy flow is zero. The magnetising current maximum value is
given by equation (15.155) ♣
555 Switched-mode and Resonant dc Power Supplies Power Electronics 556

15.8 Multiple-switch, balanced, isolated converters

The basic single-switch converters considered have the limitation of using their
magnetic components only in a unipolar mode. Since only one quadrant of the B-H
characteristic is employed, these converters are generally restricted to lower powers
because of the limited flux swing, which is reduced by the core remanence flux.
The high-power forward converter circuits shown in figure 15.15 operate the
magnetic transformer component in the bipolar or push-pull flux mode and require two
or four switches. Because the transformers are fully utilised magnetically, they tend to
be almost half the size of the equivalent single transistor isolated converter at power
levels above 100 W. Also core saturation due to the magnetising current (flux) not
being fully reset to zero each cycle, is not a major issue, since with balanced
bidirectional fluxing, the average magnetising current is zero.

15.8.1 The push-pull converter

Figure 15.15a illustrates a push-pull forward converter circuit which employs two
switches and a centre-tapped transformer. Each switch must have the same duty cycle
in order to prevent unidirectional core saturation. Because of transformer coupling
action, the off switch supports twice the input voltage, 2Ei, plus any voltage associated
with leakage inductance stored energy. Advantageously, no floating gate drives are
required.
The voltage transfer function, for continuous inductor conduction, is based on the
equivalent secondary output circuit show in figure 15.16. Because of transformer
action the input voltage is N×Ei where N is the transformer turns ratio. When a primary
switch is on, current flows in the loop shown in figure 15.16. That is
∧ ∨ N × Ei − vo
∆ iL = i L − i L = × tΤ (15.156)
L
When the primary switches are off, the secondary voltage falls to zero and current
continues to flow through the secondary winding due to the energy stored in L.
Efficiency is increased if the diode Df is used to bypass the transformer winding, as
shown in figure 15.16. The secondary winding i2R losses are decreased and minimal
voltage is coupled from the secondary back into the primary circuit. The current in the
off loop shown in figure 15.16 is given by
v
∆iL = o × (τ − tΤ ) (15.157)
L
Equating equations (15.156) and (15.157) gives the following voltage and current
transfer functions
vo I i t
= = 2 N T = 2 Nδ 0≤δ ≤ ½ (15.158) Figure 15.15. Multiple-switch, isolated output, pulse-width modulated converters:
Ei I o τ (a) push-pull; (b) half-bridge; and (c) full-bridge.
557 Switched-mode and Resonant dc Power Supplies Power Electronics 558

L Using similar analysis as for the push-pull converter in 15.8.1, the voltage transfer
function of the full bridge with a forward converter output stage, with continuous
1 N conduction is given by
Df off
+ + vo I i t
Vo = = 2 N T = 2 Nδ 0≤δ ≤ ½ (15.162)
Ei N×Ei
on
Ei I o τ
Any volts-second imbalance (magnetising flux build-up) can be minimised by
using dc block capacitance Cc, as shown in figures 15.15b and c.
The output ripple voltage is given by
∆vC ∆vo (1 − 2δ )τ 2
Figure 15.16. Equivalent circuit for transformer bridge converters based on a forward = = (15.163)
converter in the secondary.
vo vo 32 LC
In each forward converter in figure 15.15, a single secondary transformer winding and
The output voltage ripple is similar to that of the forward converter full-wave rectifier can be used. If the output diode shown dashed in figure 15.15c is
∆vC ∆vo (1 − 2δ )τ 2 used, the off state loop voltage is decreased from two diode voltage drops to one.
= = (15.159) If the output inductor is not used, conventional unregulated transformer square-wave
vo vo 32 LC
voltage ratio action occurs for each transformer based smps, where, independent of δ:
15.8.2 Bridge converters vo I i n
= = s =N (15.164)
Ei I o no
Figures 15.15b and c show half and full-bridge isolated forward converters
respectively.
15.9 Resonant dc-to-dc converters
i. Half-bridge
In the half-bridge the transistors are switched alternately and must have the same Converter switching losses may be significantly reduced if zero current or voltage
conduction period. This ensures the core volts-second balance requirement to prevent switching can be utilised. This switching loss reduction allows higher operating
saturation due to bias in one direction. frequencies hence smaller L and C components (in size and value). Also radiated
Using similar analysis as for the push-pull converter in 15.8.1, the voltage transfer switching noise is significantly reduced.
function of the half bridge with a forward converter output stage, for continuous Two main techniques can be used to achieve near zero switching losses
inductor conduction, is given by ƒ a resonant load that provides natural voltage or current zero instances for
switching
vo I i t
= = N T = Nδ 0≤δ ≤ ½ (15.160) ƒ a resonant circuit across the switch which feeds energy to the load as well as
Ei I o τ introducing zero current or voltage instances for switching.
A floating base drive is required. Although the maximum winding voltage is ½Ei,
the switches must support Ei in the off-state, when the complementary switch conducts. 15.9.1 Series loaded resonant dc-to-dc converters
The output ripple voltage is given by
∆vC ∆vo (1 − 2δ )τ 2 The basic converter operating concept involves a H-bridge producing an ac square-
= = (15.161) wave voltage VH-B. When fed across a series L-C filter, a near sinusoidal oscillation
vo vo 16 LC
current results, provided the square wave fundamental frequency is near the natural
ii. Full-bridge resonant frequency of the L-C filter. Because of L-C filter action, only fundamental
The full bridge in figure 15.15c replaces the capacitor supplies of the half-bridge current flows and harmonics of the square wave are attenuated due to the gain roll-off
converter with switching devices. In the off-state each switch must support the rail of the second order L-C filter. The sinusoidal resonant current is rectified to produce a
voltage Ei and two floating gate drive circuits are required. This bridge converter is load voltage vo/p. The output voltage is highly dependant on the frequency relationship
usually reserved for high-power applications. between the square-wave drive voltage period and the L-C filter resonant frequency.
559 Switched-mode and Resonant dc Power Supplies Power Electronics 560

Figure 15.17a shows the circuit diagram of a series resonant converter, which uses an
i (ω t ) =
(Vs
− vo / p ) − vo
× e −αt × sin ωt (15.165)
output rectifier bridge to converter the resonant ac oscillation into dc. The converter is
based on the series converter in figure 14.27b. The rectified ac output charges the dc
ωL
output capacitor, across which is the dc load, Rload. The non-dc-decoupled resistance, where
which determines the circuit Q, is account for by resistor Rc. The dc capacitor C 1 Rc 1 Rc LR
capacitance is assumed large enough so that the output voltage vo/p is maintained
ω 2 = ωo2 (1 − ξ 2 ) ωo = α= =ξ = Zo =
LR CR 2 LR 2Qs 2ωo LR CR
constant, without significant ripple voltage. Figures 15.17b and c show how the dc
output circuit can be transformed into an ac square wave in series with the L-C circuit, ξ is the damping factor. The capacitor voltage is important because it specifies the
and finally this source is transferred to the dc link as a constant dc voltage source vo/p energy retained in the L-C-R circuit at the end of each half cycle.
which opposes the dc supply Vs. These transformation steps enable the series L-C-R
resonant circuit to be analysed with square wave inverter excitation, from a dc source
(
vc (ωt ) = (Vs − vo / p ) − (Vs − vo / p ) − vo ) ωω e
o −α t
cos (ωt − φ ) (15.166)
Vs - vo/p. This highlights that the output voltage must be less than the dc supply, that is where
Vs - vo/p ≥ 0, if current oscillation is to occur. The analysis in chapter 14.3.2 is valid for
this circuit, where Vs is replaced by Vs - vo/p. The equations, modified, are repeated for tan φ = α
ω and ωo2 = ω 2 + α 2
completeness.
At the series circuit resonance frequency ωo, the lowest possible circuit impedance
results, Z = Rc. The series circuit quality factor or figure of merit, Qs is defined by
T1 T3 ω L 1 Zo
D1 D3 Qs = o = = (15.167)
Vs
LR
Rc 2ξ Rc
C Rload Operation is characterised by turning on switches T1 and T2 to provide energy from
vo/p
the source during one half of the cycle, then having turned T1 and T2 off, T3 and T4
CR Rc are turned on for the second resonant half cycle. Energy is again drawn from the
T4
T2
supply, and when the current reaches zero T3 and T4 are turned off.
D4
D2 Without bridge freewheel diodes, the switches support high reverse bias voltages, but
the switches control the start of each oscillation half cycle. With freewheel diodes the
(a) oscillations can continue independent of the switch states. The diodes return energy to
the supply, hence reducing the energy transferred to the load. Correct timing of the
+ switches minimises currents in the freewheel diodes, hence minimises the energy
T3 D3
needlessly being returned to the supply. Energy to the load is maximised. The switches
vo/p can be used to control the effective load power factor. By advancing turn-off to occur
CR LR Rc vo/p T1 D1 CR LR before the switch current reaches zero, the load can be made to appear inductive, while
Vs Vs R delaying switch turn-on produces a capacitive load effect.
The series circuit steady-state current at resonance for the H-bridge with a high circuit
Q can be approximated by assuming ωo ≈ ω, such that:
T4 D4 T2 D2
i (ω t ) =
2
×
(Vs − vo / p ) × e−αt × sin ωt (15.168)
−απ
ω LR
1− e ω

(b) (c) which is valid for the ± (Vs - vo/p) voltage loops of cycle operation at resonance. For a
high circuit Q this equation is approximately
Figure 15.17. Series resonant converter and its equivalent circuit derivation.
4
i (ω t ) ≈ × Q ×
(Vs − vo / p ) × sin ω t = 4 × (Vs − vo / p ) × sin ω t (15.169)
The series L-C-R circuit current for a step input voltage Vs-vo/p, with initial capacitor π ωo LR o
π Rc
o

voltage vo, assuming zero initial inductor current, is given by

The maximum current is
561 Switched-mode and Resonant dc Power Supplies Power Electronics 562

∧
I≈
4
×
(V
s
− vo / p )
(15.170) vo = Vs ×
Req
(15.177)
Rc π  1 
Req + Rc + j  ω LR − 
while the average current with this peak value must equal to load current, that is 
s
ω s CR 
2 ∧ 8 (Vs − vo / p ) vo / p where the H-bridge switching frequency is ωs=2π fs.
I = ×I = 2 × = (15.171) Figure 15.18 shows equation (15.177) for different circuit Q. The plot can be used to
π π Rc R
extract output voltage vo/p and H-bridge switching frequency. The output voltage is
The output voltage is obtained from equation (15.171) by isolating vo/p scaled to eliminate the coil resistance component from the total resistive voltage.
Vs
vo / p = (15.172)
8 R
1+ 2 × c
π R 1
jωs LR 1
In steady-state the capacitor voltage maxima are Rc 4
4 jωsCR (v o + v c )
1+ e −απ / ω
V π2
Vc = (Vs − vo / p ) = (Vs − vo / p ) × coth (απ / 2ω ) = − Vc
∧ ∨

Q=1 π s
1 − e −απ / ω (15.173) ¾ Req
4
≈ (Vs − vo / p ) × 2ωo / απ = × Q × (Vs − vo / p ) 4
v
π vo + vc π
2 o

The peak-to-peak capacitor voltage, by symmetry is therefore Vs Q=2

8
Vc p− p ≈ × Q × (Vs − vo / p ) ½ (a)
(15.174)
π Q=3
The energy lost in the coil resistance Rc, per half sine cycle (per current pulse) is
 2 (Vs − vo / p )
2
π /ω π / ωo 
W =∫ i 2 Rc dt ≈ ∫ 4 × × sin ωo t  Rc dt ¼
Q=4 vo + vc
0 0  π R  v o/p =
 c  (15.175) π2 R
Q =10 1+ × c
8 8 RLoad
(V − vo / p )
2
=
π ωo Rc 0
s

The relationship between the output voltage vo/p and H-bridge switching frequency, ωs,
is not a simple linear function. Because of the L-C series filter cut-off frequency ωs ≈ ½ ¾ ωs 1 1¼ 1½
ωo, only fundamental current flows as a result of the fundamental of the square-wave ωo
VH-B, which has a magnitude of 4 π Vs . A series R-C-L ac circuit at frequency ωs can be (b)
used to derive a relationship between the output voltage vo/p and ωs. The effective load Figure 15.18. Series resonant converter
resistance, from equation (15.171) becomes Req = 8 π 2 Rload such that Kirchhoff’s voltage (a) equivalent circuit and (b) normalised output voltage curves.
law for the series circuit, shown in figure 15.18a, is
4 V = i ×  R + R + j ω L − 1 
π s  eq c  R 
ωs CR  
s

  (15.176) 15.9.1i - Modes of operation - series resonant circuit

4 v =i×R The basic series converter can be operated in any of three difference modes, depending
π o/ p eq
on the switching frequency in relation to the L-C circuit natural resonant frequency. In
The equivalent output voltage is therefore given by all cases, the controlled output voltage is less than the input voltage, that is Vs - vo/p ≥ 0.
The switching frequency involves one complete symmetrical square-wave output cycle
from the inverter bridge. Waveforms for the three operational modes are shown in
figure 15.19.
563 Switched-mode and Resonant dc Power Supplies Power Electronics 564
V H-B Vs
D 1/2 T 1/2 D 1/2 T 1/2
i. fs < ½fo :- discontinuous inductor current (switch conduction 1/ 2 f ≤ tT ≤ 1/ f )
-V s
D 3/4 T 3/4 D 3/4 T 3/4 ωt If the switching frequency is less than half the L-C circuit natural resonant frequency,
lagging
power factor
as shown in figure 15.19c, then discontinuous inductor current results. This because
ic vc
once one complete L-C resonant ac cycle occurs and current stops, being unable to
reverse, since the switches are turned off when the diodes conduct and the capacitor
o voltage is less than Vs+vo. Turn-off occurs at zero current. Subsequent turn-on occurs
ωto ωt1 ωt2 ωt at zero current but the voltage is determined by the voltage retained by the capacitor.
Thyristors are therefore applicable switches in this mode of operation. The freewheel
diodes turn on and off with zero current. Since the capacitor current rectified provides
V D-B vo the load current average current, the H-bridge switching frequency controls the
ωt
output voltage. Therefore at low switching frequencies, relative to the resonance
-v o frequency, the peak resonant current will be relatively high.
(a)
ii. ½fo< fs < fo :- continuous inductor current
If the switching frequency is just less than natural resonant frequency, as shown in
Vs figure 15.19b, such that turn-on occurs after half an oscillation cycle but before a
T 1/2 D 1/2 T 1/2 D 1/2
V H-B
complete ac oscillation cycle is complete, continuous inductor current results. Switch
T 3/4 D 3/4 T 3/4 D 3/4 ωt
-V s -V s turn-on occurs with finite inductor current and voltage conditions, with the diodes
leading
ic power factor freewheeling. Diode reverse recovery losses occur and noise in injected into the
circuit at voltage recovery snap. Fast recovery diodes are therefore necessary. Switch
vc
turn-off occurs at zero voltage and current, when the inductor current passes through
o
ωt 1 ωt2 ωt
zero and the freewheel diodes take up conduction. Thyristors are applicable as
switching devices with this mode of control.
iii. fs > fo :- continuous inductor current
vo If turn-off occurs before the resonance of half a resonant cycle is complete, as shown
in figure 15.19a, continuous inductor current flows, hard switching results, and
V D-B ωt
-v o commutable switches must be used. Switch turn-on occurs at zero voltage and current
(b) hence no diode recovery snap occurs. This zero electrical condition turn-on allows
lossless turn-off snubbers to be employed (a capacitor in parallel with each switch).
For continuous inductor conduction, under light load conditions, resonant energy is
Vs
V H-B T 1/2 D 1/2 vo continuously transferred to the output stage, which tends to progressively overcharge
-v o T 3/4 D 3/4 ωt the output voltage towards the input voltage level, Vs. The charging progressively
-V s decreases as the Vs is backed off by the increasing output voltage. That is Vs - vo/p
shown in figure 15.17c tends to zero such that the effective square wave input is
ic vc reduced to zero, as will the input energy.
o
ωt 1 ωt2 ωt 15.9.1ii - Circuit variations
The number of semiconductors can be reduced by using a split dc rail as in figure
V D-B
15.20a, at the expense of halving the bridge output voltage swing VH-B to ± ½Vs.
vo Although the number of semiconductors is halved, the already poor switch utilisation
associated with any resonant converter, is further decreased. The switches and diodes
ωt
-v o
(c)
support Vs. In the full-bridge case the corresponding switch and diode voltages are both
Figure 15.19. Three modes of series load resonant converter operation Vs.
(a) fs > fo; (b) ½fo< fs < fo; and (c) fs < ½fo.
565 Switched-mode and Resonant dc Power Supplies Power Electronics 566

Voltage and impedance matching, for example voltage step-up, can be obtained by 15.9.2 Parallel loaded resonant dc-to-dc converters
using a transformer coupled circuit as shown in figure 15.20b. When a transformer is
used as in figure 15.20c, a centre tapped secondary can reduce the number of high The basic parallel load resonant dc-to-dc converter is shown in figure 15.21a and its
frequency rectifying diodes from four to two, but diode reverse voltage rating is equivalent circuit is shown in figure 15.21b. The inductor Lo in the rectified output
doubled from Vs to 2Vs. Secondary copper winding utilisation is halved. circuit produces a near constant current. A key feature is that the output voltage vo/p can
A further modification to any series converter is to used the resonant capacitor to form be greater than the input voltage Vs, that is 0 ≤ vo/p ≥ Vs.
a split dc rail as shown in figure 15.20c, where each capacitance is ½CR. In ac terms the The capacitor voltage and inductor current equations for the equivalent circuit in figure
resonant capacitors are in parallel, with one charging while the discharges, and visa 15.21b, for a constant current load Io, are
versa, such that their voltage sum is Vs. V −v
iL ( t ) = I o + ( iLo − I o ) cos ωo t + s Co sin ωo t
Zo (15.178)
V
= I o + s sin ωo t for vCo = 0 and iLo = I o
Clarge Zo
½Vs
Vs
LR vc ( t ) = Vs − (Vs − vCo ) cos ωo t + Z o ( iLo − I o ) sin ωo t
+ Co (15.179)
VH-B R = Vs (1 − cos ωo t ) for vCo = 0 and iLo = I o
VD-B Lo
½Vs CR Rc
Cs LR
Clarge (a)
LR
Vo/p
Vs VH-B R
VH-B Cp VD-B
VD-B
CR

LR
Vs Co
+ (d)
VH-B R (a)
VD-B
CR Rc LR LR
(b)
jXL
+ Vs CR CR
+
Io 4 -jXC Req 4
V V
π s
π o

½CR
LR
Vs VD-B Co
+
VH-B R (b) (c)
Figure 15.21. Parallel resonant dc-to-dc converter:
½CR Rc (a) circuit (b) equivalent ac circuit; (c) equivalent fundamental input voltage circuit;
(c) and (d) series-parallel resonant circuit stage.

The relationship between the output voltage and the bridge switching frequency can
Figure 15.20. Series load resonant converters variations: determined from the equivalent circuit shown in figure 15.21c where the output
(a) half bridge, split dc supply rail; (b) transformer couple full bridge; and (c) split resistance has been replaced by its equivalent resistance related to the H-bridge output
resonant capacitor, with a centre tapped output rectifier stage. fundament frequency magnitude, 4 π Vs . The voltage across the resonant capacitor CR is
assumed to be sinusoidal.
567 Switched-mode and Resonant dc Power Supplies Power Electronics 568
V H-B Vs
D 1/2 T 1/2
Kirchhoff analysis of the equivalent circuit in figure 15.22b gives D 1/2 T 1/2
D 3/4 T 3/4 D 3/4 T 3/4 ωt
-V s
lagging
vo 8 1 8 1 pow er factor
= × = × (15.180) iL vc
Vs π 2 XL XL π2 2 2

1− +  XL   XL 
X C Req 1 −  +   o
 X C   Req  ω to ω t1 ω t2 ωt
where the load resistance R is related to the equivalent resistance, at the switching
frequency ωs, by Req = 82 × R . Series stray non-load resistance has been neglected.
π vo
V D-B

15.9.2i - Modes of operation - parallel resonant circuit -v o

ωt

Three modes of operation are applicable to the parallel-resonant circuit, dc-to-dc (a)
converter, and waveforms are shown in figure 15.22x.
i. fs < ½fo :- discontinuous inductor current (switch conduction 1/ 2 f o ≤ tT ≤ 1/ f o )
Vs
Initially all switches are off and the load current energy stored in Lo freewheels T 1/2 D 1/2 T 1/2 D 1/2
V H-B
through the bridge diodes. T 3/4 D 3/4 T 3/4 D 3/4 ωt
-V s leading -V s
Both the inductor current and capacitor voltage are zero at the beginning of the cycle pow er factor
vc
and at the end of the cycle. Thus switch turn-on and turn-off occur with zero current iL
losses. At H-bridge turn-on the resonant inductor current increases linearly according
to i = Vs t / LR until the output current level Io is reached at time tI = LR I o / Vs , when o
ω t1 ω t2 ωt
the capacitor is free to resonant. The capacitor voltage is given by
vc ( t ) = Vs (1 − cos ωo t ) (15.181)
vo
while the inductor current is given by
V V D-B ωt
iL ( t ) = I o + s sin (ωo t ) (15.182) -v o
(b)
Zo
The resonant circuit inductor current reverses as on-switch antiparallel freewheel
diodes conduct, at which time the switches may be turned off at zero current and Vs
voltage conditions. A further inductor current reversal is therefore not possible. At V H-B T 1/2 D 1/2

the attempted reversal instant, any retained capacitor charge is discharged at a T 3/4 D 3/4 ωt
-V s
constant rate Io in the inductor Lo. The capacitor voltage falls linearly to zero at which
time the current in Lo freewheels in the output rectifier diodes. Io iL
vc
ii. ½fo < fs < fo :- continuous inductor current
When switching below resonance, the switches commutate naturally at turn-off, as o
ω t1 ω t2 ω t3 ω t4 ωt
shown in figure 15.22b, making thyristors a possibility.
Hard turn-on results, necessitating the use of fast recovery diodes. -I o

iii. fs > fo :- continuous inductor current V D-B

vo
When switching at frequencies above the natural resonance frequency, no turn-on
losses occur since turn-on occurs when a switches antiparallel diode is conducting. ωt
-v o
Turn-off occurs with inductor current flowing, hence hard turn-off occurs, with (c)
switch current commutated to a freewheel diode. In mitigation, lossless capacitive Figure 15.22. Three modes of parallel load resonant converter operation:
turn-off snubber can be used (a capacitor in parallel with each switch). (a) fs > fo; (b) ½fo< fs < fo; and (c) fs < ½fo.
569 Switched-mode and Resonant dc Power Supplies Power Electronics 570

15.9.2ii - Circuit variations LR 100µH

Q= / Rc = /1Ω = 14.6
CR 0.47µF
A parallel resonant circuit approach, with or without transformer coupling, can also be
use as indicated in figure 15.21a. The centre tapped secondary approach shown in For this high Q, the circuit resonant frequency and damped frequency will be almost
figure 15.20c is also applicable. A dc capacitor split dc rail can also be used, as shown the same, that is
for the series load resonate circuit in figure 15.20a. A further resonant circuit variation ω ≈ ωo = 1/ LR CR
is a combined series-parallel resonant stage as shown in figure 15.21d.
= 1/ 100µH × 0.47µF = 146 krad/s
Table 15.4 Switch and diode turn-on/turn-off conditions for resonant switch converters = 2π f
f = 146 krad/s /2π = 23.25 kHz
series load resonance parallel load resonance
switch diode switch diode switch diode switch diode power ii. From equation (15.171), which will be accurate because of a high circuit Q of 14.6,
on off off on on off off on factor 2 ∧ 8 (Vs − nT × vo / p ) 8 ( 340 V − ½vo / p )
I= I= 2 = 2×
zcs zcs π π Rc π 1Ω
fs<½fo zcs zcs N/A
zvs zvs = 0.81× ( 340 V − ½ vo / p )
Note that the output voltage vo/p across the dc decoupling capacitor has been referred to
zcs zcs
fs < fo hard hard leading the primary by nT, hence halved, due to the turns ratio of 1:2.
zvs zvs
The rectified resonant current provides the load current, that is
zcs zcs 1 vo / p vo / p
fs > fo hard hard lagging I= × = 2×
zvs zvs nT Rload 10Ω
vo / p
=
Example 15.9: Transformer-coupled, series-resonant, dc-to-dc converter 5
Again the secondary current has been referred to the primary. Solving the two average
The series resonant dc step down voltage converter in figure 15.17a is operated at just primary current equations gives
above the resonant frequency of the load circuit and is used with a step up transformer, vo / p
1:2 (nT = ½), as shown in figure 15.19a. It produces an output voltage for the armature I = 0.81× ( 340 − ½ vo ) =
of a high voltage dc motor that has a voltage requirement that is greater than the 50Hz 5
ac mains rectified, 340V dc, with an L-C dc link filter. The resonant circuit parameters vo / p = 456V and I = 91.2A
are L=100µH, C=0.47µF, and the coil resistance is Rc = 1Ω. That is, the load voltage is 456V dc and the load current is 456V/10Ω = 91.2A/2 =
For a 10Ω armature resistance, Rload, calculate 45.6A dc. The power delivered to the load is 4562/10Ω = 20.8kW.
i. the circuit Q and ωo
ii. the output voltage, hence dc armature current and power delivered iii. From part ii, the capacitor dc voltage requirement is at least 456V dc. The
iii. the secondary circuit dc filter capacitor voltage and rms current rating secondary rms current is
iv. the resonant circuit rms ac current and capacitor rms ac voltage 1 ∧ 1 π
v. the converter average input current and efficiency I Srms = nT × I Prms = nT × × I = ½× × ×I
2 2
P P

vi. the ac current in the input L-C dc rectifier filter decoupling capacitor 2
vii. the H-bridge square-wave switching frequency ωs, greater than ωo. = 0.555 × I = 0.555 × 91.2A
P

= 50.65A rms
Solution The primary rms current is double the secondary rms current, 101.3A rms.
i. The resonant circuit Q is
571 Switched-mode and Resonant dc Power Supplies Power Electronics 572

By Kirchhoff’s current law, the secondary current (50.65A rms) splits between the load 8 × n2 × R
(45.6A dc) and the decoupling capacitor. That is the rms current in the capacitor is vo Req π2 T Load
= =
I Crms = I Srms
2
2
− I S = 50.652 − 45.62 = 22A rms Vs  1  8 × n2 × R + R + j  ω L − 1 
Req + Rc + j  ωs LR −
That is, the secondary dc filter capacitor has a dc voltage requirement of 456V dc and a  ωs CR  π2 T Load c  s R
 ωs CR 
current requirement of 22A rms at 46.5kHz, which is double the resonant frequency
because of the rectification process. 8 × ¼ × 10Ω
226V π2
= 0.66 =
iv. The primary rms current is double the secondary rms current, namely from part iii, 340V 8 × ¼ × 10Ω + 1Ω + j  ω 100µH − 1 
 s
IPrms=101.3A rms. The 0.47µF resonant capacitor voltage is given by π2  ωs 0.47µF 
I
vcap = I Prms X c = Prms
ωo C R
101.3A 2
= = 1476V rms 0.66 =
146krad/s × 0.47µF  1 
3 + j  ωs 100µH −
The resonant circuit capacitor has an rms current rating requirement of 101.3A rms and  ωs 0.47µF 
an rms voltage rating of 1476 V rms.
 1 
1 = 1 + j 1  ωs 100µH − 
v. From part ii, the average input current is 91.2 A. The supply input power is 3 ωs 0.47µF 
therefore 340Vdc×91.2A ave = 31kW. The power dissipated in the resonant circuit Because of the high circuit Q = 14.6 and relatively high voltage transfer ratio
2
resistance Rc =1Ω is given by I Prms × Rc = 101.32 × 1Ω = 10.26kW. Note that the coil vo / Vs = 0.66 , ωs is very close to ωo, as can be deduced from the plots in figure 15.18b.
power plus the load power (from part ii) equals the input power (20.8kW+10.26kW = The output voltage control will be very sensitive to changes in the H-bridge switching
31kW). The efficiency is frequency.
output power ♣
η= × 100%
input power
20.8kW 15.9.3 Resonant-switch, dc-to-dc converters
= × 100 = 67.1%
31kW
There are two forms of resonant switch circuit configurations for dc-to-dc converters,
namely resonant voltage and resonant current switch commutation. Each type reduces
vi. The average input dc current is 91.2A dc while the resonant bridge rms current is the switching losses to near zero.
101.3A rms. By Kirchhoff’s current law, the 340V dc rail decoupling capacitor ac ƒ In resonant current commutation the switching current is reduced to zero by
current is given by an L-C resonant circuit current greater in magnitude than the load current,
I ac = I Prms
2
− I Pave
2
such that the switch is turned on and off with zero current.
ƒ In resonant voltage commutation the switch voltage is reduced to zero by the
= 101.32 − 91.22 = 44.1A ac capacitor of an L-C resonant circuit with a voltage magnitude greater than the
This is the same ac current magnitude as the current in the dc capacitor across the load output voltage, such that the switch can turn on and off with zero voltage.
in the secondary circuit, 22A, when the transformer turns ratio, 2, is taken into account. Figure 15.23a shows the basic single switch, forward, step-down voltage switch mode
vii. The voltage across the load resistance is given by equation (15.177) dc-to-dc converter. Resonant switch converters are an extension of the standard switch
mode forward converter, but the switch is supplement with passive components LR-CR
to provide resonant operation through the switch, hence facilitating zero current or
voltage switching. A common feature is that the resonant inductor LR is in series with
the switch to be commutated. Parasitic series inductance is therefore not an issue with
resonant switch converters.
573 Switched-mode and Resonant dc Power Supplies Power Electronics 574

create four full-wave resonant switch circuits (the commutation type, namely voltage or
current, is also interchanged). An important operational requirement is that the average
T1 LR CR LR
load current never falls to zero, otherwise the resonant capacitor CR can never fully
discharge when performing its zero switch current turn-off function.
Lo + Lo +
Vs CR vo Vs T1 vo
D1 Co D1 Co
- 15.9.3i Zero-current, resonant-switch, dc-to-dc converter
-
The zero current switching of T1 in figure 15.24 (15.23b) can be analysed in five
distinctive stages, as shown in the capacitor voltage and inductor current waveforms.
The switch is turned on at to and turned off after t4 but before t5.
The circuit has attained steady state load conditions from one cycle to the next. The
cycle commences, before to, with both the capacitor voltage and inductor current being
T1 (b) ½-wave ZCS (c) zero, and the load current is freewheeling through D1. The current in the output
Lo inductor Lo, is large enough such that its current, Io can be assumed constant. The
+
Vs
D1 Co vo (a) switch T1 is off.
- Time interval I
At to the switch is turned on and the series inductor LR acts as a turn-on snubber for the
(d) ½-wave ZVS (e) switch. In the interval to to t1, the supply voltage is impressed across LR since the switch
T1 is on and the diode D1 conducts the output current, thereby clamping the inductor to
zero volts. Because of the fixed voltage Vs, the current in LR increases from zero,
linearly to Io in time
CR t I = I o LR / VS (15.183)
LR LR according to
DR DR V t
iL ( t ) = s (15.184)
Lo + Lo
R LR
+
Vs vo
Vs T1 vo Time interval IIA
T1 CR D1 Co D1 Co
- - When the current in LR reaches Io at time t1, the capacitor CR and LR are free to
resonant. The diode D1 blocks as the voltage across CR sinusoidally increases. The
constant load current component in LR does not influence its ac performance since a
Figure 15.23. Dc to dc resonant switch converters:
constant inductor current does not produce any inductor voltage. Its voltage is specified
(a) conventional switch mode forward step-down converter; by the resonant cycle, provided Io > Vs / Zo. The capacitor resonantly charges to twice
(b) and (c) half-wave zero current switching ZCS resonant switch converters; and the supply Vs when the inductor current falls back to the load current level Io, at time t3.
(d) and (e) half-wave zero voltage switching ZVS resonant switch converters. Time interval IIB
Between times t3 and t4 the load current is displaced from LR by charge in CR, in a quasi
The resonant capacitor CR, can be either in a parallel or series arrangement as shown in resonance process. The resonant cycle cannot reverse through the switch once the
figure 15.23, since small-signal ac-wise the connections are the same. A well- inductor current reaches zero at time t4, because of the series blocking diode (the
decoupled supply is essential when the capacitor CR is used in the parallel switch switch must have uni-directional conduction characteristics).
arrangement, as shown in figure 15.23 part b and d. A further restriction is that a diode The time for period II is approximately
must be used in series or in antiparallel with T1 if a switch without reverse blocking
 
tII =  π − sin −1  o o   / ωo
IZ
capability is used. The use of an antiparallel connected diode changes the switching (15.185)
  Vs  
arrangement from half-wave resonant operation with reverse impressed voltage switch
commutation to full-wave resonant operation with current displacement commutation, where Z o = LR / CR , while the capacitor voltage is given by
independent of the switch reverse blocking capabilities. Reconnecting the capacitor CR R
vC ( t ) = Vs (1 − cos ωo t ) (15.186)
terminal not associated with Vs, to the other end of inductor LR in figure 15.23b-e, will
575 Switched-mode and Resonant dc Power Supplies Power Electronics 576

T1 LR Time interval III

At time t4 the input current is zero and the switch T1 can be turned off with zero
Lo current, ZCS. The constant load current requirement Io is provided by the capacitor,
+ which discharges linearly to zero volts at time t5. according to
D1 Co vo
CR Io
Vs
- (a) vC ( t ) = VC − ×t (15.187)
R R
(1 − cos ω t ) C
o II R

The time for interval III is therefore load current dependant and is given by
VC CR
tIII = × (1 − cos ωtII )
R
(15.188)
Io
T1 T1
on T1 on Time interval IV
off
Io + Vs/Zo IT1 After t5, the switch is off, the current freewheels through D1, the capacitor voltage is
zero and the input inductor current is zero. At time t1 the cycle recommences. The
interval IV, t5 to t0, is used to control the rate at which energy is transferred to the load.
Io
ICR ID1 The output voltage can be specified by either evaluating the energy from the supply,
through the input resonant inductor LR, or by evaluating the average voltage across the
t0 t1 t2 t3 t4 t5 t0
resonant capacitor CR which is filtered by the output filter Lo - Co.
By considering the input inductor energy for each shown period, from the waveforms
2Vs in figure 15.24b, the output voltage is given by
VCR
V
I IIA IIB III IV vo = s (½tI + tII + tIII ) (15.189)
INTERVALS
τ
Vs Vs where the switching frequency f s = 1/ τ .
The output voltage based on the average capacitor voltage is
1 t t t 
vo =  ∫ Vs (1 − cos ωt ) dt + ∫ Vs
4 5
dt 
τ t 1 t t5 − t4  4
t1 t2 t5 t1 (15.190)
1  Vs  3π  
VLR (b) =  × +1 × 4 3 × ( t4 − t1 ) + ½ × Vs × ( t5 − t4 ) 
τ  2π  2  
(c) The circuit has a number of features:
i. Turn-on and turn-off occur at zero current, hence switching losses are minimal.
ii. At light load currents the switching frequency may become extreme low.
IT1 IT1
Io Io ii. The basic half resonant period is given by tII = π LR CR
Io Io Io Io
iii. The capacitor discharge time is tIII ≤ 2 × Vs × CR / I o , thus the output voltage is
Vs Vs Vs Vs
load current dependant.
iv. LR and CR are dimensioned such that the capacitor voltage is greater than Vs at
time t4, at maximum load current Io.
v. Supply inductance is inconsequential, decreasing the inductance LR requirement.
vi. Being based on the forward converter, the output voltage is less than the input
Figure 15.24. Zero current switching, ZCS, resonant switch dc to dc converter: voltage. The output increases with increased switching frequency.
(a) circuit; (b) waveforms; and (c) equivalents circuits.
577 Switched-mode and Resonant dc Power Supplies Power Electronics 578

vii. If a diode in antiparallel to the switch is added as shown dashed in figure 15.23b,
reverse inductor current can flow and the output voltage is vo ≈ Vs × f s / f o . CR
Operation of the ZCS circuit in figure 15.23c, where the capacitor CR is connected in LR
parallel with the switch, is essentially the same as the circuit in figure 15.24. The DR
capacitor connection produces the result that the capacitor voltage has a dc offset of Vs, Lo +
meaning its voltage swings between ± Vs rather than zero and twice Vs, as in the circuit T1 vo
(a)
Vs VD1
just considered. Any dc supply inductance must be decoupled when using the ZVS D1 Co
-
circuit in figure 15.23c.

15.9.3ii Zero-voltage, resonant-switch, dc-to-dc converter

T1 T1
off T1 off
The zero voltage switching of T1 in figure 15.25 (15.23e) can be analysed in five Vs + I o Z o on
VCR
distinctive stages, as shown in the capacitor voltage and inductor current waveforms.
The switch is turned off at to and turned on after t4 but before t5.
The circuit has attained steady state load conditions from one cycle to the next. The Vs
cycle commences, before to, with the capacitor CR voltage being Vs and the load current VD1
Vo
Io being conducted by the switch and the resonant inductor, LR. The output inductor Lo
is large enough such that its current, Io can be assumed constant. The switch T1 is on.
t0 t1 t2 t3 t4 t5 t6 t0

Time interval I
Io Io
At time to the switch is turned off and the parallel capacitor CR acts as a turn-off
snubber for the switch. In the interval to to t1, the supply current is provided from Vs I II III IV INTERVALS
through CR and LR. Because the load current is constant, Io, due to large Lo, the IT1
capacitor charges linearly from 0V until its voltage reaches Vs in time 0
V C ICR
tI = s R (15.191) ILR
Io (b)
according to -Io
I t1 t2 t5 t6 t1
vc ( t ) = Vs − o × t (15.192)
CR
Time interval II
When the voltage across CR reaches Vs at time t1, the load freewheel diode conducts,
clamping the load voltage to zero volts. The capacitor CR and LR are free to resonant, (c)
0V
where the initial inductor current is Io and the initial capacitor voltage is Vs. The
energy in the inductor transfers to the capacitor, which increases its voltage from Vs ILR ILR ILR ILR
Io Io Io Io
to a maximum at time t2 of
v C = Vs +I o Z o (15.193) Vs Vs ID1 Vs ID1 Vs

Figure 15.25. Zero voltage switching, ZVS, resonant switch dc to dc converter:

(a) circuit; (b) waveforms; and (c) equivalents circuits.
579 Switched-mode and Resonant dc Power Supplies Power Electronics 580

The capacitor energy transfers back to the inductor which has resonated from + Io to – The circuit has a number of features:
Io between times t1 to time t3. For the capacitor voltage to resonantly return to Vs, Io > i. Switch turn-on and turn-off both occur at zero voltage, hence switching losses
Vs / Zo. Between t3 and t4 the voltage Vs on CR is resonated through LR, which are minimal.
conducts – Io at t3, as part of the resonance process. The capacitor voltage and ii. At light load currents the switching frequency may become extreme high.
inductor current during period II are given by iii. The basic half-resonant period is approximately given by t1−4 = π LR CR
vc ( t ) = Vs + I o Z o sin ωo t iv. The inductor discharge time is t III ≤ 2 × I o × LR / Vs , hence the output voltage is
(15.194) load current dependant.
iL ( t ) = I o cos ωo t
v. LR and CR are dimensioned such that the inductor current is less than zero (being
and the duration of interval II is returned to the supply Vs) at time t5, at maximum load current Io.
 V  vi. Being based on the forward converter, the output voltage is less than the input
tII =  π + sin −1 s  / ωo (15.195)
 Io Zo  voltage. Increasing the switching frequency decreases the output voltage since τ-
At the end of interval II the capacitor voltage is zero and the inductor current is t5 is decreased in equation (15.199).
Operation of the ZVS circuit in figure 15.23d, where the capacitor CR is connected in
iL ( t II ) = I o cos ωo t II (15.196) parallel with the load circuit (the freewheel diode D1), is essentially the same as the
Time interval III circuit in figure 15.25. The capacitor connection produces the result that the capacitor
At time t4 the voltage on CR attempts to reverse, but is clamped to zero by diode DR. voltage has a dc offset of Vs, meaning its voltage swings between + Vs and -Io Zo, rather
The inductor energy is returned to the supply Vs via diode DR and the freewheel diode than zero and Vs - Io Zo, as in the circuit just considered. Any dc supply inductance
D1. The inductor current decreases linearly to zero during the period t4 to t5. During must be decoupled when using the ZVS circuit in figure 15.23d.
this period the switch T1 is turned on. No turn-on losses occur because the diode DR It will be noticed that a ZCS converter has a constant on-time, while a ZVS converter
in parallel with T1 is conducting during the period the switch is turned on, that is, the has a constant off-time.
switch voltage is zero and the switch T1 can be turned on with zero voltage, ZVS.
With the switch on at time t5 the current in the inductor LR reverses and builds up, Example 15.10: Zero-current, resonant-switch, dc-to-dc converter
linearly to Io at time t6. The current slope is supply Vs dependant, according to Vs = LR
di/dt, that is The ZCS resonant dc step-down voltage converter in figure 15.24a produces an output
V voltage for the armature of a high voltage dc motor and operates from the voltage
iL ( t ) = s t + I o cos ωo tI (15.197) produced from the 50Hz ac mains rectified, 340V dc, with an L-C dc link filter. The
LR
resonant circuit parameters are LR = 100µH, CR = 0.47µF, and the high frequency ac
and the time of period III is load current dependant:
resistance of the circuit is Rc = 1Ω.
I L
tIII = o R × (1 − cos ωo t1 ) (15.198) Calculate
Vs i. the circuit Zo, Q, and ωo
Time interval IV ii. the minimum output current to ensure ZCS
At t6, the supply Vs provides all the load current and the diode D1 recovers with a iii. the maximum operating frequency, represented by the time between switch turn
controlled di/dt given by Vs /LR. The switch conduction interval IV, t6 to to, is used to on and the freewheel diode recommencing conduction, at minimum load current
control the rate at which energy is transferred to the load. iv. the average diode voltage, hence load voltage at the maximum frequency.

The output voltage can be derived from the diode voltage (shown hatched in figure Solution
15.25b) since this voltage is averaged by the output LC filter.
1 i. The characteristic impedance is given by
vo = ( Volt × second area of interval I + Volt × second area of interval IV )
τ LR 100µH
(15.199) Zo = = = 14.6Ω
1 CR 0.47µF
= (½t1 + τ − t5 ) = Vs (1 − f s ( t6 − ½t1 ) )
τ The resonant circuit Q is
581 Switched-mode and Resonant dc Power Supplies Power Electronics 582

Zo 100µH The minimum commutation cycle time is therefore 6.85+32.25+6.86 = 46µs. Thus the
Q= = /1Ω = 14.6 maximum operating frequency is 21.7kHz.
Rc 0.47µF
For this high Q, the circuit resonant frequency and damped frequency will be almost iv. The output voltage vo is the average reverse voltage of freewheel diode D1, which
the same, that is is in parallel with the resonant capacitor CR. Integration of the capacitor voltage shown
ω ≈ ωo = 1/ LR CR in figure 15.24b gives equation (15.190)
= 1/ 100µH × 0.47µF = 146 krad/s 1 t t t 
vo =  ∫ Vs (1 − cos ωt ) dt + ∫ Vs
4 5
dt 
t5  t t t5 − t 4 
= 2π f
1 4

f = 146 krad/s /2π = 23.25 kHz 1  32.25µs 6.86µs t 

340V × (1 − cos ωt ) d ωt + ∫
46µs  ∫ o
= × 340V × dt
or T = 43µ s
0 6.86µs 
1   3π  43µs 
= × 340V ×  +1 × + ½ × 340V × 6.86µs 
ii. For zero current switching, the load current must be greater than the resonant 46µs   2  2π 
current, that is
1
I o > Vs / Z o = 340V/14.6Ω = 23.3A = × [13292Vµs + 1166Vµs ] = 314.3Vdc
46µs
iii. The commutation period comprises the four intervals, I to IV, shown in figure The maximum output voltage is 314V dc. Alternatively, using the input inductor
15.24b. energy based equation (15.189):
Interval I V
vo = s (½t I + t II + III + tIV )
The switch turns on and the inductor current rises from 0A to 23.3A in a time given by τ
t I = LR ∆I / Vs 340V
= × (½ × 6.85µs + 32.25µs + 6.85µs ) = 314.8V
=100µH×23.3A/340V = 6.85µs 46µs
Intervals II and III
These two interval take just over half a resonant cycle, which takes 43µs/2 = 21.5µs to ♣
complete. Assuming action is purely sinusoidal resonance with a 23.3A offset, from 0A
to a maximum of 23.3A and down to -23.3A then from Example 15.11: Zero-voltage, resonant-switch, dc-to-dc converter
I o = Vs / Z o sin ω t for ω t > π The zero voltage resonant switch converter in figure 15.25 operates under the
following conditions:
23.3 A = −23.3A × sin ωt gives
Vs = 192V Io = 25A
t = ¾×43µs LR = 10µH CR = 0.1µF
= 32.25µs Determine
The capacitor voltage at the end of this period is given by i. the switching frequency fs for vo = 48V
VcIV = Vs (1 − cos ωt ) ii. switch average current and
iii. the peak switch/diode/capacitor voltage.
= 340V × (1 − cos 3 2 π ) iv. the minimum output current
= 340V
Interval IV Solution
The capacitor voltage must discharge from 340V dc to zero volts, providing the 23.3A i.
load current. That is 1 1
ωo = = = 1× 106 rad/s that is f o = 159.2kHz
t IV = VcIV × CR / I o LR CR 10µH × 0.1µF
= 340V×0.47µF/23.3A = 6.86µs
583 Switched-mode and Resonant dc Power Supplies Power Electronics 584

LR 10µH 15.10 Appendix: Analysis of non-continuous inductor current operation

Zo = = = 10Ω
CR 0.1µF
Operation with constant input voltage, Ei
The period of interval I is given by equation (15.191), that is In applications were the input voltage Ei is fixed, as with rectifier ac voltage input
V C 192V × 0.1µF circuits, the output voltage vo can be controlled by varying the duty cycle.
tI = s R = = 0.768µs
Io 25A In the continuous inductor conduction region, the transfer function for the three basic
The period of interval II is given by equation (15.195), that is converters is determined solely in terms of the on-state duty cycle, δ. Operation in the
discontinuous current region, for a constant input voltage, can be characterised for each
 V   192V 
t II = t3 − t1 =  π + sin −1 s  / ωo =  π + sin −1  /10 rad/s = 4.017µs
6
converter in terms of duty cycle and the normalised output or input current, as shown in
 Io Zo   25A × 10Ω  figure 15.26. Key region and boundary equations, for a constant input voltage Ei, are
The period for the constant current period III is given by equation (15.198) summarised in tables 15.5 and 15.6.
I L 25A × 10µH
t III = o R × (1 − cos ωo t1 ) =
Vs 192V
( )
× 1 − cos (106 × 0.768µs ) = 0.365µs
Operation with constant output voltage, vo
In applications were the output voltage vo is fixed, as with regulated dc power supplies,
After re-arranging equation (15.199), the switching frequency is given by
the effects of varying input voltage Ei can be controlled and compensated by varying
 vo   48V  the duty cycle.
1 −  1 − 
Vs   192V 
fs = 
In the inductor continuous current conduction region, the transfer function is
= = 157.4kHz
t5 − ½t1 ( 0.768µs + 4.017µs + 0.365µs − ½ × 0.768µs ) determined solely in terms of the on-state duty cycle, δ. Operation in the discontinuous
region, for a constant output voltage, can be characterised in terms of duty cycle and
ii. The switch current is shown by hatched dots in figure 15.25. The average the normalised output or input current, as shown in figure 15.27.
value is dominated by interval IV, with a small contribution in interval II between t5 Key region and boundary equations, for a constant output voltage vo, are summarised in
and t6. table 15.7.
Because of the invariance of power, the output current I o characteristics for each
I  ½ × t III 
IT = o × + (τ − t6 )  converter with a constant input voltage Ei , shown in figure 15.26, are the same as those
τ  1 + cos ωo tII 
 for the input current I i when the output voltage vo is maintained constant, as shown in
 figure 15.27. [That is the right hand side of each plot is the same.]
½ × 0.365µs  1 
= 25A × 157.4kHz  + − ( 0.768µs + 4.017µs + 0.365µs )  
 1 + cos (106 × 4.017µs )  157.4kHz  

= 5.17A

iii. The peak switch/diode/capacitor voltage is given by equation (15.193),

namely
v C = Vs +I o Z o
= 192V + 25A × 10Ω = 442V

iv. For proper resonant action the minimum average output current must satisfy,
I o > Vs / Z o , that is
∨ Vs 192V
Io = = = 19.2A
Zo 10Ω
♣
585 Switched-mode and Resonant dc Power Supplies Power Electronics 586

Table 15.5. Transfer functions with constant input voltage, Ei, with respect to I o Table 15.6. Transfer functions with constant input voltage, Ei, with respect to I i

Ei converter vo converter
constant step-down step-up step-up/down constant step-down step-up step-up/down
reference reference equation (15.4) (15.44) (15.74)
(15.4) (15.44) (15.74)
equation
vo vo 1 vo −δ
current vo vo 1 vo −δ current conduction =δ = =
=δ = = Ei Ei 1 − δ Ei 1 − δ
conduction Ei Ei 1 − δ Ei 1 − δ
reference equation (15.20) (15.60) (15.91)
reference
(15.21) (15.59) (15.91) vo 1
equation vo 2 LI = vo voτδ 2
vo 1 equation = 1− 2 i Ei Ei tT δ 2 =
= vo δ Ei tT
2
vo δ Eiτ 2 Ei δ τ Ei 1− Ei 2 LI i
discontinuous Ei = 1+ =− 2 LI i
current 1 + 22LI o Ei 2 LI o Ei 2 LI o
δ τ Ei vo 1
vo 4 I = Ii
normalised vo 1 normalised =1− × ∧i Ei I  1=δ2 / ∧
= vo I vo I Ei 27δ 2 I 1−δ /  ∧i2

∧ Eiτ Ei 1 = 1 + 4δ 2 / ∧ o = − 4δ 2 / ∧ o   Ii
Io = I i
 Ii 
8L 1+ 2 × ∧o Ei Ei
4δ Io Io
@δ=½ Io ∧

I i max @ δ = δ=⅔ δ=1 δ=1

 v 
4δ 2 1 − o  Io 4δ 2 Io − 4δ 2 ∧ ∧ 4 Eiτ ∧ Eiτ ∧ Eiτ
change of Io  Ei  = = Ii Ii = × Ii = Ii =
=
∧
vo ∧
vo 27 2 L 2L 2L
variable ∧
vo Io −1 Io
Io Ei Ei v
Ei δ2 o
Ii  v  Ii Ei Ii
v  vo change of variable ∧ = 27 4 δ 2  1 − o  ∧ = ∧ = −δ 2
4 o − 1 −4 Ii  Ei  I i  vo − 1 Ii
Io vo  vo  Io  Ei  Io Ei
boundary ∧
=4 1 −  = ∧
= 2  Ei 
Ei  Ei 
∧
 vo 
2
 vo 
Io Io Io
1 + 
2
   vo   vo 
 Ei   Ei  2
 − 1  E 
I i 27  vo  vo  I i  Ei  Ii
boundary ∧ = 4  1 −   = = − i 
 Ei  Ei 
∧
duty cycle I o × vo Ii
∧
vo  vo 
Ii Ii E + 1 
all with boundary ∧
v  v Ei
Ei
δ = ½ I∧ o × o  i 
Io I
δ =½ o v δ = ½ I∧ o ×  o − 1
= 4δ (1 − δ )  Ei  Ei
∧
1−
o
Io Io Ii vo
Io Ei ∧ −1
duty cycle Ii Ei Ii
δ= 4 Ii δ= ∧ × δ= ∧
27 vo Ii vo Ii
1−
Ei Ei

Ii Ii δ
boundary = 27 4 δ 2 (1 − δ ) = δ= Ii
(1 − δ )
∧ ∧ 2 ∧
Ii Ii Ii
587 Switched-mode and Resonant dc Power Supplies

Table 15.7. Transfer functions with constant output voltage, vo, with respect to I o

vo converter
constant step-down step-up step-up/down
reference equation (15.4) (15.44) (15.74)
vo vo 1 vo −δ
current conduction =δ = =
Ei Ei 1 − δ Ei 1 − δ
reference equation (15.20) (15.60) (15.91)
vo 1
vo 2 LI = vo voτδ 2
equation = 1− 2 i Ei Ei tT δ 2 =
Ei δ τ Ei 1− Ei 2 LI i
2 LI i
2 vo 1
vo 1 I v  = I v 
normalised = 1 − 2 × ∧o ×  o  Ei  I  v 2  vo
Ei 4δ E = δ 2 /  ∧o × o 
Io  i  1 − 4δ /  ∧ o ×  o    Ei 
2
Ei
 E   Io 
 Io  i  
∧

I o max @ δ = δ=0 δ=⅓ δ=0

∧ ∧ voτ ∧ 4 voτ ∧ voτ
Io Io = Io = × Io =
2L 27 2 L 2L
 v 
δ 2 1 − o  Io 27
4δ
2
Io −δ 2
Ei  = =
= 
Io ∧ 2
change of variable
∧
 vo  vo Io  vo 
∧
 vo 
2
Io  − 1  
Io  Ei  Ei
   Ei 
 Ei 
 vo 
27
4 − 1 Io −1
Io v Io  Ei  =
boundary = 1− o =
∧
 vo 
2
∧
Ei Io
1 + 
∧ 3
Io Io  vo 
   Ei 
 Ei 

Io
∧
v v vo Io
duty cycle v
δ= o Io δ= 4
27 × I∧ o  o − 1 o δ= ∧
vo E  Ei Ei
Ei
1− Io  i Io
Ei

δ = 1 − I∧ o Io
= 27 4 δ (1 − δ ) Io
2
boundary ∧ δ = 1− ∧
Io Io Io
 Power Electronics 588

( )
d2 vo 0
Ei 4 vo −1
Ei
vo

( )
−1 2
Ei vo
vo Ei
Ei
v
δ 4δ
2
o 2
E i
d =½ d =½
vo
v o
−1
2 −1
E Ei
vo 1 i
vo 1
= =
Ei 1− δ Ei 1− δ

d =¼ dis continuous d =¼

2
 vo  vo 
4 1 − Ei  
27
d =0 1 d =0
d =1
  Ei  d =1
d =¾ d =¾
vo
=δ
( )
 v 
d =½ 2 v 4δ 2 1− o E  Ei d =½
4δ 1−
27 o
 i
vo E ½ vo
=δ i
Ei
d =¼ Ei dis continuous d =¼
vo  v 
4 1− o 
Ei  Ei 
d=0   d =0
0
?

?
1 1
Ii / I i Io / I o

d =¼ d =¼
-½
v −δ −δ
o
= v o
=
E i 1− δ E i 1 −δ
− 4δ
2
-1
d =½ vo d =½
Ei
continuous
2 dis continuous
 vo  vo
 Ei  −4
−  Ei

( )
vo
 E +1  -2 vo
2

  1+
?

i
Ei
I / I
i i
vo 0
Ei constant =δ
2 - Ei constant
Ei
Figure 15.26. Characteristics for three dc-dc converters,
when the input voltage E i is held constant.
589 Switched-mode and Resonant dc Power Supplies

( ) ( )
0 vo 0
27 vo − 1 4 vo −1
4 Ei
Ei Ei

( )
3 2
vo vo
Ei Ei

4δ
2

d=½ d =½
vo
2 −1
dis continuous Ei
1 vo 1
vo
= =
Ei 1 −δ
Ei 1− δ
4 δ
2
27

( )
d=¼ d =¼
v v
o
−1 o

Ei
E i

d=0 1 d =0
d=1 d =1
d=¾ d =¾
vo
1 − vo  v 
Ei
=δ
d=½ Ei 4δ 2 1− o E 
 i
d =½
vo ½ vo

d=¼ Ei = δ  v
δ 2 1− o E 

Ei
d =¼
 i
2 vo  v 
 vo  1− o 
 E  4 Ei  Ei 
d=0  i   d=0
?

1 −δ 0
2

?
1
Io / I
( )
o 2
Ii / I i
v o

d=¼ E i d =¼
-½
v −δ v −δ
o
= o
=
E
i 1− δ E i 1 −δ
-1 − 4δ 2
d=½ vo d =½

Ei

continuous
−1 dis continuous vo
−4
( ) Ei
2
vo
1+
( )
Ei -2 2
vo
1+
Ei
v
vo constant 0 - o 0 vo constant
Ei
Figure 15.27. Characteristics for three dc-dc converte rs,
when the output voltage vo is held constant.
 Power Electronics 590

Reading list
Fisher, M. J., Power Electronics,
PWS-Kent Publishing, 1991.

Hart, D.W., Introduction to Power Electronics,

Prentice-Hall, inc, 19974.

Hnatek, E. R., Design of Switch Mode Power Supplies,

Van Nostrand Rein hold, 1981.

Mohan, N., Power Electronics, 3rd Edition,

Wiley International, 2003.

Thorborg, K., Power Electronics – in theory and practice,

Chartwell-Bratt, 1993.

http://www.ipes.ethz.ch/

Problems

15.1. An smps is used to provide a 5V rail at 2.5A. If 100 mV p-p output ripple is
allowed and the input voltage is 12V with 25 per cent tolerance, design a flyback buck-
boost converter which has a maximum switching fre quency of 50 kHz.

15.2. Derive the following design equations for a flyback boost converter, which
operates in the discontinuous mode.
∧ v 1
i i = 2 × Io (max) × o = constant tD =
Ei (min) v
f (max) o
Ei (min)
vo − E i(min) 1 I v − Ei
L = tT (min) f = = f (max) o × o
∧
i τ Io (max) vo − E i(min)
i

∧
∆Q i tT (min) ∆eo
C (min) = = ESR(max) =
i

∆eo 2 ∆eo ∧
i i

15.3. Derive design equations for the forward non-isolated converter, operating in
the continuous conduction mode.

15.4. Prove that the output rms ripple current for the forward converter in figure
15.2 is given by ∆i o / 2 3 .