AN628 - STMicroelectronics

AN628
APPLICATION NOTE
DESIGNING A HIGH POWER FACTOR SWITCHING
PREREGULATOR WITH THE L4981 CONTINUOUS MODE
by Ugo Moriconi
INTRODUCTION
Conventional AC-DC converters usually employ a full wave rectifier bridge with a simple capacitor filter to draw
power from the AC line. This "bulk" capacitor must be big enough to supply the total power during most of each
half-cycle, while instantaneous line voltage is below the DC rectified voltage. Consequentely, the line current
waveform is a narrow pulse, and the power factor is poor (0.5-0.6) due to the high harmonic distortion of the
current waveform.
If a high power factor switching preregulator is interposed between the input rectifier bridge and the bulk filter
capacitor, the power factor will be improved (up to 0.99). Increasing in addition, the RMS current capability from
the mains, reducing the bulk capacitor peak current and the harmonic disturbances.
Switching at a frequency much higher than the line's one, the preregulator draws a sinusoidal input current, in
phase with the input line voltage.
There are several way that this can be accomplished. When the output voltage is higher than the input voltage
(Vo > Vin), BOOST topology and continuos inductor current control mode are well suited to produce a good
quality input sine current waveform. The input di/dt is low because the inductor is located between the bridge
and the switch. This minimizes line noise and the line spikes will be absorbed by the inductor.
Figure 1. L4981 Block Diagram
April 2004
1/35
AN628 APPLICATION NOTE
THE L4981 PFC CONTROLLER IC
The L4981 integrated circuit is a continous mode average current controller with several specific functions for
active power factor correction. It can operate in high quality, medium/high power conversion range and provides
all the necessary features to achieve a very high power factor, up to 0.99. Thanks to the BCD technology used,
operative switching frequency higher than 200kHz can be used.
The L4981 can be used in systems with universal input mains voltage without any line switch.
This new PFC offers the alternative of synchronization working at fixed frequency (L4981A), or working in modulated frequency (L4981B) to optimize the size of the input filter. Both devices control the conversion in average
current mode PWM to maintain a sinusoidal line current without slope compensation.
MAIN FEATURES:
■
Switching frequency higher than 200 kHz.
■
Under Voltage Lockout with hysteresis and programmable turn-on threshold.
■
Overvoltage and Overcurrent Protection.
■
Precise (2%) on chip Reference externally available.
■
Input/Output Synchronization (only for L4981A).
■
Feed Forward Line and Load regulation.
■
Universal input mains.
■
Average current mode PWM.
■
High Output Current totem pole driver.
■
Low Start-up supply current.
■
Soft Start.
P.F.C. BOOST TOPOLOGY OPERATION
The operation of the P.F.C. boost converter (see fig. 2) can be summarized in the following description.
The A.C. line voltage is rectified by a diode bridge and the rectified voltage delivered to the boost converter. The
boost converter section, using a PWM switching technique, boosts the rectified input voltage to a D.C. controlled
output voltage (VO). The section consists of a boost inductor (L), a controlled power switch (Q), a boost diode
(D), an output capacitor (CO) and, obviously, a control circuitry.
Referring to the time-variable mains voltage (sine waveform), the converter produces a boost inductor average
current like the rectified input voltage, changing continuosly the duty-cycle of the active switch (Q).
The boosted D.C. voltage is controlled to a programmed value, higher than the maximum input instantaneous
voltage (VIpk).
Referring to the main currents shown in fig.2 schematic, the simplified formulae are (assuming: power efficiency
= 1; output ripple voltage = 0; high frequency inductor ripple current = 0):
1) Peak inductor (L), switch (Q) and diode (D) currents
PO
I Lpk = I Qpk = I Dpk = 2 ⋅ ----------V lpk
2) RMS inductor current
I Lrms =
2/35
PO
2 ⋅ ----------V lpk
AN628 APPLICATION NOTE
Figure 2.
L
D
~
IQ
IL
ID
IO
IC
CONTROLLER
~
Q
CO
LOAD
Cin
D94IN119
3) RMS switch current
PO
16 ⋅ V lpk
I Qrms = ----------- ⋅ 2 – ---------------------V lpk
3 ⋅ π ⋅ VO
4) Average diode current
IDavg = IO
5) RMS diode current
PO
16 ⋅ V lpk
I Drms = ----------- ⋅ ---------------------V lpk 3 ⋅ π ⋅ V O
6) Total RMS capacitor (CO) current
16 ⋅ V O
-–1
I C = I O -------------------------3 ⋅ π ⋅ V Ipk
7) RMS twice line frequency capacitor current
IO
I C ( 2f )rms = -----2
8) RMS high frequency capacitor current
16 ⋅ V O
– 1.5
I C ( hf )rms = I O -------------------------3 ⋅ π ⋅ V lpk
The figure 3 shows the above mentioned quantities, normalized to the D.C. output current (IO), plotted versus
VIpk / VO ratio. Moreover, the ILpk · ILrms normalized to IO2 value, related to the inductor energy (I2 · L), is plotted
in the diagram (dotted line). This last plot gives an idea on the heavy increase of the inductor size operating
with large input voltage range.
Obviously, in real application the efficiency is less than 100% (η < 1). The output voltage ripple, related to the
output capacitor (CO) is a parameter to be considered. The inductor high frequency current ripple (∆IL) is another parameter affected by the inductor value (L), the switching frequency (fsw) and the delivered power (PO).
3/35
AN628 APPLICATION NOTE
Figure 3.
I Lpk ⋅ I Lrms
---------------------------2
Io
Figure 4.
4/35
AN628 APPLICATION NOTE
CONTROLLER FUNCTION DESCRIPTION.
The L4981 I.C. controls the conversion process with a continuous mode average current method, using two control loops (current loop and voltage loop) see fig. 5. Moreover, several internal functions ensure high quality conversion performance.
A description of the internal blocks will be detailed in the design criteria section and pin description. However,
referring to fig. 4, here below a brief description of the main functions is done:
Multiplier block.
This block produces an output current (programming current) as a product result of four different input signals
(see fig. 13 for details). The multiplier output current, through a resistor connected to the negative side of a
sense resistor, determines the error signal to the current loop.
Figure 5.
VOLTAGE
CONTROL LOOP
LINE
CURRENT
CONTROL LOOP
POWER
STAGE
LOAD
D94IN061
Figure 6.
L
REF
Raux
11 8.5V
OSC
S
Iipk
Iaux
Q
IPK
-
R
+
Ripk
I
1,10
RS
GND
D94IN062A
Ipk
Operational amplifier blocks.
Two amplifiers allow loop control. The first one (E/A), feeds back the output voltage (VO) and delivers its output
to the multiplier block. The second (C/A), feeds back the line current and produces the reference for the PWM
section.
5/35
AN628 APPLICATION NOTE
PWM block.
This block, comparing the sawtooth produced by the oscillator, with the reference signal from the C/A output,
modulates its output signal duty-cycle. Its output, by the logic and driver sections, allows the controlled switch
(Q) to modulate the inductor current.
Logic block.
Controls the flow from the PWM and the output with the Auxiliary function signals and soft start.
Driver block.
The driver supplies the gate current to turn on and off the power switch (Q). It delivers up to 1A peak current to
allow high switching frequency applications.
Aux functions.
The Auxiliary functions allow to avoid overstress on power components of the application.
Power supply block.
This circuitry delivers the internal supply and references, recognizes the Undervoltage and Stand-by conditions
to save consumption.
P.F.C. BOOST DESIGN CRITERIA
L4981 PIN DESCRIPTION AND BIASING CIRCUITRY.
Pin 1. P-GND (Power stage ground). This pin, on the pc-board, has to be connected close the external Mosfet
source.
Pin 2. IPK (Overcurrent protection input). The current limitation is obtained with an internal comparator that holds
down the output driver when the voltage at IPK input goes down to zero. In the L4981A, to preset the IPK input
there is an internal current source (Iipk) of typically 85µA. The maximum peak current (Ipk) can be programmed
connecting (see fig. 6) a single resistor (Ripk) between this pin and the sense resistor (RS):
R S ⋅ I pk
R ipk = -----------------I ipk
In the L4981B, to preset the IPK input, an auxiliary resistor (Raux), connected from the VREF pin to the IPK pin,
is required. The maximum peak current (Ipk) can be programmed choosing (see fig. 6) the resistances Raux and
Ripk:
R S ⋅ I pk
R ipk = -----------------I aux
Where:
V VREF
I aux = ----------------R aux
Note: If used with the L4981A, the auxiliary resistor avoids that the current source spread affects the precision
of the protection simply getting an auxiliary current (Iaux) much higher than Iipk.
6/35
AN628 APPLICATION NOTE
Pin 3. OVP (Overvoltage protection input). A comparator with a precise 5.1V reference voltage and 250mV of
hysteresis, detects the overvoltage condition and turns the controller in stand-by condition (with low power consumption) and discharges the soft start capacitor. This pin (see fig. 7) has to be externally connected with a
resistive divider (Ra and Rb) to the D.C. output voltage. The divider ratio is defined by the relation:
V O + ∆V OUT
Ra
–1
-------- = --------------------------------5.1V
Rb
where: ∆VOUT is the output overvoltage limit.
Figure 7.
Pin 4. IAC (A.C. current input). This pin (see fig. 8) has to be connected through a resistor to the rectified line
voltage to drive the multiplier with a current (IIAC) proportional to the instantaneous line voltage:
VI
I IAC = --------R ac
The relation between the input alternate current (IIAC) and the output current (programming signal Imult) of the
multiplier is described at MULT-OUT section (pin8).
Figure 8.
Pin 5.CA-OUT (Current amplifier output). The CA_OUT deliveres its signal to the PWM comparator. An external
network (see fig. 9) defines the suitable loop gain to process the multiplier output and the line current signals.
To avoid oscillation problem (see fig. 10) the maximum inductor current downslope (VO/L) has to be lower than
oscillator ramp-slope (Vsrp · fsw):
VO
------- ⋅ R s ⋅ G ca ≤ V srp ⋅ f sw
L
7/35
AN628 APPLICATION NOTE
where:
Vsrp is the oscillator ramp peak-peak voltage.
Gca is the current amplifier gain.
fsw is the switching frequency.
and rewritten as:
V srp ⋅ f sw ⋅ L
G ca ≤ -------------------------------VO ⋅ Rs
Figure 9.
OSC
+
C/A
8
PWM
9
5
CA OUT
Rf
Ri'
Ri
Cf
Rs
D94IN063
Figure 10.
Rf
defines the high frequency C/A gain (1 + ----- ):
Ri
R f V srp ⋅ f sw ⋅ L
----- ≤ -------------------------------- – 1
Ri
VO ⋅ RS
To define the Cf value, it's useful to consider the current openloop gain, defined by the ratio between the voltage
across Rs and the current amplifier output signal:
v rs
G avg = ------v ca
8/35
AN628 APPLICATION NOTE
Because, in worst condition is:
Rs ⋅ VO
v rs = -----------------s⋅L
and the total variation of vca (the reference signal for PWM) is Vsrp:
Rs ⋅ VO
G avg = -----------------------------------V srp ⋅ 2π ⋅ f ⋅ L
Multiplying this Gavg by Gca and solving for the crossover frequency (f = fc), follows:
f sw
f c = ------2π
fc
To ensure a phase margin (higher than 45°), the zero frequency (fz) should be about ---- , than:
2
f sw
1
2
f z = ----------- = ------------------------------- ⇒ C f = -----------------2 ⋅ π ⋅ Cf ⋅ Rf
R f ⋅ f sw
4⋅π
Figure 12a.
Figure 11.
Gain [dB]
100
80
Gloop
Gca
RA
60
~
CA
RB
VRMS
CB
RC
7
1/V2
40
D94IN064
20
Gavg
0
-20
-40
-60
10
100
1000
f [Hz]
10000
100000
Pin 6. LFF (Load feed-forward input). This voltage input pin allows to modify the multiplier output current
proportionally to the load in order to improve the response time versus load transient. The control is
working with VLFF between 1.5V and 5.1V. If this
function is not used, the LFF pin has to be connected
to VREF pin.See also appendix A.
Pin 7. VRMS. Input to the divider (1/V2RMS), it is especially useful in universal mains applications to
compensate the gain variation related to the input
voltage change. It will be connected to an external
network (see fig. 12a) giving a voltage level proportional to the mains VRMS. The best control is reached
using a VRMS voltage level in the range between 1.5V
and 5.5V.
To avoid line current distorsion, the rectified mains
ripple (2f) level has to be reduced. A two pole filter,
with three resistors and two capacitors, setting the
lowest pole at 2Hz and the highest one at 13Hz, is
enough to get the useful voltage level reducing to 80dB the 100Hz gain.
Figure 12b.
The signal (pin 7), with the network in fig. 12a is:
9/35
AN628 APPLICATION NOTE
VRMS = 85V (110V -20%)
VRM(7) = 1.6V
VRMS = 260V (220V +20%)
VRM(7) = 5V
Gain at 2f(100Hz)
-80dB
Pin 8. MULT-OUT (Output of the multiplier). This pin deliveres the programming current (Imult) according to the
relation:
( V va – out – 1.28V ) ⋅ ( 0.8 ⋅ V LFF – 1.28V )
I mult = 0.37 ⋅ I IAC ⋅ ----------------------------------------------------------------------------------------------------------2
V RMS
where: VVA-OUT = E/A output voltage range
VLFF = voltage input at pin 6
VRMS = voltage input at pin 7
IIAC = input current at pin 4
To optimize the multiplier biasing for each application, the relation between Imult and the other input signals to
the multiplier are here reported (refer to figure 13 and see figures 13a to 13h).
Figure 13.
Figure 13a. MULTI-OUT vs. IAC (VRMS = 1.7V;
VLFF = 5.1V)
10/35
Figure 13b. MULTI-OUT vs. IAC (VRMS = 2.2V;
VLFF = 5.1V)
AN628 APPLICATION NOTE
Figure 13c. MULTI-OUT vs. IAC (VRMS = 4.4V;
VLFF = 5.1V)
Figure 13d. MULTI-OUT vs. IAC (VRMS = 5.3V;
VLFF = 5.1V)
Figure 13e. MULTI-OUT vs. IAC (VRMS = 1.7V;
VLFF = 2.5V)
Figure 13f. MULTI-OUT vs. IAC (VRMS = 2.2V;
VLFF = 2.5V)
Figure 13g. MULTI-OUT vs. IAC (VRMS = 4.4V;
VLFF = 2.5V)
Figure 13h. MULTI-OUT vs. IAC (VRMS = 5.3V;
VLFF = 2.5V)
11/35
AN628 APPLICATION NOTE
Pin 8 has to be connected through a resistor (Ri') to the negative side of Rs (see fig. 9) to sum the (IL · RS) and
the (Imult · Ri') signals. The sum result is the error signal voltage to the current amplifier non inverting input.
Ri' · Imult = Rs · IL
Pin 9. ISENSE (Current Amplifier inverting input). This pin, is externally connected to the external network described at CA-OUT (pin 5). To be noted that Ri and Ri' have the same value because of the high impedance
feedback network.
Pin 10. SGND (Signal ground). It has to be connected, to the pc-board GND, close the filtering reference capacitor.
Pin 11. VREF (Voltage reference). An internal bandgap circuitry, allows an accurate voltage reference. An external capacitor filter (from 100nF to some µF) connected to the signal ground is recommanded (see fig. 14).
This pin can deliver up to 10mA and can be used for external needs (e.g. enable for other circuits).
Figure 14.
Pin 12. SS (Soft start). This feature avoids current overload on the external Mosfet (Q) during the ramp-up of
the output boosted voltage. An internal switch discharges the capacitor if output overvoltage or VCC undervoltage are detected. An internal current generator of 100µ with the external capacitor define the soft start time
costant (see fig. 15). Because the voltage at the softstart pin acts on the E/A output (driving the multiplier with
VVAOUT = 5.1V typical voltage swing), the softstart time is defined by:
V VA-OUT
t SS = C ss ⋅ ----------------------I
SS
Figure 15.
12/35
AN628 APPLICATION NOTE
This time (tSS) depends on the application parameters (output voltage, input voltage, output capacitor value,
boost inductor size, etc.) and normally the value amounts at some tens of msec.
Pin 13. VVA-OUT (Error amplifier output). Output of the E/A that determinates the control of the boosted regulated
voltage (VO). This pin has to be connected with a compensation network to the pin 14 (see fig.16).
Figure 16.
Figure 16a.
First of all, the system does not have to attempt to regulate the twice mains frequency output voltage ripple
(∆VO) to avoid the line current distorsion. Moreover the system stability has to be ensured. The voltage open
loop gain can be splitt in two separated blocks.
The first block small signal gain, is given by the ratio between the E/A output voltage (vea) and output voltage
variation (vo) and is defined by the E/A network:
vea
1
Gea' = ---------- = -------------------------vo
s ⋅ R1 ⋅ C r
Where Gea` is the E/A gain without Rr ref. fig. 16. R2 has no effect on the error amplifier gain because the inverting input potential is fixed to VREF. The Gea can be seen also as the ratio between the error amplifier output
ripple and the imposed output voltage ripple (∆VO). The E/A output signal can swing between 1.28V to 5.1V. A
value less than 2.5% of the effective E/A output swing voltage (VVAOUT = 3.82V) could be chosen to fix the Cr.
So, the Gea defined at the output voltage ripple frequency, determinates the Cr value to ensure the 100/120 Hz
(2f) attenuation.
0.095V
Gea ≤ ------------------∆V O
∆V O
1
C r = -------------------------------------------------- ≥ Ka ⋅ ----------R1
2 ⋅ π ⋅ 2f ⋅ R1 ⋅ Gea
where: Ka = 1/60 for 50Hz and 1/72 for 60Hz mains frequency.
Lower Cr value could increase harmonic distortion.The second block (Power block) is represented by the output
filter capacitor (CO) with its own reactance (XCO), the system has to be able to compensate the total external
load variation through the E/A output response (∆Vea). The power gain transfer function (Gpw), for large variations can be written:
X CO
G pw = I O ⋅ ------------∆V ea
The total load variation (IO) to be considered is: PO(max)/VO:
13/35
AN628 APPLICATION NOTE
PO
P O ⋅ X CO
1
G pw = -------------------------- ⇒ G pw = -------------------------- ⋅ --------------V O ⋅ ∆V ea
V O ⋅ ∆V ea s ⋅ C O
The voltage open loop gain contains two poles in the origin, then stability problem can arise. Connecting the
resistor (Rr) in parallel to the capacitor Cr to shift the E/A pole from the origin to 1/(Rr · Cr), the stability is ensured.
The crossover frequency fc can be calculated by Gpw · Gea` = 1 and therefore:
fc =
PO


1
 -------------------------------------------------- ⋅  ------------------------------
2π
R1
V
⋅
∆
V
⋅
2π
⋅
C
⋅
⋅ Cr
 O
ea
O
To allow the highest DC gain maintaining a phase margin of at least 22°, the Rr maximum value is imposed as:
2.75
R r ≤ -------------------------2π ⋅ f c ⋅ C r
The output filter capacitor value (CO) is related to the output voltage filtering (see Power section design).
Pin 14. VFEED (Error amplifier input). This pin (see fig. 16), connected to the boosted output voltage through a
divider, allows the output D.C. voltage regulation. Neglecting the contribution of the E/A feedback resistor (Rr),
the 5.1V reference and the output DC voltage (VO) define the ratio between R1 and R2:
VO
R1
-------- = -----------–1
R2
5.1V
To be considered that the R1, togheter with the feedback network(see pin 13 description) define the E/A gain.
The R1/Rr ratio affects the load regulation (lower output current increases the output voltage) with the following
relation:
∆V ea ⋅ R1
∆V Omax = ------------------------Rr
where: VOmax is the maximum output voltage variation due to the E/A gain reduction and load variation.
The R1 and R2 will be chosen in the high precision class:
Pin 15. P-UVLO (Programmable supply undervoltage threshold). An internal divider (between pin 19, pin 15 and
ground) and an internal comparator with a threshold voltage of 1.28V fixes the default turn-on and turn-off 15.5V
and 10V levels of the supply section (see fig. 17). Using an external divider (RH and RL) it's possible to change
the supply thresholds: RH fixes the hysteresis, RL fixes the turn-on threshold. To design a divider for a given
supply threshold, is useful know (see fig. 17), the typical resistor value, useful to design the external divider, are:
R1 = 394k, R2 = 88k and R3 = 58k. Anyway, in fig. 17a/b a diagram with threshold values and a table, useful
for a fast choice of RH and RL are shown.
For DISABLE function see Appendix B.
14/35
AN628 APPLICATION NOTE
Figure 17.
Programmable Under Voltage
+
1.28V
19
R1
RH
UVLO
-
15
R2
P-UVLO
R3
RL
10
D94IN066A
VCCON and VCCOFF vs. RL
RH = RL * 6.8
VCC ON
VCC OFF
RH
RL
11V
10V
82kΩ
12kΩ
12V
10.1V
220kΩ
33kΩ
13V
10.5V
430kΩ
62kΩ
14V
10.8V
909kΩ
133kΩ
14.5V
10.9V
1.36MΩ
200kΩ
15V
11V
2.7MΩ
390kΩ
b)
a)
Pin 16. SYNC (In/Out synchronization). Only for L4981A, this function allows the device to be synchronized with
other circuits of a system (see fig.18a). When the device is externally synchronized, the external clock has to
satisfy these conditions: the signal amplitude must cross the threshold value (3.5V), the frequency has to be
slightly higher than that programmed by the R-C constant (see pin 18) and the pulse width has to be at least
800 nsec.
If the device has to synchronize other circuits, the signal delivered by this pin is a positive pulse of 4.6V (0.5mA)
and the pulse duration is equal to the sawtooth falltime.
The L4981B uses this pin to perform another function. If the application does not use the SYNC function, it is
preferrable to focus the EMI filtering problem using the B version. Pin 16, named FREQ-MOD in B version, allows to change the switching frequency in order to spread the energy content over a wider spectrum range.
To perform the frequency modulation (see fig.18b), the pin must be connected, through a resistor (Rfm), to the
rectified line voltage. This allows to change dinamically (cycle by cycle) the (COSC) charge and the discharge
currents that define the ramp slopes of the oscillator sawtooth. The effect of the resistor produces the frequency
change (see fig.18c) between the nominal value (fsw) and its minimum value which occurs when the input voltage reaches the peak value (VIpk). The total frequency variation (see also pin 17 and 18) can be estimated by
the formula:
V IPK ⋅ R osc
∆f sw
----------- = K ⋅ -----------------------------V
f sw
RMS ⋅ R fm
15/35
AN628 APPLICATION NOTE
where: Rfm is the programming current resistor.
K is a constant value = 0.1157 for R value in KΩ and fSW in KHz.
∆f sw
A typical 20% ----------- can be a good compromise.
f sw
Figure 18.
a)
8.5V
10I
I
Ifm
1.28V
-
1
+
+
-
17
ROSC
7
VRMS
b)
200I
1/VMRS
16
FREQ-MOD
ROSC
18
COSC
Rfm
COSC
2 POLES
FILTER
D94IN065A
Modulation Frequency Normalzed in an Half Cycle of the Mains Voltage
1
fsw
Vl
1
0.8
0.8
c)
0.4
0.4
0.2
0.2
0
0
16/35
45
90
0
135
180
Electrical degrees
AN628 APPLICATION NOTE
Pin 17. ROSC (Oscillator resistor). An external resistor connected to ground, programs the charge and the discharge currents that pin 18 (COSC) forces to the external capacitor. The reference voltage at pin 17 is 1.28V
(see fig.18a/b)
To set the charge current, the relation is:
1.28V
I c ≈ 10 ⋅ ---------------R OSC
The discharge current is defined by:
1.28V
I c ≈ 200 ⋅ ---------------R OSC
The maximum discharge current of Id = 12mA, this means a minimum Rosc value of 22KΩ.
Pin 18. COSC (Oscillator capacitor). An external capacitor (see fig 18a/b), connected between this pin and
ground, fixes the rise and fall time (tr and tf) of the sawtooth oscillator according to the previous relations (pin
17) and therefore the switching frequency. The typical ramp valley-peak voltage (Vsrp) is fixed to 5V.
The period T is defined by:
1 1
T = t r + t f = V srp ⋅ C OSC  ----- + ----
I
I 
C
d
the switching frequency is:
1
2.44
f sw = --- ≈ -----------------------------------T R OSC ⋅ C OSC
See also Fig. 19
Figure 19. Oscillator Diagram
Pin 20. GDRV (Gate driver output). This output is internally clamped to 15V (see Fig. 20), to avoid ageing
problems of the gate oxide. The output driver is normally connected to the gate of the power device
through a resistor (say 5 to 50 Ohm) to avoid overshoot and to control the dI/dt of the switch.
POWER SECTION DESIGN
Booster Inductor
The Boost Inductor design involves various parameters to be handled and there are different approaches
to define them.
Pin 19. VCC (Supply voltage input). The very low current consumption feature before the turn-on threshold is reached. The undervoltage circuitry, with the
threshold hysteresis of 5.5V typ. (see also pin 15)
and an internal clamp at 25V (typ.) ensure the IC
safety operation.
In continuous mode operation, the energy stored in
the boost inductor in each switching cycle, is not
completely transferred to the output (bulk) capacitor.
A quantity of energy is stored in the magnetic circuit,
reducing in this way the input current ripple. This minimizes the line noise and reduces the input filter size
(see fig.21).
17/35
AN628 APPLICATION NOTE
Figure 20.
Figure 21.
The energy transferred from the boost inductor to the bulk capacitor in each cycle is:
2
2
1
E/cycle = --- L ⋅ ( I Lp – I Lv ) = L ⋅ I Lt ⋅ ∆I L
2
where:
L = Boost Inductance
ILp = Inductor Peak Current (ILt + ∆IL/2)
ILv = Inductor Valley Current (ILt - ∆IL/2)
ILt = Instantaneous Line Current (ILp + ILv)/2
∆IL = Twice Inductor Current Ripple (ILp - ILV)
Because the instantaneous line current (ILt) that corresponds to the average inductor current in the cycle, draws
a full rectified (half- sinusoidal) waveform, it is useful to refer to the AC line RMS and peak parameters:
I Lpk =
where:
Irms = ILrms = PI/VIrms is the line current
PI = PO/η is the input power
η is the power yeld.
18/35
2 ⋅ I Lrms
AN628 APPLICATION NOTE
The power transferred by the inductor in each cycle
L ⋅ I Lt ⋅ ∆I L
P t = -------------------------t on
where: ton = δ / fsw and δ = (VO - VIt) / VO
For a given L, the twice ripple current ∆IL is the quantity associated to the transferred energy and can be calculated as a certain percentage of the ILpk inductor current.
V lt ⋅ ( V O – V lt )
∆I L = -----------------------------------V O ⋅ f sw ⋅ L
VO
If the maximum Vlpk value is higher than the VO/2, the maximum ∆IL occours when VIt = -------- and its value is
2
VO
∆I L ( max ) = ----------------------4 ⋅ f sw ⋅ L
If the Vlpk maximum value does not reach VO/2 voltage value, the maximum ∆IL is reduced and its value is :
V lpk ( V O – V lpk )
-.
∆I L ( max ) = ---------------------------------------V O ⋅ f sw ⋅ L
In continuous mode operation, an acceptable curent ripple level (Kr) can be considered between 10% to 35%.
∆I L
K r = ----------------2 ⋅ I Lpk
Smaller current ripple on the inductor involves smaller noise on the rectified main bus reducing the input filter
size; but the ripple reduction will impose an increase of the boost inductor.
The high voltage, the flux density and the frequency range make the standard high frequency ferrite the most
useful material in P.F.C. applications. To avoid the core saturation, related to the high permeability materials, it
is necessary built an air-gap in order to allow an adequate magnetic force range (H+Hgap).
An easy approach, is to have an approximated minimum value of core size that could be used to perform the
conversion:
∆I
Volume ≥ K ⋅ L ⋅ I Lpk ⋅  I Lpk + -------L-

2 
where : K = specific energy constant.
L = Boost inductor value in H.
The specific energy constant (K), mainly depends on the ratio between the gap length (lgap) and the effective
length (leff) of the magnetic core set and on the maximum ∆B swing. Practically
I eff
K ≈ 14 ⋅ ---------I gap
can be used to get the minimum volume of the core set in cm3. After the minimum core-set size is estimated,
the suitable type will be selected with technical and economic evaluations.
Next step will be the design of the coil parameters.
L ⋅ I Lt ⋅ ∆I L
The above mentioned formula P Ot = -------------------------t on
19/35
AN628 APPLICATION NOTE
if referred to the magnetic path, can be rewritten :
∆B
P Ot = Ae ⋅ I eff ⋅ H ⋅ ------t on
where :
Ae = effective area of the core section.
leff = effective magnetic path length.
∆B = deviated magnetic flux density.
H = magnetic field strength.
The ratio between the ferrite and the air path magnetic permeability, depends on the ferrite materials. Core materials for power application (such as B50/51), have a initial permeability value about 2500 times that of air. This
means that, above a certain air-gap length percentage, it is possible to neglect the leff (length of the core) simplifing the calculation e.g. if a 1% of air-gap length, respect to the core lenght value is used, the error introduced
is about 4%.
∆B
Rewriting  P Ot = Ae ⋅ I eff ⋅ H ⋅ -------

t 
on
∆B
P Ot ≈ Ae ⋅ I gap ⋅ H gap ⋅ ------t on
L ⋅ I Lt ⋅ ∆I L
equating to and simplifing P Ot = -------------------------t on
Ae ⋅ I gap ⋅ H gap ⋅ ∆B ≈ L ⋅ I Lt ⋅ ∆I L
Because: I gap ⋅ H gap ≈ N ⋅ I Lt and ∆B = µ 0 ⋅ ∆H
Ae ⋅ N ⋅ µ 0 ⋅ ∆H ≈ I ⋅ ∆I L
∆I L
∆H ≈ N ⋅ ---------I gap
and finally:
L ⋅ I gap
N ≈ ----------------µ 0 ⋅ Ae
This simplified relation is much easier to use than the complete one:
I gap
I eff
L
N ≈ ------ ⋅ -------------------------------------------- + ----------------2
µo 
µ
π
0 ⋅ Ae
Ae + --- ⋅ I gap


4
After N has been defined, it's necessary to check the core for saturation of the magnetic path (rated N · Imax vs.
Air-gap on ferrites databook). If the check is too close the rated limit, an increase of the lgap (gap lengh) and a
new calculation will be necessary. Copper losses RL · I2Lrms and former's winding space available will be considered for the wire selection.
20/35
AN628 APPLICATION NOTE
An auxiliary winding can be used just to get a low cost supply for the I.C. It will be a low cost thin wire coil will
be used and the number of turns is the only parameter to define.
Input Bridge
The input diodes bridge can be standard off-line, slow-recovery and low cost devices. The device selection considers just the input current (Irms) and the thermal data.
Input Capacitor
The input filter capacitor (CIN) has to sustain the input instantaneous voltage (VIt), with an imposed voltage ripple, during the turn-on (ton) time of the Mosfet.
The worst conditions will be found at the minimum rated input voltage VIrms(min).
The maximum high frequency voltage ripple (r = ∆VI / VI) has to be imposed:
I rms
C IN ≥ K r ------------------------------------------------2 ⋅ π ⋅ f sw ⋅ r ⋅ V lrms
Where: Kr is the current ripple coefficient.
r = 0.02 to 0.08.
The CIN maximum value is limited to avoid current distortion.
Output Bulk Capacitor
The choice of the output bulk capacitor (CO), mainly depends on the electrical parameters that affect the filter
performances and also on the subsequent application.
The D.C. output voltage and overvoltage, the output power and voltage ripple are the first parameters to consider in all applications. The RMS capacitor ripple current IC(2f)rms = Io/ 2 and so, the output voltage ripple
(∆VO) will be:
2
1
∆V O = I O ------------------------------------ + ( ESR )
2
( 2π ⋅ 2f ⋅ C O )
With a low ESR capacitor can be simplify:
IO
PO
- = --------------------------------------------.
C O + -------------------------------2π ⋅ 2f ⋅ ∆V O
2π ⋅ 2f ⋅ ∆V O ⋅ V O
Although the ESR, normally does not affect the output ripple parameter, it has to be considered in power losses
account both for the rectified mains frequency and the switching frequency.
If the application (i.e. computer supply) has to guarantee a specified Hold-Up time (tHOLD), the capacitance sizeing criteria will change:
The CO has to deliver the supply energy for a certain time and a specific dropout voltage.
2 ⋅ P O t HOLD
C O = -------------------------------------------------2
2
V O_min – V op_min
where:
VO_min = minimum output voltage value (normally at the maximum load conditions)
21/35
AN628 APPLICATION NOTE
Vop_min = minimum output operative voltage before the 'power fail' detection.
Power Switch
A power MOSFET is the active switch used in most application for its frequency features. It will be selected according with the output boosted voltage and the delivered power. There are two contributions for power losses
in the mosfet: conduction losses and switching losses. The on-state power losses can be calculated using the
formula:
Pon-MOS = IQrms2 Rdson
One extimation of the switching losses can be done considering two separated quantities:
1.5 1
2
10
P capacitive ≈  ------ ⋅ C oss ⋅ V o + --- C ext ⋅ V o ⋅ f sw .
3

2
Pcrossover ≈ VO · Irms · tcr · fsw + Prec
where:
Coss is the Drain capacitance at VDS = 25V.
tcr is the crossover time.
Cext is the external layout stray capacitance.
Prec is the contribution due to the diode recovery.
To reduce the crossover losses a snubber network can be used.
Booster diode
The booster diode will be selected to withstand the output voltage and current. Moreover, it has to be as fast as
possible in order to reduce the power switch losses.The STMicroelectronics Turboswitch™ diode series match
this specifications, and are expecially suitable for this application.
The diode power losses can be split in two contributions: conduction losses and switching losses. The conduction losses can be estimated by:
PDon = Vto · IO + Rd · IDrms2
where: Vto = threshold voltage
Rd = differential resistance
Sense Resistor
The sense resistor produces the signal for the current feedback loop and for the overcurrent protection circuit.
An easy criterion to choose the sense resistance is to minimize the power dissipated assuring a sufficient signal to
noise ratio. In much high power applications, it could be considered the magnetic sensing approach (see fig. 22).
Figure 22. Magnetic Sense
iD
L
iQ
Co
Rs
22/35
-Vrs
Load
AN628 APPLICATION NOTE
DESIGN PROCEDURE.
In order to fix the described concepts, here follows a brief description of a possible design flowchart referred to
a typical "low-medium range power" PFC application.
Design target specification:
– Wide range mains; VINrms = 88 Vac to 264 Vac.
– Pre-Regulated DC output voltage; VO = 400 V.
– Rated output power; PO = 200W.
The design starts fixing the operating conditions.
– The switching frequency 100kHz ///
– The 100Hz voltage ripple imposed at full load is ± 8V; this is satisfied selecting Cout = 100µF
– The Over Voltage limit is set at Vout+50V
– The maximum current ripple at nominal load has been chosen = 35%
The circuit in fig. 23 can be proposed as reference for medium range power PFC application.
Figure 23. Low-medium Power Typical Application (VO = 400V; PO = 200W)
L 0.9mH
R17
806K
1%
R6
620K
5%
R17
806K
1%
R7 360K 5% R6 620K 5%
C8
220nF
100V
R19
1.1M
5%
R19
1.1M
5%
D3
C7 2N2222
220nF
100V
R8
33K
5%
R15
10K
R14
0.5W
56
Q2
0.5W
STK2N50
D3 1N4150
DZ
22V
0.5W
C11
100µF
25V
+
D1 5TTA5060
C10
15nF
100V
R1
412K
1%
R9
910K
1%
R1
412K
1%
R9
910K
1%
D4
1N4150
R23
R20
Vo
R22
10K 5%
FUSE
Vi
88VAC to 254VAC
NTC
15
7
4
BRIDGE
4 x BY214
19
14
R12
220K
5%
C1
220nF
400V
L4981A
D2 1N4150
2
560 1%
8
5
9
R3
2.7K
5% RS
11 6
12
17
1
20
R13
STH/STW15NB50
Q1
15 5%
C12
270pF
630V
C3
R5
27K 5%
18
C2
100µF
450V
3
R11
R21
5.1K
1%
C9
330nF
13
D5
BYT
11600
1nF
R4
2.7K
5%
0.07 2W
C4
1nF
R16
24K
1%
C6
1µF
16V
R18
1.8K
4W
R2
11K
1%
R10
21K
1%
C5
1µF
16V
D93IN029C
23/35
AN628 APPLICATION NOTE
Input capacitor
The input capacitor, placed across the rectified mains, must be considered as part of the EMI filter. The advantage, in placing this part after the mains rectification, is the shunt effect for the high frequency current in order
to avoid it to flow throws the diodes of the bridge due to the poor recovery characteristic.
On the other side, the value of this capacitor must to be held as low as possible because the inherent DC voltage
content affects the harmonic distortion.
With 220nF, the high frequency is filter enough and the introduced DC level can be considered not significant
at reasonable load.
Output capacitor
For the output capacitor selection, it can be consider just the output voltage ripple.
Choosing 100µF/450Vthr 100/120Hz ripple is ± 8Vac
Instead, if the pre regulated voltage bus must ensure enough energy for Hold-up requirements (i.e. the energy
is delivered to a power supply system), the Coot value will be increased to around 180µF.
Sense resistor
The sense resistance (Rest) is selected considering both, the signal level and the power dissipation parameters.
Using ±70mΩ, the sense signal is good enough to be managed by the current loop. On the other side, the maximum power dissipation will be:
Pros = RS · (Ilrms2 + Ilhfrms2) ≤ 0.5W
Where Alarms max. = 2.50A
Power Mos
The Mosfet breakdown voltage is imposed; Bvdss ≥ Vout + Dvout + margin = 500V. The Rdson is selected taking
in to account the conduction power dissipation.
The formula for calculation is: Pon_max = Iqrms2 · Rdon
i.e. considering Ron(t) = 0.7Ω the Pon_max = 2.15 · 0.7 = 3.3W.
Adding the switching (and the capacitive) losses we can estimate 8W to 10W total power dissipation.
Boost Diode
The continuous current mode of operation, suggest using an Ultra-fast reverse recovery diode. The STMicroelectronics TURBOSWITCH™ family offers a good solution for this kind of application.
Boost Inductor
The inductor design starts defining the L value that is a function of the switching frequency and the accepted
current ripple. In this design, we suggest an inductor value L = 0.75mH that can be realized using an ET3411
gapped set-core ferrite.
24/35
AN628 APPLICATION NOTE
The results, concerning the described circuit, have been tested. And the result are shortly here reported:
Vi
f
Pi
(Vrms)
(Hz)
(W)
110
60
220
220
50
217
A-THD
H3
H5
H7
H9
VO
∆VO
PO
η
(%)
(%)
(%)
(%)
(%)
(V)
(V)
(W)
(%)
0.999
1.79
1.40
0.40
0.31
0.28
392
8
201
91.6
0.997
2.25
1.68
0.83
0.57
0.48
398
8
204
94.2
PF
DEMO-BOARD:
Design process and Evaluation results
In order to provide a powerful tool for the complete evaluation of the L4981, It is available a populated DemoBoard. The design process and the description for the demo, is here described.
The demo has been designed to operate in wide range mains and the size and is finalized for a "medium-high"
output power range.
Let us start fixing the overall target of the application.
Electrical target specification:
– Wide range mains; VINrms = 88 to 264 Vac.
– Regulated DC output voltage; VO = 400 V.
– Rated output power; PO = 360W in any mains condition.
– Target efficiency ≥ 90% in nominal load conditions.
Chosen operation conditions of the application
– The rectified mains (100/120Hz) full load voltage ripple is ± 7-8 V (peak to peak) this is achieved using
an output capacitor Cout = 220Uf/450V
– The maximum current ripple, in nominal load condition, is selected to be about 20%. This can be obtained using the boost inductor L = 0.55mH and setting the switching frequency at 100kHz. ##
– The Over Voltage Protection has been set at Vout + 58V
The demo is capable to deliver around 400W output power; anyway in order to limit the temperature, the rated
power is limited to 360W.
The schematic is shown in fig 24.
25/35
AN628 APPLICATION NOTE
Figure 24. Demo Board Circuit (VO = 400V; PO = 360W)
R14
68
Dz1
18V
B2= D1+D2
+D8+D9
C2
330n
R4
1.2M
C3
330n
F1
T15A250V
L1=0.5mE42*21*15
gap=1.9 58/6 turns
20*.2mm
R11
56k
R6
500k
R5
220k
R12
56k
BRIDGE
B1 8A
VCC
C10
150uF
R7
500k
D6
DZW06-48
L2
3u
C8
100n
R2
33k
NTC
2.5
V+ BUS=400V
D5STTA106
C11
220n
VCC
D4-STTH8R06
to220 (/40CW)
R18
6.8
2W
C14
100n
R19
750k
R20
750k
D7-STTA406
R22
750k
R23
750k
R16
220k
88 to 264 Vac
7
4
1
19
14
13
C1
330nF
400V
Cf
.22uF
600V
3
15
L4981A/ B**
RAux1
6
2
8
5
9
18
C15
220uF
450V
20
16
**
D3
10
17
12
11
Cs
330pF
R17
15
Q1+Q2#
RAux2
R21
19.6k
R10
5k
Q4
4007
TP1
R1
460
R3
2.2k
R8
17k
C6
3.3n
C9
1n
R15
24k
C12
1u
C13
1u
R24
16.9k
R25
1k
2W
R13
2.2k
R9 (RS)
50m
// 3*0.15)
# // Q1&Q2
TO220*2
STM12NM50
/ 7C/W
Some design peculiarities
A magnetic snubber solution has been adopted in order to limit/control the ramp rate of the current at the turnon edge of the Power Mos, referring to the schematic, the parts involved in this function are:
L2, D5, D6, D7, C14 and R18.
The function of the auxiliary inductor L2 is to reduce the ramp rate of the current reducing in this way the peak
current, due to the recovery of the boost diode (D4), and the associate noise emission.
The additional benefit of this magnetic snubber is also the reduction of the switching power dissipated in the
Mos. D5 and D6 needs to provide a path that allows the demagnetization of the inductor L2, clipping the negative spike thus avoiding the Breakdown activation of the booster diode D4.
D7, C14 and R18 will clamp the energy of L2 at the turn off edge of the Mos.
The inrush current limiter NTC is placed between the cathode of the boost diode and the bulk capacitor C15.
In this way, the current flowing in to the NTC limiter is the same of the boost diode well below the mains current
especially at minimum mains value.
The demo is provided with "self-supply" circuit. The proposed supply circuit, using a Graetz bridge is much
efficient anyway; it can be inadequate when the output power is reduced down to less than 5W.
On board, it has been recovered place for the Raux1 and Raux2. The two resistors will be connected in case of
L4981B version evaluation.
Here follows some comment concerning the design and the selection of the power parts of the demo.
26/35
AN628 APPLICATION NOTE
Input capacitor selection
The demo is not provided with complete dedicated EMI filter. At the input side, two parts compose the capacitor;
the first (Cf.) is placed across the AC input of the bridge and the second one (C1) is tied to the rectified mains.
The advantage, of this configuration is the minimization of the DC content in placing a low value after the mains
rectification (C1), just to filter the high frequency. The capacitor Cf placed in the AC side must be considered as
part of the EMI filter.
Output capacitor selection
For the output capacitor selection, it can be consider just the output voltage ripple.
Choosing Co = 220µF (450V), the maximum rectified mains ripple is:
πo
- = 7.3V
∆V = ---------------------------------------200 ⋅ π ⋅ V o ⋅ C o
Sense resistor selection
The sense resistance is set at 50mOhm (Rs = 3 · 0.15Ohm//) maximum power dissipation (@ 88Vac mains and
360W) will be:
PRs(max) = RS · (Ilrms2 + Ilhfrms2) = 1.04W
Where: Ilrms max. = 4.55A
Power Mos
In the selection of the power switch, it has been preferred to share the thermal dissipation in two separate
TO220 packages. This is a good solution because the size of the heath sinkers can be limited.
The breakdown voltage is imposed = 500V.
Considering the On resistance (@ Tj = 100°C) = 320 mΩ the formula for the dissipated power calculation are:
Conductive losses P_On(max) = IQ(max)2 · Rdon = 3.9 · 0.32 = 4.9W.
Adding the capacitive (about 2.5W) and the switching losses (as low as 2-3 W, thanks to the snubber) we can
estimate 10 to 12W total power dissipation at lower mains value.
Boost Diode
The 8A 600V chosen Turbuswitch fits well with the application. The power dissipated in the boost diode is about 1.4W.
Boost Inductor
The 0.55mH chosen inductor value allows a low ripple (23%) of its current; moreover, there is enough room (in
the industrialization phase) to reduce the switching frequency holding an acceptable current ripple (e.g. reducing the frequency to 75 kHz the current ripple will be held within 30%.
In this design, the coil has been realized with a gapped set-core ferrite E42*12*15.
The results that can expect, realizing the described circuit, has been tested. And the result are shortly reported
from Table 1 to Table 6.
The PCB and component Layout can be seen in figgs 25, 26 and 27 (The Gerber files of the PCB are available
on request).
27/35
AN628 APPLICATION NOTE
Figure 25. Component Layout (Dimensions 88 x 150mm)
Figure 26. P.C.B. Component Side (Dimensions 88 x 150mm)
28/35
AN628 APPLICATION NOTE
Figure 27. P.C.B. Solder Side (Dimensions 88 x 150mm)
DEMO BOARD TEST RESULTS
Table 1. Maximum power range at 110Vac
Vmains
Pout
Vout
Pin
THD
PF
Eff.
88Vac
403W
401Vdc
433W
5.1%
0.998
.93
110Vac (*)
407W
403Vdc
431W
2.2%
0.999
.945
132Vac
409W
404Vdc
430W
2.7%
0.999
.95
(*) Most significant losses balance at maximum power (110Vmains):
– Power-mos (Q1+Q2) dissipated power = 9.6W.
– Bridge (B1) dissipated power = 6.3W.
– Boost turbo-diode (D4) dissipated power = 1.6W.
– Boost inductor L1 = 2W
– Aux. Inductor L2 = 1.6W.
– NTC dissipated power = 1.1W.
– Snubber = 1.4W
29/35
AN628 APPLICATION NOTE
Table 2. Maximum power range at 220Vac
Vmains
Pout
Vout
Pin
THD
PF
Eff.
176Vac
415W
407Vdc
430W
4.2%
0.997
.965
220Vac (*)
417W
408Vdc
431W
5.8%
0.994
.967
264Vac
419W
409Vdc
431W
7.4%
0.989
.972
(*) Most significant losses balance at maximum power (220Vmains):
–
–
–
–
–
–
–
Power-mos (Q1+Q2) dissipated power = 7.1W.
Bridge (B1) dissipated power = 4W.
Boost turbo-diode (D4) dissipated power = 1.3W.
Boost inductor L1 = .6W
Aux. Inductor L2 = 0.9W.
NTC dissipated power = 0.8W.
Snubber = 0.9W
Table 3. Nominal power range at 110Vac
Vmains
Pout
Vout
Pin
THD
PF
Eff.
88Vac
366W
404Vdc
397W
5%
0.998
.92
110Vac (*)
370W
406Vdc
395W
2.2%
0.999
.94
132Vac
372W
407Vdc
394W
3%
0.999
.945
(*) Most significant losses balance at nominal power (110Vmains):
–
–
–
–
–
–
–
Power-mos (Q1+Q2) dissipated power = 9.3W.
Bridge (B1) dissipated power = 5.7W.
Boost turbo-diode (D4) dissipated power = 1.5W.
Boost inductor L1 = 1.8W
Aux. Inductor L2 = 1.35W.
NTC dissipated power = 1W.
Snubber = 1.2W
Table 4. Nominal power range at 220Vac
Vmains
Pout
Vout
Pin
THD
PF
Eff.
176Vac
378W
410Vdc
394W
4.7%
0.997
.959
220Vac (*)
381W
412Vdc
395W
6.4%
0.993
.964
264Vac
381W
412Vdc
395W
8.1%
0.987
.964
(*) Most significant losses balance at nominal power (220Vmains):
–
–
–
–
–
–
–
Power-mos (Q1+Q2) dissipated power = 6.9W.
Bridge (B1) dissipated power = 3.5W.
Boost turbo-diode (D4) dissipated power = 1.3W.
Boost inductor L1 = 0.5W
Aux. Inductor L2 = 0.83W.
NTC dissipated power = 0.8W.
Snubber = .8W
30/35
AN628 APPLICATION NOTE
Table 5. Half power range at 110Vac
Vmains
Pout
Vout
Pin
THD
PF
Eff.
88Vac
219W
420Vdc
239W
2.4%
0.999
.916
110Vac (*)
220W
421Vdc
238W
3.6%
0.999
.925
132Vac
222W
423Vdc
237W
2.7%
0.999
.937
(*) Most significant losses balance at half power (110Vmains):
–
–
–
–
–
–
–
Power-mos (Q1+Q2) dissipated power = 7.5W.
Bridge (B1) dissipated power = 3.7W.
Boost turbo-diode (D4) dissipated power = 1.1W.
Boost inductor L1 = 1.3W
Aux. Inductor L2 = 0.95W.
NTC dissipated power = 0.7W.
Snubber = 1W
Table 6. Half power range at 220Vac
Vmains
Pout
Vout
Pin
THD
PF
Eff.
176Vac
223W
424Vdc
236W
8%
0.993
.945
220Vac (*)
223W
424Vdc
235W
10.5%
0.994
.949
264Vac
223W
424Vdc
235W
15%
0.978
.95
(*) Most significant losses balance at half power (220Vmains):
–
–
–
–
–
–
–
Power-mos (Q1+Q2) dissipated power = 5.64W.
Bridge (B1) dissipated power = 1.9W.
Boost turbo-diode (D4) dissipated power = 0.88W.
Boost inductor L1 = .5W
Aux. Inductor L2 = 0.6W.
NTC dissipated power = 0.52W.
Snubber = 0.7W
Significant waveforms
Since the described application is provided with a "magnetic snubber circuitry", it is of some interest to have a
look at some switching waveform.
In figure 28, it is depicted the power drain voltage and the current measured in L2 (aux. Inductor) .
To be observed the "delay effect" of the current and the control of its slope. In fact, this circuitry allows to hardly
reducing the Voltage-Current power crossing and the ramp rate of the drain current, reducing in this way the
power dissipated inside the switch and the high frequency contents of the switching.
In figure 29, it is shown both the switching edge and pointed the recovery charge due to the boost diode and the
effect of the Voltage Clamp (D7, C14, R18).
In figure 30, the switch-off edge is magnified and pointed out the effect of the RCD snubber to ground (Cs, Q4
R25) limiting the dV/dt and the above-mentioned Voltage Clamp.
Finally, in figure 31, it is shown and pointed, in the reverse recovery region, the demagnetization effect of the
D5+D6 and its control on the second slope of the recovery itself.
31/35
AN628 APPLICATION NOTE
Figure 28.
Figure 30.
IL2-VDS waveforms at SWITCH-ON
"Magnethic snubber effect"
IL2-VDS waveforms at SWITCH-OFF and ON
Recovery Peak current
Drain voltage
VDS = 100V/div.
IL2 = 2A/div.
Drain current slope control
Figure 29.
Current flowing
in the Voltage Clamp
Figure 31.
IL2-VDS waveforms at SWITCH-OFF
Boost diode Anode voltage and IL2 Current
Drain voltage snubber RCD effect
Drain voltage
Current flowing in the voltage Clamp
0V
Negative Voltage
for L2 demagnetization
Miscellaneous:
The 360W Demo Board is replacing a previous 200W version. It is possible to order this tool quoting the order
code EVAL4981A. The board comes with a CDROM containing the inherent documentation and a special program dedicated to the ST PFC controllers (L4981 and L6561) that make it easy the design of these applications.
32/35
AN628 APPLICATION NOTE
APPENDIX A
LFF (pin 6) Function.
Since in Power Factor applications the Error Amp. compensation network has to filter the mains frequency contents, in order to reduce harmonic distortions, the crossover frequency of the loop gain must be low. This involves a poor load transient response.
An additional function (LFF) is available in L4981A/B devices. It is expecially suitable to modify the multiplier
output current, proportionally to the load, in order to improve the system response bypassing the E/A.The control
is working with VLFF voltage between 2V and 5.1V.
In fig. 32 is shown an application example to explain this function. An external OP-AMP has been used to get
the suitable signal voltage avoiding sense resistor (R1) power dissipation.
In the real application the sense resistor is often replaced by sense transformer.
Figure 32. Application example
L
~
D
IL
ID
IQ
IO
LOAD
IC
L4981
CO
Q
R1
pin6
(VLFF)
~
VREF=5.1V
10K
VCC
3.6K
R3
LM258
+
-
VO
D95IN206B
10K
R2
Design criteria:
It is avisable to ensure a minimum VLFF ≅ 2V at the minimum output current.Since the OP-AMP (LM258) Vol =
0.7V (@ 1mA), to get the minimum voltage at VLFF, 1.3V has to be added. A resistor devider tied to the reference voltage (pin 11 of the controller) shifts the output of the OP-AMP.
( 5.1V – 0.7V )
therefore 1.3V = ------------------------------------ ⋅ R3 ⇒ R3 = 1.3kΩ
R3 + 3.6kΩ
The OP-AMP supply voltage is the same used for PFC controller (VCC) and its gain if fixed in order to produce
VO = 5.1V at the maximum load (IO max).
R2
V O = R1 ⋅ I Omax ⋅  1 + ---------- = 5.1V

10k
E.g. for IOmax = 3A. R1 = 0.1Ω - 1W R2 is roughly 160KΩ.
33/35
AN628 APPLICATION NOTE
APPENDIX B
Disable
Sometimes it is useful to disable the controller. For example, in a complete system in which a PWM regulator
follows the PFC stage, at low output power it is advantageous to shutdown the PFC section to improve the overall system efficiency (stand-by / sleep mode). Likewise most of controllers, one way to do this (using L4981A/
B), is pulling down either the Soft-Start or the E/A output pin . In addition the L4981A/B can be disabled grounding the P-UVLO (pin 15) see fig 33.The P-UVLO function has been designed to program the supply thresholds
by means of an external divider (see application note for details) but it can be effectively used for this purpose
forcing a voltage below the internal reference (1.28V).Besides turning off the driver output stage this method
puts the controller in "before start-up" condition and gives the advantage of minimizing the supply consumption
of the IC.
Figure 33.
+
1.28V
R1
19
RH
UVLO
-
15
R2
R3
P-UVLO
I ≥ 1mA
RL
DISABLE
10
D95IN281B
34/35
AN628 APPLICATION NOTE
Information furnished is believed to be accurate and reliable. However, STMicroelectronics assumes no responsibility for the consequences
of use of such information nor for any infringement of patents or other rights of third parties which may result from its use. No license is granted
by implication or otherwise under any patent or patent rights of STMicroelectronics. Specifications mentioned in this publication are subject
to change without notice. This publication supersedes and replaces all information previously supplied. STMicroelectronics products are not
authorized for use as critical components in life support devices or systems without express written approval of STMicroelectronics.
The ST logo is a registered trademark of STMicroelectronics.
All other names are the property of their respective owners
© 2004 STMicroelectronics - All rights reserved
STMicroelectronics GROUP OF COMPANIES
Australia - Belgium - Brazil - Canada - China - Czech Republic - Finland - France - Germany - Hong Kong - India - Israel - Italy - Japan Malaysia - Malta - Morocco - Singapore - Spain - Sweden - Switzerland - United Kingdom - United States
www.st.com
35/35