Similar presentations:
High-Density Ac-Dc Power Supplies using Active-Clamp Flyback Topologу
1. High-Density Ac-Dc Power Supplies using Active-Clamp Flyback Topology
Ajay Hari, Bryan McCoy2. Agenda
• Introduction to active-clamp flyback operation (ACF)• ACF light-load efficiency challenge
• Introduction to the NCP1568 – Ac-Dc ACF PWM IC.
• Light load and standby solution
• Design equations for transformer selection of the ACF
• Primary and secondary component selection considerations
• Performance data of ultra-high density active-clamp flyback board.
2
3. Introduction to Active-Clamp Flyback Operation (ACF)
34. Active-Clamp Flyback
ClampCapacitor
R
Clamp
Capacitor
Load
D
Vin
Vin
Main Switch
Standard Flyback Converter w. RCD clamp
Active Clamp
Switch
Main Switch
Active-Clamp Converter
The clamp diode in a standard flyback converter is replaced by a switch
hence the name Active-Clamp Flyback or ACF.
4
Load
5. Why Active-Clamp Flyback?
Zero-Volt Switching of the FETs with Fixed-Switching FrequencyResults in high switching frequency, improves efficiency
and EMI.
Soft Increase in Secondary Current
Good for EMI
Clean Drain Waveforms Without Any Ringing
Better efficiency as the leakage energy is recycled.
Better EMI
Single-Ended Topology
Relatively simple design of magnetics compared to LLC.
Single switch/diode in the secondary.
6. Energy-Storage Mode
LleakACF works in continuous conduction mode.
Its input-to-output relationship is given by:
Lmag
Vin
Iout
Reverse
Biased
Im(on)
NS D
Vout = Vin
N P 1- D
The energy-storage mode is similar to that of a classical flyback converter:
when the main FET is on, energy is stored in the transformer.
6
7. Transition from Energy-Storage Mode to Power-Delivery Mode
LleakVclamp =
Vin D
1-D
Cclamp
Body
Diode
Vin
Lmag
Iout
SW
Vsw
Tcharge
When FET turns off, the lump capacitor on the SW node
is linearly charged at a rate given by Tcharge
Tcharge =
Clump (Vin +Vclamp )
Im(peak)
8. Power-Delivery Mode
In this mode, Lleak resonates with clampcapacitor (Cclamp). The resonant
frequency is given by:
Lleak
Fres =
Cclamp
Lmag
Isec
Vin
1
2p L leak Cclamp
The primary resonant current is
given by:
Ires =Im cos ( wt )
The magnetizing current during the (1-D)
phase is given by:
The difference between the primary resonant current
and the magnetizing current flows in the secondary
8
Toff
Imag =Vclamp
L mag
9. Transition from Power-Delivery Mode to Energy-Storage Mode
LleakCclamp
Lmag
Isec
Vin
Clump
Vsw
When the clamp FET turns off, Lleak resonates with Cclamp. For the main FET to
get ZVS, following condition has to be satisfied
2
L leak I2pri >Clump VSW
10. Leakage Inductance Needed for ZVS
For universal design, leakage inductanceneeded to get ZVS increases in a
parabolic fashion.
Leakage Inductance (µH)
14
12
10
8
Increasing leakage & tightly controlling
the spread add cost
6
4
2
0
100
150
300
250
200
Input Voltage (V)
350
400
Assuming Clump = 220 pF, constant Ipeak = 1 A,
85 V to 265 V rms (universal input)
10
Additional resonant inductor is an alternative,
but inductor adds cost & volume
11. ZVS Phenomenon – 1
During Tdis1 (shaded region),Lleak resonates with Cclamp.
Vsw
Tdis1
Vvalley
Tdis2
Isec
p
Tdis1 =
2
Imag
0A
11
The time it takes for the resonance between
leakage inductance and lump capacitance
to reach its valley point is 1/4th of a
resonant period. Therefore:
L leak Clump
Vvalley =Imag(peak)
Ivalley
L leak
Clump
12. ZVS Phenomenon – 2
VswTdis1
Vvalley
Tdis2
During Tdis2 (shaded region), negative
magnetizing current starts to discharge the
clamp capacitance
The time it takes to discharge the lump
capacitance is given by:
Isec
Tdis2 =Clump
Imag
0A
12
Ivalley
Vsw -Vvalley
Ivalley
13. Fixed-Frequency Operation
Fixed frequency OperationMagnetizing current in ACF is in CCM.
Frequency
Imag
Load
13
0A
0A
As the load current decreases, the
valley point of the magnetizing current
decreases.
14. Variable-Frequency Operation
FrequencyImag
Load
14
Variable frequency Operation
0A
As the load current decreases, increasing the
frequency minimizes Imag and reduces the
conduction losses.
Ideally, the valley of the magnetizing current
needs to be maintained constant for ZVS.
0A
15. Light-Load Efficiency & Standby Power Challenge
Light-Load Efficiency&
Standby Power Challenge
15
16. Light-Load Efficiency Requirements
European Code of Conduct, Ver. 5, Tier 2 poses stringent efficiency standards atlight-load condition
For a 60-W design, 4-point average (25%, 50%, 75%, and 100% average)
efficiency needs to be > 88% for full load and 78% for 10% load.
16
17. Standby Power Standard
US DoE standards are equally stringentMost of the brand name OEMs require to pass
stringent Tier-2 standard
DoE: Department of Energy
17
18. ACF Specific Light-Load Challenges
Magnetizing current is in CCM.Frequency modulation results in high-frequency operation at light
load
Classical frequency foldback is not possible to implement when
magnetizing current is in CCM
18
19. DCM Operation
LleakMain Gate
Cclamp
Lmag
Isec
Imag
Iout
OFF
Vin
Vsw
Holding active-clamp FET off, DCM operation can be implemented in ACF.
This allows magnetizing current to enter DCM: frequency foldback can be implemented
19
20. Introduction to NCP1568
Ac-Dc PWM Controller for ACF20
21. Introduction to NCP1568
Control Scheme• Adaptive ZVS frequency modulation allows variable Vout operation
• Integrated adaptive dead-time
• Peak-current-mode control
DCM & Light-Load Operation
• Optional transition to DCM mode
• Frequency foldback with 31-kHz minimum frequency clamp
• Quiet skip eliminates audible noise
• Standby power < 30 mW
HV Startup
• 700-V HV startup JFET
• Integrated sensing of HV SW node for optimum ZVS
• Brownout and X2 discharge inbuilt.
21
1 HV
SW 16
2 NC
NC 15
3 NC
4 FLT
5 RT
6 DTH
NCP1568
NC 14
ATH 13
ADRV 12
VCC 11
7 FB
LDRV 10
8 CS
GND 9
22. Frequency Movement vs. Load
NCP1568Fmax=4.2*F
Active Clamp Mode
Frequency
Transition Mode
NCP1568 features a combination of
nonlinear & linear foldback schemes
DCM Mode
Frequency Foldback
The lower the frequency at light load,
the higher the efficiency
F
F/2
25 kHz
Clamp
Skip
VDTH VATH
22
FB α Iload
23. Clamp Capacitor Challenge
LleakCclamp
Rclamp
Lmag
Vin
Isec
Iout
OFF
VCLAMP_DCM>VCLAMP_ACF
Leakage energy is not recycled in DCM and is dissipated
in the clamp resistor (RClamp)
23
24. Transition from DCM to ACF
Active-clamp FET can be soft-started to discharge the clamp capacitor slowly.Leading-edge modulation of active-clamp FET allows the main FET to achieve ZVS
24
25. DCM Operation Determination
NCP1568 can be configured to operate inpure ACF mode and pure DCM mode.
Efficiency can be plotted in both ACF and
DCM to determine optimal transition
points.
NCP1568 uses the feedback information to
transition from ACF to DCM or vice-versa.
25
26. Key Components Selection
Transformer Design & Key Equations26
27. Design Specifications
DescriptionMin
Input Voltage
Max
Unit
85
265
V rms
Line Frequency
47
63
Hz
Min Output Voltage
4.75
5
5.25
V
Max Output Voltage
19
20
21
V
Output Current
0
3.0
A
Target Full Efficiency @ 115, 230 V rms
Frequency ACF
Max Power
27
Typ
93
100
%
400
kHz
60
W
28. Turns Ratio Selection
Turns ratio can be calculated by the following formula assuming Dmax=0.5.Dmax Vin(min)
N PS =
(1-Dmax ) Vout(max)
Turns ratio should be calculated at the lowest input voltage while delivering maximum
power
For this design, rounded turns ratio is 6.
28
29. Minimum On-Time
Minimum on-time needs to be calculated at worst case duty ratio to ensure that the controller candeliver the pulses
N PS Vout(max)
Tmin1 =
( NPS Vout(max) + Vin(max) ) Fmax
N PS Vout(min)
Tmin2 =
( NPS Vout(min) + Vin(max) ) Fmin
NCP1568 has a minimum on-time of 200 ns. The calculated on-times of the above equations are
600 ns and 700 ns respectively.
If the min on-time is < 200 ns, the turns ratio needs to be adjusted and the process iterated
29
30. Valley Current for ZVS
In order to determine the inductance value, valley current is needed.To calculate the valley current, the capacitance lumped at the SW node can be
expressed as follows:
Co(er)Q3
Clump =Co(er)Q1 +Co(er)Q2 +
N 2PS
Main FET
Output capacitance
Active Clamp FET
Synchronous Rectifier FET
Output capacitance
The above capacitances can be approximated from the FET datasheet.
Considering an ac-dc power supply, a 600-V FET for primary and a 120-V type for
secondary have been selected resulting in a Clump of 220 pF
30
Co(er) is the nonlinear capacitance averaged along a given VDS
31. Inductance Calculation
Inductance can be calculated as follows:Vin(min) Dmin
L mag =
æ I
æ
out(max)
2 FSW(min) æ
æ 1-D N - Ivalley æ
æ
(
)
æ
æ
min
PS
Where Dmin is the minimum duty cycle given by:
Vout(min) N PS
Dmin =
Vout(min) N PS +Vin(min)
For this design, the above formula results in a magnetizing inductance of 120 µH
31
32. Inductance vs. Required Valley Current for ZVS
Inductance (µH)200
180
160
140
120
100
80
0
0,1
0,2
0,3
0,4
0,5
0,6
Required Valley Current (A)
As the required valley current for ZVS decreases, the inductance falls.
32
33. Core Selection
A RM8LP core has been selected forthis low-profile and high-density
design.
33
RM8LPLoss =PLossV ´ VolumeRM8LP
Assuming a 200-mT Bmax operating at
400 kHz results in a core loss of 1.8 W.
34. Primary and Secondary Turns
The primary and secondary turns can be calculated from the following formulae:æ I
æ
out(max)
L mag æ
æ 1-D N - Ivalley æ
æ
(
)
æ
æ
min
PS
NP =
DB A e
NP
NS =
N PS
This results in a primary turns of 23.
Since turns ratio is 6, 24 turns are selected for primary turns and 4 for secondary turns
A flux density, ∆B, of 0.2 T & Ae of 65 mm2 have been assumed for this design.
34
35. Clamp Capacitor Selection
Clamp capacitor should be selected atworst-case off-time i.e., lowest frequency
and minimum D
Clamp capacitor should be selected such
that it resonates 1/4th of the resonant
period at worse case off-time.
æDmin æ
1
Cclamp = æ
ææ
2
æFmin æ 0.5 L leak p
2
The above equation results in 330 nF.
Ceramic capacitors are selected for clamp After derating, a 660 nF is selected
capacitors. Standard derating should be
followed (voltage and rms current).
35
36. RMS Current Formulae
The primary and secondary FET selection criterion is no different than withstandard flybacks.
The active-clamp FET voltage rating is same as main FET.
The clamp and secondary FETs see different current waveforms than standard
flyback. Their formulae are noted below
1- Dmin
IAC(RMS) = I PK ´
6
36
Isec(RMS) =
2Pout
Vout ´ 2 (1- Dmin )
37. 60-W UHD-Board Performance
3738. Simplified Schematic
Secondary side is similar to any standard flyback topology.38
39. Frequency Modulation w. Load
1A2.25 A
1.7 A
As the load current decreases, the negative current is minimized & kept constant
leading to low conduction losses
39
40. Fixed Frequency vs. Frequency Modulation
115 V rms, 1.5-A loadFixed Fsw of 231 kHz
40
115 V rms, 1.5-A load
Frequency modulation, Fsw = 260 kHz
41. Frequency Movement w. Vout & Vin
Frequency Movement w. Vout & Vin20 V
20 V
15 V
15 V
12 V
12 V
9V
9V
5V
5V
90 V rms
265 V rms
Frequency movement is similar to QR flyback switching in 1st valley
41
42. Frequency vs. Load Current
Switching Frequency (kHz)115 Vac Switching Frequency vs. Load Frequency
450
Active-Clamp
Mode
400
350
300
20V
250
15V
200
12V
150
9V
100
5V
50
Discontinuous
Mode
42
3
2,8
2,6
2,4
2,2
Load Current (A)
2
1,8
1,6
1,4
1,2
1
0,8
0,6
0,4
0,2
0
0
43. DCM SW waveforms
20 V20 V
15 V
12 V
9V
15 V
12 V
9V
5V
5V
90 V rms
43
265 V rms
44. NCP1568 USB PD 65-W UHD Demonstration Board
Featured Devices:NCP1568 ACF Controller
NCP51530 Half-Bridge Driver
NCP4305 SR Controller
Full Load Efficiency: 93.4% @ 115 V rms (20 V/3.0 A)
93.6% @ 230 V rms (20 V/3.0 A)
Transformer Type:
Power Density:
Board Dimensions:
RM8 LP
30 W/in3 or 17 W/cm3
1.66” x 1.78” x 0.70” or
4.2 cm x 4.5 cm x 1.7 cm
45. UHD Board Performance
Achieving a full-load efficiency of 93.5% at a 60-W outputPrimary FETs running at 83 C
45
46. NCP1568 Demonstration Board Efficiency
Output Voltage (V)88
86
84
82
80
78
76
74
72
70
115 VAC
rms
rms
230 VAC
Limit
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
230 VAC
rms
rms
115 VAC
Limit
10% Load Efficiency vs.
Output Voltage
Efficiency (%)
94
92
90
88
86
84
82
80
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Efficiency (%)
4 Point Average Efficiency vs.
Output Voltage
Output Voltage (V)
47. Key Takeaways
ACF results in ZVS for both main and active-clamp FETs.High-frequency operation while achieving high efficiency is possible.
DCM transition is needed to pass stringent regulatory standards.
Elimination of heat sinks is possible with ACF topology.
Power density while employing ACF is 2 to 3 times that of a standard ac-dc supplies
Industry standard super-junction FETs yield excellent results up to 400 kHz.
47