Power Supply Design Seminar

Practical EMI Considerations for Low-Power

AC/DC Supplies

Reproduced from
2020 Texas Instruments Power Supply Design Seminar SEM2400
TI Literature Number: SLUP400
© 2020 Texas Instruments Incorporated

Power Seminar topics and online power training modules are available at: ti.com/psds
Practical EMI Considerations
for Low-Power AC/DC Supplies

Bernard Keogh
Joe Leisten
Electromagnetic interference (EMI)
• I built my 65 W adapter prototype and resolved all functional issues
• I ran first-pass EMI scan – and my design failed – badly! ~100 dBµV
– How can I fix the EMI?
– Where do I start?
• This presentation will show how to get to a result like this,
without necessarily adding a big EMI filter

48 dB

Before After

2
Agenda
• Introduction to EMI testing
• What causes EMI
• Differential-mode vs common-mode EMI
• EMI mitigation options
• Analyzing the transformer
• Troubleshooting & debug
• 65 W USB-PD example design using active clamp flyback (ACF) topology

3
Conducted emissions (CE) standards
• Summary of main product standards for conducted emissions

Product Sector CISPR Standard EN Standard FCC Standard

Automotive CISPR 25 EN 55025 --
Multimedia CISPR 32 EN 55032 Part 15
ISM CISPR 11 EN 55011 Part 18
Household appliances, electric CISPR 14-1 EN 55014-1 --
tools and similar apparatus
Lighting equipment CISPR 15 EN 55015 Part 15/18

Reference:
Timothy Hegarty, “An overview of conducted EMI specifications for power supplies,” http://www.ti.com/lit/wp/slyy136/slyy136.pdf

4
 How conducted EMI is measured

• Equipment under test (EUT) placed on

non-conductive table

• Horizontal & vertical ground planes

– Or screened room

• EUT powered through line impedance

stabilization network (LISN)

• Measure high-frequency (HF)

emissions from LISN

[1] EN55022, 2010, “Information technology equipment – Radio disturbance characteristics – Limits and methods of measurement”

5
 LISN – Line impedance stabilization network
• Presents stable, consistent & repeatable line source impedance
• Separation of power source noise current for measurement
• Low frequency power current passes straight through from AC source
• “Total” noise levels measured separately on L1 (live) and L2 (neutral)

EMI Receiver

** Functional equivalent circuit of a LISN, not a complete schematic **

6
EMI receiver – Built-in detector types

7
Component parasitics
Ideal

• Parasitic elements are the

L
dominant cause of EMI issues
ZL
Practical
ESR
L
• EMI noise is coupled & propagated
CPAR
through parasitic elements:
freq
Frequency
– Capacitive coupling
– Inductive coupling Practical

Ideal
C
• EMI filter performance is dominated by RLEAK ZC
C
parasitic elements at higher frequency:
ESR
– Parasitic capacitance of inductors
– Parasitic inductance of capacitors ESL
freq
Frequency

8
LISN + EMI filter + power supply
LISN EMI Filter Power Supply
50 µH
L1

+
LDM
CBULK

~
~
CX1
LCM

–
CX2
50 µH
L2
0.1 µF 0.1 µF
1 µF 1 µF CY2
CY1

50 Ω 50 Ω

EARTH

LISN – measures noise EMI filter – limits noise Power supply – generates the
that gets to the LISN noise
(CM – red, DM – blue) (parasitic cap – red dotted)

9
DM EMI filter and current path
LISN EMI Filter Power Supply
50 µH
L1

+
LDM
CBULK

~
~
CX1
LCM

–
CX2
50 µH
L2
0.1 µF 0.1 µF
1 µF 1 µF CY2
CY1

50 Ω 50 Ω

EARTH

• DM EMI filter limits DM noise that gets to the LISN

– X-caps divert current away from LISN, keep local to power supply
– DM choke high impedance reduces size of current flowing to LISN
10
CM EMI filter and current path
LISN EMI Filter Power Supply
50 µH
L1

+
LDM
CBULK

~
~
CX1
LCM

–
CX2
50 µH
L2
0.1 µF 0.1 µF
1 µF 1 µF CY2
CY1

50 Ω 50 Ω

EARTH

• CM EMI filter limits CM noise that gets to the LISN

– Y-caps divert current away from LISN, keep local to the power supply
– CM choke high impedance reduces size of current flowing to LISN
11
Why care about differential-mode (DM) EMI?
• DM noise conducts to the AC utility supply network
• Long AC distribution cables – act as good dipole antenna
• Will inadvertently radiate switching noise and interfere with radio communications
– (E.g., noise @ ~100 MHz will affect FM radio)

12
How is DM EMI generated?
• Switching ripple current produces ripple voltage across ESR (& ESL)
• Ripple voltage is the DM noise that needs to be attenuated/filtered
+

CBULK
~
~
–

VAC VBULK IPRI

DM EMI
Voltage

13
Mitigation options for DM noise
• Include EMI at the design phase
– Make design & component choices to minimize DM EMI signal amplitude
– Chose frequency, inductance, etc. to minimize PK-PK ripple current
– Choose capacitors with low ESR to minimize PK-PK ripple voltage
– Good PCB layout important to minimize EMI
• Design sufficient DM LC filter to reduce the ripple that gets onto the AC line input
LISN EMI Filter Power Supply
50 µH
L1
+

LDM
CBULK
~
~

CX1
LCM
–

CX2
50 µH
L2
0.1 µF 0.1 µF
1 µF 1 µF CY2
CY1

50 Ω 50 Ω

EARTH

14
 DM filter design methodology 2.63 A

• Measure, simulate or calculate time-domain current

waveform -0.33 A

2.0 µs

5.5 µs
• Fourier analysis of time-domain switching current
– Convert waveform into harmonic components

• Establish required attenuation at each frequency

– To get sufficient margin below the required EMI limit

• Design the DM filter to achieve required attenuation

– Need to check all frequencies of interest
– Typically limited by lowest frequency inside
measurement band
– Typically EMI starts at 150 kHz for AC/DC PSU
to meet EN55022 or EN55032

15
DM filter choke practical considerations
• Choke requires high attenuation over wide bandwidth:
– Load current amplitude typically several amps
– At 50 dBµV, current in LISN 50 Ω resistor only ~6.3 µA

• Beware inductance roll-off with DC-bias

– Must not saturate to be effective – needs high current rating
– Consider the peak line current for non-PFC – high crest factor

• Switching power stage has fast changing magnetic fields

– Beware filter bypassing & noise coupling
Practical
• Parasitic capacitance across DM inductor very important
ESR
– Reduces effectiveness, especially at high frequency L

• Example: To filter 300 kHz component, typically set LC freq. ~30 kHz CPAR
– Expect ~40 dB attenuation at 300 kHz (double-pole Þ 40 dB/decade)
– With parasitic cap Þ more like 30 dB attenuation only – even worse at higher frequency

16
Why care about common-mode (CM) EMI?
• Again, AC distribution cables and output load cables act as good uni-polar antenna
• CM noise will radiate from the cables and interfere with radio communications

17
How is CM EMI generated?
• Switching voltage across parasitic capacitance causes CM current flow to EARTH
+
~
~
–

VAC

CM EMI
Z
Voltage

EARTH

18
How is CM EMI generated?
• Switching voltage across parasitic capacitance causes CM current flow to EARTH
• CM noise also radiated to other circuit nodes
+
~
~
–

VAC

CM EMI
Z
Voltage

EARTH

19
Observing the time-domain CM signal at the output
• Useful debug technique – ball-park indication of CM performance
– Remove Y-cap temporarily (maximize signal)
– Power EUT through LISN, with resistor loads
– Wind several turns of wire around the load cables to
create capacitive sensing coil (pickup coil)
– Connect scope EARTH lead to LISN EARTH
– Connect scope tip to sensing coil
– Scope plot shows how much CM is coupled to output

20
Interpretation of time-domain CM signal
• Will see “switch-node” shaped waveform –
coupled to output

• Large PK-PK amplitude Þ bad CM noise

– Will require significant CM filtering to suppress
– Result from ACF example with 100 dBµV EMI
2 V/div, 2 µs/div

• Small PK-PK amplitude Þ good CM noise

– “Balanced” structure giving low CM
– Will require much smaller CM filter

• Residual HF “spikes” Þ should only need small

HF CM choke 200 mV/div, 10 µs/div

21
Mitigation options for CM noise
1. Shielding:
– Reduce flow of HF current to EARTH

2. Cancellation:
– Arrange transformer and power stage for
balanced CM

LISN EMI Filter Power Supply

3. Filtering: L1
50 µH

+
LDM

– Increase impedance of the EARTH return ~

~
CX1
LCM
–

CX2

path
50 µH
L2
0.1 µF 0.1 µF
1 µF 1 µF
– Provide alternative routes for the HF current CY1 CY2

50 Ω 50 Ω

EARTH

22
1. CM mitigation by shielding 1
2
1. Shielding inside the transformer
– Internal shields between pri & sec

2. Shielding outside the transformer

– GNDed flux-band

3. Shielding of noisy circuit nodes 4 3

– GNDed heatsinks over/around high-voltage
switching nodes

4. Shielding of EMI filter from switching cct.

– GNDed shields/enclosures around filter

23
Transformer internal shielding
• Shield added to keep most of CM current local to primary
• Shield is 1-turn winding Þ lower induced voltage, less voltage across parasitic
capacitance between shield & sec Þ less CM current flows
• Shield must be thin (< 50 µm) Þ minimize induced eddy current loss
– Eddy currents get very significant as FSW increases

24
2. CM mitigation by cancellation/balance
• Single-ended topologies – can add explicit
additional cancellation elements
– Add auxiliary (AUX) transformer winding
– AUX voltage proportional to CM waveform
– Arrange AUX polarity for opposite phase

– Capacitor to inject cancelling current, ICM2, to

balance CM current from primary, ICM1
– Injection capacitor explicit physical component
added to design
– Or can use parasitic capacitance, e.g., CS-AUX,
part of transformer structure

25
3. CM mitigation by filtering
• CM filter uses high-impedance CM chokes and low-impedance Y-capacitors
• CM choke limits the flow of CM current from EARTH through the LISN
• Y-cap provides low impedance to keep CM current local to primary GND and away from LISN

LISN EMI Filter Power Supply

50 µH
L1 +
LDM ~ CBULK
~

CX1
LCM
–

CX2
50 µH
L2
0.1 µF 0.1 µF
1 µF 1 µF CY2
CY1

50 Ω 50 Ω

EARTH

26
CM filter choke practical considerations
• Frequency response of core material
– High-µ cores Þ high L value @ low freq, but roll off fast at higher freq
– High-freq cores Þ low-µ, smaller L value @ low freq, better vs freq
– Sometimes need to use 2 CM chokes, 1 for LF & 1 for HF Split-Wound*
• Split-wound vs bifilar-wound toroid
– Split-wound popular, lower cost, “free” DM choke from leakage field
– Bifilar Þ 1-side insulated wire, higher cost, but better noise immunity
• Parasitic input-output cap – multi-layer windings
– Parasitic cap depends on number of turns & layers CPAR Bifilar-Wound*
• High CPAR input-output cap Þ worse @ HF L
– Less layers Þ lower L, but also lower CPAR
– Sectional bobbins – used to reduce CPAR
L

*CM choke 3-D images reproduced with permission of Wurth Elektronic

Multi-Section
27
CM filter choke – Impact of CPAR
• Split-wound 2-layer: CPAR

– 25T, 5.1 mH L
– Excess pass margin @ LF
– Low pass margin @ 20 MHz L
Split-Wound 2L
• Bifilar-wound 1-layer:
– 14T, 1.1 mH
– Low input-output CPAR
– Lower L at LF, but better at HF
Bifilar-Wound 1L – Better balance across frequency span

• Split-wound 1-layer:
– 14T, 1.4 mH
– Similar low input-output CPAR
Split-Wound 1L – Similar result as bifilar-wound

28
CM choke example
• Initial split-wound choke had issues due to asymmetric noise coupling
from transformer
– Shows up as big difference in EMI on L vs N

*CM choke 3-D image reproduced with permission of Wurth Elektronic

29
 CM choke example
• Changed to bifilar-wound choke, much
better EMI result; less difference L vs N
• Much better noise immunity

12 dB

*CM choke 3-D image reproduced with permission of Wurth Elektronic

30
Transformer CM noise analysis – PMP21479 ACF
• Initial design – interleaved flyback transformer construction, no internal shielding
• Same transformer used for initial test with poor 100 dbµV EMI result

31
Transformer CM noise analysis – PMP21479 ACF
• Circuit connections to the ACF power stage
• Note the secondary low-side rectifier – causes inverted secondary winding polarity

32
Transformer CM noise analysis – PMP21479 ACF
• Note that primary and secondary waveforms are inverse of each other
• This increases the CM voltage
– Caused by low-side rectifier

VOUT

VBULK CCLMP
DSEC

QCLMP VAUX
DAUX

QMAIN

33
Transformer CM balance – PMP21479 ACF
• Add CM balance auxiliary layer (purple) in-between inner PRI (noisier) to SEC interface:
– Fill layer completely – acts as shield between PRI & SEC
– Add turns to create CM balance, inject current to balance other PRI-SEC interface

• NOTE: this example shows one way to add CM balance

• But there are many different ways to achieve the same CM result
34
Transformer CM balance – PMP21479 ACF
• PRI, SEC & AUX bias connections same as before
• CM auxiliary layer starts at PRI GND, and winds in SAME direction as SEC

VOUT

VBULK CCLMP
DSEC

QCLMP VAUX
DAUX
QMAIN

35
Transformer CM balance – PMP21479 ACF
• Waveforms – PRI & SEC same as before
• CM AUX – same phase as secondary – but amplitude increased to compensate for outer
PRI

VOUT

VBULK CCLMP
DSEC

QCLMP VAUX
DAUX
QMAIN

36
Transformer CM balance – PMP21479 ACF
• CM AUX – need to adjust number of turns to balance the CM nulling
• Waveforms showing:
A: Slightly under-compensated A: 0.2 V/div, 5 µs/div

B: Slightly over-compensated
C: “Just right”

VOUT C: 0.2 V/div, 10 µs/div

VBULK CCLMP
DSEC

QCLMP VAUX
DAUX
QMAIN B: 0.2 V/div, 5 µs/div

37
 Transformer “housekeeping” best practices
• Center-leg air-gaps only – outer legs will radiate
• “Noisier” windings on inside of layer structure
• Tie ferrite core to local primary GND
• Flux-band to minimize stray coupling
– GNDing & flux-banding can give >10 dB improvement!
• Interleaving trade-offs Flux-Band +
– Lower leakage inductance EMI Shield
~15 dB
– Higher pri-sec parasitic capacitance, higher CM
• Be aware of internal construction
– Average CM voltage across the pri-sec capacitance
– Arrange winding layers to minimize voltage difference
• Minimizes CM current
38
PMP21479 65 W ACF USB-PD – Final result
• Improve transformer structure, add CM balance/shield winding layer
• Add transformer grounding & flux-banding + EMI shield
• Improve CM choke to bifilar-wound type
• Improve output capacitor location, improve PCB orientation
– Largely “low-cost” improvements
– Small efficiency penalty (eddy loss in cancellation layer)

39
Summary & conclusions
• Consider EMI right from the start – inherent part of the power supply design
• Minimize DM & CM EMI noise at source
• DM filter can be designed/calculated/simulated more easily than CM
• CM balance is important – as much as practically possible
• Debug to establish if EMI issue is CM or DM or both
• Assess CM performance in time-domain – compare different transformers
• For isolated PSU, transformer is most important component
– Internal construction details, CM balance, shielding, parasitic capacitance, housekeeping
• EMI filter components – be aware of parasitics and HF effects
• PCB layout and component placement – be aware of stray coupling paths

40
References
• “Understanding and Optimizing Electromagnetic Compatibility in Switchmode Power Supplies,”
Bob Mammano & Bruce Carsten, 2002 TI Power Supply Design Seminar.
– http://www.ti.com/lit/slup202
• “Flyback transformer design considerations for efficiency and EMI,”
Isaac Cohen & Bernard Keogh, 2016/7 TI Power Supply Design Seminar.
– http://www.ti.com/lit/slup338
• “Input EMI Filter Design for Offline Phase-Dimmable LED Power Supplies,”
James Patterson & Montu Doshi, 2012/3 TI Power Supply Design Seminar.
– http://www.ti.com/lit/slup298
• “Designing low-EMI power converters for industrial & automotive systems,”
Perry Tsao, David Baba & JP Fung, 2016/7 TI Power Supply Design Seminar.
– http://www.ti.com/lit/slup362
• “Understanding Noise-Spreading Techniques and Their Effects in Switch-Mode Power Applications,”
John Rice, Dirk Gerhke & Mike Segal, 2008/9 TI Power Supply Design Seminar.
– http://www.ti.com/lit/slup269
• “65W Active clamp flyback with Si FETs reference design for a high power density 5-20V AC/DC adapter,”
PMP21479, Brian King, 2019.
– http://www.ti.com/tool/PMP21479
41
APPENDIX – BACK-UP SLIDES
Transformer flux-band detailed results – 115 V

• EMI shield only (over switch-node

and between EMI filter and
transformer)
• NO flux-band, ferrite core floating
• Biggest improvement ~5-8 MHz

43
Transformer EMI shield detailed results – 230 V

• EMI shield only (over switch-node

and between EMI filter and
transformer)
• NO flux-band, ferrite core floating
• Biggest improvement ~3-8 MHz

44
 Transformer flux-band detailed results – 115 V

• EMI shield only (over switch-node and

between EMI filter and transformer)
• Add flux-band, connected to local
primary GND
• Much more significant reduction from
150 kHz to ~4 MHz

45
 Transformer flux-band detailed results – 230 V

• EMI shield only (over switch-node and

between EMI filter and transformer)
• Add flux-band, connected to local
primary GND
• Much more significant reduction from
150 kHz to ~4 MHz
– Especially for AVERAGE, which is much
tougher at 230 V

46
 IMPORTANT NOTICE AND DISCLAIMER

TI PROVIDES TECHNICAL AND RELIABILITY DATA (INCLUDING DATASHEETS), DESIGN RESOURCES (INCLUDING REFERENCE
DESIGNS), APPLICATION OR OTHER DESIGN ADVICE, WEB TOOLS, SAFETY INFORMATION, AND OTHER RESOURCES “AS IS”
AND WITH ALL FAULTS, AND DISCLAIMS ALL WARRANTIES, EXPRESS AND IMPLIED, INCLUDING WITHOUT LIMITATION ANY
IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT OF THIRD
PARTY INTELLECTUAL PROPERTY RIGHTS.
These resources are intended for skilled developers designing with TI products. You are solely responsible for (1) selecting the appropriate
TI products for your application, (2) designing, validating and testing your application, and (3) ensuring your application meets applicable
standards, and any other safety, security, or other requirements. These resources are subject to change without notice. TI grants you
permission to use these resources only for development of an application that uses the TI products described in the resource. Other
reproduction and display of these resources is prohibited. No license is granted to any other TI intellectual property right or to any third
party intellectual property right. TI disclaims responsibility for, and you will fully indemnify TI and its representatives against, any claims,
damages, costs, losses, and liabilities arising out of your use of these resources.
TI’s products are provided subject to TI’s Terms of Sale (www.ti.com/legal/termsofsale.html) or other applicable terms available either on
ti.com or provided in conjunction with such TI products. TI’s provision of these resources does not expand or otherwise alter TI’s applicable
warranties or warranty disclaimers for TI products.

Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265
Copyright © 2020, Texas Instruments Incorporated