Alfred Hesener Fairchild Semiconductor Europe
www.fairchildsemi.com
Agenda
• Introduction • • Re Regu gula lati tion onss an and d sta stand ndar ards ds fo forr EMI EMI • Me Meas asur ureement an and d so source cess of of EMI EMI • Conducted EMI • Radiated EMI • EM EMII as as int integ egra rall par partt of of the the de desi sign gn fl flow ow • Conclusion
Company Confidential
Page 2
www.fairchildsemi.com
Agenda
• Introduction • • Re Regu gula lati tion onss an and d sta stand ndar ards ds fo forr EMI EMI • Me Meas asur ureement an and d so source cess of of EMI EMI • Conducted EMI • Radiated EMI • EM EMII as as int integ egra rall par partt of of the the de desi sign gn fl flow ow • Conclusion
Company Confidential
Page 2
www.fairchildsemi.com
Introduction EMI more and more complex • Increa Increasi sing ng powe powerr densi density, ty, fast faster er switc switchin hing, g, highe higherr curren currents ts are are caus causin in more more EMI EMI-r -rel elat ated ed iss issue uess • Cond Conduc ucte ted d / radi radiat ated ed EMI EMI • Furt Furthe herr chan change gess comp compli lica cati ting ng thi thing ngss • New semico semicondu nducto ctorr switc switches hes are faster faster • New topolo topologie giess (e.g. (e.g. Quas Quasii-re reson sonant ant)) • How to ac eve a “ro ust” es gn? • Embed Embed EMI EMI into into the the desi design gn flo flow w from from the the begi beginni nning ng • Emit Emit low low EMI EMI levels levels to meet meet regulati regulations ons (don’t (don’t dist disturb urb other other applications nearby) EMI compliance • Work Work pro prope perl rly y (be (be self self-c -com ompl plia iant nt)) Robustness Company Confidential
Page 3
www.fairchildsemi.com
Agenda
• Introduction • • Re Regu gula lati tion onss an and d sta stand ndar ards ds fo forr EMI EMI • Me Meas asur ureement an and d so source cess of of EMI EMI • Conducted EMI • Radiated EMI • EM EMII as as int integ egra rall par partt of of the the de desi sign gn fl flow ow • Conclusion
Company Confidential
Page 4
www.fairchildsemi.com
Different types of EMI and their characteristics “Reduce emission of source”
Emitter
Capacitive Coupling
Coupling
Inductive Coupling
Coupling
“Reduce sensitivity of receiver”
Receiver
signals in the circuit
system
Typically <30MHz ny no sy s gna n system (RC) filtering Company Confidential
“Reduce transmission in the system”
Medium-high frequencies e
arge Metal shield
Typically > 30MHz arge Magnetic shield Page 5
High frequencies as sw c ng Electromagnetic shield www.fairchildsemi.com
Agenda
• Introduction • • Regulations and standards for EMI • Measurement and sources of EMI • Conducted EMI • Radiated EMI • EMI as integral part of the design flow • Conclusion
Company Confidential
Page 6
www.fairchildsemi.com
Regulations and standard for EMI EN550xx and EN61000 most important •
Two main considerations:
•
• Limit the amount of emission which a given application generates • Define minimum immunit levels a iven a lication must tolerate EN550xx – the “EMI” norm (class A = “consumer”, class B = “industrial”) •
CISPR11, EN55011 for industrial, medical, scientific applications
•
CISPR13, EN55013 for consumer applications
•
,
or ome app ances, power oo s, nvo v ng mo on con ro
•
CISPR15, EN55015 for lighting equipment
•
CISPR22, EN55022 for computing applications
•
CISPR16, EN55016 defines the measurement method
•
Many applications being tested against a “mix” of different norms (e.g. EN55022 for frequencies >150kHz, EN55015 for frequencies <150kHz)
•
EN61000 – the “PFC” norm (equipment classes see next page) th
•
= .
•
EN61000-3-2 for applications < 16A
•
EN61000-3-12 for applications with 16A…75A
•
EN61000-4-7 defines the measurement and evaluation method - -
.
. .
-
•
Many further standards exist, dealing with more specialized applications
•
EN61000 specifies maximum harmonic currents, not a power factor
Company Confidential
. .
Page 7
www.fairchildsemi.com
Regulations and standard for EMI Equipment classes for EN61000 Class
Equipment
Power
3phase equipment, household appliances, , , equipment, everything not B, C or D
Comment
Limit values are defined as absolute values Limit values are defined as absolute values
,
Limit values defined as harmonic
C
Lighting
< 25W
D
Personal Computer, Monitor, Television
600W
Company Confidential
Page 8
Limit values defined only for 3rd and 5th harmonic, relative to first harmonic
input power www.fairchildsemi.com
The power factor Simulation results •
Simulation shows input and bus cap voltage, and current spikes in the input
•
High dI/dt illustrates significant harmonic content
•
, • EN61000 considers harmonics to 2kHz/2.4kHz – this would be a pretty large filter if realized with passive components • Attenuation of this filters’ com onents for hi her fre uencies conducted EMI would be low, due to potentially high parasitic capacitance, and it may not help with CM noise at all
Limit values for EN61000 class D
Company Confidential
Page 9
www.fairchildsemi.com
Agenda
• Introduction • • Regulations and standards for EMI • Measurement and sources of EMI • Conducted EMI • Radiated EMI • EMI as integral part of the design flow • Conclusion
Company Confidential
Page 10
www.fairchildsemi.com
Measurement and sources of EMI Conducted EMI test setup
Line Impedance Stabilizer Network (“LISN”): - Defined impedance for noise voltage measurement - Blocking the noise coming from the grid
Company Confidential
Page 11
www.fairchildsemi.com
Measurement and sources of EMI Conducted EMI limits Vertical: Amplitude in dbuV Horizontal: Frequency in MHz Solid blue line: EN55011/22 limits for average Solid red line: EN55022 limits for quasipeak Black spectrum line: average measurement values
Frequency Limit (dbuV) Limit (V) Comment 9kHz ... 50kHz 110 316mV EN55011 Quasipeak 50kHz ... 150kHz 90 ... 80 32mV ... 10mV EN55011 Quasipeak , 66 ... 56 2mV ... 0.63mV with log (frequency) 150kHz ... 500kHz EN55022 B, Average; linearly falling with 56 ... 46 0.63mV ... 0.2mV log (frequency) 56 630uV EN55022 B, Quasi-peak 0.5MHz ... 5MHz 46 200uV EN55022 B, Average 60 1mV EN55022 B, Quasi-peak 5MHz ... 30MHz 50 316uV EN55022 B, Average
Company Confidential
Page 12
Frequency range
Bandwidth (-6dB)
9kHz ... 150kHz
200 Hz
150kHz ... 30MHz
9 kHz
30MHz ... 1GHz
120kHz
www.fairchildsemi.com
Conducted EMI Differential and common mode noise •
In most cases, two different noise voltages will appear at nodes L and N • Separate into differential (“DM”) and common mode (“CM”) noise • Different filtering required for both noise types!
•
Differential mode noise appears out of phase at the nodes •
•
“ ”
Common mode noise appears in phase at both nodes • Noise current flows via ground and back through the lines (“2”) L
N
DM noise current
2
Ground
Parasitic
CM noise current
Company Confidential
Page 13
www.fairchildsemi.com
Measurement and sources of EMI Conducted EMI as result of switching •
The main switching action will cause a current flow into / out of the bulk cap, at the main switching frequency • This current flow causes a noise voltage to appear at the input • Typical values are ESR max = 1.9Ω, ESLtyp = 20nH • Im edance minimum is ESR will increase at hi h fre uencies EMI is primarily a result from parasitic elements in the circuit
Company Confidential
Page 14
www.fairchildsemi.com
Conducted EMI Different filter types Filter type
Balanced
Unbalanced
Pi filter 18 db / oct 60 db / dec T filter 18 db / oct 60 db / dec er 12 db / oct 40 db / dec (Calculation of component values is explained later) Company Confidential
Page 15
www.fairchildsemi.com
Conducted EMI Common mode vs differential mode • For common mode noise, the line to line capacitors do not help • Onl the inductors contribute but t
icall the are too small
• Introduce a common mode choke • Designed for (large) leakage inductance to provide DM filter function
Choke (with leakage inductance)
line cap
Example of a 200W power supply input stage with a two-stage CM choke Company Confidential
Page 16
www.fairchildsemi.com
Conducted EMI Calculation of the filter components
Design impedance
Input data
Line frequency
f Line
Minimal RMS voltage
Vmin
Maximum RMS load current
Imax
Lowest switching frequency
fswmin
Attenuation
Determine required attenuation level er fre uenc from simulation or measurement
Calculate the component values
Determine suitable filter topology and cutoff frequency so attenuation oals are met with a mar in of 6...10dB (but f cut > 10* f Line) Company Confidential
Page 17
www.fairchildsemi.com
Conducted EMI Simulation and results
• Simulation for compliance: Noise generation and filter attenuati are mostl determined b arasitic elements in the circuit
• Noise generation: Leakage inductance, ESR, ESL, capacitiv coupling (to ground) • Attenuation: Core frequency response, capacitive coupling • Most simulators allow to set parasitics for all passive
• Using a behavioural model for the noise (current) source is a good approximation
• Simulation for function and robustness: Very complex – better t design accordingly, test a prototype, implement fixes in final cir
Conducted EMI Example values for parasitics
Inductor
Parallel capacitance e.g. 50pF for 1mH
Capacitor
Series resistance e. . 1.9 Ohm for 100uF Series inductance e.g. 20nH for 100uF
Transformer
Leakage inductance e.g. 10uH for 200uH (prim) Parasitic capacitance . .
CM choke
Leakage inductance e.g. 300uH for 10mH aras t c capac tance e.g. 100pF for 10mH
Company Confidential
Page 19
www.fairchildsemi.com
Conducted EMI Simulation circuit example Bus cap Load ±0.05A
100kHz
Input voltage 230V / 50Hz
Company Confidential
CM filter
DM filter (T type) Page 20
Parasitic
www.fairchildsemi.com
Conducted EMI Simulation results without filter EN55022 limits (quasi-peak)
with filter EN55022 limits (quasi-peak)
Company Confidential
Page 21
www.fairchildsemi.com
Agenda
• Introduction • • Regulations and standards for EMI • Measurement and sources of EMI • Conducted EMI • Radiated EMI • EMI as integral part of the design flow • Conclusion
Company Confidential
Page 22
www.fairchildsemi.com
Radiated EMI What generates it? •
•
Magnetic EMI is caused by changing currents:
Vnoise =
R M R S + R M
dI *M*
Current (di/dt)
R S
Vnoise + Vmeas
dt M
•
Coupling factor M depends on: •
stance, area an or entat on o t e disturbing magnetic loops
• Magnetic absorption between the loops • Current risetime • Impedance of the receiver Company Confidential
Page 23
www.fairchildsemi.com
Radiated EMI Main contributions to radiated EMI •
Avoid high dI/dt – move to softer (slower switching) or zero-current switching
•
•
“Reduce emission of source”
Reduce the coupling factor M between the magnetic loops
•
Or entat on o t e current oops s ou
•
The current loop areas should be made as small as possible
•
Increase the distance between the emitting current loop and the loop ickin u the noise ener transfer ro ortional to ower of 3
• •
, which elements will only see current flow in one part of the cycle – these elements are very likely to be in a current loop with high dI/dt e ort ogona , not para e .
“Reduce transmission ”
Magnetic shielding
Make the signal processing nodes in the system as low-impedance as possible
•
urrent- ase s gna trans er
•
Add additional resistors to Ground at sensitive
•
Differential signaling
Company Confidential
Page 24
“Reduce sensitivity of receiver”
www.fairchildsemi.com
Radiated EMI How to measure radiated EMI
Company Confidential
Page 25
www.fairchildsemi.com
Radiated EMI How to identify “hot spots” • Use a two-channel scope • • Connect the H-field probe to a probe amplifier (if necessary) and to the second channel (proper termination required) • Use the main switching signal as a trigger signal • “Wander” around the PCB to identify areas of large emission, then • Take (static) pictures of the critical field signals to determine fre uenc and ualit factor this can be used to identif the elements of the resonant tanks)
Company Confidential
Page 26
www.fairchildsemi.com
Transformers radiating magnetic fields High leakage inductance == leakage field
ot core as t e sma est e
oro w expose core emits more than it should Company Confidential
not surpr s ng
core – etter t an
e er o move a r gap o center leg (may increase AC losses) Page 27
core t g ter w n ng
E core – stronger field due to leakage inductance
www.fairchildsemi.com
Radiated EMI - various issues (incomplete list…..) • • • • • • • • • •
Leakage inductance fields External field of air gaps Diode reverse recovery ater a s os ng t e r amp ng Caps becoming inductive Inductors becoming capacitive secondar side chokes ickin u magnetic noise Ringing between parallel caps Ringing between parallel rectifiers
Company Confidential
Page 28
www.fairchildsemi.com
Example 1: 70W QR flyback supply 18MHz peak from transformer
Company Confidential
Page 29
www.fairchildsemi.com
Example 1: 70W QR flyback supply Two different diodes in the snubber
Yellow: Drain voltage of main MOSFET Blue: Magnetic field at snubber
• Fast snubber diode gives faster rise / fall times and lower losses • Slow diode with much larger Qrr shows significant magnetic EMI • Impact can not be seen in the node voltage – need to investigate
Company Confidential
Page 30
www.fairchildsemi.com
Example 2: 300W CCM PFC board Strong EMI event at turn-off Yellow: Drain voltage of MOSFET Blue: Magnetic field
At MOSFET
At Diode
Inside inductor
•
, ~ Coss (780pF) and the parasitic inductance of the PCB and package (20nH), well damped, after which the inductor ringing takes over • Smaller EMI spike at the (SiC) diode shows ringing at similar frequency, indicating that this is im osed b the ower MOSFET in this case the e uivalent char e of the diode is 100x smaller so the contribution is too) • Long ringing tail of the inductor shows the energy flowing between the inductor and its parasitic capacitance • Field is stron distributed air a and the tail lasts for 800ns hi h • Ringing frequency is ~9MHz, parasitic cap ~20pF (estimated) effective inductance is reduced by 40x at this frequency! (core material) Company Confidential
Page 31
www.fairchildsemi.com
Example 3: 400W interleaved PFC PCB layout
Company Confidential
Page 32
www.fairchildsemi.com
Example 3: 400W interleaved PFC Main difference between two boards
“
”
Company Confidential
“
”
10dB difference in magnetic field peak intensity Page 33
www.fairchildsemi.com
Example 4: 200W LLC power supply Turn-off of main LLC stage
• Small EMI fields around the converter, most at leakage inductor (gapped core) which itself has small leakage resonance at 22.7MHz • 70pF and 0.7uH leakage • Not visible in the node voltage
ns
v
Red: Magnetic field at leakage ind. Pink: Phase node voltage
Company Confidential
Page 34
www.fairchildsemi.com
Agenda
• Introduction • • Regulations and standards for EMI • Measurement and sources of EMI • Conducted EMI • Radiated EMI • EMI as integral part of the design flow • Conclusion
Company Confidential
Page 35
www.fairchildsemi.com
EMI as part of the design flow Design steps
Write the
Select the
• Consider impedance
• EN550xx • EN61000 • Time for EMI testing • EMI filter
Company Confidential
Calculate the
PFC
Page 36
• Make circuit nodes low impedance (esp. control loop) • Avoid high di/dt and dv/dt
www.fairchildsemi.com
EMI as part of the design flow Design steps Simulate the design
• Simulate with a LISN
Build a prototype
Test the prototype
• Try to be close to final -
filter) to predict noise nents, so the coupling • Use behavioural model and radiated EMI can for the load to save be tested simulation time • Minimize high-current loop area • Chose filter topology • Minimize node area with for needed attenuation levels simulate a ain hi h dv/dt • Put realistic values for • Leave some space at the parasitic elements input to put a EMI filter • Use an impedance analyzer o measure yp ca com ponents and put these in Company Confidential
Page 37
• After checking the function, EMI testing to see the “real” conducted noise • Check CM noise on a grounded metal plate (worst case) • Perform first radiated EMI tests to identif critical spots in the circuit • Compare with simulation results and calibrate the mo e s
www.fairchildsemi.com
EMI as part of the design flow Design steps Add the EMI filter
• Build the EMI filter into the rotot e and erform full functional test again • Check if EMI filter impedance and possible reso• Perform pre-compliance testing again to see if the measured attenuation matches calculation
Company Confidential
Design the final version
Test the final version
• Final implementation will • After full functional testing, chan e the noise “si nature” erform re-com liance of conducted DM and CM testing especially and high as well as radiated EMI and low line conditions • (Alternate source) compoover full load range may have different parasitics, so the EMI behaviour may change – need to add appropriate margins
Page 38
components (including different vendors) • Build several prototypes and check if the noise results are repeatable
www.fairchildsemi.com
Agenda
• Introduction • • Regulations and standards for EMI • Measurement and sources of EMI • Conducted EMI • Radiated EMI • EMI as integral part of the design flow • Conclusion
Company Confidential
Page 39
www.fairchildsemi.com
Conclusion
• There is no silver bullet! • • Assess implications early in the design cycle, and prepare • The later in the design cycle the problem is detected, the more expensive it is to fix • Use topologies and control ICs that create less noise to begin with • LLC, QR y ac , PSR
[1]
Didier Bozec, David Cullen, Les McCormack, John Dawson, Bryan Flynn: An investigation into the EMC emissions from switched mode power supplies and similar switched electronic load controllers operating at various loading conditions (IEEE Symposium on Electromagnetic Compatibility, Santa Clara CA, August 2004)
[2]
Bruce Carsten: Application note for H-field probe (http://bcarsten.com)
[3]
Jonathan Harper: Electromagnetic compatibility design for power supplies (Fairchild Semiconductor power seminar series 2004/2005)
[4]
Richard Lee Ozenbaugh: EMI filter design (CRC, Nov 2000)
[5]
Christophe Basso: Switch-Mode Power Supplies SPICE Simulations and Practical Designs“, McGraw-Hill, 2008
Company Confidential
Page 40
www.fairchildsemi.com