EMC Insights and Solutions:

DC/DC Converter EMC Troubleshooting

Livonia, MI EMC Lab

April 2022
 Presenter Intro: Mark Malik
• Senior FAE supporting automotive OEM and Tier 1
customers in the Midwest region, based out of
Livonia MPS office
• 20 years of experience in semiconductor field
support, automotive/military hardware design and
automotive OEM system level design
• Recently supporting OEM, Tier 1/2 customers with
analog hardware design: power supplies, SERDES,
motor control, LED control
• Passion for electronics and solving problems (that
sometimes do not exist)
• Outside of work: married with children,
snowboarding, electronics and other DIY activities
Agenda

Approach to EMC Failure Troubleshooting

Tools for EMC Troubleshooting

Basics on Making Electrons Behave

Examples with MPS Evaluation Boards

Inductor Impact on EMC

Structured Troubleshooting:
• Which EMC test is failed by DUT? (Radiated/Conducted Emission)

• Or Immunity?

• What does the test set-up look like?

• Housing: metal or plastic?

• Are cables connected to the system?

• At which frequency does the DUT exceed the limits?

• Is it possible to identify the source clock(s)?

• Where are the DC/DC converters placed in the system?

Structured Troubleshooting:
• Are filter elements placed at each cable?

• What is the distance from DC/DC to the cable/connector?

• Identify and mark all high dV/dt and dI/dt circuit nodes!

• Check the routing of those circuit nodes for potential coupling!

• Use snap-on ferrites on cables, try to distinguish CM and DM noise!

• Place shielding over DC/DC converter block

Tool Set for Troubleshooting:
• Copper Foil: Used to create a shielding around certain areas or
devices, modify ground. Low cost and versatile.

• Snap-on Ferrites: Used to create impedance to high frequency signals

travelling along a cable. Low cost, not as versatile.

• Field Probes: Used to identify areas of relatively high field strength

on a circuit board. Note: identifying areas of high field strength may
does not guarantee finding the source of the EMC problems, but
those areas are much more likely to be causing problems than not.
Medium cost and versatile.

• Current Probes: Can be used in conjunction with spectrum analyzer

to look at differential and common mode currents. Can be useful to
distinguish between common/differential modes. High cost, not as
versatile.
Use Cu-Foil
Start with DC/DC and Coil..

.... to part of the PCB

Use Snap-on Ferrite

MPQ43xx
Field Probes
H and E Field probes

They can be used with

Scope, Spectrum Analyzer
or EMC Receiver
Basics on EMC
• A moving charge generates an electromagnetic field –
remember the Right Hand Rule
• Charges moving in opposite directions create opposing fields
which cancel
• An electromagnetic field applied to a conductor causes free,
charged particles to move
• Rise and fall times affect frequency content and amplitude –
5ns versus 10ns can be ~100 times different at 100MHz
• Drift Velocity of electrons in 18AWG wire with 1A current is
µ𝒎
approximately 74
𝒔

• Emissions: 20dBµV → 200nA into 50Ω, Immunity: 0.2V  200nA into 1MΩ
• If every charge that leaves a source gets back to the source
using the smallest possible path/loop, it will be a boring day
in the ALSE chamber
Basics on EMC
Case #1 – For every charge moving to the right there is a charge moving to the left

Case #2 – There is an unequal amount of charge moving in either direction

Case #3 – There are charges travelling in the same direction

Case #4 – There is equal charge flow in both directions, but the paths are different
Refresh: Buck Converter Voltage and
Current Waveforms
Simplified ☺
IQ1 𝒅𝒊
high Vsw TON
𝒅𝒕
Vin 𝒅𝑽
ΔIL
IOut high
𝒅𝒕

Vout
TOFF
t
ICin 𝒅𝒊
high t
𝒅𝒕 IOut
Vin
𝒅𝑽
Cin high  𝑬 − 𝑭𝒊𝒆𝒍𝒅
𝒅𝒕
t
Q1 IOut Increase with voltage
SW Vout Increase with antenna size
and height above PCB
𝒅𝒊
IQ2 high 𝒅𝒊
𝒅𝒕 Q2 high  𝑯 − 𝑭𝒊𝒆𝒍𝒅
IOut 𝒅𝒕
Increase with current
Increase with loop size
and height above PCB
return current

….and Boost Converter is just an mirror image.

Refresh: Boost Converter Voltage and
Current Waveforms
Simplified ☺
IQ1 𝒅𝒊
high Vsw TON
𝒅𝒕
Vin 𝒅𝑽
ΔIL
IOut high
𝒅𝒕

Vout
TOFF
t
ICin
Cout 𝒅𝒊
high t
𝒅𝒕 IOut
Vin
𝒅𝑽
Cin high  𝑬 − 𝑭𝒊𝒆𝒍𝒅
𝒅𝒕
t
Q1 IOut Increase with voltage
SW Vout Increase with antenna size
and height above PCB

IQ2 high
𝒅𝒊
𝒅𝒕
IIn Q2 high
𝒅𝒊
 𝑯 − 𝑭𝒊𝒆𝒍𝒅
IOut 𝒅𝒕
Increase with current
Increase with loop size
and height above PCB
return current
CE FM Band Common Mode Noise
Buck converter 13V input running at 2MHz
Drives ~260mV at 100MHz at SW
OEM AV Limit: 12dBµV which is 4µV at 50Ω input of EMI receiver
Current flowing in this AC loop at 100MHz has to be below 80nA

SW-node (NOISE SOURCE)

𝑽𝑽𝑫𝒓𝒊𝒗𝒆 Parasitic capacitance
𝒊𝑨𝑪 = From SW to GND plane
DC/DC circuit
(𝒁𝑪𝒂𝒑 + 𝒁𝑮𝑵𝑫 + 𝒁𝑪𝒂𝒃𝒍𝒆 + 𝒁𝑬𝑴𝑰 𝑹𝒆𝒄𝒆𝒊𝒗𝒆𝒓 ) PCB (4 Layer)

cable
Z(cable)
Common mode current on both cables
AN &
EMI Z(Capacitance)
Receiver input
Resistance: 50Ω

Z (Ground plane)
5 cm Dielectric between DUT and GND (Table plate)
Identify the source:
Are discrete frequency lines above the limit or “mountains”?
Try to identify the source.

Sometimes the sources is obvious. ..and sometimes not …

f of harmonics is equal to fundamental

Identify the source:
Which clocks are used in the system?
Measure exact frequencies of all clocks.
Create a table with the clocks, harmonics and mixing products.

f1 [MHz] f2 [MHz] f2 - f1 [MHz] f1+f2 [MHz] f3 [MHz] f3 + f1 [MHz] f3 - f1 [MHz]

Fsw 0.489 1.93 1.441 2.419 20.03 20.519 19.541
2x 0.978 3.86 2.882 4.838 40.06 41.038 39.082
3x 1.467 5.79 4.323 7.257 60.09 61.557 58.623
4x 1.956 7.72 5.764 9.676 80.12 82.076 78.164
5x 2.445 9.65 7.205 12.095 100.15 102.595 97.705
6x 2.934 11.58 8.646 14.514 120.18 123.114 117.246
7x 3.423 13.51 10.087 16.933 140.21 143.633 136.787
8x 3.912 15.44 11.528 19.352 160.24 164.152 156.328
9x 4.401 17.37 12.969 21.771 180.27 184.671 175.869
10x 4.89 19.3 14.41 24.19 200.3 205.19 195.41
11x 5.379 21.23 15.851 26.609 220.33 225.709 214.951
12x 5.868 23.16 17.292 29.028 240.36 246.228 234.492
13x 6.357 25.09 18.733 31.447 260.39 266.747 254.033
Try to find the path:
• Increase Input Filter capacitors by 2x
• Increase Input Filter Coil by 2x

• What is the distance between Buck CIN and Filter?

• Is an OFF Board Filter effective?

Large boards with many DC/DC power supplies:
Several DC/DC Buck at one large Power Rail
IQ1 𝒅𝒊
high
𝒅𝒕
ΔIL
This can be the main System IOut
Input like 12V or 24V.
t

Buck 1 High di/dt input current

PCB traces have CIN1 of Buck 1, splits to
few mΩ and nH CIN1 to CIN4, according
trace impedance between
System Connector CIN2
Buck 2 DC/DC CINx.
Same for the other Buck 2-4

Buck 3 Potential radiation and

CIN3
EMC Filter coupling in other circuits
possible!

Buck 4
CIN4
Several DC/DC Buck at one large Power Rail
IQ1 𝒅𝒊
high
Intermediate rail, ΔIL
𝒅𝒕

like 5V or 3.3V “POL” on a IOut

large Board
t

Buck 1 High di/dt input current

PCB traces have CIN1 of Buck 1, splits to
few mΩ and nH CIN1 to CIN4, according
System trace impedance between
Connector CIN2
Buck 2 DC/DC CINx.
DC Same for the other Buck 2-4
DC

Primary Buck 3 Potential radiation and

EMC Filter CIN3
Buck coupling in other circuits
possible!
Part of Buck1
input current will
Buck 4
flow in Primary buck CIN4
Cout. Between those circuit nodes
high di/dt current will flow
Several DC/DC Buck at one large Power Rail
IQ1 𝒅𝒊
Place small Coil of 100nH…1uH high
𝒅𝒕
Between individual DC/DC ΔIL
IOut

t
Coil needs to have
much higher L as PCB Buck 1
CIN1
trace between DC/DC

System Connector CIN2

Buck 2 This configuration forces
the high di/dt of each
Buck to flow in its local
CIN !
CIN3
Buck 3
EMC Filter

CIN4 Buck 4
MPS Example #1: 5A Buck Fsw=2MHz
R6 is reducing CE CM
Noise at 100MHz by ~3/5dB
R6=4.7 Ω R-Bst
Changed L3 from XAL5030-1µH
to XAL4020-1µH
5V 1 Ω

MPQ43xx

Option to set
Fsw
Top Layer
Input Filter placed on bottom side
“T” Capacitor Versus “I” Capacitor Design
MPQ43xx 5V 5A Buck 2.2MHz with SSFM
Input Filter on Bottom
Input filter has around 50dB
damping at 2.2MHz

MPQ43xx
Input L, C Filter – simplified 1-Stage vs. 2-Stage
Input EMC Filter: Single-Stage vs. Two-Stage =Enter your parameter

Single Stage Fundamental 1st Harm 2nd Harm 3rd Harm 4th 5th
Fsw: 2.20MHz 4.40 6.60 8.80 11.00 13.20
Omega-Fsw 13.821/µs 27.65 41.47 55.29 69.11 82.94
L_single: 0.33µH
XL: 4.56Ohm 9.12 13.68 18.25 22.81 27.37
C-effective: 0.70µF
XC: 0.10Ohm 0.052 0.034 0.026 0.021 0.017
Damping -33.09dB -44.99 -52.00 -56.99 -60.86 -64.03

Two stage filter

design:
1st L: 0.10µH
XL: 1.38Ohm 2.76 4.15 5.53 6.91 8.29
1st C: 0.60µF
XC: 0.121Ohm 0.060 0.040 0.030 0.024 0.020
Damping 1 -21.91dB -33.42 -40.36 -45.32 -49.18 -52.33

2nd L: 0.10µH
XL: 1.38Ohm 2.76 4.15 5.53 6.91 8.29
2nd C: 0.40µF
XC: 0.181Ohm 0.090 0.060 0.045 0.036 0.030
Damping 2: -18.73dB -29.99 -36.88 -41.82 -45.67 -48.82

Total Damping: -40.65dB -63.40 -77.23 -87.13 -94.84 -101.16

Simplified calculation: 20 log (Xc/(Xc + XL)) at Fsw, 2x Fsw etc…

• Murata has a filter simulator as part of their SimSurfing Suite

• Circuit simulation can also be done
Reduced Input Filter for 2MHz with SSFM
This filter has ~ 33dB damping at 2.2MHz

CIN3/4= 470nF 0603

L1: BLM or short 330nH 0805
CE Test without and with reduced Input Filter
Around 30dB at 2.2MHz
Filter works!
~50dB at 4x Fsw

Not good enough in FM band

Top Layer - modification
Cout 1210 should be rotated by 90°

Also rotate the coil, to verify SoW!

XAL5030 1µH in both directions

Start of Winding at SW is 2-3dB better

Replace XAL5030-1µH by XAL4020-1µH

smaller coil is slightly better 1-2dB

Add 4.7 Ω Into BST Circuit
Effective above 90MHz

Average below 10dBµV now!

MPS Example #2

MPQ3414B 2.2MHz
MPQ4314 45/50V SYNC Boost
4A Sync. Buck
3.3V to 5V local Boost
for CAN (330mA; 1.66W)

13V to 3.3V System Buck, 4A

Large 15 Ω load
MPQ314B Boost Schematic
SW has 5Vpkpk with
fast dV/dt
Booster Output
has high di/dt
~ 1Apkpk
Circuit Modification
As large load resistor (Antenna!) is directly connected to Booster Output,
the first modification was an L,C output filter.

NO EFFECT !  Second modification:

Cout with Semi-Symmetric
placement

Small 330nH coil

was placed here.
1µF MLCC here
Original MPS EV Board and modified board
Initial MAPI4020 IRMS=8A and 8mΩ AT2010-R68 2x1.6x1mm
4x4x2mm size 41mΩ IRMS=3.5A; ISAT=4.9A

Cout with
Semi-Symmetric
placement

MAPI4020 680nH
MPL-AT2010-R68
CE Average Test with OEM Limit

MAPI4020-680nH
Chip-Coil 2010-680nH 9dB
10dB
7dB
7.5dB

76MHz to 108MHz; BW=120kHz

Power Inductors
Power Inductor share the same schematic symbol…

…but the construction is very different!

And also the EMC performance differs a lot!

MPS Case #3: More about Power Inductor and
EMC Tests with MPS 5A Buck and different 10µH Coil
EMC
Panasonic; TDK RLF and SLF; Toko D104/124; Epcos & Vishay IHLP4040

Result with RLF dot at Switch

Dot on coil indicates Start of Winding (SoW)

More about Power Inductors and EMC
Molded inductor WE LHMI 10mm x 10mm

Log. f scale
More about Power Inductors and EMC
TDK SLF12575 SoW at Switch

Reduction from
22dBµV/m to 6dBµV/m
More about Power Inductors and EMC
Epcos Coil with MnZn Core SoW at SW (Pink) and SoW at Vout (green)
The core of this MnZn coil is conductive, this might be the reason for better
Result with SoW at Vout.

6dB difference
More about Power Inductors and EMC
Is this an effective
intentional radiator?

What about now?

More about Power Inductors and EMC
4.5mm Height
End of Winding side
Contact metal with two
clip into airgap for
mechanical robustness

Even if S o W is at
Contact metal acts as antenna for SW, plate will cause
E-field from SW-node dV/dt trouble

Example Drawings from Murata, TDK and ABC

More about Power Inductors and EMC
More about Power Inductors and EMC

SW-node contact plate removed:

Winding soldered directly to PCB.
6 dB lower emissions at 1MHz
In Monopole Antenna Test

4.5mm height contact plate act as E-Field antenna for SW-node high
dV/dt.
Optimum coil should have SW-contact at the bottom.
RE Monopole Test 0.1MHz to 30MHz
Peak
Blue is original Coil
Pink is with removed metal at SW
RE Monopole Test 0.1MHz to 30MHz

Average
Green is original Coil
Pink is with removed metal at SW

6 dB lower at 1MHz !!!

MPS Example #5:
Isolated DC/DC; 5W, Only CE Average is shown

EMC Input filter has NO Effect!

Fsw needs 30dB damping
45dB
MPS Example #5:
• Additional tests with Snap-ON Ferrite: No Improvement!
• Test with off Board EMC Filter: No Improvement!
• Copper Foil around circuit: some improvement!
• Add “Y”-Capacitor 2.2nF between primary and secondary
GND shows strong improvement
• Additional circuit and Layout changes mainly on secondary
needed.
Conclusion:
• Check EMC filter structure for effectiveness
• Review PCB layout for high di/dt circuit nodes:
Loops have to be minimized!
• Review layout and components connected to high
dV/dt nodes:
High dV/dt area should be small and low profile!
• Try different inductors. Usually smaller and lower
profile types radiate less.
• If distance between DC/DC and cable/connector is to
small, use a local shield on top of DC/DC circuit.
 Q&A
On-Demand Webinars: www.monolithicpower.com/webinars

rev 20180504