Updated! Debugging EMI Problems Using a Digital Oscilloscope

06/2009 | Fundamentals of DSOs | 1 Nov 2010 | Scope Seminar – Signal Fidelity | 1

1

Debugging EMI Using a Digital Oscilloscope

Dave Rishavy Product Manager - Oscilloscopes

Debugging EMI Using a Digital Oscilloscope

2

l Background – radiated emissions l Basics of near field probing l Frequency domain analysis using an oscilloscope

l FFT computation l Dynamic range and sensitivity l Time gating l Frequency domain triggering

l EMI debugging process l Measurement example

Background – radiated emissions

Basic Principles: Radiated Emissions The following conditions must exist ı An Interference source A sufficiently high enough disturbance level in a frequency range that is relevant for RF emissions ı A Coupling mechanism transmits disturbance signals from the interference source to the emitting

element ı An Emitting element (Antenna) capable of radiating the energy produced by the source into the far field

4



5

Interference sources ı Fast switching signals within

digital circuits Single-ended (asymmetrical) data

signals Switched mode power supplies -

harmonics Differential data signals with

significant common mode component

6

ı High-order harmonics decrease at 20 to 40 db/decade

ı Structures on the PC board can begin to resonate at harmonic frequency

Inference Sources: Differential Mode RF Emissions

ı Emission results when signal and return are not routed together ı Near field probe can detect

this by positioning within the loop – position of probe is critical

7

General steps to help reduce Differential Mode RF emissions ı Reduction of the loop area (i.e. closer routing of the forward and

return conductors) ı Reduction of the current in the conductor loop (if possible without

impacting the circuit operation.) ı Reduce the rise/fall times for the transmitted data signals ı Use filtering to eliminate higher-frequency signal components

(limit the disturbance spectrum.)

8

Inference Sources: Common Mode RF Emissions ı Common problem in multi-

layer PC boards ı Caused by parasitic

inductance in return path or asymmetrical transmission ı External cable acts as an

antenna ı Rule of thumb for line length

as an antenna: λ/10 not critical λ/6 critical

9

Common Mode RF Emissions

10

Best Possible Differential mode transmission

Undesired parasitic capacitance in return path

Unbalanced parasitic terminating impedances

General steps to help reduce common-mode RF emissions ı Reduce the RFI current ICM by optimizing the layout, reducing the

ground plane impedances or rearranging components ı Reduce higher-frequency signal components through filtering or

by reducing the rise and fall times of digital signals ı Use shielding (lines, enclosures, etc.) ı Optimize the signal integrity to reduce unwanted overshoots

(ringing)

11



12

Coupling Mechanisms

ı Three coupling paths: Direct RF emissions from the source, e.g. from a trace or an individual

component RF emissions via connected power supply, data or signal lines Conducted emission via connected power supply, data or signal lines

ı Coupling Mechanisms Coupling via a common impedance Electric field coupling – parasitic capacitance between source and antenna Magnetic field coupling – parasitic inductance between source and antenna Electromagnetic coupling – far field coupling (greater than 1 wavelength)

13



14

Emitting Elements (Antennas)

ı Unintentional antennas in electronic equipment Connected lines (power supply, data/signal/control lines) Printed circuit board tracks and planes Internal cables between system components Components and heat sinks Slots and openings in enclosures

ıMain factor is the length of the antenna with respect to the

wavelength of the interference. Rule of thumb – antennas with length less than λ/10 are not critical

15

Basics of near field probing

17

Near Field Definition

ı Sources with Low Voltage, but high current predominantly generate magnetic fields (e.g. terminated high speed signals)

ı Sources with High Voltage, but low current predominantly generate electrical fields (e.g. unterminated signals)

Near field Transition Far field

E fie

ld

H fi

eld

Distance from DUT r

Wav

e im

peda

nce

r = 1.6m for f > 30 MHz

18

EMI Debugging Example Oscilloscope and accessories

R&S ® RTO R&S ® RTE

Nearfield probes R&S ® HZ-15

E- and H-field

30 MHz – 1 GHz Applicable from 100 kHz

Optional: R&S ® HZ-16

preamplifier

Current clamp R&S ® EZ-17

Magnetic and Electrical Near-Field Probes

19

ı Basically the probes are antennas that pickup the magnetic & electric field variation ı The output Depends on the position & orientation of the probe

20

H-Field Probe

ı Maximum response with probe parallel with current and closest to the current carrying conductor

ı Traces with relatively high current, terminated wires and cables

Current flow

H field

Vo

21

E-Field Probe

ı Maximum response with probe perpendicular with current and closest to the current carrying conductor

ı Traces with relatively high voltage: unterminated Cables, PCB traces to high impedance logic (tri-state outputs of logic IC’s)

Current flow

E field

Vo

Frequency domain analysis using an oscilloscope

Using an Oscilloscope for EMI Debugging

ıBenefits Wide instantaneous frequency coverage Overlapping FFT computation with color grading Gates FFT analysis for correlated time-frequency analysis Frequency masks for triggering on intermittent events Deep memory for capture of long signal sequences

ıLimitations Dynamic range No preselection No standard-compliant detectors (i.e CISPR)

23

Important Scope-Parameters for EMI Debugging

Parameter Description

Record length Ensure that you capture enough

Sample rate >2x max frequency, start with 2.5 GS/s for 0 – 1 GHz frequency range

Coupling 50 Ω for near-field probes (important for bandwidth)

Vertical sensitivity 1 – 5 mV/div is usually a good setting across full BW

Color table & persistence Easily detect and distinguish CW signals and burst

FFT – Span / RBW Easy to use “familiar” interface, Lively Update

Signal zoom & FFT gating

Easily isolate spurious spectral components in time domain

24

25

Frequency Domain Analysis FFT Basics

ı NFFT Number of consecutive samples (acquired in time domain), power of 2 (e.g. 1024)

ı ∆ fFFT Frequency resolution (RBW) ı tint integration time ı fs sample rate FFT

sFFT N

ftf ==∆int

1

Integration time tint NFFT samples input for FFT

FFT

Total bandwidth fs NFFT filter output of FFT

FFTf∆ts

FFT as Basis for EMI Debugging with Oscilloscopes Conventional FFT Implementation on a Scope

26

t

Time Domain

Record length T

Windowing FFT

Data acquisition

Zoom (f1…f2) f

Frequency Domain

∆t = 1/Fs

f

Display f2 f1

Fmax = Fs/2

S(f) S(f) x(t)

∆f = 1/T

f1 f2

FFT on the Rohde and Schwarz RTE and RTO Spectrum Analyzer Use Model

ı Use model: Frequency domain controls time domain Time domain parameters automatically changed as

necessary

ı Downconversion FFT (DDC) zooms into frequency range before FFT Largely reduced record length, much faster FFT

Time Domain

t

Record length T

Data acquisition

Fs=2Β

x(t)

Frequency Domain

f

Display f2 f1

Β=f2-f1

S(f)

∆f = 1/T

Windowing FFT

27

HW Zoom (DDC)

NCO

Decim- ation

LP

Zoom happens here – before the FFT

500 MHz center, 10 MHz span: Fs = 1 GS/s vs 20 MS/s

Measurement Consideration Gated FFT in the RTE and RTO Practical Time-Frequency Analysis

50% overlap (default setting) Gated FFT:

|---------------------------------- One complete Time-Domain capture ----------------------------|

The Key to unraveling the time domain

28

FFT Gating

29

Measurement Consideration: Sensitivity Ability to detect weak Signals EMI tends to be weak and near field probes have low gain, the oscilloscope needs to be able to detect small signals over its full bandwidth

Low Noise and High Sensitivity at Full Bandwidth

1mV/div

30

Signal to Noise and ENOB Higher ENOB => lower quantization error and higher SNR => Better

accuracy

l Thermal noise is proportion to BW. l An FFT bin is captures a narrow BW proportional to 1/

NFFT

l Noise is reduced in each bin by a factor of l The limit approaches sum of all non-random errors.

(Measurement induced errors are still present)

FFTf∆

∗

FFTN1log10 10

31

Measurement Consideration: Signal to Noise

>80 dB

32

ı Mask Tool ı 6 dB EMI filter? Not critical for precompliance, will change results only slightly.

Measurement Consideration: Limit Lines

33

Upper for limit line usage Mask definition in units of FFT

Upper region mask acting as limit line

Stop-on mask violation setting is very useful!

Measurement Consideration: Frequency Mask Triggering

34

EMI debugging process

The Problem: isolating sources of EMI

ı EMI compliance is tested in the RF far field Compliance is based on specific allowable power levels as a function of

frequency using a specific antenna, resolution bandwidth and distance from the DUT

No localization of specific emitters within the DUT ı What happens when compliance fails? Need to locate where the offending emitter is within the DUT Local probing in the near field (close to the DUT) can help physically locate

the problem Remediate using shielding or by reducing the EM radiation

ı How do we find the source? Frequency domain measurement Time/frequency domain measurement Localizing in space

36

E) Nearfield probe to localize the interferer source

D) Interferer current measurement to find out the coupling type

B) „Know your DUT“: List of potential interferer sources

37

A) Far-field measurement C) Reference measurement without DUT

Source Frequency

Clock frequency e.g. 25 MHz + Multiples

Ethernet PHY e.g. 125 MHz + Multiples

Voltage converter / power adapter

broadband

…

EMI compliant testing / Test lab EMI Debugging / R&D

F) Analysis of counter-measures

EMI Debugging Procedure Analysis steps

Observe the Spectrum While Scanning With a Near-Field Probe

38

I) General Approach ı Wide Span scan – fundamental of interfering signals are usually lower than 1GHz,

a span of <1GHz is sufficient as a start


39


a span of <1GHz is sufficient as a start ı Identify abnormal spurious or behavior and its location while moving the probe

around


40


a span of <1GHz is sufficient as a start ı Identify abnormal spike or behavior and its location while moving the probe around ı Narrow down to smaller span and RBW, change to smaller probe for better analysis

Measurement Example

Measurement Example – IP Phone

42

Example: IP-phone Situation ı IP-phone components Complex processor unit DDR2 memory Ethernet Layer 2 Switch 2 x Gigabit Ethernet PHYs

Several DC/DC Converters SPI-Interface to keyboard module Analog circuits (loudspeaker, Microphone)

ı Failed in EMI compliant test

43

Frequency (MHz)

Level (dBµV/m)

Limit (dBµV/m)

Margin (dB)

Height (cm)

Azimuth (deg)

Polarization

248.68 41.20 47.50 6.30 0.0 157.00 HOR 250.00 44.50 47.50 3.00 0.0 293.00 HOR

375.00 52.30 47.50 -4.80 0.0 359.00 HOR

Far Field Test Result

44

RFI Current Measurement 375 MHz Spur

Peak detect separates intermittent interference

45

ı Additionally detected emissions on following frequencies: CW: 375 MHz Broadband: 250 MHz

46

200 MHz 425 MHz

250 MHz 375 MHz

Example: IP-phone Current-probing on interface lines

Example: IP-phone Nearfield-probing for source localization Lokalisierung

47

DC/DC-converter no. 2 No significant emission

Nearfield spectrum in the area of the processor module; among others 375 MHz interferer

Correlating Time and Frequency Domains

48

Example: IP-phone Results ı Interferer signal detected on interface lines The interferer is probably transferred via common-mode coupling

ı Interferer sources localized DC/DC converter no. 1 Processor module respectively LAN PHY interfaces

ı Analysis of layout and implementation of counter measures.

49

Debugging EMI Using a Digital Oscilloscope Summary ı If we can measure something in the far field, it must have an electric and

magnetic near field source.

ı The conditions required for a radiated emission allow us insight to track down a source and mitigate potential interferers.

ı EMI Debugging with an Oscilloscope enables correlation of interfering signals with time domain while maintaining very fast and lively update rate.

ı The combination of synchronized time and frequency domain analysis with advanced triggers allows engineers to gain insight on EMI problems to isolate and converge the source and solution quickly.

ı Please see this shortcut to our application note for additional information: http://goo.gl/rvpfCK

50

Technology

Updated! Debugging EMI Problems Using a Digital Oscilloscope