Top Banner
Practical EMC Issues in Large Practical EMC Issues in Large Experiments Experiments Georges Blanchot Georges Blanchot January 2009 January 2009 [email protected] [email protected] PH-ESE Seminar 1
52

Practical EMC Issues in Large Experiments Georges Blanchot January 2009 [email protected] PH-ESE Seminar 1.

Dec 27, 2015

Download

Documents

August Richard
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Practical EMC Issues in Large Practical EMC Issues in Large ExperimentsExperiments

Georges BlanchotGeorges BlanchotJanuary 2009January 2009

[email protected]@cern.ch

PH-ESE Seminar

11

Page 2: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

22

Introduction to EMCIntroduction to EMC

GroundingGrounding

Back end EMC requirements.Back end EMC requirements.

Front-end EMC requirements.Front-end EMC requirements.

How to deal with EMI couplings.How to deal with EMI couplings.

Selected good design practices.Selected good design practices.

Selected TopicsSelected Topics

Page 3: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

33

Why do we need to care about this thing !?Why do we need to care about this thing !?

““EMC = Electro Magnetic Compatibility”EMC = Electro Magnetic Compatibility”Means:Means:

Ability of a system to operate as required in presence of electro Ability of a system to operate as required in presence of electro magnetic disturbances.magnetic disturbances.

and also:and also:

Ability of a system to operate without compromising the normal Ability of a system to operate without compromising the normal operation of other systemsoperation of other systems

The compatibility between systems is achieved:The compatibility between systems is achieved: Defining methods and tools to quantify the disturbances.Defining methods and tools to quantify the disturbances. Defining limits that will ensure the good operation of all Defining limits that will ensure the good operation of all

systems.systems.

Introduction to EMCIntroduction to EMC

Page 4: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Interconnection of SystemsInterconnection of Systems

44

AC

DC

AC

DC

FE FE

FrameFrame

ELMB

• Disturbances from AC mains• Conducted noise: via cables• Radiated noise: via E/M fields• Couplings between cables• Grounding cables

Conducted EMI

ESD

Radiated Coupling

Near Field Coupling

Page 5: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

EMC in the Design ProcessEMC in the Design Process

55

Time

Cos

t

Development Production

Design corrections

Prototype corrections

Cabling and filtering

On site actions

Impr

ovem

ent

Simple Design Fixes

Produce new prototypes

Patches

Page 6: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Some reference documentsSome reference documents

66

Non exhaustive list:Non exhaustive list:1.1. Conducted emissions and radio disturbances:Conducted emissions and radio disturbances:

1.1. CISPR11: ISMCISPR11: ISM

2.2. CISPR22: ITECISPR22: ITE

2.2. AC Mains:AC Mains:1.1. Harmonic current: Harmonic current: EN-61000-3-2, EN-61000-3-4.EN-61000-3-2, EN-61000-3-4.

2.2. Immunity levels (industrial env.): Immunity levels (industrial env.): EN-61000-6-2EN-61000-6-2

3.3. Emission levels (industrial env.): Emission levels (industrial env.): EN-61000-6-4EN-61000-6-4

4.4. Fast transient/burst immunity: Fast transient/burst immunity: EN-61000-4-4EN-61000-4-4

5.5. Surge immunity : Surge immunity : EN-61000-4-5EN-61000-4-5

6.6. Voltage dips immunity: Voltage dips immunity: EN-61000-4-11EN-61000-4-11

3.3. LHC-EM-ES-0001 rev 2 (EDMS 113154): LHC-EM-ES-0001 rev 2 (EDMS 113154): Main Parameters of the LHC Main Parameters of the LHC 230/400V distribution system.230/400V distribution system.

Page 7: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

77

What to expect from a ground wire?What to expect from a ground wire?

A ground wire is an inductor of typically 1uH/m that carries about A ground wire is an inductor of typically 1uH/m that carries about 20 mA.20 mA.

ZL = 6 ZL = 6 ΩΩ/m at 1 MHz./m at 1 MHz.• It develops typically 125 mV/m at 1 MHz and 20 mA.It develops typically 125 mV/m at 1 MHz and 20 mA.

ZL = 125 ZL = 125 ΩΩ/m at 20 MHz with currents that /m at 20 MHz with currents that should be should be below 1 mA.below 1 mA.• It develops again 125 mV/m at 20 MHz and 1 mA.It develops again 125 mV/m at 20 MHz and 1 mA.• 20 MHz is typically the peak of susceptibility of our front-ends.20 MHz is typically the peak of susceptibility of our front-ends.

Given this:Given this: Long ground cables are unable to sink high frequency noise currents Long ground cables are unable to sink high frequency noise currents

without developing common mode voltages.without developing common mode voltages.• Grounding cables will often fail to reduce a front-end noise unless they are Grounding cables will often fail to reduce a front-end noise unless they are

kept short (<< 10 cm).kept short (<< 10 cm). Large structures, with interconnected conductors and frames, offer Large structures, with interconnected conductors and frames, offer

much lower impedances: they are better references for systems.much lower impedances: they are better references for systems.

About the GroundingAbout the Grounding

Page 8: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Development of CM through Development of CM through Grounding cableGrounding cable

88

Page 9: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

But grounding is still neededBut grounding is still needed

Electrical Safety:Electrical Safety: To carry fault or short circuit currents without To carry fault or short circuit currents without

developing hazardous voltages within systemsdeveloping hazardous voltages within systems

Protection against ESD:Protection against ESD: Carry the ESD charge away from system Carry the ESD charge away from system

(without developping hazardous voltages).(without developping hazardous voltages).

To enable shielding feature of frames, boxes, To enable shielding feature of frames, boxes, etc.etc.

99

Page 10: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

EMC by DesignEMC by Design

If grounding is not useful to insure the If grounding is not useful to insure the requested performance, what is the way to requested performance, what is the way to follow?follow?

1010

Good design practices!Good design practices!

Design taking into account all the disturbances Design taking into account all the disturbances present in real life.present in real life.

Minimize disturbances sent to others.Minimize disturbances sent to others.

Page 11: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

1111

The back end interfaces between the electrical AC The back end interfaces between the electrical AC network and the front-end systems.network and the front-end systems.

Usually commercial off-the-shelve equipment: EU EMC Usually commercial off-the-shelve equipment: EU EMC regulations apply!regulations apply!

Power SuppliesPower Supplies

Computing equipmentComputing equipment

Pumps, Motors, SwitchesPumps, Motors, Switches

InterfacesInterfaces

Back-End EMC RequirementsBack-End EMC Requirements

Page 12: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Tolerance to AC disturbancesTolerance to AC disturbances

1212

Microcuts, voltage dips, overvoltages, non sinusoidal waveforms: they occur many times per day, every day in all control rooms of all experiments.

The behaviour of any equipment is determined by immunity tests with specialized equipment, according to an international standard.

The pool is equipped with tools to characterize system immunity according to IEC standards.

Event #2 at 04-03-2007 12:31:41.450Pre-trigger

Event Details/Waveforms

12:31:41.4504-03-2007

Sunday

12:31:41.46 12:31:41.47 12:31:41.48 12:31:41.49

-300

-200

-100

0

100

200

300

Vo

lts

A V

DEMO

Dran-View 6.4.01 Exclusive Demo

Schaffner Modula Test System

Events recorded in ATLAS Control

Rooms

Page 13: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Effect of AC disturbancesEffect of AC disturbances

1313

A back end LVPS in a LHC experiment: immunity test in lab.

Voltage drop of 5% induces output overvoltage and transients on DC line.

Voltage drop

Output transient and overvoltage

Page 14: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Emission of Harmonic CurrentEmission of Harmonic Current

1414

Transformer Losses.Transformer Losses. Heat up the transformers and Heat up the transformers and

reduce their lifetime.reduce their lifetime. To cope with harmonic current, To cope with harmonic current,

transformers must be derated transformers must be derated (=cost).(=cost).

Increased Neutral Current.Increased Neutral Current. Voltage Distortion. Voltage Distortion.

Distortion due to voltage drops Distortion due to voltage drops caused by harmonic currents.caused by harmonic currents.

AC mains received by other AC mains received by other equipment is distorted and can equipment is distorted and can cause malfunctions.cause malfunctions.

Low frequency Interferences.Low frequency Interferences. Harmonic current cause induction Harmonic current cause induction

noise.noise. Noise below few kHz is hard to Noise below few kHz is hard to

filter, need to use regulators.filter, need to use regulators. Degraded Power FactorDegraded Power Factor

2.IRPLoss

The harmonics in the load current drawn from the mains is a source of disturbance for the electrical distribution network and for other users

Harmonic Current

Source

Distorded line Voltage

Harmonic Current

Victim

Injected Harmonic

Current

fP

fP

Hysteresis

tsEddyCurren

2Extra Losses

Page 15: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Harmonic CurrentsHarmonic Currents

1515

In practice, the neutral RMS current can double at In practice, the neutral RMS current can double at most the phase RMS current.most the phase RMS current.

Limit set by the facility.Limit set by the facility.

Increased Neutral Current:

1

23.3

j jN IIRMS

Remedies are either passive filters for simple cases, or active power factor correction (APFC) for complex harmonic current sources such as power converters.

User gets charged for bad power factor.No explicit limit.

Degraded Power Factor:

.

1001..

2

%11

THD

IV

PF

RMSRMS

Tot

Total Harmonic Distortion:

1001

2

2

%

I

ITHD h h

Each harmonic current is a source of extra lossesLimit set by the facility.Limits specified in EMC standards.

Limits on each harmonic content is defined in IEC-61000-3-2, it is achievable using PFC circuits or filters. Line diode bridge rectifiers cannot be used without PFC.

Page 16: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Inrush CurrentsInrush Currents

1616

Inrush currents in rack equipment can pose difficulties to the electrical network (tripping computing farms): PFC and passive dampers help to smooth down the startup.Inrush, harmonics, voltage distortions can all be measured with Power Xplorer available at the Pool.

DAQ Rack in ATLAS

Page 17: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Noise emission/immunityNoise emission/immunity

Noise emissionNoise emission Limits on conducted noise (common mode) on all Limits on conducted noise (common mode) on all

IO and power ports.IO and power ports. Limits on radiated noise (EM fields): cannot be Limits on radiated noise (EM fields): cannot be

tested at CERN (only qualitative near field).tested at CERN (only qualitative near field). ImmunityImmunity

Incoming noise degrades the performance:Incoming noise degrades the performance:• This can be tested and quantified at CERN.This can be tested and quantified at CERN.

All these are common with front-end All these are common with front-end requirements and will be described on next requirements and will be described on next slides.slides.

1717

Page 18: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

1818

Front-end = Sensing DeviceFront-end = Sensing Device

Always custom made circuits, EMC performance will depend Always custom made circuits, EMC performance will depend of the designer experience and of its awareness of the of the designer experience and of its awareness of the

EMC problemsEMC problems Sensitive to conducted noise from its back end Sensitive to conducted noise from its back end

Noise current that enters through ports and cables.Noise current that enters through ports and cables. Is translated in larger RMS in data.Is translated in larger RMS in data. Setup dependent.Setup dependent.

Sensitive to couplings from neighbors (systems and Sensitive to couplings from neighbors (systems and cables)cables)

Near field: electric, magnetic.Near field: electric, magnetic. Source of conducted and radiated noise as wellSource of conducted and radiated noise as well

Can compromise neighboring systems.Can compromise neighboring systems.

Front-End EMC RequirementsFront-End EMC Requirements

Page 19: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Common and Differential ModesCommon and Differential Modes

1919

Ground Plane

LoadDC / DC

IDM

IDM

Ground Plane

Load

ICM

ICMDC / DC

Few pF Few pF

Differential Mode:The functional current carried by a wire and its intentional return: the electromagnetic

field is mostly contained within the

cable

Common Mode:A non functional

current carried by a set of wires and some

surrounding conductive structures or elements:

the electromagnetic field is mostly

contained between the cable and the

conductive structures

Small Loop

Big Loop

Page 20: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

EMI SourcesEMI Sources

2020

Radiated Noise from system is small because at f=40MHz λ=7.5m that is easily shielded by the system faraday cages and enclosures.

Radiated Noise from cables comes mainly from CM noise (far field* from electrically short cables).

EC

S

L

d

ED

ID

ID

S

L

d

IC

IC

Differential Mode: the far fields are opposed and cancel each other

Common Mode: the far fields add up.

The contribution of CM current to EMI is typically more than 3 orders of magnitude stronger than the contribution of the same DM current. * Far field region starts at a distance d = λ/6, i.e. 1 m at 40

MHz.

Need to control the sources of CM noise:

Switched power circuits and converters.

Digital circuitry.

CM coupling across cables.

Page 21: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

LVPS Conducted NoiseLVPS Conducted Noise

2121

DC-DCConverter

ISL

Ground Plane

+-

Shielded Cable

+-

Shielded Cable

+

-L1

L2

Power Supply

+-

L2

LISN

L1

Current Probe

Current Probe

Ground plane

- Reference return path for CM currents

- Cables lay on the plane.

- Source, load and filters earthed to the plane.

LISN: Line Impedance Stabilization Network

- Calibrated, standardized filter.

- Filters noise from bulk LVPS.

- Provides reference impedance seen by the converter towards its source of power, over the whole frequency range of interest.

ISL: Impedance Stabilized Load

- Calibrated load:

- DM (load).

- CM (output to earth plane).

LISN

ISL

Splitter

DC/DC

A dominant source for FE noise is the power supply. It can be characterized on a reference setup that is independent of the front-end and of the primary source.

Page 22: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Line Impedance Stabilization Network Line Impedance Stabilization Network (LISN)(LISN)

2222

10 kHz 100 kHz 1 MHz 10 MHz 100 MHz

dB W

LISN ESH3-Z6 Transfer Impedance

-5

10

30

20

0

FREQ

40

Calibration

Probe

L1,L2

P/L

T/E

C2

L1

R1

SL1C1

R2

LISN

- Provides a standardized voltage measurement of the symmetric and asymmetric noise between line and earth (ICM + IDM).

- The impedance is calibrated from 100 kHz to 100 MHz.

- Above 1 MHz: Z=50 ohms.

- To measure accurately the CM current only, a calibrated current probe is used. Alternatively, a CM/DM splitter can be used (less accurate).

TDMCMLISN ZIIV )(

V(LISN)ICM+DM

L1

L2

2N:N

R

R

MD

MC

R

R

Current probe CM/DM splitter

Page 23: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Instruments and Tools for conducted Instruments and Tools for conducted EMI measurementsEMI measurements

2323

Conducted EMI.Conducted EMI.

EMI Receiver (9kHz to 3GHz)

Current Probes and Injectors

Receiver

DM CM

Receiver

DM CM

Receiver

DM

CM

CMII 2 CMDM III DMII 2)(a )(c)(b

150 kHz 30 MHz

dBµV dBµV

1 PKCLRWR

SGL

RBW 9 kHzMT 100 msPREAMP OFFAtt 10 dB

PRN

1 MHz 10 MHz

0

10

20

30

40

50

60

70

80

90

100

1

Marker 1 [T1 ] 55.66 dBµV 4.642000000 MHz

Date: 11.AUG.2004 11:25:10

9kHz --- 100 MHz

dB

µA

Page 24: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

CM Noise MeasurementsCM Noise Measurements

2424

Example: output common mode noise for 2 custom prototypes using identical discrete components (commercial driver + switches). Only the design of the PCB and the passive components differ

Frequency (Hz) Frequency (Hz)

Noi

se (

dBuA

)

“Reference” level based on Class A limit from CISPR11 converted to current on a given impedance (Careful: this is NOT a real limit)

Page 25: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Importance to limit the noise currents in Importance to limit the noise currents in experimentsexperiments

CM currents (on all cables) must be contained under a limit that CM currents (on all cables) must be contained under a limit that is reasonably set under 100uA in the sensitivity band of our front-is reasonably set under 100uA in the sensitivity band of our front-ends.ends.

CM currents above few mA will definitively collapse front-ends.CM currents above few mA will definitively collapse front-ends.

Patches are always difficult to put in place.Patches are always difficult to put in place.

The sensitivity of FE is determined with susceptibility tests.The sensitivity of FE is determined with susceptibility tests.

2525

Page 26: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Typical LimitsTypical Limits

2626

Conducted emissions limits in LHC experimentsConducted emissions limits in LHC experiments- - Extension of CISPR11 Class A Group 1:Extension of CISPR11 Class A Group 1:

- to all power links in experiment zones- to all power links in experiment zones- to high voltage and data links- to high voltage and data links

- Conversion of limits in terms of dB- Conversion of limits in terms of dBμμA that can be easily measured A that can be easily measured on site:on site:

- Extension of frequency:- Extension of frequency:- up to - up to 100 MHz to cover the LHC clock and its first harmonic.100 MHz to cover the LHC clock and its first harmonic.

Frequency Band

[MHz]

Class A Group 1

dBμV into 50Ω/50μH dBμA

QPK AVG QPK AVG

0,15 – 0,50 79 66 45 32

0,50 - 5 73 60 39 26

5 - 30 73 60 39 26

30 -100 - - 39 26

= 89 uA

Page 27: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Measuring SusceptibilityMeasuring Susceptibility

2727

I/O PortFront-End

Electronics

RF Generator

RF Amplifier

Spectrum Analyzer

CM DM

The injected signal is provided by a RF generator that delivers a low distortion single The injected signal is provided by a RF generator that delivers a low distortion single frequency (swept).frequency (swept).

CM or DM currents are injected inductively on the tested port using an injection current CM or DM currents are injected inductively on the tested port using an injection current probe.probe.

The injected current must be monitored with a calibrated probe and an accurate The injected current must be monitored with a calibrated probe and an accurate spectrum analyzer.spectrum analyzer.

The injection of current is not very effective at low frequencies: amplifiers are often required.

The injected currents usually vary from few uA to few mA.

The frequency of interest ranges from 100 kHz up to 100 MHz: beyond this, radiated couplings take place.

Page 28: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Measuring SusceptibilityMeasuring Susceptibility

2828

ProbePower Link

Data Links

AC

Analyzer

Ground plane

CM Noise

LVPSSystemUnderTest

DAQ/DCS

RF GeneratorRF Amplifier

EMI Couplings will degrade the noise performance of the system:The relationship between EMI coupling and resukting noise is determined with immunity tests.

Online/Offline Data

Porcessing

Parameters can be:• Hit rate (Hz)• ENC (pC)• Vrms, Irms• Any combination of those, chi2, linearity coefficients, etc…

The system must define the

parameter that sets the quality of

its data.

SystemI (dBuA)f (Hz) <Xn>

Page 29: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Susceptibility exampleSusceptibility example

2929

Two identical detector systems:Detector A is powered through a 15 meter shielded cable.Detector B is powered through 15 meter unshielded cable.The power supply is common.The DAQ is common.The noise evaluation parameter is set as hit rate recorded by counters.

Low Voltage Power Supply

5VDC11A

High Voltage Power Supply

3kV @ μA

LV Box

LV Box

HV Split

Filter

Filter

Detector A

Detector B

PE Hit Rate Counter

SHIELDED

UNSHIELDED

ATLAS MDT Prototype (2004)

Page 30: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Susceptibility CurveSusceptibility Curve

3030

Unshielded Detector B Hit Rate at 5 MHz

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000 6000 7000

5 MHz Common Mode Current [uA]Hi

t Rat

e In

crea

se

Shielded Detector A Hit Rate at 5 MHz

0

10

20

30

40

50

60

0 1000 2000 3000 4000 5000 6000 7000

5 MHz Common Mode Current [uA]

Hit R

ate

Incr

ease

GOODThe shielding improves the EMI immunity

of the front end electronics

BADThe use of an

unshielded power cable has a direct impact on the EMI

immunity of the front end

electronics

Threshold: 5mA doubles the hit rate.One channel is particularly sensitive to CM.Few mA are sufficient to screw up the best low noise front end electronics by more than one order of magnitude; it is also very easy to pick up few mA of noise.

Threshold: 2mA doubles the hit rate.

Page 31: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Another example: ATLAS ALFAAnother example: ATLAS ALFA

3131

64PMT

MAROC2 (ASIC)

Preamp Discri

ALFA-R (FPGA)

64

Front-End Configuration

Readout System

Stream Register3

LVL1A

CLK40

SPI

DATA BUS

Motherboard

5V

12V

To DAQ

Common mode currents are injected in the Common mode currents are injected in the 12V input port first, after in the 5V port, with 12V input port first, after in the 5V port, with magnitudes up to 10 mA in the ferquency magnitudes up to 10 mA in the ferquency range between 150 kHz and 30 MHz. range between 150 kHz and 30 MHz.

The 12V powers exclusively the The 12V powers exclusively the motherboard, which is fully digital. It was motherboard, which is fully digital. It was found to be insensitive to the injected current. found to be insensitive to the injected current.

The 5V powers the front-end chips (MAROC, The 5V powers the front-end chips (MAROC, FPGA), with analog circuitry. It was found to FPGA), with analog circuitry. It was found to be sensitive to the injected current.be sensitive to the injected current.

Page 32: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

ALFA Conducted SusceptibilityALFA Conducted Susceptibility

3232

Refer to: “MAROC: Multi Anode Readout Chip”, S. Blin, TWEPP 2007.

The front-end chip is configured at nominal gains and the susceptibility is measured for different thresholds (DAC) in the transition region of the S curves.

3d plot

The sensitive DAC range at nominal gain

is found to be between 88 and 94.

The noise hit rate is a function of current, frequency and DAC

At a given threshold, the maximum noise current permitted is established for every critical frequency

Injected current: 10 mA on 5V input.

Hit rate: up to 40% for all pixels.

Frequency peaks: 13 Mhz, 22-30 MHz

Page 33: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Susceptibility to radiated fieldsSusceptibility to radiated fields

3333

Qualitative tests to evaluate susceptibility to E Qualitative tests to evaluate susceptibility to E field (capacitive coupling) and H field field (capacitive coupling) and H field (inductive coupling) are easy to perform.(inductive coupling) are easy to perform.

Accurate and quantitative measurements are Accurate and quantitative measurements are much more difficult and require more much more difficult and require more specialized infrastructure.specialized infrastructure.

Page 34: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Example: TOTEM near field susceptibilityExample: TOTEM near field susceptibility

3434

The susceptibility of systems to the magnetic field emitted by inductors of power converters is a major concern. System tests were carried out on TOTEM, with a coil driven by an amplified RF source. The coil is accurately positioned above the detector, the bondings and the ASICs and the induced noise is analyzed from the test DAQ.

538 nH air core, 1A.

Distance to center (mm)

Field (uT)

4 1560

9 88

14 19

19 6.4

24 2.9

100 uT at 5 mm from edge

Coil edge

Page 35: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

H Field Susceptibility ExampleH Field Susceptibility Example

3535

The TOTEM system showed noise sensitivity increasing with the frequency:

• System not able anymore to extract correct S curves parameters.

• The test was made at constant dB/dt: (I*f = constant).

Inductor focused obliquely on the bonding

Corrupted S curve beyond this point

Page 36: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

E Field Susceptibility ExampleE Field Susceptibility Example

3636

Large plate cap. coupling Small plate cap. coupling. Spot cap. Coupling (wire end).

The system showed also sensitivity to capacitive coupling (electric field):

• 3.4V/1MHz: signal equivalent to the one present on the inductor wires.

• Exposed areas develop large noise.

Page 37: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Want better tools?Want better tools?

3737

ETS-Lindgren Near Field Probing Kit

Provides calibrated measurement of emitted fields, but can be used as EMI sources as well!

Agilent Near Field Probing Kit

Have a different sensitive geometry, to be used for calibrated measurements

Nexus B field Scan

Locate EMI spots, evaluate shields

Page 38: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

3838

How to mitigate the coupling effects?How to mitigate the coupling effects?

A front-end can be designed with some degree of A front-end can be designed with some degree of immunity against conducted or radiated noise.immunity against conducted or radiated noise.

It can be connected to a compatible power supply It can be connected to a compatible power supply (emits CM currents lower than the acceptable limit of (emits CM currents lower than the acceptable limit of the front-end system).the front-end system).

Still, unexpected couplings often come from unknown Still, unexpected couplings often come from unknown neighbors, in particular within the cable trays.neighbors, in particular within the cable trays. The only protection is to SHIELD.The only protection is to SHIELD.

Dealing with EMI CouplingsDealing with EMI Couplings

Page 39: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

A wonderful world!A wonderful world!

3939

Page 40: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Shielding against E FieldShielding against E Field

4040

Equivalent circuit: Shielding against external E field

G R

IGS

• A current is coupled into the shield through stray capacitances– Capacitance to shield– Capacitance to ground

• Performant shield is only achieved if current can flow easily– E field shielding is achieved by grounding one side only.– Connectors must be metallic, all around bonded to the shield,

without straps, capacitors or resistors.– Enclosure must be well bonded to the equipotential network.

vGS

vGE

Page 41: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

About grounding both endsAbout grounding both ends

4141

r, l

AC/DC Conv.Load

CouplingICM

EquipotentialEquipotential bound

Shield connected on both ends

Grounding of the shield: on both ends

• Reduced common mode loop.– The CM current is forced to return in the shield (mutual inductance)– CM emissions are reduced radically.

• Reduced electric field– Contained within the shield

• Magnetic field shielding– The inductively coupled current is cancelled in the live conductor by mutual inductance.

•Electric field shielding is improved- Two paths instead of one!

Page 42: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Shield is return path for CM CurrentShield is return path for CM Current

4242

Equivalent circuit: mutual inductance forces CM current to return into the shield

101

102

103

104

105

106

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Shield return path for CM current

[Hz]

I shie

ld/I cm

MjRI

I

ShieldCM

Shield

1

1

ICM

21 LLM

Ishield

Page 43: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Shielding for B fieldShielding for B field

4343

GGRILj SHI

Equivalent circuit: shielding against external B field.

101

102

103

104

105

106

10-5

10-4

10-3

10-2

10-1

100

101

Inductive Shielding

[Hz]

IW/I

G

One end groundedBoth ends grounded GRS ILj

GGS ILjWI

GGRILj

SF =

1

Page 44: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Shielding EffectivenessShielding Effectiveness

4444

The effectiness to shield depends of the shield construction: RSH, LSH, coverage.

It can be measured on a section of 1 meter of cable injecting a reference shield current as shown.

Optical coverages greater than 85% are always recommended.

ZT = transfer impedance <-> shielding effectiveness

IshVl

Figure : Current probe test setup: grounded connectors (C), injection probe (1), current probe (2).

50Ω

C

DSO1

RF Gen.2

C

Vsh

Ish

Vl sh

lT I

VZ

Figure : Transfer impedance of a shielded cable.

Page 45: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Transfer ImpedanceTransfer Impedance

4545

Example of 2 shielde dpower cables from CERN stores.

The shield effectiveness starts degrading at 100 kHz!

The corner frequency depends in particular of the shield inductnace and of its connection to the plane: pigtails add inductance and therefore degrade the shielding effectiveness.

Page 46: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

When susceptibility is When susceptibility is discovered latediscovered late

4646

Unnamed detector suffered from LVPS noise during commissioning.

BADNoise shows up once using the production power

supplies

GOODHowever… it looked so good in the lab!

Page 47: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Adding ChokesAdding Chokes

4747

106

107

108

-40

-20

0

20

40

60

80

Hz

dB A

No ChokeWith ChokeLimit

- Bulky common mode chokes provide the best attenuation of switch frequency noise.

- Very hard to add on site: it is much better to insert them at the beginning of the design.

- Limitation: magnetic field of experiments.

Before

After

Before After

Page 48: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Adding ferritesAdding ferrites

4848

104

105

106

107

108

-40

-20

0

20

40

60

80

Hz

dB A

One ferriteTwo ferritesLimit

- Last resource!

- Compromised in presence of magnetic field.

- Works better at few tens of MHz in the sensitivity range of typical front-end amplifiers.

Page 49: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Other filtersOther filters

4949

Power plane (VO2)Ground Plane (VO1)Earth bonding (GND)

Earth bonding (CM)

Whatever filter is used, make sure that the CM current never enters in

your system

Page 50: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

5050

How can I apply all this to my designs?How can I apply all this to my designs?

Recommended PracticesRecommended Practices

1. Always prefer back end systems that are compliant with EU EMC regulations: LV directive, EMC directive, CE mark.

2. Look for PFC in back end equipment.

3. Never leave floating any conductive frame, chassis, structure, box- Interconnect them and tie them to earth with the shortest connection- This is necessary for safety, also it radically improves the protection

against EMI couplings.- Earth loops are never a hazard: they are not part of active circuits.

4. Use shielded data and power cables whenever possible.- It is almost always better to earth the shield to the frames on both

ends.- Shields are tied to earth via frames, not to active return paths.

Page 51: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

5151

Recommended PracticesRecommended Practices

5. Fit the power lines with common mode filters in the back end side.

6. Reference your front-end system to the closest ground connection.

7. Design with an eye on EMC aspects:- Measure the noise emissions of prototypes.

- This leaves a chance to correct easily before production/installation.- Measure the susceptibility of prototypes.

- It brings invaluable information that is then used to specify power supplies and cables.

8. Design for the real world!

Page 52: Practical EMC Issues in Large Experiments Georges Blanchot January 2009 Georges.Blanchot@cern.ch PH-ESE Seminar 1.

Thanks for your attention!Thanks for your attention!

5252