Debugging EMI Using a Digita Oscioscope 06/2009 Nov 2010 Fundamentas Scope Seminar of DSOs Signa Fideity 1 1 1
Debugging EMI Using a Digita Oscioscope Background radiated emissions Basics of near fied probing EMI debugging process Frequency domain anaysis using an oscioscope FFT computation Dynamic range and sensitivity Time gating Frequency domain triggering Measurement exampe 06/2009 Nov 2010 Fundamentas Scope Seminar of DSOs Signa Fideity 2 2 2
The Probem: isoating sources of EMI EMI compiance is tested in the RF far fied Compiance is based on specific aowabe power eves as a function of frequency using a specific antenna, resoution bandwidth and distance from the DUT No ocaization of specific emitters within the DUT What happens when compiance fais? Need to ocate where the offending emitter is within the DUT Loca probing in the near fied (cose to the DUT) can hep physicay ocate the probem Remediate using shieding or by reducing the EM radiation How do we find the source? Frequency domain measurement Time/frequency domain measurement Locaizing in space 06/2009 Nov 2010 Fundamentas Scope Seminar of DSOs Signa Fideity 3 3 3
Basic Principes: Radiated Emissions The foowing conditions must exist ı An interference source exists that generates a sufficienty high disturbance eve in a frequency range that is reevant for RF emissions (e.g. fast switching edges) ı There is a couping mechanism that transmits the generated disturbance signas from the interference source to the emitting eement ı There is some emitting eement that is capabe of radiating the energy produced by the source into the far fied (e.g. a connected cabe, sots in the encosure or a printed circuit board that acts as an antenna) March 2013 EMI Debugging with the RTO 4
Interference sources ı Fast switching signas within digita circuits Singe-ended (asymmetrica) data signas Switched mode power suppies - harmonics Differentia data signas with significant common mode component ı Reativey ow votage swing of signas makes them sensitive to externa EMI (e.g. SMPS) high-order harmonics decrease at 20 to 40 db/decade Structures on the PC board can begin to resonate at harmonic frequency March 2013 EMI Debugging with the RTO 5
Differentia Mode RF Emissions ı Emission resuts when signa and return are not routed together ı Near fied probe can detect this by positioning within the oop position of probe is critica ı Mitigate by routing signa and ground coser, reducing signa current or decreasing sew rate March 2013 EMI Debugging with the RTO 6
Common Mode RF Emissions ı Common probem in mutiayer PC boards ı Caused by parasitic inductance in return path or asymmetrica transmission ı Externa cabe acts as an antenna ı Rue of thumb for ine ength as an antenna: λ/10 not critica λ/6 critica March 2013 EMI Debugging with the RTO 7
Genera steps to hep reduce common-mode RF emissions ı Reduce the RFI current ICM by optimizing the ayout, reducing the ground pane impedances or rearranging components ı Reduce higher-frequency signa components through fitering or by reducing the rise and fa times of digita signas ı Use shieding (ines, encosures, etc.) ı Optimize the signa integrity to reduce unwanted overshoots (ringing) March 2013 EMI Debugging with the RTO 8
Couping Mechanisms ı Three couping paths: Direct RF emissions from the source, e.g. from a trace or an individua component RF emissions via connected power suppy, data or signa ines Conducted emission via connected power suppy, data or signa ines ı Couping Mechanisms Couping via a common impedance Eectric fied couping parasitic capacitance between source and antenna Magnetic fied couping parasitic inductance between source and antenna Eectromagnetic couping far fied couping (greater than 1 waveength) March 2013 EMI Debugging with the RTO 9
Emitting Eements (Antennas) ımain types of unintentiona antennas in eectronic equipment ı Connected ines (power suppy, data/signa/contro ines) ı Printed circuit board tracks and panes ı Interna cabes between system components ı Components and heat sinks ı Sots and openings in encosures March 2013 EMI Debugging with the RTO 10
Magnetic and Eectrica Near-Fied Probes ı Basicay the probes are antennas that pickup the magnetic & eectric fied variation ı The output Depends on the position & orientation of the probe 11
H-Fied Probe H fied Vo Current fow Maximum response with probe parae with current and cosest to the current carrying conductor Traces with reativey high current, terminated wires and cabes 06/2009 Nov 2010 Fundamentas Scope Seminar of DSOs Signa Fideity 12 12 12
E-Fied Probe Vo E fied Current fow Maximum response with probe perpendicuar with current and cosest to the current carrying conductor Traces with reativey high votage: unterminated Cabes, PCB traces to high impedance ogic (tri-state outputs of ogic IC s) 06/2009 Nov 2010 Fundamentas Scope Seminar of DSOs Signa Fideity 13 13 13
Basic EMI Debug Process Noise from power suppy Unknown broadband noise peak CW Emission March 2013 EMI Debugging with the RTO 14
Using an Oscioscope for EMI Debugging ı Benefits ı Wide instantaneous frequency coverage ı Overapping FFT computation with coor grading ı Gated FFT anaysis for correated time-frequency anaysis ı Frequency masks for triggering on intermittent events ı Deep memory for capture of ong signa sequences ı Limitations ı Dynamic range ı No preseection ı No standard-compiant detectors (i.e CISPR) March 2013 EMI Debugging with the RTO 15
Important Scope-Parameters for EMI Debugging Parameter Record ength Sampe rate Couping Vertica sensitivity Coor tabe & persistence FFT Span / RBW Signa zoom & FFT gating Description Ensure that you capture enough >2x max frequency, start with 2.5 GS/s for 0 1 GHz frequency range 50 Ω for near-fied probes (important for bandwidth) 1 5 mv/div is usuay a good setting across fu BW Easiy detect and distinguish CW signas and burst Easy to use famiiar interface, Livey Update Easiy isoate spurious spectra components in time domain 06/2009 Nov 2010 Fundamentas Scope Seminar of DSOs Signa Fideity 16 16
Frequency Domain Anaysis FFT Basics FFT t s f FFT Integration time t int N FFT sampes input for FFT Tota bandwidth f s N FFT fiter output of FFT N FFT f FFT t int f s Number of consecutive sampes (acquired in time domain), power of 2 (e.g. 1024) Frequency resoution (RBW) integration time sampe rate f = 1 = f FFT t N int s FFT 17
FFT as Basis for EMI Debugging with Oscioscopes Conventiona FFT Impementation on a Scope Time Domain t = 1/F s F max = F s /2 Frequency Domain x(t) S(f) S(f) t Data acquisition Windowing FFT f 1 f 2 f Zoom (f 1 f 2 ) f 1 f 2 Dispay f Record ength T f = 1/T Disadvantages: Time domain settings define frequency domain Zoom in frequency domain does not give more detais Correated Time-Frequency Anaysis not possibe 21.03.2013 18
FFT on the RTO Spectrum Anayzer Use Mode ı Use mode: Frequency domain contros time domain Time domain parameters (record-ength / samping Time Domain x(t) Zoom happens here before the FFT! 500 MHz center, 10 MHz span: Fs = 1 GS/s vs 20 MS/s F s =2Β Data acquisition t HW Zoom (DDC) NCO LP rate) automaticay changed as necessary ı Downconversion FFT (DDC) zooms into frequency range before FFT Largey reduced record ength, much faster FFT 500 MHz center, 10 MHz span: 1 GHz vs 20 MHz Decimation samping frequency Windowing FFT S(f) Β=f 2 -f 1 f 1 f 2 Dispay f Frequency Domain Record ength T f = 1/T 21.03.2013 19
What if we combine time and frequency domain? Overap FFT comes into pay Record ength 10GS/s 18.96ns/div 5 us/div 500 1898 ksampes FFT 1 FFT 2 FFT N Max frame count imit N = N max Frame coverage up to here ~440 1 FFTs (persistance disabed) Advantages: Anayze time-dependent spectrum Conventiona (non-overapping) FFT oses information due to windowing overapping aows to capture everything Limit No of frames to ensure fast FFT processing Note: FFT processing starts from the eft! 21.03.2013 20
Overapping FFT Computation 21
Gated FFT in the RTO Practica Time-Frequency Anaysis Gated FFT: 50% overap (defaut setting) ---------------------------------- One compete Time-Domain capture ---------------------------- Key Feature for EMI Debug! 21.03.2013 22
FFT Gating 23
Signa to Noise and ENOB Higher ENOB => ower quantization error and higher SNR => Better accuracy Therma noise is proportion to BW. An FFT bin is captures a narrow BW proportiona to 1/ N FFT Noise is reduced in each bin by a factor of 10 og10 The imit approaches sum of a non-random errors. (Measurement induced errors are sti present) 1 N FFT f FFT 24
Signa to Noise >80 db 25
Noise Figure ı RTO Noise Figure Vertica Settings 1mV/div, 50 Ω Enabe FFT Use RMS detector Set center frequency Set RBW to e.g. 1 MHz Set unit to dbm NF = Output noise Input noise RMS Power dbm/rbw - (-174 dbm/hz + 10xog10 (RBW/Hz) ) = = -98 dbm 60 db + 174 dbm = 16 db 21.03.2013 Product Champ Training 26
Frequency Mask Triggering 27
Lab 1: Working with FFT s Objective: Learn how to make frequency domain measurements using an FFT on an oscioscope ı Any rea waveform can be produced by adding sine waves 28 28
Measurement Exampe IP Phone 29
Far Fied Measurement Faiure at 375 MHz 30
RFI Current Measurement 375 MHz Spur Peak detect separates intermittent interference 31
Identifying Couping Using Near Fied Probes 32
Correating Time and Frequency Domains 33
Debugging EMI Using a Digita Oscioscope Summary ı The modern oscioscope with hardware DDC and overapping FFT is capabe of far more than a traditiona oscioscope ı EMI Debugging with an Oscioscope enabes correation of interfering signas with time domain whie maintaining very fast and ivey update rate. ı The combination of synchronized time and frequency domain anaysis with advanced triggers aows engineers to gain insight on EMI probems to isoate and converge the soution quicky. ı Power Suppy design choices have a arge impact on EMI emissions, frequency and time techniques can hep unrave the mystery. 34