Comparison of Time Domain Scans and Stepped Scans in EMI Test Receivers
Matthias KellerRohde & Schwarz GmbH & Co.KGMunich
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 2
Comparison of Time Domain Scans and Stepped Scans i n EMI Test Receivers
l Conventional EMI measurementsl FFT-based Measuring Receiverl Level Accuracyl Consideration of measurement timesl New insights for EMI diagnosisl Conclusion
Contents
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 3
Conventional EMI measurements
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 4
l Type of disturbance signal is often unknown(narrowband, broadband)
l Timing of disturbance signal must be considered(CW, intermittent)
l Measurements to commercial standards require quasipeak weighting (and sometimes CISPR-average )
l Measurement procedures are very time-consuming(e.g. radiated emission tests with mast and turntab le)
� Use of time saving methods and procedures� Quasipeak weighting only when necessary
Challenges of EMI testing
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 5
Conventional solution for test time reduction(commercial standards)
Preview: Signal detection in the frequency
range with peak-(average-)weighting
Data reduction: (acceptance analysis)
���� frequency list
Final measurement: QP-(CISPR-average-)
weighting only of frequencies in list
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 6
FFT-based Measuring Receiver
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 7
Receiver with broadband FFT applied at the baseband
Principle of FFT-based measuring receiver (1)
Display
Receiverpreselectionand mixer
widebandIF Filter
IF=90 MHz
ADC FFT Detectors ScreenDisplay
Superheterodyne receiver with broadband FFT of the IF
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 8
Bw signal 1 GHzDynamic range > 90 dB ADC
fs > 2 GHz
ADC word length ≥ 24 Bit
ADC word length ≥ 16 Bit
fs > 60 MHz ����
Baseband sampling
IF sampling
Principle of FFT-based measuring receiver (2)
requires split of the frequency range into subranges
Mixer
fIF
IF filter
fLO
Localoscillator
Bw signal 26 GHzDynamic range > 90 dB ADC
e.g. 30 MHz
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 9
Principle of FFT-based measuring receiver (3)
�
�
�
Frequency domain
Split the measured frequency range into
consecutive frequency intervals
Time-domain
Sample the frequency interval with high
sampling rate
F(s) f(t)
Fast-Fourier transformation
Transform the signals from time domain to
frequency domain
Frequency domain
Merge the spectra of all frequency blocks
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 10
Modern Preselector Design (1)
with preselection
withoutpreselection
ı Frontend is overloaded without preselectorı No reliable overload indication
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 11
Modern Preselector Design (2)
E1pk
t
E(t)
t
E(t)
E2pk
t
E(t)
E3pk
voltage ratio E1pk / E2pk = BW pulse / BW preselector
Reduction of Pulse Voltage to prevent overload of Mixer, ADC, IF amps
RF Input Mixer
fIF
RBW
Localoscillator
Preselector Bandwidth
150 kHz to 100 MHzDisplay
e.g. 120 kHz
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 12
Modern Preselector Design (3)
l Overall design is important for dynamic range
l Preselector is only one part
l ADC length is important
l More bits allow for more bandwidth
l Each extra bit doubles the voltage ratio
l Double voltage range allows double bandwidth
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 13
l Disturbance signals are not periodically(mixture of periodic signals , intermittent signals and noise )
l The observation time T meas cannot be set as multiple integer of the period T 0 of the disturbance signal
l In this case signal distortions appearl Leakage effect: signal spectrum becomes wider, additional
spectral components appear
Leakage effect (1)
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 14
l A limited observation time partly distributes the spectral energy of the signal over a wide frequency range� The main lobe of the signal spectrum
becomes wider� Additional spectral components
appear (side lobes)� The main lobe amplitude is reduced
l CISPR 16 requires asuppression of the additionalspectral components (side lobes)of at least 40 dB
Leakage effect (2)
Magnitude of the transfer function of a rectangular window function
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 15
Sampling
Time f
A "Scan"
Sampling
Time
Time
f
Afs/2fbin
Leakage effect (3)
Main lobe ⇔⇔⇔⇔ Measurement BandwidthRemoval of side lobes!
Main lobe must match filter characteristic requirem ents(CISPR, MIL-STD)
Time
?
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 16
Leakage effect (4)Windowing
Flattop
HanningRectangular
Gaussian
A
1
Gaussian40 dB
The width of the main lobe and the amplitude of the side lobes depend on the type of the window function
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 17
CISPR 16-1-1 tolerance mask forIF selectivity
Gaussian-type window (1)
FFT-based scan:
Frequencybin
Stepped frequency
scan:IF
bandwidth
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 18
l Gaussian window function���� Effective reduction of the leakage effect
l The Fourier transformation of a Gaussian function in the time-domain results in a Gaussian function in the frequency domain
l The characteristic of the measurement bandwidth (select ivity)defined by CISPR 16-1-1 or MIL-STD-461 matches with a Gaussian filter function
l Measurement bandwidth requirements are fulfilled
Gaussian-type window (2)
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 19
t
A
t
A
Rectangular window
Gaussian window
t
AGuassian window with overlapping
Continuous overlapping:Short-time FFT (STFFT)
T1 T2 T3 T4
FFT of intermittent signals (1)
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 20
FFT of intermittent signals (2)
Short-time FFT (STFFT) of a pulse spectrum using Ga ussian-type windowing with different overlapping factors
25%
75% 90%Reference:
0%
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 21
Detector weighting functions
FFT n f
f
f
f
Bins
FFT n+1
FFT n+2
FFT n+3
FFT n+4
FFT n+5
FFT n+6
FFT n+7
f
f
f
f
Continuous overlapping of windows in time domaint
A
t
A
Continuous overlapping of w
indows in tim
e domain
parallel for all 'bins'e.g. 16384 x Quasipeak
Result per 'bin' after dwell time
Weighting per 'bin' (e.g. QP)
tD
well tim
e
Dw
ell time
Video Voltage per 'bin'
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 22
Technical realization
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 23
Leakage effectSide lobe suppression of R&S ESR (CISPR bandwidth 9 kHz)
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 24
Measurement Bandwidth 9 kHz
Selectivity for CISPR band B (measurement bandwidth 9 kHz)
Frequency in MHz
Am
plitu
de in
dB
µV
Tolerance mask
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 25
Level Accuracy
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 26
Level Accuracy R&S ESR (1)Time Domain Scan versus Stepped Scan
Pulse with 4.00 µs Pulse period, 0.10 µs Pulse width Detector Quasi Peak
Yellow Trace: Time-Domain-ScanGreen Trace: Stepped Scan
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 27
Level Accuracy R&S ESR (2)Time Domain Scan versus Stepped Scan
y-axis: level difference, x-axis: frequency
-0,50
-0,40
-0,30
-0,20
-0,10
0,00
0,10
0,20
0,30
0,40
0,50249000
1000500
1749750
2499000
3250500
3999750
4749000
5500500
6249750
6999000
7750500
8499750
9249000
10749750
11499000
12250500
12999750
13749000
14500500
15249750
15999000
16750500
17499750
18249000
19000500
21750000
22499250
23250750
24000000
24749250
25500750
26250000
26999250
27750750
TD-Scan vs. Stepped ScandB
Hz
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 28
Consideration of measurement times
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 29
l Annex B of CISPR 16-2-1 to 16-2-3 specifies minimum sweep times or the fastest scan rates
l Quasi-peak detection results in unacceptably long m easurement times
Frequency range Peak detector Quasipeak detector
A 9 to 150 kHz 100 ms/kHz: 14.1 s
20 s/kHz: 2 820 s = 47 min
B 0,15 to 30 MHz 100 ms/kHz: 2.985 s
200 s/MHz: 5 970 s = 1 h 39 min
C/D 30 to 1 000 MHz 1 ms/MHz: 0.97 s
20 s/MHz: 19 400 s = 5h 23 min
Minimum scan times for commercial standards (CISPR)
… If the level of disturbance is not steady, the reading on the measuring receiver is observed for at least 15 s for each measurement.
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 30
… For equipment that operates such that potential emissions are produced at only infrequent intervals, times for frequency scanning shall be increased as necessary to capture any emissions.
Minimum scan and sweep times for military standards (MIL-STD-461E/F)Bandwidth and measurement time requirements
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 31
Comparison of measurement times (CISPR)R&S ESR
Measurement TimesFrequency Range Detector,
Dwell Time,Measurement BW(Number of Points)
Stepped Scan Time Domain Scan
CISPR Band B150 kHz – 30 MHz
Pk, 100 ms, 9 kHz (13.267)
22 min 117 ms
CISPR Band B150 kHz – 30 MHz
QP, 1 s, 9 kHz (13.267)
3.6 h 2 s *
CISPR Band C/D30 MHz – 1 GHz
Pk, 10 ms, 120 kHz (32.334)
5 min, 23 s 630 ms
CISPR Band C/D30 MHz – 1 GHz
Pk, 10 ms, 9 kHz (431.000)
71 min, 50 s 850 ms
CISPR Band C/D30 MHz - 1 GHz
QP, 1 s, 120 kHz (32.334)
~ 9 h 80 s *
* incl. 1 s settling time per FFT segment
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 32
How to select the right Measurement Time (1)Pulse modulated carrier, 12 ms pulse period – Stepped Scan
Even 10 ms
measurement time
yields a closed trace
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 33
How to select the right Measurement Time (1)Pulse modulated carrier, 12 ms pulse period – Stepped Scan
Even 10 ms
measurment time
yields a closed trace
Zooming in reveals
gaps in the trace
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 34
How to select the right Measurement Time (2)Pulse modulated carrier, 12 ms pulse period – Time Domain Scan
Closed trace with 12 ms
measurement time
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 35
How to select the right Measurement Time (2)Pulse modulated carrier, 12 ms pulse period – Time Domain Scan
Closed trace with 12 ms
measurement time
Gaps in trace with 10 ms
measurement time
Important:
Measurement time ≥
signal period
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 36
How to select the right Measurement Time (3)Pulse modulated carrier, 12 ms pulse period – Spectrum Analyzer Zero Span
Zero span display in
spectrum analyzer
measures signal
period
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 37
New insights for EMI diagnostics
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 38
Scan Spectrogram
l Spectrogram Gapless Spectrogram of Quasi peak trace.Quasi peak values of whole CISPR Band B is processed in real time.
EUT is a laptop power supply. Different load conditions change the spectrum over time.
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 39
Persistance Spectrum (1)
l The trace color shows how often a signal occurs at a specific frequency and level
l ⇒⇒⇒⇒ Spectral histogram
Virtual table and result display
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 40
Persistance Spectrum (2)
Clear write display (yellow trace) Max hold display (blue trace) of a broadband interferer (windshield wiper motor) with conventional spectrum analysis
The same disturbance signal in persistence spectrum mode : A second pulsed disturbance signal is hidden by the broadband noise and not detectable by conventional spectrum analysis
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 41
Frequency Mask Trigger (FMT)R&S ESR EMI Test Receiver
Frequency mask editor:
The spectrum display is continuously updated.
Draw the mask by touch screen or enter points numerically.
Mask
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 42
l Commercial standardsl DUTs with rapidly changing disturbance characteristic (e.g. lighting equipment)or drifting interferersl Intermittent disturbances requiring long dwell timesl Fieldstrength measurements with EUT positioning (mast/turntable)
l Automotivel DUTs with very short duty cycle (e.g. starter, power windows)l Use of small IF bandwidths due to low limit lines
l Military/Aerospace (A&D) l EUT cycle time in emissions testing is made practicall Probability of intercept for transient emissions increased
Applications and User Benefit
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 43
l The time needed for an EMI measurement (diagnosis/ precompliance/certification) is significantly reduced with the FFT-based time-domain scan
l The measurement uncertainty of the time-domain scan is equivalent to the uncertainty of the stepped frequency scan
l The method is applicable for EMI compliance measurements to CISPR 16-1-1 ed. 3
l In EMI diagnosis realtime functions offer new and powerful measurement and analysis capabilities
Conclusion
Comparison of Time Domain Scans and Stepped Scans in
EMI Test Receivers | 44