Improving the accuracy of EMI emissions testing James Young Rohde & Schwarz
Q&A Who uses what for EMI? Spectrum Analyzers (SA) Test Receivers (TR) CISPR, MIL-STD or Automotive? Software or front panel? Novice, Capable, Fluent or Expert?
Overview EMI Equipment comparison Spectrum Analyzers (SA) Test Receivers (TR) Making accurate measurements RF / IF overload and Preselection EMI Detectors and Filters Preamps, where and when
Spectrum Analyzers for EMI
Spectrum Analyzers for EMI Level [dbµv] 120 100 80 60 40 20 0-20 30M 50M 70M 100M 200M 300M 500M 700M 1G SA setup for EMI Example - Set Start freq, stop freq, RBW and detector from standard - Span of 1G-30MHz = 970 MHz / 1000 = 1 MHz resolution - x samples in RBW are stored, 500 or 1000 are displayed - samples within RBW analyzed or weighted by the detector - QP integrates voltage in RBW, applies CISPR weighting
Spectrum Analyzers for EMI Question - What was wrong with the previous setup? Answer 30M 50M 70M 100M 200M 300M 500M 700M 1G - Frequency and amplitude accuracy depend on many samples falling within each measurement bin (also called display pixel). - Next bin will be 1 MHz away (nearly 10 x 120 khz)! - Frequency resolution much too course for EMI - Solution?
Level [dbµv] 120 100 80 60 40 20 0 Spectrum Analyzers for EMI -20 30M 50M 70M 100M 200M 300M 500M 700M 1G Example Revisited - Span of 970 MHz / 1001 = 1 MHz resolution - If using 120 KHz RBW, CISPR recommends 60 KHz bin points (17 x finer than 1 MHz) - Solution: subrange in the span
Spectrum Analyzers for EMI Frequency Accuracy of SA - SA frequency resolution is far too course for EMI without sub-ranging the CISPR span - SA frequency accuracy when exploring peaks influenced by Span, RBW, VBW, marker accuracy Amplitude Accuracy of SA - 6 db (EMI) filters - QP and AVE detector times are observed - Data correction for system transducers - EUT specific timing issues are considered - Subranges set properly for sample # - RF and IF stages are not overloaded
Spectrum Analyzers for EMI Conclusion SA lacks 2 control parameters for EMI - STEP SIZE between measurement bins - DWELL TIME at each measurement bin - Lacks dynamic range and overload protection (later slides) Sub-ranging and zero span is an attempt to make SA measure like TR
Test Receivers for EMI
Test Receivers for EMI Receivers for EMI - Frequency Span (start / stop) - RBW Filter and detector - DWELL TIME at each measurement point - FREQUENCY INCRIMENT (step size independent of RBW) - TR adjusts sample x and bins depending on span - x is often 16,000 100,000+ - Span / x = frequency resolution
Sample Points Example 120 100 80 60 40 20 0 30M 50M 70M 100M 200M 300M 500M 700M 1G Level [dbµv] 57.00 50.00 40.00 30.00 20.00 10.00 0.00 79.94M 85M 90M 95M 100M 109.63M Frequency [Hz] Example: TR (green) vs. SA (blue) sample points
Samples and Bins Freq error samples RBW RBW Peak Lost Integration of samples detector Peak detector QP peak peak QP QP Ave Ave RMS Display BIN RMS Display BIN Bin amplitude is detector value, Bin freq is reported in center SA has 1001 bins, TR accesses 100,000 bins (from memory)
Receiver Measurements CISPR 16 recommends Step size RBW / 2 Frequency Range 6-dB Bandwidth Step Size (<1/2 RBW) # of meas Bins 150 khz to 30 MHz 9 khz 4 khz 7,463 30 MHz to 1 GHz 120 KHz 50 KHz 19,400 VS 80M 85M 90M 95M 100M 110M Frequency [Hz]
Test Receivers for EMI Conclusion TR incorporates EMI control parameters - STEP SIZE between measurement bin - DWELL TIME at each measurement bin - # of measurement bins as necessary for accuracy Time Penalty? - Time dependant on detector and EUT, not measurement speed of instruments
Best Instrument for EMI? Pros and Cons of Each - SA is faster for initial preview - SA can also be used for RX and TX measurements - TR has little use outside EMI, expensive unit for one use - SA sub-ranging negates any speed advantage over TR for EMI - SA frequency / amplitude accuracy easily skewed by improper settings and interpretation
Which one to use? Use SA or TR to develop hit list Use SA or TR for maximization Att 30 db AUTO RBW 120 khz MT 1 s PREAMP ON FREQUENCY 931.9200000 MHz LEVEL QPK dbµv 10 20 30 40 50 60 70 80 90 TR is optimum for final (dwell time, step size and auto attenuator) dbµv 100 MHz 200 MHz 300 MHz 400 MHz 500 MHz 600 MHz 700 MHz 800 MHz 900 MHz 60 50 FCC15RB 40 1 PK CLRWR 30 20 10 0-10 -20 30 MHz 1 GHz Date: 8.SEP.2003 14:13:12 SGL
Making accurate measurements Overload protection Detectors for EMI RBW Filters for EMI Preamps
Accurate Measurement Dynamic range of SA / TR is ~160 db 160 db = 10e8 or 8 orders of magnitude EMC engineers don t know what signals they are looking for initially Accuracy killers - Overloads (RF and IF) - Incorrect detector settings - Preamplifiers improperly used - Improper RBWs
Input 1st mixer RF Overdrive RF: Watch for harmonics of large signals - Use attenuator to set mixer input level Max Input Level Ref Level Mixer Level RF Attenuation sets mixer input level
RF Overload Example MARKER 1 RBW 3 MHz 496 MHz VBW 10 MHz Ref 0 dbm 1 * Att 10 db SWT 10 ms Marker 1 [T1 ] 3.03 dbm 496.000000000 MHz 1 SA AVG 0-10 -20 Marker 2 [T1 ] -48.80 dbm 996.000000000 MHz Marker 3 [T1 ] -48.60 dbm 1.500000000 GHz A -30-40 -50 2 3-60 -70-80 -90-100 Center 1 GHz 200 MHz/ Span 2 GHz Date: 23.JUN.2004 20:33:37 Example: amplified signal at 500 MHz
Input 1st mixer IF Gain IF Overdrive IF: Watch for overload flags - Use Ref Level to set IF Gain Ref Level - Set IF gain using ref Level Mixer Level Ref Level sets IF Gain
IF Overload Example RF ATTENUATION RBW 3 MHz 10 db VBW 10 MHz Ref 0 dbm * Att 10 db 1 SWT 5 ms Marker 1 [T1 ] 6.13 dbm 502.053410569 MHz 0 OVLD 1 AP CLRWR -10-20 A -30-40 -50-60 -70-80 -90-100 Center 502.0534106 MHz 99.02160153 MHz/ Span 990.2160153 MHz Date: 23.JUN.2004 20:55:20 Example: +6 dbm Pulse
Preselection Filtering Preselector is a tracking RF filter - ALL RF power (noise & signals) go into mixer - high amplitude signals outside displayed span can influence amplitude and may be aliased Cure: preselect filtering of signals before RF or IF - SA may not warn of RF or IF overdrive - IF overload won t show on display - Signals outside display ruin amplitude reading
Preselection Simple Pre-selection Mixer stage V(f) V(f) B 1 B 2 f
Overdrive and Preselection MARKER 1 RBW 3 MHz 1.000272533 GHz VBW 10 MHz Ref 0 dbm * Att 20 db SWT 10 ms Marker 1 [T1 ] -43.41 dbm 1.000272533 GHz 0-10 Marker 2 [T1 ] -54.36 dbm 1.501204189 GHz A 1 PK * CLRWR -20-30 -40 1-50 2-60 -70-80 -90-100 Start 400 MHz 141.6135906 MHz/ Stop 1.816135906 GHz Date: 8.SEP.2003 13:03:46 Example: +10 dbm signal at 500 MHz causing overdrive
Overdrive and Preselection START FREQUENCY RBW 3 MHz 600 MHz VBW 10 MHz Ref 0 dbm * Att 20 db SWT 10 ms 0-10 Marker 1 [T1 ] -43.41 dbm 1.000272533 GHz Marker 2 [T1 ] -54.58 dbm 1.501204189 GHz A 1 PK * CLRWR -20-30 -40 1-50 2-60 -70-80 -90-100 Start 600 MHz 121.6135906 MHz/ Stop 1.816135906 GHz Date: 8.SEP.2003 13:04:13 Example2: Moving overload off screen won t clear overload
Overdrive and Preselection MARKER START FREQUENCY 1 RBW 3 MHz 1.000272533 600 MHz GHz VBW 10 MHz Ref 0 dbm * Att 20 db SWT 10 ms Marker 1 [T1 ] -43.41-62.14 dbm 1.000272533 GHz 0-10 Marker 2 [T1 ] -54.36-54.58-63.57 dbm 1.501204189 GHz A 1 PK * CLRWR -20-30 -40 1 1 PS -50 2 2-60 1 2-70 -80-90 -100 Start 400 600 MHz 141.6135906 121.6135906 MHz/ Stop 1.816135906 GHz Date: 8.SEP.2003 13:04:47 13:03:46 13:04:13 Example3: Activating Preselector clears overload by filtering the fundamental
PRF, Dwell time and Detectors EUT / Detector dwell time requirements - Must capture worst case emissions of EUT - Cycle time, modulation, and pulse repetition frequency may require extended dwell in each subrange (QP,AVE, RMS) - Test Receivers include this parameter -Since tuning each bin, can dwell as long as necessary - Spectrum Analyzers have a workaround -Zero Span is a way to trick SA into dwelling at tuned bin -Must do calculation since overall sweep time is controlled -Dwell time = sweep time / measurement bins
Detector Settings Quasipeak???? QP is an attempt to quantify a signals Impact on a radio receiver (annoyance) - Factors: amplitude, frequency, pulse repetition frequency Quasipeak restrictions - Dwell time (per step or subrange) at least 1 second - PRF issues increase dwell time requirements
Detector Settings Peak Detector - Peak gives worst case - Safest detector: Often fast enough to see most signals even if instrument settings are not optimum Average or RMS Detector - Average detector used above 1 GHz for FCC and CE tests - RMS for ultra wide band, dwell time critical for integration - dwell for at least 100 ms at each bin for proper integration - Watch RBW (1MHz or 120 KHz above 1GHz?)
Detector Settings Example Ref -20 dbm -20 Att 10 db RBW 120 khz VBW 1 MHz SWT 175 ms Marker 1 [T1 ] -49.64 dbm 280.000000000 MHz -30 A 1 RM * CLRWR -40-50 1-60 -70-80 -90-100 -110-120 Start 0 Hz 50 MHz/ Stop 500 MHz Date: 23.JUN.2004 21:44:15 Example: -40 dbm signal at 280 MHz measures with RMS
1 RM AV * CLRWR Detector Settings Example Ref -20 dbm -20-30 -40 Att 10 db RBW 120 khz VBW 1 MHz SWT 175 ms Marker 1 [T1 ] -49.64-56.89 dbm 280.000000000 MHz A -50-50 1 1-60 -60-70 -70-80 -80-90 -90-100 -100-110 -110-120 -120 Start 0 Hz Start 0 Hz 50 MHz/ 50 MHz/ Stop 500 MHz Stop 500 MHz Date: 23.JUN.2004 21:44:41 Date: 23.JUN.2004 21:44:15 Example2: Same -40 dbm but with Ave and same settings
1 RM AV QP * CLRWR Detector Settings Example Ref -20 dbm -20-30 -40 Att 10 db RBW 120 khz VBW 1 MHz SWT 175 ms Marker 1 [T1 ] -49.64-56.89-76.57 dbm 280.000000000 MHz A -50-50 1 1-60 -60-70 -70 1-80 -80-90 -90-100 -100-110 -110-120 -120 Start 0 Hz Start 0 Hz 50 MHz/ 50 MHz/ Stop 500 MHz Stop 500 MHz Date: 23.JUN.2004 21:44:41 21:45:04 Date: 23.JUN.2004 21:44:15 Example3: Same -40 dbm but with QP and same settings
Transducers and PRF Transducer correction - CISPR needed a way to eliminate effects of variability in chambers, antennas and cable losses - Standards require normalization of these effects to compare results to limit line - Chamber ruled by Normalized Site Attenuation (NSA) - Transducers and connections used correction factors derived from actual calibrations
Resolution Bandwidth Filters 97 Ref Lvl 97 db V RBW 100 khz VBW 10 MHz SWT 5 ms RF Att 20 db Unit db V 3dB RBW 90 80 A 70 6dB RBW 60 50 1MAX 2VIEW IN1 1MA 2MA 40 30 20 10 0-3 Center 100 MHz 100 khz/ Span 1 MHz
Preamps Preamps: When are they needed? - Generally needed above 7 GHz - Below 7 GHz ONLY if stringent limit line or long cables Preamps: Where to put them? - Preamps amplify signal and noise - Real goal of preamp is PRESERVE signal to noise ratio (best at the antenna) - Low level signals require amplification at antenna, before signal is subjected to path loss of cables
Preamps Preamp S/N Ratio example S N S N S N Preamp near EMI receiver Common Cables loss S N Preamp at antenna ½ db per meter at 1GHz 3 db per meter at 40 GHz S N S N
Preamps Preamps - Noise Figure equation states the 1st gain/loss encountered has the most impact on s/n ratio - Must have high gain & low noise figure (amp contributed noise) - 1-18 GHz gain around 30 db with NF of 3.2 db or less - 18-26 GHz gain around 35 db with NF of 3.0 db or less - 26-40 GHz gain around 50 db with NF of 2.8 db or less Noise contributed by preamp (NF) S N RF Thermal Noise (N0)
Preamps Know your Preamp overload range and behavior - Must keep preamp in linear region - Know the preamp range (what is the max signal input without compression or damage) - Watch RF input levels to 1 st mixer system check if unsure (variable attenuator) - Non-wave guide antennas have a direct connection to the FET, watch static discharge
Preamps
Conclusion TR vs. SA Span Subranges step size resolution dwell time overloading preselection RF & IF Overloads RF -> Harmonics Attenuator IF -> If overload flag Ref Level Preamps amplify at antenna know linear region below 1gz system check (attenuator) static