An Introduction to Free-Field Measurements of Wireless Devices in Reverberation Chambers
Kate A. Remley, Group Leader Metrology for Wireless Systems What is a Reverberation Chamber? . A shielded, highly reflective free-field test chamber What you can do with one: . Create known fields (EMC/susceptibility) . Radiated emissions and power (CW and modulated-signal) . Antenna parameters (G, efficiency, etc.) You can also do communication system tests . Receiver sensitivity . Throughput . EVM, BER, etc. . DUT needs realistic channel
1 BER BER .01 32.8328 kHzkHz HighLow BERBER
.01 0 45 90 135 180 225 270 315 360 Paddle angle (degrees) -40 NIST measurements of prototype 4G -60 NISTMIMO measurements cellular telephone of wireless antennas router -80 -25 dBm -35 dBm -45 dBm -55 dBm -65 dBm -25 dBm -35 dBm -45 dBm -55 dBm
-100 Receivedpower (dBm) Receivedpower (dBm) 0 45 90 135 180 225 270 315 360 Paddle position (degrees) Fields in a Metal Box (A Shielded Room)*
. In a metal box, the fields have well defined modal distributions. . Some locations have very high field values . Some locations have very low field values
* With thanks to Chris Holloway Fields in a Metal Box with Large, Rotating Scatterer (Paddle)
Frozen Food
• The paddle changes locations where the high and low field values occur • After one mode-stirring sequence, all locations in the chamber will have experienced nearly the same collection of field maxima and minima A “Statistical” Test Chamber . Quantities measured in the reverberation chamber are averaged over a mode-stirring sequence . Randomize fields with . Mode-stirring paddles . Changing physical position . Using multiple antennas: various locations, polarizations Reverberation Chambers Come in All Shapes and Sizes Lowest frequency of operation, uncertainties determined by chamber size, wall loss
NASA: Glenn Research Reverberation Chamber Center (Sandusky, OH) with Moving Walls Chamber Electrical Characteristics Constructive and destructive interference for each
-10 mode-stirring sample All Frequencies, one angle PCS Band, four angles -20
2 -30
• Frequency domain (|S | ): (dB) 21 2 -40 Reflections create multipath |S21|
-50
Reverberation Chamber -60 Measurement 1 1.5 2 antenna Frequency (GHz)
Mode-stirring paddle . Time domain (power
Reference delay profile): Decay time antenna of reflections depends on Device Vector Network RF under RF Analyzer chamber reflectivity abs test abs P1 P2 Platform Original Applications
. Radiated Immunity . Antenna efficiency . components . Calibrate RF probes . large systems . RF/MW Spectrograph . Radiated Emissions . absorption properties . Shielding . Material heating . cables . Biological effects . connectors . Conductivity and . enclosures material properties Wireless Applications . Multipath environments . Standardized over- . Rayleigh, Rician multipath the-air test methods channels: with/without . Radiated power of channel emulators mobile wireless devices . Time response: power . Receiver sensitivity delay profile, delay spread . Large-form-factor and . Channel models body-worn devices (with . Biological effects of phantoms) modulated-signal . Public-safety emergency exposure equipment . Gain from multiple antenna systems . TX or RX diversity . MIMO Cellular Wireless: Over-the-Air Tests Required . Network providers assess performance of every wireless device model on their network: . Total Radiated Power (TRP) . Total Isotropic Sensitivity (TIS) . OTA testing traditionally done in anechoic chambers
Reverberation chambers can be used as well! Modulated Signals in RCs: What is different from EMC Testing? . Receiver needs a realistic “frequency flat” channel . Loading required: Add RF absorber to chamber . Coherence bandwidth should match DUT design . Spatial uniformity decreases . Position stirring required Reverberation Chamber: Real Channel: Loading Slows Variations Slow Variations with Frequency R 9 -20 -15 sUNLOAD ~ 10 dB TX1 to RX6 R |H(f)|2 (mean) = -58 dB -20 10 -30 2 -25 Mean |H(f)| R5 -40 2 s ~ 5 dB Mean |N(f)| -30
2
(dB) 2 (dB) Std Dev |H(f)| -35
-50 | 2 21 T1 R12 |S -40
|H(f)| T s -60 3Loaded = LOAD -45 ~ 5 dB -70 -50 Unloadedbroader ChamberT2 CBW Loaded Chamber: 7 abs -55 -80 0 1 2 3 4
730 740 750 760 770 780 790 800 Bandwidth (MHz) Frequency (MHz)
Cellular Device Testing: How is it done?
. Reference measurement provides: . Chamber loss: Transfer function Gref of chamber . Spatial uniformity of averaged fields in chamber . Rotating platform: ReverberationReverberation Chamber Chamber Measurement Gref =
DeviceDevice Vector Network RFRF underunder RFRF Analyzer absabs absabs testtest Base station P1 P2 PlatformPlatform emulator Total Radiated Power from Cell Phones Data from CTIA working group shows good agreement between anechoic and reverberation chambers
TRP Comparison, Free Space, Slider Open, W-CDMA Band II
Cell phone 5.00 4.50 AC RC 4.00 Anechoic Lab A Band II 3.50 Anechoic Lab B Band II 3.00 Anechoic Lab C Band II Reverb Lab A Band II 2.50 Reverb Lab B Band II 2 dB 2.00
Reverb Lab C Band II Relative TRP Relative 1.50 Reverb Lab D Band II 1.00 Reverb Lab E Band II 0.50 0.00 Low Mid High Reference Channel +/-2 dB is threshold TRP Comparison, Free Space, Slider Open GSM850
5.00 4.50 AC RC 4.00 Anechoic Lab A GSM850 3.50 Anechoic Lab B GSM850 3.00 Anechoic Lab C GSM850 • Cell phone Reverb Lab A GSM850 2.50 Reverb Lab B GSM850 2.00
testing: ca. 2010 Reverb Lab C GSM850 RelativeTRP 2 dB 1.50 Reverb Lab D GSM850 • But in 2016… 1.00 Reverb Lab E GSM850 0.50 0.00 Low Mid High Channel The Machine-to-Machine Revolution . By 2019: 11.5 billion mobile devices (world population 7.6 B)* . M2M/IoT growing faster than smart phones . 2014: 495 million . 2019: 3 billion
* Source Cisco Testing Large Form-Factor Devices . Integrated antennas: test of entire device required . Reverberation chamber: now only option for SISO tests . Device placement not critical within chamber . Relatively low cost . OTA test issues: Large, lossy DUTs Loading Decreases Spatial Uniformity Loading helps with demodulation but introduces other “nonideal” effects
RelativePDP for mean different power loading at Monopole12 at Monopole2 -20Unloaded 1 RF 2 RF 3 RF 4 RF 5 RF 6 RF -70 Unloaded RF Absorbers Unloaded 18 TwoBoxes 18 -22-75 Metallic Boxes TwoBoxes Theoretical StDev FourBoxesFourBoxes SixBoxesSixBoxes -24-8017 17 EightBoxesEightBoxes TenBoxesTenBoxes -26-85 16 TwelveBoxesTwelveBoxes16 OneRF -28 OneRF -90 TwoRFTwoRF 15 ThreeRF 15 -30 ThreeRF -95 FourRFFourRF FiveRF -3214 FiveRF 14 -100 SixRFSixRF
-34 (dB) Power Relative Decreases spatial Relative mean power (dB) (dB) power mean Relative -10513 13 uniformity -36-110 12 IncreasesDecreases chamber decay loss time 12 -38-115 (can(replicates calibrate out)real world) 11 11 -40Unloaded 2 Bx 4 Bx 6 Bx 8 Bx 10 Bx 12 Bx 1500-120 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 0 1 2 Frequency3 4(MHz) 5 6 7 8 s increasesTime with(s) loading S. van de Beek, et al., Characterizing large-form-factor (in percent) devices in a reverberation chamber, EMC Europe 2013. Loading and Position Stirring go Hand in Hand . Spatial lack of uniformity a necessity: . unstirred energy . correlated samples: paddle position, location, frequency . Industry uses position, polarization, source stirring to improve estimate of DUT performance
• Chamber reference for PCS channel 9262 (1.850-1.854 GHz) • VNA : 9 spatially uncorrelated positions • Combinations of all sets of data Set-up: Absorber Placement • Standing on floor • Lying on floor • Stacked
Considerations: • Exposed absorber surface area • Exposed metal surfaces
• Proximity to antennas 18 Comparable Loading, Different Uncertainty PCS band measurement (~1950 MHz) . Load chamber for approximately the same CBW . Chamber loss approximately the same as well . 휎 is higher when absorbers lie on floor 퐺ref . Less exposed metal surface . Higher proximity effect
Distributed on Distributed on Stacked Floor: Standing Floor: Lying CBW (MHz) 3.13 3.32 3.48 No. abs 3 7 4
퐺ref (dB) -29.46 -29.69 -29.78 휎 (dB) 퐺ref 0.15 0.30 0.35 Set-up: Stirring Sequence is Important . Stirring mechanisms influence results differently . Each chamber will have a different “mix” of optimal stirring
13 7 x steps steps y
Cart 3 (dashed) Cart on ground 1
Cart 2 x
y
Measure effects of paddle angle and antenna position at three locations in a loaded chamber Measured and modeled uncertainty at the three locations
K.A Remley, R.J Pirkl, H.A Shah, and C.-M. Wang, "Uncertainty from choice of mode-stirring technique in reverberation-chamber M = antenna positions measurements," IEEE Trans. Electromagnetic Compat., vol. 55, no.6, pp. 1022-1030, Dec. 2013. N = paddle angles Set-up: Antenna Placement is Important Unstirred energy: increased K factor, reduced spatial uniformity Antenna placement guidelines: Relationship between . Orient away from each other antennas and absorber . Cross polarize is also important . Aim toward stirrers DUT: Unknown pattern? 3 Cell Band 2 TX/RX pairs PCS Band 1
0 Dir/Omni -1
-2 Omni/Dir K K Factor (dB) -3
-4 Dir/Dir Omni/Omni -5 1 2 3 4 Antenna Configuration Good Set-up = Good Results . Must account for . placement and amount of RF absorber . number, type, and correlation of mode-stirring samples . antenna type and placement . Good comparison between chambers: throughput vs. input power Lab Good Nominal Bad AC1 -100.50 -99.00 -94.20 lab1, R1 lab1, R1 lab1, R2 lab1, R2 lab1, R3 AC2 -102.80 -100.00 -94.20lab1, R3 lab2, R1 lab2, R1 lab2, R2 lab2, R2 lab2, R3 RC1 -103.58 -100.29 -93.40lab2, R3 lab3, R1 lab3, R1 lab3, R2 lab3, R2 lab3, R3 RC2 -101.46 -98.22 -92.12lab3, R3 RC3 -101.93 -98.66 -90.91 Two devices tested in three different reverberation chambers Spread+/- 1.54 From1.04 MOSG1406041.65 Results show good repeatability and From MOSG131207 comparison with anechoic methods TRP for Large M2M Device
. Wireless, solar-powered trash compactor . TRP measured for W-CDMA signal (BW = 3.84 MHz) . Loading: stacked absorbers . Coherence bandwidth: Verified >3.84 MHz (4.42 MHz)
. Antenna proximity effect: No effect at 1 λ (at fc) . Reference: Nine locations, DUT one location
abs abs Agreement with anechoic chamber: . 0.2 dB, PCS band (1.850 GHz to 1.995 GHz) . 1.95 dB, Cell band (800 MHz to 900 MHz) DUTDUT OTA Tests to Model Multipath Environment
Office Corridor Oil Refinery
Apartment Building
Automobile Plants Subterranean Tunnels NIST channel measurements: Standards development for electronic safety equipment such as firefighter beacons Channel Measurements: Denver High Rise
18
21 15 17 19 14 16 13 20 12 11 10 8 9 7 6 5 3 4
North
1 West 2 East Receiving Transmitting antenna antenna South VNA measurement test locations are in pink tripods 62 inches
VNA
Port 1 Port 4
Ground Plane
Fiber Optic Fiber Optic Transmitter Receiver 200 m RF Optical RF Optical Optical Fiber Replicate Environment in Reverberation Chamber • Add RF absorbing material to “tune” the decay time of the chamber • Distributed multipath (reflections) matched by chamber’s decay profile
Time response of channel replicated in chamber
0 1 absorber: rms=187 ns -5 -10 3 absorber: rms=106 ns -15
) -20 7 absorber: =66 ns
B rms
d (
-25
e
l
i f
o -30
r
P
y -35
a l
e -40
D
r
e -45
w o
P -50 -55 Reverberation chamber with -60 Large office biulding -65 absorbing material rms=59 ns -70 0 50 100 150 200 250 300 350 400 450 500 time (ns) Emulating Other Reflective Environments . Oil Refinery Emulating Other Reflective Environments Automobile Factories . 0
-5
-10
-15
-20 o delay-spread=207ns * mean-delay=252.4ns o delay-spread=123ns * mean-delay=130.3ns -25 o delay-spread= 69ns * mean-delay=71.8ns o delay-spread=195ns * mean-delay=147.0ns -30 o delay-spread=199ns * mean-delay=161.6ns
o delay-spread=261ns * mean-delay=275.3ns PowerDelay (dB) Profile -35 1 absorber 3 absorbers -40 7 absorbers Jefferson North Plant, 10m -45 Jeffersonassembly North Plant, 50m Jefferson plantNorth Plant, 100m -50 0 100 200 300 400 500 600 700 800 900 time (ns)
0 o delay-spread=277ns * mean-delay=290.5ns 1 absorber o delay-spread=122ns * mean-delay=130.1ns 3 absorbers o delay-spread= 67ns * mean-delay=71.2ns 7 absorbers o delay-spread=168ns * mean-delay=137.8ns -5 o delay-spread=175ns * mean-delay=142.7ns flint metal, drg-drg, 10m o delay-spread=128ns * mean-delay=105.8ns flintstamping metal, drg-drg, 50m o delay-spread=173ns * mean-delay=176.1ns flint metal, drg-drg, 80m flint metal,plant drg-drg, 110Bm -10
-15
PowerDelay (dB) Profile -20
-25
-30 0 200 400 600 800 1000 1200 1400 1600 1800 2000 time (ns) Urban Canyon Multipath Effects
-90 TX1 to RX5 -110 = 115 ns -100 rms Noise Threshold = -116 dB -110 -120
-120 -130 -130 -140
PDP (dB) PDP -140 PDP (dB) PDP -150 -150 TX1 to RX9 -160 = 39 ns -160 rms Noise Threshold = -116 dB R -170 -170 9 0 200 400 600 800 1000 1200 0 500 1000 Delay (ns) Delay (ns) R10 Line of Sight Non Line of Sight R5
T R T1 1212 •Measurements made in T3 Denver urban canyon 2009 T2 •Channel characterization: LOS and NLOS Replicating Clustered Multipath in 1 fitted simulated data Reverberation Chamber 0.9 measured Data Denver
0.8
0.7 Shortest path 0.6
0.5 PDP TX 0.4
0.3
0.2
0.1
0 0 100 200 300 400 500 600 700 800 900 1000 RX Delay [ns] • Blue: Mean of 27 NLOS Welton Street measurements Clusters of RX RX RX 1 2 3 exponentially • Red: RC + channel Parking lot t
e emulator e r t
distributed signals S
h t • Dashed: Exponential 7 received off of 1 model buildings TX Transmitter Site Glenarm Pl Channel Models Used for Standardized OTA tests . Outdoor-to-indoor channel model for 700 MHz . 8 environments, hundreds of measurements . “NIST Model” included in 3GPP reverberation- chamber-based test methods
0 700 MHz, measured Excess tap delay Relative power [dB] 4900 MHz, measured -5 700 MHz, fit [ns] 4900 MHz, fit 0 0.0 -10 40 -1.7 120 -5.2 -15 180 -7.8 210 -9.1 -20
260 -11.3 PDP (dB)PDP -25 350 -15.2 RMS DS: 80 ns -30 Discrete version of the “NIST Model” @ 700 MHz for anechoic-chamber measurements -35 0 100 200 300 400 500 600 Delay (ns) D.W. Matolak, K.A. Remley, C.L. Holloway, and C. Gentile, “Outdoor-to-Indoor Channel Dispersion and Power-Delay Profile Reverberation chamber can easily Models for the 700 MHz and 4.9 GHz Bands,” IEEE Antennas replicate diffuse multipath and Wireless Propagat. Lett., vol. 15, 2016, pp. 441-443. Millimeter-Wave 5G Wireless Concepts: Wireless for “5G” . Massive MIMO . Tiered spectrum (licensed and unlicensed) . Millimeter-wave Left: 10,000 paddle positions at frequencies1 antenna position Right: 100 paddle positions at 100 antenna positions
.LowGoal Uncertainty is uncertainty ≈ 1 % or Required:1 u = + 퐾2 ≤ 0.0001 . High 푁carrier frequencies . HighVery low modulation K factor required bandwidths Stay tuned … Reverberation Chambers for Wireless Test
horn antenna
mode stirrers
wireless device antenna
absorber Some issues: . Angle-of-arrival information lacking . Advanced transmission, multiple antenna systems . Test methods (CTIA, 3GPP groups) . Instantaneous channel can be problematic for receiver (even if mean characteristics are OK) . Field non-uniformity increases with loading and loading is often required for receiver tests . Testing devices with repeaters is difficult Reverberation Chambers for Wireless Test Some benefits: . Capable of simulating key characteristics of many multipath environments for the testing of wireless devices . For OTA test reverberation chambers are: . Accurate – uncertainties on par with anechoic methods . Able to provide realistic distributed power delay profile . Suitable for testing diversity and MIMO gain (due to multipath) . Cost effective . Space efficient The “NIST Model” for building penetration: 8 environments 0 700 MHz, measured RMS DS: 80 ns @ 700 MHz 4900 MHz, measured -5 700 MHz, fit Excess tap delay Relative power [dB] 4900 MHz, fit [ns] -10 0 0.0 40 -1.7 -15 120 -5.2 180 -7.8 -20
210 -9.1 PDP (dB)PDP -25 260 -11.3 350 -15.2 -30 D.W. Matolak, K.A. Remley, C.L. Holloway, and C. Gentile, “Outdoor-to-Indoor Channel Dispersion and Power-Delay Profile -35 0 100 200 300 400 500 600 Models for the 700 MHz and 4.9 GHz Bands,” IEEE Antennas Delay (ns) and Wireless Propagat. Lett., vol. 15, 2016, pp. 441-443. Watch this space for more information on over-the-air testing with reverberation chambers