a 140 ghz mimo transceiver in 45 nm soi cmos · 4/22/2020 bcicts 2018 san diego, ca slide 1 (paper...
TRANSCRIPT
BCICTS 2018 San Diego, CA4/22/2020 Slide 1
(Paper 15.3)
A 140 GHz MIMO Transceiver in
45 nm SOI CMOS
Arda Simsek1, Seong-Kyun Kim2, and Mark J. W. Rodwell1
1University of California, Santa Barbara, USA2Teledyne Scientific and Imaging, USA
BCICTS 2018 San Diego, CA
Acknowledgements
4/22/2020 Slide 2
Contact support: NSF Giganets program.
IC Fabrication: Global Foundries. 45nm SOI
BCICTS 2018 San Diego, CA
Motivation
4/22/2020 Slide 3
Large available spectrum at mm-waves
Shorter wavelength – small IC, antenna arrays
Massive # of parallel channels – multiple independent beams
BCICTS 2018 San Diego, CA
Applications and Challenges
4/22/2020 Slide 4
High capacity mobile communications
Problems: Solutions:
High attenuation Phased arrays
High blockage Mesh networks & beam steering
BCICTS 2018 San Diego, CA
Direct conversion receiver
140 GHz LNA, double balanced passive mixer
LO distribution through two x9 multipliers from common LO port
4-Channel MIMO Receiver
4/22/2020 Slide 5
1.69 mm x 1.76 mm
BCICTS 2018 San Diego, CA
Direct conversion transmitter
140 GHz PA (same with LNA), I/Q Gilbert Cell Active Mixer
LO distribution thru two x9 multiplier from common LO port
4-Channel MIMO Transmitter
4/22/2020 Slide 6
1.67 mm x 1.76 mm
BCICTS 2018 San Diego, CA
45 nm CMOS SOI
Design Details – Transistor Modeling
4/22/2020 Slide 7
Gate
Series Gate Feeding
Gate
Gate
Dra
in (
M2)
So
urc
e (M
2)
Gate
(M2)Dra
in (
M2)
So
urc
e (M
1)
Parallel Gate Feeding
0.0 0.1 0.2 0.3 0.4 0.50
1
2
3
4
5
6
7
8
PGF
SGF
Current Density (uA/um)
MA
G (
dB
)
2.4
2.6
2.8
3.0
3.2
3.4
3.6
3.8
4.0
Total width = 1um x 30 fingers
No
ise
Fig
ure
(d
B)
Similar performances (MSG, NF)
SGF is hard to extract the gate inductance
PGF - Source can directly connect to ground
BSIM 𝑓𝑚𝑎𝑥 > 400 GHz, measured 𝑓𝑚𝑎𝑥 = 213 GHz
(30x1um single gate contact device) (Inac et al, 2014)
BCICTS 2018 San Diego, CA
45 nm CMOS SOI
Design Details – Transistor and Matching Modeling
4/22/2020 Slide 8
Cgd cancellation
: MOM Cap
M1 ground plane
Cgd cancellation – better isolation & gain
PEX
CA, PC, M1 parasitic
(capacitance, resistance)
EM Simulation
M2 – LB interconnection parasitic
(capacitance, resistance, inductance)
BSIM
Transistor core
BSIM + PEX + EM simulation
(BSIM & PEX from the foundry, HFSS for EM)
VDD: M2-M3-C1-C2
Gate bias: M2
UB-LB
M1 ground plane
BCICTS 2018 San Diego, CA
Design Details – Low Noise Amplifier (LNA)
4/22/2020 Slide 9
3-stage differential CS amplifier
Cgd cancellation
Transformers for matching networks and sing.-to-diff.
315 x 170 um2 without pads
~ 415 x 370 um2 with pads
BCICTS 2018 San Diego, CA
Design Details – Down-conversion Mixer
4/22/2020 Slide 10
Double balanced passive mixer
Better linearity and low flicker noise
Pseudo-differential TIA
Total width of each transistor (TIA) = 30 µm
RF
LO_Q
LO_I
TIA
16
18
20
22
24
26
28
2
3
4
5
6
7
8
0 5 10 15 20 25
Ga
in (
dB
)
No
ise
Fig
ure
(dB
)
Frequency (GHz)
BCICTS 2018 San Diego, CA
Design Details – x9 Multiplier
4/22/2020 Slide 11
X3 (140 GHz) X3 (47 GHz)
LO Input stage
Inverter based single-to-differential
conversion & 3rd harmonic generation
47 GHz and 140 GHz triplers use
similar topology with LNA
80 mA @ 1V
BCICTS 2018 San Diego, CA
Design Details – Up-conversion Mixer / DA
4/22/2020 Slide 12
BB input stage
IQ modulator
Active double balanced Gilbert cell design
Lower conversion loss
Digital baseband inputs - Limited to QPSK
DA is same with LNA
-15
-10
-5
0
5
10
6
8
10
12
14
16
18
20
-30 -25 -20 -15 -10 -5 0
Ou
tpu
t P
ow
er
(dB
m)
Ga
in (d
B)
Input Power (dBm)
Output P1dB = 4.3 dBm
Output power: 5 dBm
P1dB = 4.3 dBm
BCICTS 2018 San Diego, CA
LNA
Measurements – Circuit Blocks
4/22/2020 Slide 13
0
5
10
15
20
25
-30
-20
-10
0
10
20
120 130 140 150 160 170 180
Gain_Sim
NF_Sim
Gain_Meas
S11_Sim
S22_Sim
S11_Meas
S22_Meas
Gain
(d
B)
S11
/S2
2 (d
B)
Frequency (GHz)
-6
-4
-2
0
2
4
6
140 145 150 155 160
Measured #1 (VDD: 1 V)
Measured #2 (VDD: 1 V)
Measured #1 (VDD: 1.1 V)
Measured #2 (VDD: 1.1 V)
SimulatedOu
tpu
t P
ow
er
(dB
m)
Frequency (GHz)
x9 LO Multiplier
41 mA @ 1V
Peak Gain = 19.9 dB @ 145 GHz
3-dB BW = 10GHz
98 mA @ 1V
Peak output power = 1.5 dBm
@ 148 GHz
BCICTS 2018 San Diego, CA
Measurements – Receiver Channel
4/22/2020 Slide 14
-5
0
5
10
15
20
0 5 10 15 20
IC1-Ch.1IC1-Ch.4IC2-Ch.1IC2-Ch.4
Co
nvers
ion
Ga
in (
dB
)
BB Frequency (GHz)
Fixed LO @ 145.9 GHz0
5
10
15
20
135 140 145 150 155 160 165 170
IC1-Ch.1IC1-Ch.4IC2-Ch.1IC2-Ch.4
Co
nvers
ion
Ga
in (
dB
)
RF Frequency (GHz)
Fixed BB @ 100 MHz
18 dB conversion gain
12 GHz 3-dB BW
Narrow-band notch in RF response - limits the data rate
163 mA + 109 mA + 223 mA = 495 mA @ 1V
BCICTS 2018 San Diego, CA
Measurements – Transmitter Channel
4/22/2020 Slide 15
3-dB modulation bandwidth ~ 6 - 8 GHz
Total transmitter output power: -2 dBm with 1 V supply,
3 dBm with 1.2 V supply @ 145 GHz
161 mA + 94 mA + 208 mA = 463 mA @ 1V
-20
-15
-10
-5
0
5
0 2 4 6 8 10 12 14 16
IC1-Ch.1IC1-Ch.4IC2-Ch.1IC2-Ch.4
Re
lati
ve S
ide
ban
d
Po
wer
(dB
m)
Modulation Frequency (GHz)
Fixed RF @ 146.01 GHz
-10
-8
-6
-4
-2
0
2
4
140 145 150
VDD = 1 V
VDD = 1.1 V
VDD = 1.2 V
Tx
Ou
tpu
t P
ow
er
(dB
m)
Carrier Frequency (GHz)
LO
LO+BBLO-BB
BCICTS 2018 San Diego, CA
Measurements – Link Experiment
4/22/2020 Slide 16
Measurement Setup
Wired link: G-band (140-220 GHz) waveguide
probes, ~1 meter waveguide connection and a
waveguide attenuator (20 dB)
common 16.6 GHz LO subharmonic drive signal
mm-wave carrier is at 16.6 x 9 = 149.4 GHz
BCICTS 2018 San Diego, CA
State of the Art Transceivers
4/22/2020 Slide 17
UC Berkeley
(2014)
Tel Aviv Uni.
(2016)
UCSD
(2014)
Chalmers Uni.
(2016)
This Work
(UCSB)
Technology65nm
CMOS28nm CMOS
45nm
CMOS
250nm
InP DHBT
45nm
CMOS
Freq. [GHz] 240 102-128 155 110-170 140
Gain [dB] 25 36-39 23 26 18
NF [dB] 15# 8.4-10.4 20* 9.5 5.5*
Pdc [mW]260
(1 TRx)
51
(1 Rx)
345
(1 TRx)
357
(1 TRx)
958
(4 TRx)
Area [mm2]2
(1 TRx)
0.89
(1 Rx)
3.92
(1 TRx)
3.64
(1 TRx)
5.91
(4 TRx)
Integration Full Rx Front End Tx/Rx Tx/Rx Tx/Rx
*: simulated#: calculated from measurerements
BCICTS 2018 San Diego, CA
Conclusion and Future Work
4/22/2020 Slide 18
A 140 GHz QPSK transceiver front-end
with four channels per IC to support MIMO
One channel transceiver measurements
Initial link experiment results up to 800 Mb/s
Work in Progress
Wireless link measurements
with higher data rate
Antenna and package designs
4-channel MIMO receiver hub experiment.
Baseband
ProcessingPackaged 4-channel Rx
BCICTS 2018 San Diego, CA
Backup Slides
4/22/2020 Slide 19
BCICTS 2018 San Diego, CA
0
5
10
15
20
25
-30
-20
-10
0
10
20
120 130 140 150 160 170 180
Gain_SimGain_Meas
S11_SimS22_SimS11_MeasS22_Meas
Gain
(d
B)
S11/S
22 (d
B)
Frequency (GHz)
0
5
10
15
20
25
-30
-20
-10
0
10
20
135 140 145 150 155 160
Gain_High-ResGain_Low-Res
S11_High-ResS22_High-ResS11_Low-ResS22_Low-Res
Gain
(d
B)
S11/S
22 (d
B)
Frequency (GHz)
3-dB bandwidth: 11 GHz (138 – 149 GHz)
Power consumption: Trap-rich 42.4 mW @ 1 VLow-res. 41.7 mW @ 1 V
Substrate: High (trap-rich) vs Low resistivitySimulation vs Measurements (trap-rich)
LNA Measurements
BCICTS 2018 San Diego, CA
IIP3: -5.84 dBm, Input P1dB: -13.87 dBm
-40 -30 -20 -10 0-20
-10
0
10
20
30
Gain
Output Power
Gain = 19.07 dB
Ou
tpu
t P
ow
er
(dB
m),
Gain
(d
B)
Input Power (dBm)
Input P1dB = -13.87 dBm
-40 -30 -20 -10 0-80
-60
-40
-20
0
20
Ou
tpu
t P
ow
er
(dB
m)
Input Power (dBm)
IIP3 = -5.84 dBm
140 GHz with 10 MHz spacing
LNA Linearity Simulations
BCICTS 2018 San Diego, CA
TIA Breakout measurements
Power consumption- Simulation 12 mA @ 1 V- Measurement 11.9 mA @ 1 V
In/output feeding lines are not de-embedded in the measurementSimulation results doesn’t include the feeding linesNotch due to the bias network (next slide)
5
10
15
20
25
30
0 10 20 30 40 50
SimulatedMeasured
S(2
,1)
(dB
)
Frequency (GHz)
*TIA: total width of each transistor: 30 umSimulation: IdealMeasurement: VDD without50 Ω termination
Simulation: IdealMeasurement: VDD with50 Ω termination
5
10
15
20
25
30
0 10 20 30 40 50
SimulatedMeasured
S(2
,1)
(dB
)
Frequency (GHz)
5
10
15
20
25
30
0 10 20 30 40 50
Single-endedDifferential
S(2
,1)
(dB
)
Frequency (GHz)
BCICTS 2018 San Diego, CA
Multiplier chain (x9) measurement setup
Harmonic mixer(OML,WR-05)
20 dB attenuation
WR-05 probeLO Input
Harmonic mixer:- Harmonic number: 16, IF @ 404.4 MHz
Ref 0 dBm
1 AP
CLRWR
A
EXTM IX G
EXT
2 kHz/Center 140.13 GHz Span 20 kHz
*
*
3DB
RBW 200 Hz
VBW 200 Hz
SWT 1 s
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
1
Marker 1 [T1 ]
-19.94 dBm
140.130000000 GHz
Date: 31.MAR.2017 12:46:00
Ref 0 dBm
1 AP
CLRWR
A
EXTM IX G
EXT
2 kHz/Center 153 GHz Span 20 kHz
*
*
3DB
RBW 200 Hz
VBW 200 Hz
SWT 1 s
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
1
Marker 1 [T1 ]
-16.72 dBm
153.000000000 GHz
Date: 31.MAR.2017 12:54:26
Ref 0 dBm
1 AP
CLRWR
A
EXTM IX G
1 MHz/Center 144 GHz Span 10 MHz
*
*
3DB
RBW 2 kHz
VBW 2 kHz
SWT 2.5 s
EXT
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
1
Marker 1 [T1 ]
-22.95 dBm
144.000000000 GHz
Date: 31.MAR.2017 13:05:54
Ref 0 dBm
1 AP
CLRWR
A
EXTM IX G
100 MHz/Center 144 GHz Span 1 GHz
EXT
*
3DB
RBW 10 kHz
* VBW 10 kHz
SWT 10 s
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
1
Marker 1 [T1 ]
-22.41 dBm
144.000000000 GHz
Date: 31.MAR.2017 13:07:09
BCICTS 2018 San Diego, CA
Receiver Simulations
31
31.5
32
32.5
33
0 2 4 6 8 10 12
Co
nvers
ion
Ga
in (
dB
)
Frequency (GHz)
LO @ 139 GHz
4
6
8
10
12
14
16
18
105
106
107
108
109
NF
ds
b (
dB
)
Frequency (Hz)
Integrated NFdsb (dB, 100kHz - 1 GHz) = 5.53
NFdsb = 5.5 dB
5
10
15
20
25
105
106
107
108
109
Temp. = 27 o
Temp. = 85 o
NF
ds
b (
dB
)Frequency (Hz)
Integrated NFdsb (dB, 100kHz - 1 GHz) = 5.53
NFdsb @ 1 GHz = 5.5 dB
Integrated NFdsb (dB, 100kHz - 1 GHz) = 6.2
NFdsb @ 1 GHz = 6.1 dB
30
30.5
31
31.5
32
32.5
33
0 2 4 6 8 10 12
Temp. = 27 o
Temp. = 85 o
Co
nv
ers
ion
Ga
in (
dB
)
Frequency (GHz)
LO @ 139 GHz
BCICTS 2018 San Diego, CA
Receiver measurements
-5
0
5
10
15
20
0 5 10 15 20
#1-Ch.1#1-Ch.4#2-Ch.1#2-Ch.2
Co
nvers
ion
Ga
in (
dB
)
IF Frequency (GHz)
Fixed LO @ 136.9 GHz-5
0
5
10
15
20
0 5 10 15 20
#1-Ch.1#1-Ch.4#2-Ch.1#2-Ch.4
Co
nvers
ion
Ga
in (
dB
)
IF Frequency (GHz)
Fixed LO @ 145.9 GHz
0
5
10
15
20
135 140 145 150 155 160 165 170
#1-Ch.1#1-Ch.4#2-Ch.1#2-Ch.4
Co
nvers
ion
Ga
in (
dB
)
RF Frequency (GHz)
Fixed IF @ 100 MHz
1 AP
CLRWR
A
Ref 0 dBm Att 25 dB
Center 100 MHz Span 100 MHz10 MHz/
EXT
*
3DB
RBW 10 kHz
* VBW 10 kHz
SWT 1 s
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
1
Marker 1 [T1 ]
-9.48 dBm
100.000000000 MHz
Date: 6.APR.2017 18:35:15
1 AP
CLRWR
A
Ref 0 dBm Att 25 dB
EXT
Center 16.21111111 GHz Span 10 kHz1 kHz/
*
3DB
RBW 100 Hz
* VBW 100 Hz
SWT 2 s
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
1
Marker 1 [T1 ]
-60.93 dBm
16.211111095 GHz
Date: 6.APR.2017 18:37:29LO feedthroughIF spectrum: span 100 MHz
BCICTS 2018 San Diego, CA
-30
-25
-20
-15
-10
-5
0
5
130 135 140 145 150
Ou
tpu
t P
ow
er
(dB
m)
Frequency (GHz)
RF-L: -0.9 dBm,RF-U: -0.7 dBm,
LO: -1.3 dBm
RF @ 1 GHz
-30
-25
-20
-15
-10
-5
0
5
10
130 135 140 145 150
Ou
tpu
t P
ow
er
(dB
m)
Frequency (GHz)
5 dBm
-15 dBm
RF @ 1 GHz
Transmitter Simulations
0 V quadrature inputLO feedthrough: Q-mixer is unbalancedLower output power ~ 6 dB
BCICTS 2018 San Diego, CA
Transmitter measurements
BB @ 0.1 GHz, LO @ 160 GHz BB @ 0.2 GHz, LO @ 160 GHz
Measurement setup: BB-I input is open, all channels are in the on state
BIAS VDD (V)Current (mA)
Chip1 Chip2
Driver amp. 1 160 161
Mixer 1 89 94
Multiplier chain 1 204 208
BCICTS 2018 San Diego, CA
Transmitter measurements
-25
-20
-15
-10
-5
0
0 2 4 6 8 10 12 14 16
Chip1-Ch.1Chip1-Ch.4Chip2-Ch.1Chip2-Ch.4
Ou
tpu
t P
ow
er
(dB
m)
BB Frequency (GHz)
Fixed RF @ 146.01 GHz-20
-15
-10
-5
0
0 2 4 6 8 10 12 14 16
Chip1-Ch.1Chip1-Ch.4Chip2-Ch.1Chip2-Ch.4
Ou
tpu
t P
ow
er
(dB
m)
BB Frequency (GHz)
Fixed RF @ 150.01 GHz
-14
-12
-10
-8
-6
-4
0 1 2 3 4 5 6 7 8
Chip1-Ch.1Chip1-Ch.4Chip2-Ch.1Chip2-Ch.4
Ou
tpu
t P
ow
er
(dB
m)
BB Frequency (GHz)
Fixed LO @ 146 GHz-14
-12
-10
-8
-6
-4
-2
0
2
140 142 144 146 148 150 152 154 156
Chip1-Ch.1Chip1-Ch.4Chip2-Ch.1Chip2-Ch.4
Ou
tpu
t P
ow
er
(dB
m)
RF Frequency (GHz)
Fixed BB @ 0.01 GHz