Modeling, measurement and suppression of

substrate noise in mixed-signal ICs

Ming Shen
Email: [email protected]
22 April 2010

Technology Platforms Section,

Department of Electronic Systems, Aalborg University
 Outline

• Introduction
• Impact of substrate noise on UWB LNA
• Modeling of substrate noise
• Measurement of substrate noise
• Isolation and suppression of substrate noise
• Conclusion and assignment

© Aalborg University / Technology Platform Section 1 / 35

 Outline

• Introduction
• Background
• Topics

• Impact of substrate noise on UWB LNA

• Modeling of substrate noise
• Measurement of substrate noise
• Isolation and suppression of substrate noise
• Conclusion and assignment

© Aalborg University / Technology Platform Section 2 / 35

Introduction

Background
• There is the need for efficient and
low cost implementations of System-
on-Chips
• Switching noise is generated by
digital circuits
• The switching noise propagates and
spreads through the substrate
• The performance of the sensitive
analog/RF circuits sharing the
common die could be deteriorated
by the substrate noise

© Aalborg University / Technology Platform Section 3 / 35

Introduction

Substrate noise in Ultra-wideband circuits

• Frequency range: 3.1 GHz - 10.6 GHz
• Bandwidth: > 500 MHz
• EIRP: < -41.3 dBm/MHz
For a UWB system using 4 GHz with a band width of 500 MHz, at the
distance of 10 meters

Fress Space Loss(dB) = 32.45 + 20Log10F(MHz) + 20Log10D(km)

= 32.45 + 72 – 40
= 64.45 (dB)

The power of the received signal: -41.3+27- 64.45 = -78.75 dBm

The power of the noise in band: -174+87 = -87dBm
SNR=8.25 dB

Substrate noise is one of the most severe noise issues for ultra
wide band systems implemented in CMOS processes
© Aalborg University / Technology Platform Section 4 / 35
Introduction
Topics
• Impact evaluation of substrate noise in UWB circuits
• Modeling and characterization of substrate noise
• Measurement of substrate noise
• Isolation and suppression of substrate noise

© Aalborg University / Technology Platform Section 5 / 35

 Outline

• Introduction
• Impact of substrate noise on UWB LNA
• Modeling of substrate noise
• Measurement of substrate noise
• Isolation and suppression of substrate noise
• Conclusion and assignment

© Aalborg University / Technology Platform Section 6 / 35

 Impact of substrate noise on UWB LNA

Test
chip

Digital noise block

• The digital block consisting of inverter

chains with different stages and sizes is
used as the noise source
• A 1-5 GHz UWB LNA is the victim circuit
• Noise detector array is for substrate
noise measurements
UWB LNA
© Aalborg University / Technology Platform Section 7 / 35
 Impact of substrate noise on UWB LNA

-30
• Substrate noise in the UWB Output signal due to RFin
-40 of -50 dBm at 3.88 GHz
frequency band could be significantly -50

higher than the thermal noise floor -60

-70

• The noise figure at the harmonics of -80

-90
the substrate noise is obviously -100
0 1 2 3 4 5 6 7 8 9 10
increased Frequency [GHz]
Measured PSD at RF output when
digital block is on
Clock for DC supply for
digital block digital block

RF in RF output

DC supply for Chip

UWB LNA

Test PCB Measured NF when digital block is

on and off
© Aalborg University / Technology Platform Section 8 / 35
 Impact of substrate noise on UWB LNA

-30

The PSD-magnitude of the -40

-50
substrate noise is increased -60

when the clock of the digital -70

-80
circuit is increasing -90
-100
3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5
Frequency [GHz]

fclock = 10 MHz
-30 -30
-40 -40
-50 -50
-60 -60
-70 -70
-80 -80
-90 -90
-100 -100
3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5
Frequency [GHz] Frequency [GHz]

fclock = 30 MHz fclock = 50 MHz

© Aalborg University / Technology Platform Section 9 / 35

 Impact of substrate noise on UWB LNA

-30
-40 Intermodulation
Apart from the harmonics of the -50

clock frequency, intermodulation -60

-70
components are also generated -80
-90
-100
3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5
Frequency [GHz]

fRF = 3.86 GHz

-30 -30
-40 -40
Intermodulation Intermodulation
-50 -50
-60 -60
-70 -70
-80 -80
-90 -90
-100 -100
3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5 3.5 3.6 3.7 3.8 3.9 4.0 4.1 4.2 4.3 4.4 4.5
Frequency [GHz] Frequency [GHz]

fRF = 3.87 GHz fRF = 3.88 GHz

© Aalborg University / Technology Platform Section 10 / 35
 Outline

• Introduction
• Impact of substrate noise on UWB LNA
• Modeling of substrate noise
• Low and high level SPICE simulations
• Analytical approaches

• Measurement of substrate noise

• Isolation and suppression of substrate noise
• Conclusion and assignment

© Aalborg University / Technology Platform Section 11 / 35

 Modeling of substrate noise

Low level SPICE simulations

Simulation circuit including

extracted substrate network

Measured and simulated substrate

noise in time domain

• Based on extracted substrate networks

• Suitable for small scale circuits
• Not for large scale circuits
7-stage inverter chain used
as noise source
M. Van Heijningen et al. “Analysis and experimental verification of digital substrate noise generation for
epi-type substrate” IEEE Journal of solid-state circuit vol. 35 no. 7 Jul. 2000 p. 1002-1008.

© Aalborg University / Technology Platform Section 12 / 35

 Modeling of substrate noise

High level SPICE simulations

Simplified simulation circuit

Simplified simulation circuit

• Based on simplified coupling networks

• Suitable for large scale circuits
• Extraction of the current source is needed Switching noise source

C. Soens. et al. “Modeling of Substrate Noise Generation, Isolation, and Impact for an LC-VCO and a Digital
Modem on a Lightly-Doped Substrate ” IEEE Journal of solid-state circuit vol. 41 no. 9 Sep. 2006 p. 2040-2051.

© Aalborg University / Technology Platform Section 13 / 35

 Modeling of substrate noise

Analytical model for simple ciruits

Vsub w  H (w)  I s w

Noise coupling mechanism in mixes-signal IC

Noise coupling equivalent network

M. Shen, T. Tian, J. H. Mikkelsen, and T. Larsen, “An Analytical Model for Switching Noise in Lightly
Doped Substrates," Microelectronics Journal, Pending Review.
© Aalborg University / Technology Platform Section 14 / 35
 Modeling of substrate noise

Analytical model for simple ciruits

Vsub w  H ( w)  I s w

Vdd VDD V -|V | 90%VDD

Vin (t), Vout (t)

DD tp
Vout t
In/Out put voltage
Ip t Vopsl
Vin t
Vin t Ich t 0
0 tpon tpsl tf ttw t
Vout t
Ipsat

Ip (t) [A]
In t
CL Switching current

0
0 tpon tpsl ttw t

M. Shen, T. Tian, J. H. Mikkelsen, and T. Larsen, “An Analytical Model for Switching Noise in Lightly
Doped Substrates," Microelectronics Journal, Pending Review.

© Aalborg University / Technology Platform Section 15 / 35

 Modeling of substrate noise

Analytical model for simple ciruits

Test chip

M. Shen, T. Tian, J. H. Mikkelsen, and T. Larsen, “An Analytical Model for Switching Noise in Lightly
Doped Substrates," Microelectronics Journal, Pending Review.

© Aalborg University / Technology Platform Section 16 / 35

 Modeling of substrate noise

Analytical model for simple ciruits

fclock = 20 MHz fclock = 35 MHz fclock = 50 MHz

• Provide insight into the generation and propagation of substrate noise

• It is suitable for simple digital circuits such as big size buffers
• SPICE simulation is avoided
• Not for large scale circuits
M. Shen, T. Tian, J. H. Mikkelsen, and T. Larsen, “An Analytical Model for Switching Noise in Lightly
Doped Substrates," Microelectronics Journal, Pending Review.

© Aalborg University / Technology Platform Section 17 / 35

 Modeling of substrate noise

Statistical model for large scale ciruits

• Statistical analysis is applied to switching noise

source
• Markov chain is used to model the synchronous digital
circuit.
• It is suitable for large scale circuits.
• The transfer function and switching noise source need
to be extracted

MA Mendez, D. Mateo, A. Rubio, and JL Gonzalez. Analytical and Experimental Verification of Substrate
Noise Spectrum for Mixed-Signal ICs. IEEE Trans. Circuits Syst. I, Reg. Papers vol. 53 no. 8 Aug. 2006
p.1803–1815.

© Aalborg University / Technology Platform Section 18 / 35

 Modeling of substrate noise

Statistical model for large scale ciruits

Discrete components due to the

mean of the switching current
waveform.

Continuous components due to

the random departure from the
mean of the switching current
waveform.

MA Mendez, D. Mateo, A. Rubio, and JL Gonzalez. Analytical and Experimental Verification of Substrate
Noise Spectrum for Mixed-Signal ICs. IEEE Trans. Circuits Syst. I, Reg. Papers vol. 53 no. 8 Aug. 2006
p.1803–1815.

© Aalborg University / Technology Platform Section 19 / 35

 Outline

• Introduction
• Impact of substrate noise on UWB LNA
• Modeling of substrate noise
• Measurement of substrate noise
• Isolation and suppression of substrate noise
• Conclusion and assignment

© Aalborg University / Technology Platform Section 20 / 35

 Measurement of substrate noise

DC measurement

• The purpose is to verify EM simulation

results

• The DC potential or current are

measured

• Can obtain the resistances between

contacts with different parameters,
such as contact size and distance

R. Gharpurey and R. G. Meyer, “Modeling and Analysis of Substrate Coupling in ICs”, IEEE
Journal of Solid State Circuits, vol. SC-31, n. 3, pp. 344–353, March 1996.

© Aalborg University / Technology Platform Section 21 / 35

 Measurement of substrate noise

Transient measurements

• Specific ‘sensor’ is used

• General purpose is to verify substrate coupling model or noise spectrum
• It is often a band limited measurement

M. Van Heilningen, J. Copiet, P. Wambacq, S. Donnay, M. Gngels, and I. Bolsens, “Analysis and
experimental verification of digital substrate noise generation for epi-type substrates,” IEEE J. Solid-State
Circuits, vol. 35, no. 7, pp. 1002–1008, Jul. 2000.

© Aalborg University / Technology Platform Section 22 / 35

 Measurement of substrate noise

Frequency domain measurements

• Specific ‘sensor’ is used

• Can avoid noise coupled through package

R. Gharpurey, “A Methodology for Measurement and Characterization of Substrate Noise in High-Frequency

Circuits”, in Proc. IEEE Custom Integrated Circuit Conference, pp. 487–491, May 1999.

© Aalborg University / Technology Platform Section 23 / 35

 Measurement of substrate noise

Differential noise sensor

• Specific ‘sensor’ is used

• Wide band measurement

H. Dai, K. W. Knepper, “Differential Sensing of Substrate Noise in Mixed-Signal 0.18-μm BiCMOS Technology”,
IEEE Electron Device Letters, pp. 898–901, Aug. 2008.

© Aalborg University / Technology Platform Section 24 / 35

 Measurement of substrate noise

Small signal measurements

Test structure
Measurement fixture

 Treat substrate coupling as two-port problem, with one port acting as a

noise injector and the other as a noise sensor
 Generally used to verify substrate coupling model
 It is possible to make wide band measurements up to GHz
 G-S-G probe pads also can be used to measure frequency spectrum

M. Shen, T. Tian, J. H. Mikkelsen, and T. Larsen, "A Measurement Fixture Suitable for Measuring
Substrate Noise in the UWB Frequency Band," Journal of Analog Integrated Circuits and Signal
Processing, vol. 58, no. 1, pp. 11-17, Jan. 2009.

© Aalborg University / Technology Platform Section 25 / 35

 Measurement of substrate noise

Small signal measurements

Test structure

Measured Magnitudes of S21

Equivalent circuit of the test structure

 Parasitic effects due to the ground-substrate capacitive coupling

© Aalborg University / Technology Platform Section 26 / 35

 Measurement of substrate noise

Small signal measurements

Improved measurement fixture

Measured Magnitudes of S21

Test chip

 Parasitic effects is reduced by connecting the ground to substrate

© Aalborg University / Technology Platform Section 27 / 35

 Outline

• Introduction
• Impact of substrate noise on UWB LNA
• Modeling of substrate noise
• Measurement of substrate noise
• Isolation and suppression of substrate noise
• Conclusion and assignment

© Aalborg University / Technology Platform Section 28 / 35

 Isolation and suppression of substrate noise
Passive methods

• Physical separation
• Guard rings
• N-well and Deep N-well
• Silicon on insulator P+ Guard ring

N-Well Deep N-Well P+ Guard ring+DNW

Hao-Ming Chao, Wen-Shen Wuen, and Kuei-Ann Wen. An Active Guarding Circuit Design for Wideband Substrate
Noise Suppression. IEEE Trans. Microwave Theory and Techniques, vol. 56 no. 11 p. 2609–2619, Nov. 2008.
© Aalborg University / Technology Platform Section 29 / 35
 Isolation and suppression of substrate noise
Passive methods

• Wide band suppression

• Easy to implement
• Uncertainty of isolation level
• High cost

© Aalborg University / Technology Platform Section 30 / 35

 Isolation and suppression of substrate noise
Active cancellation

The substrate noise at the

receiver is reduced by the
anti-phase signal injected
at the injection ring

(a) Top view, and (b) cross section

view of the active cancellation design
Hao-Ming Chao, Wen-Shen Wuen, and Kuei-Ann Wen. An Active Guarding Circuit Design for Wideband Substrate
Noise Suppression. IEEE Trans. Microwave Theory and Techniques, vol. 56 no. 11 p. 2609–2619, Nov. 2008.

© Aalborg University / Technology Platform Section 31 / 35

 Isolation and suppression of substrate noise
Active cancellation

• Substrate noise suppression

level of 9 dB up to 250 MHz is
achieved
• Guard rings are still needed
• Sensitive to layout and time delay
in the active guard ring circuit

Hao-Ming Chao, Wen-Shen Wuen, and Kuei-Ann Wen. An Active Guarding Circuit Design for Wideband Substrate
Noise Suppression. IEEE Trans. Microwave Theory and Techniques, vol. 56 no. 11 p. 2609–2619, Nov. 2008.

© Aalborg University / Technology Platform Section 32 / 35

 Isolation and suppression of substrate noise
Passive methods Active methods

• Wide band suppression • Has the potential of complete

• Easy to implement cancellation

• Uncertainty of isolation level • Design of active guard ring

circuit
• High cost
• Sensitive to layout and time
delay in the active guard ring
circuit

© Aalborg University / Technology Platform Section 33 / 35

 Conclusion

• Substrate noise is an important issue in mixed-signal ICs

• Analytical modeling method provides more insight than
SPICE simulations
• Passive isolation structures are good options for wide
band applications, but expensive
• Active cancellation has the potential of complete
cancellation, but sensitive to layout and time delay

© Aalborg University / Technology Platform Section 34 / 35

 Assignment

Evaluate the resistance RAB between the P+ contacts A and B shown in the
figure below. w, L and t are the width, length and thickness of the lightly
doped P substrate, respectively. The radius of the P+ contacts A and B is ra.
ρis the resistivity of the substrate. d is the distance between A and B. The
values of w, L and t are greatly larger than that of ra. The thickness of the
P+ contacts is assumed to be zero. (clue can be found in the reference at
the bottom)

S. Kristiansson, F. Ingvarson, and K. O. Jeppson. Compact Spreading Resistance Model for Rectangular Contacts
on Uniform and Epitaxial Substrates. IEEE Trans. Electron Devices, vol. 54 no. 9 Sep. 2007, p2531–2536.

© Aalborg University / Technology Platform Section 35 / 35