Chapter 5 Circuit Designs and Implementation
5.2 Simulation Results
For comparison reason, the structure and layout of NMOS with deep n-well used as varactor in VCO2 is different from that of NMOS with deep n-well used as transistor in the given 0.25um CMOS process. Therefore, there is no model for simulation. The simulated results of VCO1 and VCO3 are summarized in Table 5.1 and Table 5.3 respectively. It should be noted that the simulated phase noise of VCO1 is lower than that of VCO3. The simulation tool is Agilent Advance Design System.
5.3 Layout Designs
Fig. 5.5 shows the layout of VCO1. The total area is 1110um x 1000um. The arrangement of the component devices is spiral inductors, all transistors, and varactors respectively from the top to the bottom of the layout. The symmetry of the layout is well considered. In order to conform the specific layout rules of the on-wafer measurement in National Nano Device Laboratories (NDL), the RF GSG pads of output signals are arranged on left and right side, the RF GSG pads of input signal used as substrate noise are arranged on top side, the PGPPGP DC pads are arranged on bottom side, and the single DC pad on top side is connected to the ground node of the spiral inductors.
Fig. 5.6 shows the layout of VCO2. The total area is 1110um x 1000um. The arrangement of the layout is similar to that of VCO1.
Fig. 5.7 shows the layout of VCO3. The total area is 1110um x 1000um. The arrangement of the component devices is spiral inductors, all transistors, and varactors respectively from the top to the bottom of the layout. The symmetry of the layout is well considered. In order to conform the specific layout rules of the on-wafer measurement in National Nano Device Laboratories (NDL), the RF GSG pads of output signals are arranged on left and right side, and the PGPPGP DC pads are arranged on bottom side.
5.4 Measurement Results
The output spectrums and the phase noise of the VCO circuits are measured using Agilent E4407B spectrum analyzer. The RF signal injected into the substrate as substrate noise is provided by signal generator.
Fig. 5.8 shows the measured output spectrum of VCO1 at 1.9358-GHz oscillation frequency. Fig. 5.9 shows the measured and simulated oscillation frequency of VCO1 versus the control voltage Vcont. Fig. 5.10 shows the measured
phase noise of VCO1 at 1.9368-GHz carrier frequency. The phase noise at 1-MHz offset from the carrier is -93.39dBc/Hz. The phase noise at 100-KHz offset from the carrier is estimated at about -70dBc/Hz according to Fig. 5.10.
Fig. 5.11 shows the measured output spectrum of VCO2 at 2.548-GHz oscillation frequency. Fig. 5.12 shows the measured oscillation frequency of VCO2 versus the control voltage . Fig. 5.13 shows the measured phase noise of VCO2 at 2.5493-GHz carrier frequency. The phase noise at 1-MHz offset from the carrier is -98.56dBc/Hz. The phase noise at 100-KHz offset from the carrier is estimated at about -85dBc/Hz according to Fig. 5.13.
Vcont
Fig. 5.14 shows the measured output spectrum of VCO3 at 2.2327-GHz oscillation frequency. Fig. 5.15 shows the measured and simulated oscillation frequency of VCO3 versus the control voltage . Fig. 5.16 shows the measured phase noise of VCO3 at 2.4333-GHz carrier frequency. The phase noise at 1-MHz offset from the carrier is -102.82dBc/Hz. The phase noise at 100-KHz offset from the carrier is estimated at about -80dBc/Hz according to Fig. 5.16.
Vcont
The measured and simulated results of VCO1, VCO2, and VCO3 are summarized in Table 5.1, Table 5.2, and Table 5.3, respectively. It should be noted that the measured phase noise is much higher than the simulated phase noise. This problem is attributed to the extra noise on DC pads generating by current switching through the parasitics of DC probes. However, the DC voltage sources in the simulator are ideal and noiseless. Thus, the extra noise on DC pads deteriorates the measured phase noise. The parasitics of DC probes also results in the shift of the measured oscillation frequency from the simulated oscillation frequency. This problem can be improved by adding bypass capacitors between DC pads and ground pads in the layouts. As shown in Table 5.1 and Table 5.3, the simulated phase noise of VCO1 is lower than that of VCO3 because VCO1 has fewer constituent devices and
thus less noise source. However, the measured phase noise of VCO1 is higher than that of VCO3. This problem is attributed to the higher VCO gain of VCO1. The higher VCO gain makes VCO1 more sensitive to the extra noise on DC pads. Table 5.1 shows that the frequency tuning range of VCO1 is 548MHz, at 1.3-V control voltage range. Table 5.3 shows that the frequency tuning range of VCO3 is 505MHz, at 2.5-V control voltage range.
In VCO1, at 2.499-GHz oscillation frequency, the measured output spectrum without substrate noise injection and the measured output spectrum with substrate noise injection are shown in Fig. 5.17.
In VCO2, at 2.5713-GHz oscillation frequency when the deep n-wells of NMOS varactors are biased at 2.5V, the measured output spectrum without substrate noise injection and the measured output spectrum with substrate noise injection are shown in Fig. 5.18. At 2.5692-GHz oscillation frequency when the deep n-wells of NMOS varactors are biased at 0V, the measured output spectrum without substrate noise injection and the measured output spectrum with substrate noise injection are shown in Fig. 5.19. At 2.5686-GHz oscillation frequency when the deep n-wells of NMOS varactors are floating, the measured output spectrum without substrate noise injection and the measured output spectrum with substrate noise injection are shown in Fig. 5.20. It is noted here that the oscillation frequency shifts when the deep n-wells are biased at different conditions (without substrate noise injection). Therefore, the shift of oscillation frequency doesn’t result from substrate noise injection. When power supply is off, the measured output spectrum with substrate noise injection is shown in Fig. 5.21. This shows that the injected substrate noise travels through the substrate and then couples to the output pads.
A 2.6-GHz signal with 0-dBm power provided by signal generator is injected into the substrate as substrate noise. It is obvious that “noise folding” happens when
substrate noise is injected. As mentioned in section 4.2.2, the magnitudes of the components at and depends on the noise shaping properties of VCO circuits. Noise shaping property is relative to the quality factor of LC tank. VCO1 and VCO2 differ in the type of varactor. Thus, they have different noise shaping properties.
ωn 2ω0−ωn
In VCO2, the magnitudes of the components at 2ω0 −ωn in Figs. 5.18, 5.19, and 5.20 are almost equal. This shows that different bias conditions of the deep n-wells have little influence on the magnitudes of the components at and . As shown in Fig. 3.4, deep n-well biased at 2.5V has best isolation capability, and floating deep n-well has worst isolation capability. However, the difference of their measured is small at high frequency (>1GHz).
ωn n
0 ω
2ω −
S21
LO+ LO-Bias Tee Bias Tee
Vcont 1.25V VDD=
b GND
V = GND
V b =
On-chip
Mp Mp Mp Mp
var var
Ls Ls
Fig. 5.1 The circuit topology of VCO1 and VCO2.
Fig. 5.2 The top view and physical dimension of spiral inductor.
VDD VDD VDD VDD
Vcont
p+ guard ring
p+ ring
GND RF
signal
Fig. 5.3 The circuit topology of VCO1 and VCO2 with substrate noise injection, and the Bias Tees are not drawn.
DD 2.5V
V =
b 2.5V V = b 2.5V
V =
Vcont
Bias Tee Bias Tee
LO+
LO-On-chip
Mp Mp
Ls Ls
var var
Mn Mn
Mn Mn
Fig. 5.4 The circuit topology of VCO3.
Vcont VDD
Bias Tee VDD
LO+ Bias Tee
VDD
LO-VcontGND VDD GND
n-type MOS varactors
transistors spiral inductors
Fig. 5.7 The layout of VCO3.
Fig. 5.8 The measured output spectrum of VCO1 at 1.9358-GHz oscillation frequency.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.9
2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7
Oscillation frequency [GHz]
Vcont [V]
Measurement Simulation
Fig. 5.9 The measured and simulated oscillation frequency of VCO1 versus the control voltageVcont.
Fig. 5.10 The measured phase noise of VCO1 at 1.9368-GHz carrier frequency.
Fig. 5.11 The measured output spectrum of VCO2 at 2.548-GHz oscillation frequency.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 2.54
2.55 2.56 2.57 2.58 2.59 2.60 2.61 2.62 2.63
Oscillation frequency [GHz]
Vcont [V]
Fig. 5.12 The measured oscillation frequency of VCO2 versus the control voltageVcont.
Fig. 5.13 The measured phase noise of VCO2 at 2.5493-GHz carrier frequency.
Fig. 5.14 The measured output spectrum of VCO3 at 2.2327-GHz oscillation frequency.
0.0 0.5 1.0 1.5 2.0 2.5 2.2
2.3 2.4 2.5 2.6 2.7 2.8 2.9
Oscillation frequency [GHz]
Vcont [V]
Measurement Simulation
Fig. 5.15 The measured and simulated oscillation frequency of VCO3 versus the control voltageVcont.
Fig. 5.16 The measured phase noise of VCO3 at 2.4333-GHz carrier frequency.
2.499GHz (fo) -1.683dBm
(a)
2.6GHz (fn) 2.499GHz
(fo) 2.398GHz
(2fo-fn) -49.23dBm
-22.57dBm -1.723dBm
(b)
Fig. 5.17 In VCO1, at 2.499-GHz oscillation frequency, (a) the measured output spectrum without substrate noise injection and (b) the measured output spectrum with substrate noise injection.
2.5713GHz (fo) -2.817dBm
(a)
2.6GHz (fn) 2.543GHz
(2fo-fn)
2.5713GHz (fo)
-21.69dBm -32.28dBm
-2.97dBm
(b)
Fig. 5.18 In VCO2, at 2.5713-GHz oscillation frequency when the deep n-wells of NMOS varactors are biased at 2.5V, (a) the measured output spectrum without substrate noise injection and (b) the measured output spectrum with substrate noise injection.
2.5692GHz (fo) -2.741dBm
(a)
2.6GHz (fn) 2.5382GHz
(2fo-fn) 2.5692GHz (fo)
-21.01dBm
-33.28dBm
-2.61dBm
(b)
Fig. 5.19 In VCO2, at 2.5692-GHz oscillation frequency when the deep n-wells of NMOS varactors are biased at 0V, (a) the measured output spectrum without substrate noise injection and (b) the measured output spectrum with substrate noise injection.
2.5686GHz (fo) -2.785dBm
(a)
2.6GHz (fn) 2.5371GHz
(2fo-fn)
2.5686GHz (fo)
-21.75dBm -33.68dBm
-2.72dBm
(b)
Fig. 5.20 In VCO2, at 2.5686-GHz oscillation frequency when the deep n-wells of NMOS varactors are floating, (a) the measured output spectrum without substrate noise injection and (b) the measured output spectrum with substrate noise injection.
-22.63dBm
2.6GHz
Fig. 5.21 In VCO2, when power supply is off, the measured output spectrum with substrate noise injection.
Table 5.1 A summary of the measured and simulated results of VCO1.(*estimated) Simulation Measurement
Power supply 1.25V 1.25V
Control voltage 0~1.3V 0~1.3V
Frequency range 2.187~2.706GHz 1.936~2.484GHz
Tuning range 519MHz 548MHz
Phase noise@100KHz -106.86dBc/Hz * -70dBc/Hz Phase noise@1MHz -126.86dBc/Hz -93.39dBc/Hz
VCO bias current 9mA 11mA
Table 5.2 A summary of the measured results of VCO2.(*estimated) Measurement
Power supply 1.25V
Control voltage 0~1.3V
Frequency range 2.548~2.615GHz
Tuning range 67MHz
Phase noise@100KHz *-85dBc/Hz
Phase noise@1MHz -98.56dBc/Hz
VCO bias current 11mA
Table 5.3 A summary of the measured and simulated results of VCO3.(*estimated) Simulation Measurement
Power supply 2.5V 2.5V
Control voltage 0~2.5V 0~2.5V
Frequency range 2.242~2.84GHz 2.233~2.738GHz
Tuning range 598MHz 505MHz
Phase noise@100KHz -93.24dBc/Hz * -80dBc/Hz Phase noise@1MHz -120.66dBc/Hz -102.82dBc/Hz
VCO bias current 8mA 8mA
Chapter 6
Conclusions and Future Works
6.1 Conclusions
Isolation capabilities of two different structures for substrate noise are compared. One structure is n-well on substrate, and the other structure is p-well surrounded by deep n-well and n-well. The measured results show that they have almost equal isolation capabilities from 1GHz to 10GHz.
Three 2.4GHz LC VCOs are realized in a 0.25-um CMOS process. VCO1 (n-type MOS varactor) and VCO2 (NMOS varactor with deep n-well) differ only in the type of varactor. The influences of substrate noise coupling to the varactors on the output spectrums of VCO1 and VCO2 are investigated. The measured output spectrums show that “noise folding” happens when a RF signal provided by signal generator is injected into the substrate as substrate noise. VCO1 (PMOS-only topology) and VCO3 (complementary topology) differ in the type of circuit topology.
In simulation, the DC voltage sources are ideal and noiseless. VCO3 has higher simulated phase noise than VCO1 due to more transistors and thus more noise sources.
However, VCO3 has lower measured phase noise than VCO1 due to smaller VCO gain and thus less sensitivity to the extra noise on DC pads. The extra noise on DC pads is generated by current switching through the parasitics of DC probes.
6.2 Future works
At higher frequency, it is more difficult to isolate substrate noise by wells and guard rings. Therefore, substrate noise coupling is inevitable. The measured results show that “noise folding” happens when substrate noise couples to the constituent
devices of LC VCOs. If the frequency of substrate noise is close to the oscillation frequency of LC VCO, the phase noise at small offset is deteriorated. However, the influence of substrate noise coupling on the output spectrums of LC VCOs can be decreased by increasing the overall quality factor of LC tank. As mentioned in section 4.2.2.1, the magnitude of the unwanted components generated through “noise folding”
depends on the noise shaping property of the VCO. The noise shaping properties of LC VCOs is determined by the overall quality factor of LC tank. The higher the overall quality factor of LC tank, the sharper the noise shaping is. Thus, on chip inductors and varactors with high quality factor are required to suppress the influence of substrate noise coupling.
References
[1] B. Razavi, RF Microelectronics, Prentice Hall PTR, pp. 248, 1998.
[2] B. Razavi, Design of Analog CMOS Integrated Circuits, McGRAW-HILL, pp. 574, 2001.
[3] J. Maget, M. Tiebout, and R. Kraus, “Influence of novel MOS varactors on the performance of a fully integrated UMTS VCO in standard 0.25-um CMOS technology,” IEEE J. Solid-State Circuits, vol. 37, no. 7, pp. 953-958, July 2002.
[4] F. Svelto, P. Erratico, S. Manzini, and R. Castello, “A metal-oxide-semiconductor varactor,” IEEE Electron Device Letters, vol. 20, pp. 164-166, Apr. 1999.
[5] A.-S. Porret, T. Melly, C.-C. Enz, and E.-A. Vittoz, “Design of high-Q varactors for low-power wireless applications using a standard CMOS process,” IEEE J.
Solid-State Circuits, vol. 35, no. 3, pp. 337-345, Mar. 2000.
[6] P. Andreani, and S. Mattisson, “On the use of MOS varactors in RF VCO’s,” IEEE J. Solid-State Circuits, vol. 35, no. 6, pp. 905-910, June 2000.
[7] K. Molnar, G. Rappitsch, Z. Huszka, and E. Seebacher, “MOS varactor modeling with a subcircuit utilizing the BSIM3v3 model,” IEEE Trans. Electron Devices, vol.
49, no. 7, pp. 1206-1211, July 2002.
[8] H. Cho, and D.-E. Burk, “A three-step method for the de-embedding of high frequency S-parameter measurements,” IEEE Trans. Electron Devices, vol. 38, no. 6, pp. 1371-1375, June 1991.
[9] H. Cho, and D.-E. Burk, “Improved three-step de-embedding method to accurately account for the influence of pad parasitics in silicon on-wafer RF test-structures,”
IEEE Trans. Electron Devices, vol. 48, no. 4, pp. 737-742, Apr. 2001.
[10] B. Razavi, RF Microelectronics, Prentice Hall PTR, pp. 207, 1998.
[11] B. Razavi, “A study of phase noise in CMOS oscillators,” IEEE J. Solid-State
Circuits, vol. 31, pp. 331-343, Mar. 1996.
[12] B. Razavi, Design of Analog CMOS Integrated Circuits, McGRAW-HILL, pp.525-529, 2001.
[13] N. Fong, J.-O. Plouchart, N. Zamdmer, D. Liu, L. Wagner, C. Plett, and G. Tarr,
“A low-voltage multi-GHz VCO with 58.7% tuning range in SOI CMOS,” in Proc.
IEEE Custom Integrated Circuits Conf, pp. 423-426, 2002.
[14] E. J. Baghdady, R. N. Lincoln, and B. D. Nelin, “Short-term frequency stability:
Characterization, theory, and measurement,” Proc. IEEE, vol. 53, pp. 704-722, July 1965.
[15] L. S. Cutler and C. L. Searle, “Some aspects of the theory and measurement of frequency fluctuations in frequency standards,” Proc. IEEE, vol. 54, pp. 136-154, Feb.
1966.
[16] D. B. Leeson, “A simple model of feedback oscillator noises spectrum,” Proc.
IEEE, vol. 54, pp. 329-330, Feb. 1966.
[17] A. Hajimiri and T. H. Lee, “A general theory of phase noise in electrical oscillators,” IEEE J. Solid-State Circuits, vol. 33, no. 2, pp. 179-194, Feb. 1998.
[18] E. A. M. Klumperink, S. L. J. Gierkink, A. P. van der Wel, and B. Nauta,
“Reducing MOSFET 1/f noise and power consumption by switched biasing,” IEEE J.
Solid-State Circuits, vol. 35, no. 7, pp. 994-1001, July 2000.
[19] S. Levantino, C. Samori, A. Bonfanti, S. L. J. Gierkink, A. L. Lacaita, and V.
Boccuzzi, “Frequency dependence on bias current in 5-GHz CMOS VCOs: impact on tuning range and flicker noise upconversion,” IEEE J. Solid-State Circuits, vol. 37, no.
8, pp. 1003-1011, Aug. 2002.
[20] N. Fong, J.-O. Plouchart, N. Zamdmer, D. Liu, L. Wagner, C. Plett, and G. Tarr,
“A 40 GHz VCO with 9 to 15% tuning in 0.13um SOI CMOS,” in Symp. VLSI Technology Dig. Tech. Papers, pp. 186-189, June 2002.