Digitally-Controlled RF Passive Attenuator in 65 nm
`CMOS for Mobile TV Tuner ICs
`
`Ahmed Youssef and James Haslett
`Electrical and Computer Engineering Department
`University of Calgary
`Alberta, Canada
`
`Edward Youssoufian
`Newport Media Inc.
`Lake Forest, California, USA
`
`Abstract—A novel VHF/UHF passive attenuator linearization
`circuit suitable for mobile TV applications has been designed
`and implemented in 65 nm CMOS technology. The proposed
`attenuator has a wide gain range of 48 dB that can be digitally
`programmed in 3 to 6 dB steps. At every gain setting, the input
`and output of the attenuator are matched to 50 Ω to facilitate its
`integration into mobile TV tuners.
`
`INTRODUCTION
`I.
`Mobile TV is one of the latest features to be added to cell
`phones and other hand-held devices. The low cost, low power,
`and small size demands of this application have pushed
`researchers to use nanometer CMOS technologies in designing
`high performance tuner chip sets. The bulky RF filters (i.e.,
`SAW filters) usually used in traditional TV-can tuners to
`suppress far-away interferer blockers are thus not an option for
`these integrated tuners. This results in tightening the linearity
`requirement of the RF front-end needed for mobile TV
`reception, and hence demands innovative design techniques to
`adhere to the low power necessities for this application [1].
` The RF-AGC (Automatic gain control) technique has
`been proposed recently in the literature as one of the low
`power solutions that can help mobile TV receivers achieve
`their stringent linearity requirements [2]-[4]. Decreasing the
`RF gain at large input signal levels helps the receiver pass
`larger signals without any degradation in the output SNR
`(Signal-to-Noise Ratio). Although there are many mechanisms
`to vary the RF gain in receivers, the efficacy of any given
`mechanism depends on the amount of the dynamic range that
`can be achieved while decreasing the RF gain.
`This paper proposes an RF attenuator linearization circuit
`used to vary the RF gain of mobile TV receivers while
`maximizing their dynamic range. The paper describes a
`passive attenuator designed, implemented in 65 nm CMOS
`technology and characterized in the lab. Additionally, a 5 bit
`linear thermometer decoder [5] integrated in the same test chip
`is used to program the gain of the attenuator. The decoder sets
`the gain value according to the signal level received at the
`attenuator input. Also, an on-chip programmable matching
`network is used to provide a stable 50 Ω input resistance
`
`RFin
`
`LNA
`
`(a)
`
`RF Attenuator
`
`RFin
`
`LNA
`
`(b)
`
`Figure 1. RF gain control through a) a variable gain LNA or through b) RF
`programmable passive attenuator.
`
`to the mobile TV antenna for the entire gain range.
`This paper is organized as follows. Section II discusses the
`advantages of using passive gain control over active gain
`control (i.e., Variable Gain (VG) LNA) to vary the RF gain of
`a mobile TV receiver. Section III presents the proposed RF
`attenuator design and demonstrates some practical issues dealt
`with in its integration with the rest of the mobile TV system.
`Measurement results are given in Section IV, and finally
`Section V draws the conclusions.
`
`II.
`
`PASSIVE GAIN CONTOL VERSUS ACTIVE
`GAIN CONTROL
`There are several ways to achieve gain control in RF front-
`ends. Fig. 1a shows a VG-LNA used to control the RF gain,
`while Fig. 1b shows a programmable passive attenuator used
`to control the RF gain. Both techniques are capable of
`preventing a receiver from clipping at large input signal levels
`and, in theory, either one can be used to boost the linearity of a
`mobile TV tuner. However, the difference between them
`becomes clear when the receiver dynamic range (DR) is taken
`into consideration. Having the attenuator control (passive
`control) the RF gain results in a DR value that is far superior
`to
`that
`achieved when
`gain
`is
`controlled
`by
`a VG - LNA ( active control ), especially at the higher
`
`978-1-4244-5309-2/10/$26.00 ©2010 IEEE
`
`1999
`
`INTEL 1409
`
`

`

`R
`
`2R
`
`4R
`
`8R
`
`16R
`
`32R
`
`64R
`
`128R
`
`128R
`
`vcont7
`
`128x
`vcont6
`
`64x
`vcont5
`
`32x
`vcont4
`
`16x
`vcont3
`
`8x
`vcont2
`
`4x
`vcont1
`
`vcont0
`
`enable
`
`2x
`
`1x
`
`1x
`
`Clipping level for the
` LNA and the attenuator
`
`Attenuator IIP3
`
`LNA IIP3 could limit the
`receiver IIP3
`
`-10
`
`0
`
`Y
`
`10
`
`20
`
`30
`
`40
`
`X
`
`Dynamic range decreases
`nearly 25 dB in case of
`LNA gain control method
`
`Binary
`Weighted
`Attenuator
`
`Sensitivity
`IIP3
`IIP3
` Sensitivity
`
`Sensitivity
`
`Gain, dB
`
`60
`
`30
`
`0
`
`-20
`
`-30
`
`-60
`
`-90
`
`-120
`
`Power, dBm
`
`Figure 2. Simulation results show the impact of using the active gain control
`method versus the passive gain control on a receiver dynamic range.
`
`attenuation (lower gain) settings.
`The simulation results shown in Fig.2 illustrate the impact
`of using passive versus active gain control on receiver DR. In
`this simulation, the DR is assumed to be limited by third order
`nonlinearity (IIP3). When passive gain control is used, the
`clipping level and the system IIP3 improve by one dB for
`every one dB increase in the attenuation, and the dynamic
`range value is preserved. However, at certain gain settings (Y
`in Fig. 2), the system IIP3 will be limited by the attenuator
`IIP3 and therefore the DR will start to decrease. Passive gain
`control results in higher DR value than active control due to
`the fact that LNAs are generally less linear than attenuators,
`and thus DR decreases much earlier when gain is controlled
`by a VG-LNA (X in Fig.2). Therefore, using the attenuator to
`control the RF gain in this case maximizes the receiver
`dynamic range.
`
`III. RF PROGRAMMBLE PASSIVE ATTENUATOR
`The design of an RF attenuator suitable for use in mobile
`TV applications presents several challenges. Such an
`attenuator has to achieve certain characteristics so that it can
`protect the RF performance of a mobile TV receiver in the
`presence of interferer blockers as high as 0 dBm. Typically, it
`should provide from 40 dB to 50 dB gain range in steps of 3-6
`dB [6]. Also, it should have a 50 Ω input impedance to allow
`maximum power to transfer from the antenna or to provide the
`right termination for the GSM SAW filter. Additionally, it
`should provide a 50 Ω output matching so that it would not
`affect the LNA noise figure [7]. The input and output
`matching should remain constant throughout the entire gain
`range of the attenuator.
`A. Binary Weighted Passive Attenuator
`The design of the binary weighted attenuator network is
`shown in Fig. 3. The value of R = 50 Ω was chosen so that the
`output matching to the LNA would be verified for every
`attenuation setting. There are eight control bits (vcont7-
`vcont0) to program the attenuator for different gain settings.
`
`RFin
`
`2R
`
`2R
`
`Matching
` Network
`
`R
`
`m0
`
`M1
`
`M2
`
`m1
`
`enable
`
`M3
`
`C1
`
`C2
`
`C3
`
`A<4:0>
`
`Control Logic
`
`RFout
`
`50 Ω
`LNA
`Zin
`
`
`
`Figure 3. RF passive attenuator with the input matching network.
`
`
`These control bits can be set in a thermometer code fashion in
`order to achieve 6 dB attenuation steps. The highest gain
`setting (-6 dB) is when all control bits are set HIGH and the
`lowest (-48 dB) is when they are set LOW. An enable bit is
`included to activate the attenuator path when it is
`necessary to improve the receiver linearity. All bits control
`the gates of NMOS switch transistors. Switch sizes were
`selected to minimize the off-state capacitance while still
`providing a sufficiently small resistance in the on-state
`compared to the resistance being switched. Although the
`binary weighted attenuator achieves the required gain range
`and also achieves the output matching requirement, it still
`needs to provide the matching to the mobile TV antenna.
`A programmable matching network (shown in Fig. 3) was
`added at the input of the attenuator to provide the required
`input matching at different gain settings. This network can be
`programmed by two bits called m0 and m1 (see Table I).
`Adding the matching network modifies attenuation values of
`the binary weighted attenuator. However, as shown in Table I
`the attenuation step of 3 dB to 6 dB can still be achieved.
`
`2000
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`vcont
`0
`x
`1
`1
`1
`1
`1
`1
`1
`1
`0
`
`
`Gain
`(dB)
`OPEN
`-6
`-9
`-13
`-18
`-24
`-30
`-36
`-42
`-48
`
`
`TABLE I. THE GAIN VALUES OF THE RF ATTENUATOR INCLUDING THE INPUT MATCHING
`Decoder
`Thermometer Decoder Output
`Input
`
`
`vcont
`7
`x
`1
`0
`0
`0
`0
`0
`0
`0
`0
`
`
`vcont
`6
`x
`1
`1
`0
`0
`0
`0
`0
`0
`0
`
`
`vcont
`5
`x
`1
`1
`1
`0
`0
`0
`0
`0
`0
`
`
`vcont
`4
`x
`1
`1
`1
`1
`0
`0
`0
`0
`0
`
`
`vcont
`3
`x
`1
`1
`1
`1
`1
`0
`0
`0
`0
`
`
`vcont
`2
`x
`1
`1
`1
`1
`1
`1
`0
`0
`0
`
`
`vcont
`1
`x
`1
`1
`1
`1
`1
`1
`1
`0
`0
`
`
`enable
`
`0
`1
`1
`1
`1
`1
`1
`1
`1
`1
`
` A
`
`
`<4:0>
`9
`8
`7
`6
`5
`4
`3
`2
`1
`0
`
`
`
`
`
`
`
`
`
`
`
`Input
`Matching
`
`
`m0
`m1
`
`0
`0
`0
`0
`0
`1
`1
`1
`1
`1
`
`0
`0
`0
`1
`1
`1
`1
`1
`1
`1
`
`
`
`
`
`
`
`
`
`RESISTIVE
`DIVIDER
`
`Figure 4. Die photograph of the proposed RF attenuator.
`
`TABLE II. MEASUREMENT RESULTS SUMMARY.
`Technology
`65 nm CMOS
`0.05 mm2
`Die area
`Power supply
`1.2 V
`Noise Figure
`48 dB
` (NF)
`5.8 dB
`fc_low
`100 MHz
`Gain steps
` ~ 3-6 dB
`S11 for all gain settings
`<-12 dBm
`S22 for all gain settings
`<-13 dBm
`max
`In-band
`+25.3 dBm
` (IIP3)
`min
`+23 dBm
`
`max
`min
`
`
`B. The Attenuator Bandwidth
` Integrating the designed attenuator linearization circuit
`with the LNA requires some design modifications to avoid
`any undesirable interactions between the two. Connecting the
`attenuator to the LNA might create different DC paths for the
`LNA through the MOS switches M1, M2, and M3 (shown in
`Fig. 3) which might result in severe degradation of its
`performance. Therefore, AC coupling capacitors (C1, C2, and
`C3) are added to the attenuator to avoid any disturbances in
`the operating points of the LNA devices. It is of note that the
`capacitance values of these capacitors would set the lower
`frequency limit of the attenuator, which would be the VHF
`frequency for the mobile TV applications.
`
`IV. MEASUREMENTS RESULTS
`The proposed RF attenuator was fabricated in 65 nm
`CMOS technology. The MOS switches were designed taking
`their “on resistance” and their parasitic capacitance into
`consideration. The N-well-based MOS cap was used to
`implement the attenuator AC coupling caps to save the
`receiver die area. To support the VHF band, 70 pF and 30 pF
`capacitances were chosen for the attenuator (C3) and the
`matching network caps (C1&C2) respectively. The RF
`attenuator die photo is shown in Fig. 4. The fabricated chip
`consumes 0.05 mm2 of silicon.
`The attenuator die was characterized in the lab. The HP
`8753D network analyzer was used
`for S-parameter
`measurements. The output matching of the attenuator to a 50
`Ω resistance was tested by measuring S22. Fig. 5 shows the
`measured S22 for different gain code settings across the UHF
`band. Measured values of S22 were less than -13 dB for all
`gain settings across the UHF band. The same test was repeated
`to evaluate the attenuator input matching to a 50 Ω source
`resistance. Measured values of S11 were less than -12 dB for
`all gain settings across the UHF band as shown in Fig. 6.
`Noise analysis was conducted using the noise mode of the
`Agilent E4408B spectrum analyzer. Fig. 7 shows the NF
`measurements of the attenuator for all gain settings across the
`UHF band. The loss of the SMA connectors and the coax
`cable ranging from 0.7 dB to 1.2 dB was removed from the
`measurement. The NF measurements agree with
`the
`attenuation values reported in Table I.
`
`
`One of the most critical measurements for the attenuator is
`the third order nonlinearity (IIP3) since the receiver DR might
`be limited by this value at lower gain settings. A two-tone test
`was conducted by applying two tones that were spaced by 4
`MHz. Fig. 8 illustrates a two-tone test measurement for one of
`the attenuator gain settings (-6 dB gain mode). The IIP3 is
`calculated to be +25 dBm. The same test was conducted for all
`gain settings of the attenuator. It was noticed that the IIP3
`values degraded by 1 dB to 2 dB at lower gain settings (shown
`in Fig. 9). The measurement results of the RF passive
`attenuator are summarized in Table II.
`
`V. CONCLUSION
`A novel RF attenuator linearization circuit has been
`proposed to overcome the shortcomings of having the VG-
`LNA alone control the mobile TV front-end gain. The
`attenuator designed in 65 nm CMOS technology enables a low
`power, highly linear, wide dynamic range front-end realization
`with low noise figure at sensitivity level. The attenuator
`design can be scaled to any application that requires a wide
`dynamic range RF front-end.
`
`2001
`
`

`

`
`
`0
`
`-5
`
`-10
`S11,
`(dB)
`-15
`
`-20
`
`-25
`
`-30
`400.00
`
`480.00
`
`640.00
`560.00
`Frequency, MHz
`
`720.00
`
`800.00
`
`code 0
`code 1
`code 2
`code 3
`code 4.0
`code 5
`code
`code 7
`code 8
`code 9
`code 10
`code 11
`code 12
`code 13
`code 14
`
`
`
`
`Figure 6. Measured input matching (S11) for different gain code settings.
` The frequency was swept from 400 MHz to 800 MHz.
`
`
`
`IP3 ~ + 25.5dBm
`
`0
`
`10
`Input Level, dBm
`
`20
`
`30
`
`20
`
`0
`
`-20
`
`-40
`
`-60
`
`Tone Levels [input referred], dBm
`
`-80
`
`-10
`
`
`
`
`
`
`
`Figure 8. Linearity measurement for (-6 dB) gain mode.
`
`REFERENCES
`[1] T. Shinyaitos, O. AlsushiSakail, M. Okazaki, M. Nagsumim, A.
`Saitow, K. Kioi, and M. Koutani, “A Digital TV Receiver RF and BB
`Chipset with Adaptive Bias-Current Control for Mobile Applications,”
`IEEE International Solid-State Circuits Conference, San Francisco, pp.
`212-213, February 2007.
`[2] Y.J. Kim, J.W. Kim, V.N. Parkhomenko, D. Baek, J.H. Lee, E.Y. Sung,
`and I. Nam, “A Multi-Band Multi-Mode CMOS Direct-Conversion
`DVB-H Tuner,” IEEE International Solid-State Circuits Conference,
`San Francisco, pp. 2504-2513 February 2006.
`[3] M. Gupta, S. Lerstaveesin, D. Kang, and B.S. Song, “A 48-to-860MHz
`CMOS Direct-Conversion TV Tuner,” IEEE International Solid-State
`Circuits Conference, San Francisco, pp. 2013-2024, February 2007.
`[4] V. Peluso, Y. Xu, P. Gazzerro, Y. Tang, L. Liu, Z. Li, W. Xiong, and
`C. Persico, “A Dual-Channel Direct-Conversion CMOS Receiver for
`Mobile Multimedia Broadcasting,” IEEE International Solid-State
`Circuits Conference, San Francisco, pp. 2524-2533, February 2006.
`[5] G.L. Radulov, P.J. Quinn, P.C.W. van Bee, J.A. Hegt, and A.H.M. van
`Roermund, “A Binary-to-Thermometer Decoder with Built-in
`Redundancy for Improved DAC Yield,” IEEE International Symposium
`on Circuits and Systems, Kos, Greece, pp. 1414-1418, May 2006
`[6] A. Youssef, A. Ismail, and J. Haslett, “A sub – 2 dB Noise Figure LNA
`in 65 nm CMOS for Mobile TV Applications,” IEEE Radio and
`Wireless Symposium, Louisiana, Jaunuary 2010. “in press”.
`[7] A. Youssef and J. Haslett, Nanometer CMOS RFICs for Mobile TV
`Application, Springer Publisher, 2010.
`
`2002
`
`S22,
`(dB)
`
`0
`
`-5
`-10
`-15
`-20
`-25
`-30
`-35
`400.0
`
`code0
`code 1
`code 2
`code 3
`code 4
`code 5
`code 6
`code 7
`code 8
`code 9
`code 10
`code 11
`code 12
`code 13
`code 14.
`
`480.0
`
`560.0
`640.0
`Frequency, MHz
`
`720.0
`
`800.0
`
`
`Figure 5. Measured output matching (S22) for different gain code settings.
`The frequency was swept from 400 MHz to 800 MHz.
`
`
`5
`
`10
`
`15
`
`Gain Codes
`
`50
`
`40
`
`30
`
`20
`
`10
`
`0
`
`0
`
`NF, dB
`
`Figure 7. Measured NF for different gain settings.
`
`
`
`26
`
`25.5
`
`25
`IP3, dBm
`
`24.5
`
`24
`
`23.5
`
`23
`
`22.5
`
`1
`
`2
`
`3
`
`5
`4
`Code
`
`6
`
`7
`
`8
`
`Figure 9. IIP3 measurement for different gain modes.
`
`
`
`
`
`