Induced Gate Noise - Stanford Technology CAD Home Page

tcad.stanford.edu

Induced Gate Noise - Stanford Technology CAD Home Page

Induced Gate Noise(Physical Origin)At Low FrequencyGateSourceDrainABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


Induced Gate Noise(Physical Origin)At High FrequencyGateSourcecapacitivecouplingDrainABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversityNoticeable i g2iff > f t /10


Induced Gate Noise (Continue)(Impact on Noise Figure)MOSFET * NF 50ΩPower-Constrained LNAÆ ¼Å [dB]151051/f Noisew/ ¾ InducedGateDrainNoiseThermalNoise w/o ¾ Noise Figure [dB]54321Á =5mA = 4GHz Ò =50Åw/ ¾ w/o ¾ .010 5 10 6 10 7 10 8 10 9 10 10Frequency [Hz].00.5 1 1.5Gate Bias [V]ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity* by A.J. Scholten et al. (IEDM, 1999)


Induced Gate Noise (Continue)(SPICE Models)van der ZielBSIM3BSIM4 ¾ ¾ ¾ Ô ¾ Ê ×Ú ¾ Ê×❂ Solid Physical Basis❂ de facto Standard ❂ Released in Mar.2000❂ No Induced Gate Noise ❂ New ApproachABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


BSIM4 Noise Model(Modeling Strategy)van der ZielBSIM4 ¾ ¾ Ô ¾ Ú ¾ Ü❂ Connected to Gate❂ Connected to SourceABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversityR x >>R s is assumed


BSIM4 Noise Model(Modeling Strategy)van der ZielBSIM4 ¾ ¾ Ô ¾ Ú ¾ Ü❂ Connected to Gate❂ Frequency Dependent❂ Connected to Source❂ Frequency IndependentABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


BSIM4 Noise Model(Modeling Strategy)van der ZielBSIM4 ¾ ¾ Ô ¾ Ú ¾ Ü❂ Connected to Gate❂ Frequency Dependent❂ Correlated with i d2(3 Noise Parameters)❂ Connected to Source❂ Frequency Independent❂ Uncorrelated with i dp2(2 Noise Parameters)ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


BSIM4 Noise Model (Continue)(Arguments)• BSIM4 is easy to implement.• What is the physical relationship to the vander Ziel model ?• How to extract the parameters ?• What is the underlying assumption ?• How small is the discrepancy ?ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


BSIM4 Noise Model (Continue)(Physical Relationship to van der Ziel)van der ZielBSIM4Ô Ú Ü ×ÓÖØ ×ÓÖØ ×ÓÖØ ×ÓÖØFrom ×ÓÖØ ´½½ · ½¾µÚ ÜABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


BSIM4 Noise Model (Continue)(Physical Relationship to van der Ziel)van der ZielBSIM4Ô Ú Ü ×ÓÖØ ×ÓÖØ ×ÓÖØ ×ÓÖØFrom ×ÓÖØ ´½½ · ½¾µÚ ÜFrom ×ÓÖØABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity ´¾½ · ¾¾µÚ Ü · Ô


BSIM4 Noise Model (Continue)(Physical Relationship to van der Ziel)Exact FitÚ ¾ Ü ¾ ½½ · ½¾ ¾ ¾ Ô ¾ ¾½ · ¾¾ ¾ Ú ¾ ÜABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


BSIM4 Noise Model (Continue)(Physical Relationship to van der Ziel)Ú ¾ ÜExact Fit ¾ ½½ · ½¾ ¾ ¾ Ô ¾ ¾½ · ¾¾ ¾ Ú ¾ ÜDiscrepancy £ ¾ ´ ¾½ · ¾¾ µ £ ´ ½½ · ½¾ µ £ £ ℄3214 x 10−23 Frequency [GHz]nMOS 100/0.25ÎË=1.5VÎË=1.5VBSIM4van der Ziel0is assumed in BSIM4 0 2 4 6.ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


Model Comparison Method(Parameter Extraction Procedure)Measured Data[ÆÑÒ,ÊÒ,ÓÔØ]Extract van der Ziel[ ¾ , £ , ¾ ] Extract BSIM4[Ú ¾ Ü, ¾ Ô ]Transformation[ ¾ , £ , ¾ ] BSIM4Transformation[ÆÑÒ,ÊÒ,ÓÔØ] ZielTransformation[ÆÑÒ,ÊÒ,ÓÔØ] BSIM4ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversityComparison


Model Comparison Method (Continue)(Noisy Two-Port Theory)Cascade Connection (Chain Parameter)ÚÒÒNoiselessTwo-Port¸¾¿ ½½ ½¾ ¾½ ¾¾ ½ ¾ Ý ½ · ½¾¿ ¸ ÚÒÚ £ Ò Ú Ò £ ÒABabcdfghiejkl Stanford ÒÚ £ Ò Ò £ Ò¾ÊÑÒ ½¿Ò¾Ê Ò ÓÔØ£ ¾ÌÑÒ ½Integrated Circuits Lab ¾Ê Ò ÓÔØ Ê Ò ÓÔØ ¾Center for Integrated SystemsUniversity❂ Measured 4-Noise Parameters


Model Comparison Method (Continue)(Noisy Two-Port Theory)½Parallel Connection (Y Parameter)NoiselessTwo-Port¾ ¸¾ ¿ Ý ½½ Ý ½¾Ý ¾½ Ý ¾¾ ½ · ¾¾¿ ¸ ½ £ ½ ½ £ ¾ ¾ £ ½ ¾ £ ¾¾Ì ℄ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity❂ van der Ziel Model


Model Comparison Method (Continue)(Noisy Two-Port Theory)Series Connection (Z Parameter)Ú ½ Ú ¾ ¸NoiselessTwo-Port ¾ Þ ½½Þ½¾Þ¾½ Þ¾¾¿ ½ · ¾¾ Ú ½Ú £ ¾¿ ¸ Ú½Ú £ ½ Ú ¾Ú £ ½ Ú ¾Ú £ ¾¾Ì℄ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


Validation for Device(4 Noise Parameters)NF minR nMinimum Noise Figure [dB].1.41.210.80.60.40.2Measuredvan der ZielBSIM400 2 4 6Frequency [GHz]nMOS 100/0.25ÎË=1.5VÎË=1.5VNoise Resistance [Å].15010050Measuredvan der ZielBSIM400 2 4 6Frequency [GHz]nMOS 100/0.25ÎË=1.5VÎË=1.5VABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


Validation for Device (Continue)(4 Noise Parameters)G optB optOptimum Conductance [mS].2.521.510.5Measuredvan der ZielBSIM400 2 4 6Frequency [GHz]nMOS 100/0.25ÎË=1.5VÎË=1.5VOptimum Susceptance [mS].0−1−2−3−4−5Measuredvan der ZielBSIM4−60 2 4 6Frequency [GHz]nMOS 100/0.25ÎË=1.5VÎË=1.5VABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity


Validation for Circuits(Shunt-series Feedback Amplifier)Noise Figure [dB]108642 = 100MHzÁ = 100mA Ò = ÓÙØ =50Åvan der ZielBSIM4Ê × ÒÊ ÓÙØÊ Ä0.5 1 1.5 2 2.5ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity.~00Gate Bias [V]Æ ¸ ÆÑÒ · ´ × ÓÔص ¾ ÊÒConstant × ¾ ÊÒ××SameY s >>Y opt


Validation for Circuits (Continue)(1/g m Termination Amplifier)Noise Figure [dB]108642Ò= 50Å = 100MHzvan der ZielBSIM4Ê × Ò0.5 1 1.5 2 2.5.0Gate Bias [V]ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity❂ By the same reason.❂ Validates a non-groundedsource electrode condition.


Validation for Circuits (Continue)(Inductive Degeneration Tuned Amplifier) ÒÊ ×Ä 12Á =5mA = 4GHz10 Ò =50Å86Ä ×42van der ZielBSIM400.5 1 1.5 2 2.5Noise Figure [dB].14Gate Bias [V]ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity❂ Y opt is comparable to Y s .❂ Another non-groundedsource electrode case.


Validation for Circuits (Continue)(Tuned Amplifier with Cascode Stage)5ž ÒÊ Ä × Å½Ä ×Noise Figure [dB]4321Á =5mA = 4GHz Ò =50ÅΠ׎ =0.7Vvan der ZielBSIM4.00.6 0.8 1 1.2Gate Bias of Å ¾ [V]ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity❂ More noticeable discrepancy.


Conclusions• First independent comparison between BSIM4and van der Ziel models for induced gate noise.• BSIM4 reproduces the van der Ziel model butintroduces errors in the correlated term.• In practical circuits, noticeable errors arise forvery low gate bias conditions only.• The two models are equivalent in most practicalcircuits.ABabcdfghiejkl StanfordIntegrated Circuits LabCenter for Integrated SystemsUniversity

More magazines by this user
Similar magazines