Micro Electro-Mechanical Systems (MEMS) sensors are nowadays the main solution to build low-power microphones, occupying reduced area though featuring high-level performances. The integration of multiple MEMS microphones on portable smart devices is needed to obtain high sound quality, and this imposes the scaling of microphone package, achieved through both minimizing the sensor and scaling the readout electronics area. Current circuital solutions for audio sensors readout mainly employ switched circuits, to implement techniques such as Chopping, to reduce low frequency flicker noise strongly affecting low frequency applications such as audio systems, or transconductance amplifiers, which can reach high gain and linearity, at the expenses of increased area and power consumption. In this thesis different design solutions to build up sensor interface amplifiers for MEMS microphone readout circuits are explored. The designs exploit a scaled technology 55nm technology node, which represents a fair trade-off between production costs and achievable area reduction. The here presented topologies are two: the Super Source Follower (SSF), in both PMOS and NMOS input device versions, and the Differential Difference Amplifier (DDA). The first circuits (SSFs) are the natural evolution of the most fundamental circuital topology through which it is possible to implement constant charge readout method for capacitive microphone. The second topology (DDA) adds complexity and elegance to the whole microphone system, achieving low current consumption and versatile gain programmability, although linearity and noise have still to face a thorough optimization to perform beyond current products. However, both circuital solutions reported in this work offer considerable area and power advantages, and results to be competitive in terms of performance with respect to current solutions, although exploiting a deep sub-micron technology node as the 55nm could result disadvantageous for the design of high-performance analog cells. In fact, the choice of deep sub-micron technology nodes is fundamental for the realization of scaled and complex mixed-signal devices, but introduces critical aspects for mixed-signal interfaces, as high-level performance with low supply voltage is required. The main challenges for audio circuits are to achieve high-linearity, with Total Harmonic Distortion (THD) reaching the value of 10% (Acoustic Overload Point, AOP) for amplitude values over 130dBspl, and to show output integrated noise levels within the audio frequency range (from 20Hz to 20kHz) below -108dBV(A). Results of silicon test-chip measurements show that these specifications are mainly achieved by the proposed circuits.
I sensori basati su sistemi Micro Elettromeccanici (MEMS) sono oggigiorno la soluzione piu usata per costruire microfoni a bassa potenza, occupando area ridotta e dimostrando performance di alto livello. L’integrazione di diversi microfoni MEMS su dispositive smart portatili é necesaria per ottenere alta qualitá di suono e questo impone di scalare il package del microfono, il che é ottenuto sia minimizzando il sensore, sia scalando l’area dell’elettronica di readout. Le soluzioni circuitali attuali per costruire il readout di sensori audio usano tecniche come il Chopping, per ridurre il rumore Flicker che é fortemente presente in applicazioni a bassa frequenza come i sistemi audio, o amplificatori a transconduttanza, che possono raggiungere alti guadagni e alta linearitá a spese di incrementare area e consumo di potenza. In questa tesi vengono presentate diverse soluzioni di design per costruire amplificatori di circuiti di reaodut per interfacce di microfoni MEMS. I design sfruttano una tecnologia scalata CMOS a 55nm, che rappresenta un buon compromesso tra costi di produzione e riduzione di area ottenibile. Le topologie circuitali presentate nella tesi sono due: il Super Source Follower (SSF), in due diverse versioni, una con dispositivo di ingresso PMOS e una con dispositivo di ingresso NMOS, e il Differential Difference Amplifier (DDA). La prima topologia (SSF) é la naturale evoluzione della topologia piu semplice per implementare un readout a carica costante per microfoni capacitivi. La seconda topologia (DDA) aggiunge complessitá al sistema, ma permette di ottenere corrente ridotta e programmabilitá di guadagno, nonostante la linearitá e il livello di rumore ne risentano e debbano affrontare un’ulteriore ottimizzazione per avere performace migliori dei prodotti attuali. Entrambe le soluzioni circuitali presentate in questo lavoro offrono un considerevole vantaggio in termini di area e potenza e risultano essere competitive in termini di performance rispetto alle soluzioni attuali, nonostante l’uso di una tecnologia a canale corto cfome la 55nm possa risultare svantaggiosa per il design di circuiti analogici. La scelta di una tecnologia scalata é fondamentale per la realizzazione di dispositivi complessi e miniatirizzati, ma introduce aspetti critici nella progettazione analogica, poiché devono essere garantite performance di alto livello con tensioni di alimentazione limitate. Le principali sfide per i circuiti audio sono: raggiungere alta linearitá, con distorsione totale armonica (THD) che raggiunga il 10% (Acoustic Overload Point, AOP) per valori di ampiezza superiori a 130dBspl; avere livelli di rumore di uscita integrato in banda audio (da 20Hz a 20kHz) inferiori a -108dBV(A). I risultati di misura di test-chip mostrano che queste specifiche sono raggiunte dai circuiti progettati.
(2024). Low Noise MEMS Microphone Interfaces in 55nm CMOS Technology. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2024).
Low Noise MEMS Microphone Interfaces in 55nm CMOS Technology
BENEDINI, FEDERICA
2024
Abstract
Micro Electro-Mechanical Systems (MEMS) sensors are nowadays the main solution to build low-power microphones, occupying reduced area though featuring high-level performances. The integration of multiple MEMS microphones on portable smart devices is needed to obtain high sound quality, and this imposes the scaling of microphone package, achieved through both minimizing the sensor and scaling the readout electronics area. Current circuital solutions for audio sensors readout mainly employ switched circuits, to implement techniques such as Chopping, to reduce low frequency flicker noise strongly affecting low frequency applications such as audio systems, or transconductance amplifiers, which can reach high gain and linearity, at the expenses of increased area and power consumption. In this thesis different design solutions to build up sensor interface amplifiers for MEMS microphone readout circuits are explored. The designs exploit a scaled technology 55nm technology node, which represents a fair trade-off between production costs and achievable area reduction. The here presented topologies are two: the Super Source Follower (SSF), in both PMOS and NMOS input device versions, and the Differential Difference Amplifier (DDA). The first circuits (SSFs) are the natural evolution of the most fundamental circuital topology through which it is possible to implement constant charge readout method for capacitive microphone. The second topology (DDA) adds complexity and elegance to the whole microphone system, achieving low current consumption and versatile gain programmability, although linearity and noise have still to face a thorough optimization to perform beyond current products. However, both circuital solutions reported in this work offer considerable area and power advantages, and results to be competitive in terms of performance with respect to current solutions, although exploiting a deep sub-micron technology node as the 55nm could result disadvantageous for the design of high-performance analog cells. In fact, the choice of deep sub-micron technology nodes is fundamental for the realization of scaled and complex mixed-signal devices, but introduces critical aspects for mixed-signal interfaces, as high-level performance with low supply voltage is required. The main challenges for audio circuits are to achieve high-linearity, with Total Harmonic Distortion (THD) reaching the value of 10% (Acoustic Overload Point, AOP) for amplitude values over 130dBspl, and to show output integrated noise levels within the audio frequency range (from 20Hz to 20kHz) below -108dBV(A). Results of silicon test-chip measurements show that these specifications are mainly achieved by the proposed circuits.File | Dimensione | Formato | |
---|---|---|---|
phd_unimib_789509.pdf
embargo fino al 31/01/2027
Descrizione: Tesi di Benedini Federica - 789509
Tipologia di allegato:
Doctoral thesis
Dimensione
14.18 MB
Formato
Adobe PDF
|
14.18 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.