Nowadays the world consumption of electrical energy is continuously increasing. More than an half of the produced electricity is consumed by electric motors. In order to cope with the increase in electricity consumption, the use of variable speed motor drives is promoted by energy efficiency regulations in most countries. These motors are able to consume only as much electricity as the application actually needs. They can do this by exploiting inverterization: variable speed drive motors are driven by power semiconductor switches connected in the inverter-leg configuration. They are alternatively switched on and off by the gate driver to generate a pulse width modulated signal on motor phase nodes used to provide the required current to the load. High voltage gate drivers are usually designed to have one single output current level used to charge/discharge the parasitic gate of external power switches to turn them on and off. Driving with a switching system is very efficient, but the dissipation, even if small, is always present and must be taken into account. In recent years, superjunction (SJ) technology has revolutionized the industry of high voltage power devices significantly improving the overall performance/cost ratio of power conversion. SJ devices are able to overcome the trade-off between breakdown voltage and on resistance, better known as the Silicon Limit. However, they are characterized by a non linear Miller capacitance, which causes the switching speed transient to start with very high dV/dt and to finish with a long slow-tail in the last few volts. In standard gate driver, the io+ driving capability is selected to find the best compromise between two opposite constraints: limiting the power dissipation with a fast dV/dt while satisfying conducted and radiated emission constraints with a slow dV/dt. With SJ devices it is not possible to meet this trade-off by simply selecting a fixed io+ value. To drive power switches with high efficiency a new driving strategy to make the whole system working in the optimum self-adjusting operating point for the fast dV/dt portion and to avoid the final slow-tail is presented. The aim of this PhD project is to propose a simple closed-loop solution where no large bandwidth neither discrete elements are required and the non-linearities of the power switching devices are compensated. In this way a controlled dV/dt transient can be achieved optimizing the trade-off between switching losses and conducted and radiated emission constraints. To do so, a linear integrated HV capacitor connected between the low side and the floating high side of the gate driver is used as sensing element. The current required by the charging and discharging of this capacitor during the active switching event is proportional to the dV/dt slope. The gate current is then changed cycle-by-cycle accordingly to the slope detected to reach the target io+ value needed by the power switches and forced in real-time to a very high value to cancel the slow-tail effect. Two silicon were taped-out. A first test chip to validate the sensing circuit and to prove the effectiveness of the core idea. The measurements were very promising: most of the circuit worked as expected. For this reason, a second tape-out was made integrating the sensor in a gate driver environment with some minor modifications to improve the performance and to fix some minor bugs detected with the bench evaluation of the first silicon. The measurements related to the second silicon confirmed the effectiveness of the proposed driving technique for hard-switching inverters stages, even though the device needs further development and validation before it can be widely employed. The details of the circuit design and the complete measurement evaluation of both the two test chips will be deeply discussed in the PhD thesis.

Al giorno d'oggi il consumo mondiale di energia elettrica è in continuo aumento e più della metà dell'elettricità prodotta è consumata dai motori elettrici. Per far fronte a questo aumento, l'uso di motori a velocità variabile è fortemente incentivato nella gran parte dei paesi dalle normative sull'efficienza energetica. Questi motori sono in grado di consumare solo la quantità di elettricità necessaria. Questo è possibile grazie all’inverterizzazione: i motori a velocità variabile sono pilotati da switch di potenza collegati nella configurazione inverter-leg. Per fornire la corrente richiesta dal motore, gli switch di potenza vengono accesi e spenti alternativamente dal gate driver per generare un segnale modulato ad ampiezza di impulso nei nodi di fase del motore. I gate driver sono generalmente progettati per avere un singolo livello di corrente di uscita che viene utilizzato per caricare/scaricare le capacità parassite sulla gate dei power switch per accenderli/spegnerli. Un sistema di questo genere è molto efficiente, ma la dissipazione, anche se piccola, è sempre da considerare. Negli ultimi anni, la tecnologia superjunction (SJ) ha rivoluzionato l'industria dei dispositivi di potenza ad alta tensione migliorando il rapporto prestazioni/costi della conversione di potenza. I dispositivi SJ sono in grado di superare il Silicon Limit: il tradeoff tra tensione di breakdown e resistenza di stato on. Tuttavia, sono caratterizzati da una capacità di Miller non lineare, che fa sì che l'evento di switching inizi con dV/dt molto elevati e termini con un lunga slow-tail nelle ultime decine di volt. In un gate driver standard, il valore della corrente di gate io+ viene scelto come compromesso tra due vincoli opposti: limitare la dissipazione di potenza con un dV/dt veloce e soddisfare i vincoli di emissione condotta e radiata con un dV/dt lento. Con i dispositivi SJ non è quindi possibile usare un valore fisso di io+. Per usare i power switch in modo efficiente, viene presentato un sistema di controllo in grado di autoregolare la pendenza dV/dt, e di eliminare la slow-tail finale. Lo scopo di questo progetto di dottorato è quello di proporre una soluzione semplice ad anello chiuso dove non sia richiesta una grande larghezza di banda nè siano coinvolti elementi discreti e le non linearità dei power switch siano compensate. Si può ottenere in questo modo un dV/dt controllato che riduca la dissipazione e rispetti i vincoli di emissione condotta e radiata. Per fare ciò, come elemento di sensing viene utilizzata una capacità lineare integrata ad alta tensione collegata tra il low side e l'high side del gate driver. La corrente richiesta dalla carica e scarica di questa capacità durante l'evento di commutazione è proporzionale alla pendenza dV/dt. La corrente di gate viene quindi modificata di ciclo in ciclo in base alla pendenza rilevata per raggiungere il valore target di io+ richiesto dall'applicazione e viene forzata in tempo reale a un valore molto alto per eliminare la slow-tail negli ultimi volt. Sono stati prodotti due prototipi. Un primo chip di prova per convalidare il circuito di sensing e dimostrare l'efficacia dell'idea alla base. I risultati di misura si sono rivelati promettenti: la maggior parte del circuito ha funzionato come previsto. Per questo motivo è stato prodotto un secondo chip integrando il sensore in un gate driver con alcune piccole modifiche per migliorarne le prestazioni e per correggere alcuni difetti minori rilevati con la valutazione a banco del primo silicio. I risultati di misura relativi a questo secondo silicio hanno confermato l'efficacia della soluzione ad anello chiuso proposta, anche se prima di poter essere ampiamente utilizzato, il dispositivo necessiterà di ulteriori sviluppi e validazioni. I dettagli del progetto del circuito e la valutazione completa della misura di entrambi i prototipi saranno ampiamente discussi nel corso di questa tesi di dottorato.

(2022). Closed-loop dV/dt control solution for monolithic high voltage gate drivers. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2022).

Closed-loop dV/dt control solution for monolithic high voltage gate drivers

AROSIO, MARTINA
2022

Abstract

Nowadays the world consumption of electrical energy is continuously increasing. More than an half of the produced electricity is consumed by electric motors. In order to cope with the increase in electricity consumption, the use of variable speed motor drives is promoted by energy efficiency regulations in most countries. These motors are able to consume only as much electricity as the application actually needs. They can do this by exploiting inverterization: variable speed drive motors are driven by power semiconductor switches connected in the inverter-leg configuration. They are alternatively switched on and off by the gate driver to generate a pulse width modulated signal on motor phase nodes used to provide the required current to the load. High voltage gate drivers are usually designed to have one single output current level used to charge/discharge the parasitic gate of external power switches to turn them on and off. Driving with a switching system is very efficient, but the dissipation, even if small, is always present and must be taken into account. In recent years, superjunction (SJ) technology has revolutionized the industry of high voltage power devices significantly improving the overall performance/cost ratio of power conversion. SJ devices are able to overcome the trade-off between breakdown voltage and on resistance, better known as the Silicon Limit. However, they are characterized by a non linear Miller capacitance, which causes the switching speed transient to start with very high dV/dt and to finish with a long slow-tail in the last few volts. In standard gate driver, the io+ driving capability is selected to find the best compromise between two opposite constraints: limiting the power dissipation with a fast dV/dt while satisfying conducted and radiated emission constraints with a slow dV/dt. With SJ devices it is not possible to meet this trade-off by simply selecting a fixed io+ value. To drive power switches with high efficiency a new driving strategy to make the whole system working in the optimum self-adjusting operating point for the fast dV/dt portion and to avoid the final slow-tail is presented. The aim of this PhD project is to propose a simple closed-loop solution where no large bandwidth neither discrete elements are required and the non-linearities of the power switching devices are compensated. In this way a controlled dV/dt transient can be achieved optimizing the trade-off between switching losses and conducted and radiated emission constraints. To do so, a linear integrated HV capacitor connected between the low side and the floating high side of the gate driver is used as sensing element. The current required by the charging and discharging of this capacitor during the active switching event is proportional to the dV/dt slope. The gate current is then changed cycle-by-cycle accordingly to the slope detected to reach the target io+ value needed by the power switches and forced in real-time to a very high value to cancel the slow-tail effect. Two silicon were taped-out. A first test chip to validate the sensing circuit and to prove the effectiveness of the core idea. The measurements were very promising: most of the circuit worked as expected. For this reason, a second tape-out was made integrating the sensor in a gate driver environment with some minor modifications to improve the performance and to fix some minor bugs detected with the bench evaluation of the first silicon. The measurements related to the second silicon confirmed the effectiveness of the proposed driving technique for hard-switching inverters stages, even though the device needs further development and validation before it can be widely employed. The details of the circuit design and the complete measurement evaluation of both the two test chips will be deeply discussed in the PhD thesis.
BASCHIROTTO, ANDREA
MORINI, SERGIO
gate driver; controllo dV/dt; anello chiuso; riduzione EMI; superjunction
gate driver; dV/dt control; closed-loop; EMI reduction; superjunction
ING-INF/01 - ELETTRONICA
English
27-gen-2022
FISICA E ASTRONOMIA
34
2020/2021
embargoed_20250127
(2022). Closed-loop dV/dt control solution for monolithic high voltage gate drivers. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2022).
File in questo prodotto:
File Dimensione Formato  
phd_unimib_784479.pdf

embargo fino al 27/01/2025

Descrizione: Closed-loop dV/dt control solution for monolithic high voltage gate drivers
Tipologia di allegato: Doctoral thesis
Dimensione 16.48 MB
Formato Adobe PDF
16.48 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.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/10281/355848
Citazioni
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
Social impact