Lowering injection amplitude in sensorless control by means of current oversampling

Abstract

Sensorless control approaches for low and zero speed determine the rotor position by injecting high frequency voltage signals. Current measurement noise and quantization necessitate a minimum amount of injection amplitude, which implies an acoustic noise emission that for many applications is unacceptable. This paper proposes to sample the current with the maximum A/D conversion frequency while processing the data with an FPGA. By means of a recursive least squares method and a algorithmic extension, high precision current slopes are calculated from low injection voltage. Experimental results show that the injection magnitude can be reduced by factor 7 to 10 compared to standard PWM centered current measurement.

Silent and parameter independent hybrid sensorless control for SPMSM based on current oversampling

“…where Y Σ is the mean admittance, which according to [14] for low magnitude triangular injection must be considered a…”

“…The entries of this matrix are estimated using the vectorial mean admittance equation from [14] Ŷ Σx…”

“…The Arbitrary Injection algorithm [15] extracts this information from the current response to any active voltage with sufficient magnitude, which at standstill is ensured by means of a slight injection (without prescribed pattern). Arbitrary Injection has been realized with oversampling based current slope data in [14] before.…”

“…According to [14] switching transients disable the exploitation of the active switching states for low injection magnitude. Therefore the linear regression is only applied to the passive states, which in this condition amount almost the complete period T s and yield very accurate regression results ξ s 0 , ξ s 0 , i s o0 and i s o0 .…”

“…The Arbitrary Injection scheme [14] is naturally free of machine parameters. In order to gain a completely machine parameter independent hybrid scheme an SNR based combination technique is derived, that manages the phase over between the HF and the FM method and adjusts the injection magnitude for sufficient estimation quality.…”