The resolution of a spinning rotor gauge (SRG) depends primarily on the precision of measurement of the rotor deceleration rate and on the stability of the rotor residual drag. The ultimate resolution of the SRG can be achieved only when the ambient conditions allow stable residual drag. Our measurements of the residual drag were done in a sealed vacuum system with glass and stainless steel SRG thimbles and a small appendage ion pump. The system was placed in a thermostat together with the SRG suspension head. The suspension head was fixed to the laboratory wall virtually vibration free. The imprecision of the deceleration rate measurements is determined by the rotor frequency and the integration time. Old versions of SRG controllers were limited to the integration time of 30 seconds and the rotor operation frequency was preset to the narrow window of 405 to 415 Hz. Some new versions of controllers allow the extension of the integration time to 60 seconds and the operation of the rotor at frequencies up to 800 Hz. In the present study we varied the integration time and rotor frequency and observed the statistical distributions of sequential readings of the rotor residual drag. We found some instabilities of the gauge immediately after the re-acceleration of the rotor. Such instabilities were not reported previously and can affect the accuracy of pressure measurements. We also found that operation of the rotor at higher speeds increases the frequency dependence of the residual drag.