I don't see that in the datasheet.
A different chip the DRV8811 in "Section 7.7 Timing Requirements" specifies a maximum of 1mS. (Time from leaving sleep til a step input is accepted.) But the DRV8811 has a separate sleep pin unlike the DRV8835 which requires the Vcc pin to be switched off to enter "sleep".
My application is ultra low power, so I will duty cycle power to the DRV8835 Vcc pin to save energy while my whole system is sleeping.
Related questions about where does the chip draw current from if pin Vcc is off. It must be from pin Vm, power supply to the motor, since that is the test condition in the datasheet (Vcc 0V, Vm 5V). If you also switch Vcc to Vm, why would the DRV8835 draw any current at all? (Instead of the specified max 95nA with Vm at 5V.) Would the time to wake be any different if Vcc and Vm were switched? (Its no big deal, 95nA is not much, the mosfet to switch Vcc and Vm could leak on that order, and a sleeping MSP430 mcu is drawing say 500 nA.)
I am prototyping with a breakout board from Polulu, which has circuitry between Vcc and pin Vm to protect Vcc from back EMF transients (I think.) Another question is: if you are careful to switch IN1 and IN2 in the right order so that you brake before applying the next step (where braking configures the H-bridge to shunt the back EMF through a section of the ground rail), is it possible you could eliminate the protection circuitry on pin Vm? (I'm not looking for a definitive answer, I can just experiment.)
The application uses a tiny stepper motor, say the size used for camera focusing.