DRV8243-Q1: whether Register only control is feasible?

Hi Shinya,

Thanks for your reply!

Currently they plan to use 10kHz PWM frequency, is it ok?

In addition, you don't answer some questions. How to connect IN1, IN2 and DRVOFF if the customer just use register only control? Float?

Thanks & Best Regards,

Sherry

Hi Shinya,

Thanks for your reply! But I don't understand below description. Could you please help clarify it? In my understanding, if PWM frequency is 20kHz, then MCU must keep sending at least two command in 50us which is more strict. 

Shinya Morita said:

10KHz PWM should have two toggle in 100us. Each SPI comment is ~2us period with 10MHz clock. MCU mush keep sending at least two command in 100us.

Why this delay will impact the PWM frequency and duty cycle? For example, 10KHz PWM should have two toggle in 100us, and duty cycle is 50%. Each toggle needs to add 2us SPI command time, so high time of PWM will be 48us+2us, and low time of PWM is 48us, so the PWM frequency and duty cycle is changed, right? If so, the higher the frequency, the lower the accuracy. Because if PWM frequency is 20kHz,  the high time is 23us+2us and the low time is 23us.

Shinya Morita said:

SPI communication delay makes less accuracy of PWM frequency and duty cycle.

Thanks & Best Regards,

Sherry

Hi Shinya,

Can you give me the reply today? Thanks!

Thanks & Best Regards,

Sherry

Hi Sherry,

Sherry Liang said:

My customer hopes to use Register only control method to drive BDC. According to below description, I think it is feasible, and the register only control method can achieve the exactly same function with pin only control method, right? If they use this method, they can float IN1, IN2 and DRVOFF pins, right? Is there any risks here?

In addition, how to connect nFault pin if not used?

I think it depends on the application of the motor control -- simple like running a fan or high performance like 10bit pwm servo control.

I would only use SPI_pin control  for S_EN_IN1, S_PH_IN2 only for simple motor application as it requires the host processor to write twice to the SPI_IN register per every pwm cycle. Imagine for a typical servo closed-loop control, I would use 10bit PWM and 20Khz pwm, then every the host needs to write 2 times every 50us to the SPI_IN. This doesn't sound too bad, but consider this worst case: the PID control loop needs to output a pwm value of 1 (100% is 0x3FF for 10bit pwm), so the host first write the bit to 1 at time t1, then it needs to toggle the bit to zero at t2, which is 97ns after t1. The host needs to run a Timer counter interrupt every 97ns to trigger a SPI write. Do you want to do this?

I don't know why TI engineers didn't add a PWM register for the SPI version, either 10bit or 8bit register, then the driver chip just has an internal pwm generator taking the 8bit or 10bit input via the SPI. This helps off load the host from writing the SPI_bit .

Yes, IN1, IN2 and DRVOFF can be float as they have internal pull up or pull down.

Since nFault is an output, you don't need to do anything if you want to read the nFault status using SPI.

Brian

Hi Ryan,

If my understanding is correct, then below description is wrong. The higher the frequency, the lower the accuracy, then 10kHz is more accurate than 20kHz., is it correct?

Shinya Morita said:

If PWM operation and PWM frequency is faster (e.g 20KHz), SPI communication delay makes less accuracy of PWM frequency and duty cycle.

Thanks & Best Regards,

Sherry

Hi Brian,

I am not very clear about below clarification. How does 97ns come from? Is it included into 2us SPI command replay?

Brian Dang said:

but consider this worst case: the PID control loop needs to output a pwm value of 1 (100% is 0x3FF for 10bit pwm), so the host first write the bit to 1 at time t1, then it needs to toggle the bit to zero at t2, which is 97ns after t1. The host needs to run a Timer counter interrupt every 97ns to trigger a SPI write. Do you want to do this?

Thanks & Best Regards,

Sherry