SPI VS UART For Minimal Data Transfer & Sleep

[snoopy20]

Hi devs,

I'm using the NRF52810 to occasionally reprogram a PIC16F15 series chip on boot using SPI. It's working well.

Once the NRF main routine is entered I periodically need to transfer a single byte to the PIC (TX only no confirm) using a timer and have a choice between SPI or UART.

The PIC16F supports both but UART has an advantage in that it's so slow I can sleep the PIC CPU, wake-up on INT and still capture the byte. SPI always needs the PIC clock running.

As such I'm leaning towards UART. My question is when UART EasyDMA is triggered by a timer via PPI will the sd_sleep routine get in the way? I also read UART RX takes 2.3mA of power which I certainly want to avoid, but I assume it can be shut off for TX only. Is there any other aspects I should be aware off before implementing?

Essentially what I'd like to do is for PPI to enable UART, transfer, then disable UART. If I can't enable/disable without CPU intervention and it adds quite a current penalty then the merits over sticking with SPI are less apparent.

Thanks,

Andrew

[snoopy20]

It looks like I'm sticking with SPI, there are no power advantages over UART from what I can see. This assumes the transaction is triggered by PPI via a timer and afterwards the CPU tries to enter sleep mode (the HF won't turn off until the SPI transaction is over, I think).

[Simonr]

Hi

UART with EasyDMA will draw quite a bit of current as you say, so I think I would lean towards SPI in this case as well. If you, by sd_sleep routine, mean the nRF's sleep, then that will indeed be interrupted if you're triggering UART I believe, so you won't be able to stay asleep while the UART is running. Indeed, the SPI will keep the HF clock running until it is finished.

Best regards,

Simon

[snoopy20]

For some reason I can't close based on your answer, but yes I'm sticking with SPI. There is another advantage, the slave device (in my case a PIC16) can use the pin clk and fully Sleep until the buffer is read.