nRF52 DK - SPI with PPI

RE: nRF52 DK - SPI with PPI

juulee — Tue, 13 Dec 2022 21:26:35 GMT

Wow, your advice helped me a lot!

Now it works. Thank you so much

RE: nRF52 DK - SPI with PPI

Einar Thorsrud — Tue, 13 Dec 2022 07:32:11 GMT

Hi,

I see. You set up the transaction one time in your DAC_init() where you call nrf_drv_spi_xfer() with the NRF_DRV_SPI_FLAG_HOLD_XFER flag. This will configure everything for a transaction that can be triggered by PPI, which it looks like you have done correctly. It will only happen once, though. To repeat it you need to set up a new transaction with nrf_drv_spi_xfer() before triggering it with PPI again. I would suggest doing this in the SPI interrupt/event handler, so that you always set up a new transaction when the previous has completed.

(It is also possible to set up several transactions in a row that can be triggered by PPI without CPU intervention using array list, which is described in this thread).

RE: nRF52 DK - SPI with PPI

juulee — Mon, 12 Dec 2022 18:05:55 GMT

Hi, it was my mistake in the text, My target is 25us but actually it looks like 20us.
I connected nRF52-DK with DAC, and I measured DAC output by oscilloscope.

From oscilloscope, I could see that value is updated every ~20us (but not accurately 20us), and it maintained similar value although I changed TIMER_DAC_TIMEOUT_US to 100 or 150.

I think it is because of "nrf_drv_spi_transfer(&spi, m_tx_buf, m_length, m_rx_buf, 0)" in spi_event_handler function. Because after spi transfer, event_handler is called and there is transfer function again, so this loop continues without PPI I guess...

But when I erase "nrf_drv_spi_transfer(&spi, m_tx_buf, m_length, m_rx_buf, 0)" in spi_event_handler, I could not see any outputs (SPI transfer)...

RE: nRF52 DK - SPI with PPI

Einar Thorsrud — Mon, 12 Dec 2022 14:05:14 GMT

Hi,

I see that in the code you have set TIMER_DAC_TIMEOUT_US to 25 while in the text you write 20 us. Is that the issue? If not, can you elaborate on how the "time scale is not correct"? Please elaborate on your measurments and what you have found when debugging.

[quote user=""]I think PPI doesn't work, maybe?[/quote]

PPI itself should not fail as it is quite simple and the timer part of this triggering the SPI transfer for every compare0 event seems correct, so I would look elsewhere first.