[nRF51422] SPI Slave solution

RE: [nRF51422] SPI Slave solution

ovrebekk — Thu, 12 Nov 2015 11:39:59 GMT

If you can do that it would definitely be the best ;)

RE: [nRF51422] SPI Slave solution

wizath — Tue, 10 Nov 2015 13:12:25 GMT

Thanks for answer, the easiest way will be to change the master side transmission.

RE: [nRF51422] SPI Slave solution

ovrebekk — Tue, 10 Nov 2015 12:44:31 GMT

The SPI is 8 bits only unfortunately. In other words you have to find some way to clock in the last 6 bits (so you get at least 16 bits in total), otherwise the data won't be written into the RX buffer. Theoretically you could do this from the nRF51 as well, and use another additional pin to feed clocks into the SPI, but you would have to find some way to switch between feeding the clock from the SPI master or the nRF51. If you can add 6 additional clock pulses on the master side that should be easier to implement.

RE: [nRF51422] SPI Slave solution

wizath — Mon, 09 Nov 2015 16:38:32 GMT

Thanks for answer, now I'm using two shorted pins and it works.

But there is my next question. I'm using spi_slave example from Keil, and trying to read ~10 bits. The issue is that the SPI amountrx reg always shows 1, so the remaining 2 got lost? Is there any way to capture all the transmission? My buffers are uint8_t arrays.

The chip select pin is toggled about 1ms after the end of transmission.

RE: [nRF51422] SPI Slave solution

ovrebekk — Mon, 09 Nov 2015 14:47:49 GMT

Unfortunately there is no way to manually start an SPI slave transaction without the physical chip select line.
The most direct way to do this would be to connect a different GPIO to CSN physically, and set/clear the output pin from your software.

This could be automated through the task/event system by using a GPIOTE channel to control the output pin.

Best regards
Torbjørn