SPI doesn't work in nrf52810 while emulating nrf52810 in nrf52832DK was successful

[MnP_Junho]

Our project is to read the gyro (lis3dh) by using SPI communication.

Since the infocenter says that using nrf52832 for developing nrf52810 is okay.

We currently developed codes by using nrf52832DK for nrf52810.

SPI was successful in the emulation and we have tried to transfer the code to the actual nrf52810 chip as the infocenter explained and correcting some bugs

emulating and transferring process:

https://infocenter.nordicsemi.com/index.jsp?topic=%2Fcom.nordic.infocenter.sdk5.v15.0.0%2Fnrf52810_user_guide.html&cp=4_0_0_5_0

correcting bugs: SPIM0 peripheral

https://devzone.nordicsemi.com/f/nordic-q-a/35145/crashing-during-the-spi-transfer-on-nrf52810

We thought that mapping the pins would be enough but it was not.

In emulation, reading the sensor was successful but in the actual nrf52810 environment, SPI reads only '0xff'

For double checking, blinking LED and BLE NUS was successful but the spi communication was failed.

How can we solve this problem?

Does the transferring procedure has some bug?

[awneil]

MnP_Junho said:

How can we solve this problem

The first step is to see what's actually happening on the wires; eg, using an oscilloscope.

And use the debugger to see what's happening in your code.

[MnP_Junho]

Thanks. Here is a schematic of our code

#define SPI_INSTANCE	  0 /**< SPI instance index. */
static const nrf_drv_spi_t spi = NRF_DRV_SPI_INSTANCE(SPI_INSTANCE);  /**< SPI instance. */
static volatile bool spi_xfer_done;  /**< Flag used to indicate that SPI instance completed the transfer. */

void spi_init(void)
{
nrf_drv_spi_config_t spi_config = NRF_DRV_SPI_DEFAULT_CONFIG;
spi_config.ss_pin = 16;
spi_config.miso_pin = 15;
spi_config.mosi_pin = 14;
spi_config.sck_pin = 12;
spi_config.mode = NRF_DRV_SPI_MODE_3;
spi_config.frequency = NRF_DRV_SPI_FREQ_4M;
APP_ERROR_CHECK(nrf_drv_spi_init(&spi, &spi_config, spi_event_handler, NULL));
}

uint8_t SPI_Read(uint8_t cmd, uint8_t cmd_len, int delay)
{
memset(m_tx_buf,0,3);
memset(m_rx_buf,0,4);

m_tx_buf[0]=cmd;
spi_xfer_done = false;
APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, m_tx_buf, cmd_len,m_rx_buf, cmd_len));
while(!spi_xfer_done); // sending command

m_tx_buf[0]=0xFF;
spi_xfer_done = false;
APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, m_tx_buf, cmd_len,m_rx_buf, cmd_len));
while(!spi_xfer_done);
return m_rx_buf[0]; // reading recieved data
}

uint8_t LIS3DH_ReadID(void)
{
uint8_t ret, cmd;

cmd=0x8F;
ret=SPI_Read( cmd, 1,DELAY_MS);

nrf_delay_ms(DELAY_MS);
return m_rx_buf[0];
}

main{

spi_init();

LIS3DH_ReadID();

	NRF_LOG_INFO("[HHC] LIS3DH ID = 0x%x.", sensorId); //returns 0xff  only..

} 

[MnP_Junho]

These are capture of osciloscope of nrf53820

SCK and CS:

we used nrf_gpio_pin_set and nrf_gpio_pin_clear to handle the cs 

SCK and MOSI : 

We send the read command (0x8f and 0xff) but there is no response from the sensor

These are signals of nrf52810 emulation

SCK and MISO (successful to read 0x33)

SCK and MOSI (sending 0x8f and 0xff)

Even if the same source code, there are some difference in SCK signal

I can't understand how to deal with it 

[awneil]

Plug a USB stick in here:

and get proper screenshots from the scope!

[MnP_Junho]

IN nrf52810 emulation on nrf52832DK

SCK and MOSI (sending 0x8f and 0xff)

SCK and MISO (successful to read 0x33)

In actual nrf52810

SCK and MOSI (sending 0x8f and 0xff)

there are a gap between sending 0x8f and 0xff

mosi signal sending 0x8f for 8 clocks are shown below

 

SCK and MISO

there are no response from the sensor

[awneil]

That's so much clearer - isn't it?!

Does your scope really only have 2 channels? Would be a lot easier to have SCK, MOSI, and MISO all together on the same screen!

So your transmission looks OK, although the nRF52810 is adding a big gap between to 2 bytes. That shouldn't matter.

But the MISO is not working in the nRF52810 case - it looks like something is pulling it down to ground?

I seem to remember there was another thread recently with a similar problem ... ?

Check your hardware.

Try disconnecting MISO from the nRF52810, and measure at the Slave.

Do the same with the nRF52832, for comparison.

[awneil]

awneil said:

I seem to remember there was another thread recently with a similar problem

Here:

https://devzone.nordicsemi.com/f/nordic-q-a/47335/problem-using-digital-io-as-miso

don't really see a clear diagnosis of the problem, though.

Like I said, it looks like something is pulling it down to ground - have you tried another pin?

[MnP_Junho]

Yes I treid it.

It seems they cannot solved the same problem in the thread TT

[awneil]

MnP_Junho said:

Yes I treid it.

What did you try?

What was the result?

[MnP_Junho]

Thanks for your support I'll do it tommorrow.

I'd tried another pin but it doesn't work. It seems very strange that spi has problem in nrf52810 whereas emulation was quite fine.

I assumed that porting emulated code into nrf52810 is easy because nrf52832 and nrf52810 is quite same architecture as said in the infocenter.

To my knowledge, it is not necessary to use pull up or pulldown in spi.

[awneil]

You need to try disconnecting MISO from the nRF, and monitoring at the Slave - that way you can tell if it's the Slave that's not sending, or the nRF that's clobbering it.

You need to know that to know how to proceed.

MnP_Junho said:

I assumed that porting emulated code into nrf52810 is easy because nrf52832 and nrf52810 is quite same architecture

Indeed.

It seems to me that the nRF52810 can be more trouble than it's worth - unless you're using huge volumes where every penny really counts.

MnP_Junho said:

it is not necessary to use pull up or pulldown in spi

Correct.

[Luccas]

In my case (https://devzone.nordicsemi.com/f/nordic-q-a/47335/problem-using-digital-io-as-miso) using an ADC pin or connecting a low series resistor (100ohms) solved the MISO problem. But as commented I didn't discovered why.