Problem with SPI communication read/write for MS5540C.

RE: Problem with SPI communication read/write for MS5540C.

Simonr — Mon, 09 Sep 2019 10:55:14 GMT

Hi Vishal

The wait_for_low_DOUT(); and wait_for_falling_edge_on_DOUT Hugh is describing should be very similar to your wait_for_spi_event(); I believe. Instead of the while loop waiting for the SPI transfer to be done, you should set it to wait for D OUT to go low. You should probably set the delay to a bit more than 35ms as well. Could you try with a 40-50 ms delay to see if that's successful?

Best regards,

Simon

RE: Problem with SPI communication read/write for MS5540C.

vishal — Fri, 06 Sep 2019 13:49:52 GMT

Hi,

If I change delay 200ms to 35ms then sensor is not getting correct value. If I add 200ms delay then it should fine and working and got corrected value of D1 and D2.

What I need to do in wait_for_low_DOUT(); and wait_for_falling_edge_on_DOUT(); functions.

Currently my program is working with added 200ms delay but is this good way.

Thanks....

RE: Problem with SPI communication read/write for MS5540C.

hmolesworth — Sun, 01 Sep 2019 00:41:26 GMT

One last thing, the more I look at the timing diagrams (not as clear as I would like) the more I see CPHA 1 and CPOL 0 to match the sensor timing. In the Nordic encoding this is NRF_DRV_SPI_MODE_1, ie. not Mode_0:

  spi_config.mode = NRF_DRV_SPI_MODE_1; // CPOL = 0, CPHA = 1

RE: Problem with SPI communication read/write for MS5540C.

hmolesworth — Fri, 30 Aug 2019 17:02:50 GMT

First I would not wait for the spi event, or use such a long delay, since the 33mSec conversion starts while the spi transfer is finishing up.

Change:
  nrf_delay_ms(200);
  wait_for_spi_event();
To:
  nrf_delay_ms(33 and a bit more); // 33 mSec conversion time
  wait_for_low_DOUT();             // DOUT goes low when data ready
or just:
  wait_for_falling_edge_on_DOUT(); // DOUT goes low when data ready

Next the SPI transfer is already in progress, so I would suggest the first 2 bytes of received data are the bytes making up the 16-bit value in D1 (say), not the 2nd and 3rd bytes. For there to be no issues with Tx presenting unwanted '1's while reading out Rx, which could cause erroneous operation, I would be inclined to make the transfer specifically send '0' on Tx while reading in the 2 Rx bytes

// Ensure trailing Tx bits are all '0'
spi_config.orc = 0x00;

RE: Problem with SPI communication read/write for MS5540C.

vishal — Fri, 30 Aug 2019 06:56:32 GMT

Thanks for your great explanation. Now I have modified my program snippet here is below:

unsigned int read_cal_word(const uint8_t *address, const uint8_t add_len) {
  word = 0;
  memset(m_rx_buf, 0, m_length);
  reset_sensor(); /*Before read calibration word need to reset sensor module*/
  spi_xfer_done = false;
  APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, address, add_len, m_rx_buf, sizeof(m_rx_buf)));
  wait_for_spi_event();
  NRF_LOG_FLUSH();
  NRF_LOG_INFO("RX Buff_byte1:%X byte2:%X", m_rx_buf[2], m_rx_buf[3]);
  word = (m_rx_buf[2] << 8) | m_rx_buf[3]; /*Combine received 2 bytes from sensor*/
//  NRF_LOG_DEBUG("Return Calibration word = %u", word);
  return word;
}

\\ Below function use for read D1 and D2
unsigned int read_sensor_value(const bool check_add) {
  sensor_val = 0;
  static uint8_t       add_pressure[] = { 0x0F, 0x40 }; // D1 command
  static uint8_t       add_temp[] = { 0x0F, 0x20 }; // D2 command       
  uint8_t dummy_byte = 0x00;
  memset(m_rx_buf, 0, m_length);
  reset_sensor(); /*Before read calibration word need to reset sensor module*/
  spi_xfer_done = false;
  if (check_add == true) { /*Check here address of read sensor value*/
  APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, add_pressure, sizeof(add_pressure), NULL, 0));
  } else {
  APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, add_temp, sizeof(add_temp), NULL, 0));
  }
  nrf_delay_ms(200);
  wait_for_spi_event();
  NRF_LOG_FLUSH();
  spi_xfer_done = false;
  APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, &dummy_byte, sizeof(dummy_byte), m_rx_buf, sizeof(m_rx_buf)));
  nrf_delay_ms(200);
  wait_for_spi_event();
  NRF_LOG_FLUSH();
  sensor_val = (m_rx_buf[1] << 8) | m_rx_buf[2];
  NRF_LOG_DEBUG("Sensor value MSB:%X LSB:%X return:[%u]", m_rx_buf[1], m_rx_buf[2], sensor_val);
  return sensor_val;
}

But problem is the value of D1= 58048 and D2= 63552 which is out of range of max value of sensor.

Here is my Logs:

<info> app: Level Detector started.
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  FE                     |.       
<info> app: RX Buff_byte1:D7 byte2:9C
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  FE 01 9F 8C            |....    
<info> app: RX Buff_byte1:9F byte2:8C
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  FE                     |.       
<info> app: RX Buff_byte1:C6 byte2:2E
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  FE 01 DA 5D            |...]    
<info> app: RX Buff_byte1:DA byte2:5D
<info> app: coefficients- C1: 27598 C2:2973 C3:873 C4:792 C5:638 C6:12
<info> app: Wake up sensor
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  21 E3 40               |!.@     
<info> app: Sensor value MSB:E3 LSB:40 return:[58176]
<info> app: compensated pressure value [D1]: 58176
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  3D F9 40               |=.@     
<info> app: Sensor value MSB:F9 LSB:40 return:[63808]
<info> app: compensated temperature value [D2]: 63808
<info> app: Compensated pressure in mbar: -669
<info> app: Water level: 0cm
<info> app: Real Temperature in C: 208
<info> app: Wake up sensor
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  21 E2 C0               |!..     
<info> app: Sensor value MSB:E2 LSB:C0 return:[58048]
<info> app: compensated pressure value [D1]: 58048
<info> app: Reset Sensor
<info> app:  Received:
<info> app:  3D F8 40               |=.@     
<info> app: Sensor value MSB:F8 LSB:40 return:[63552]
<info> app: compensated temperature value [D2]: 63552
<info> app: Compensated pressure in mbar: -1069
<info> app: Water level: 0cm
<info> app: Real Temperature in C: 207

How I know the measured value is corrected, because the temperature is not getting correct.

Thanks....

RE: Problem with SPI communication read/write for MS5540C.

hmolesworth — Thu, 29 Aug 2019 16:09:32 GMT

A few suggestions, if I may. First the expected command byte 0x1D is 0001'1101 in binary, which corresponds with the data sheet description of "D1" backwards. However, confusingly the actual binary code in the example for "D1" is 10-bit sequence 111'1010'000. Since you are using MSB first transfers, that requires a 10-bit sequence, say 0xF4'0, or the 16-bit sequence 000000'111'1010'000 which is 0x03, 0xD0 although the extra received data would require re-aligning. Using postfix of unwanted clocks rather than prefix would give 111'1010'000'000000 or 0xF4, 0x00. Extra 0x00 bytes would read out the data.

D2 would then be the 16-bit sequence 000000'111'1001'000 which is 0x03, 0xC8, or postfix 111'1001'000'000000 which is 0xF2, 0x00.

The 21-bit Reset sequence would be 101010101010101000000'000, which with MSB First is the 3-byte sequence 0xAA, 0xAA, 0x00. This could perhaps better be preceded with the 3 unwanted '0' bits, giving 000'101010101010101000000 or 0x15, 0x55, 0x40. This is what you are using I see.

I think my starting tests would therefore use the following:

static uint8_t reset_seq[]     = { 0x15, 0x55, 0x40 };       // 000'101010101010101000000 or 0x15, 0x55, 0x40
static uint8_t address_word1[] = { 0xF4, 0x00, 0x00, 0x00 }; // 111'1010'000'000000 or 0xF4, 0x00.
static uint8_t address_word2[] = { 0xF2, 0x00, 0x00, 0x00 }; // 111'1001'000'000000 or 0xF2, 0x00
or
static uint8_t reset_seq[]     = { 0x15, 0x55, 0x40 };       // 000'101010101010101000000 or 0x15, 0x55, 0x40
static uint8_t address_word1[] = { 0x0F, 0x40, 0x00, 0x00 }; // 0000'111'1010'000'00
static uint8_t address_word2[] = { 0x0F, 0x20, 0x00, 0x00 }; // 0000'111'1001'000'00

So not the same as your code

There is one other issue worth noting, which will break the code if anything other than 4-byte data arrays are used. Have a look at these tests, which pass:

static uint8_t address_word1[] = { 0x1D, 0x50, 0x00, 0x00 };       /**< TX buffer. */
const uint8_t *address_p = address_word1;
STATIC_ASSERT(sizeof(address_p) == 4, "Misunderstanding on meaning of sizeofaddress_p)");
STATIC_ASSERT(sizeof(address_p) == 4, "Misunderstanding on meaning of sizeof(address_p)");
STATIC_ASSERT(sizeof(address_word1) == 4, "Error in sizeof(address_word1)");

Now try using something other than a 4-byte command:

static uint8_t address_word5[] = { 0x1D, 0xA0, 0x00, 0x00, 0x99 };       /**< TX buffer. */
const uint8_t *address_p5 = address_word5;
STATIC_ASSERT(sizeof(address_p5) == 4, "Misunderstanding on meaning of sizeofaddress_p)");
STATIC_ASSERT(sizeof(address_p5) == 4, "Misunderstanding on meaning of sizeof(address_p)");
STATIC_ASSERT(sizeof(address_word5) == 5, "Error in  sizeof(address_word5)");

These tests also pass .. but in the code there is a bug:

unsigned int read_cal_word(const uint8_t *address) {
  spi_xfer_done = false;
//blah-blah
  APP_ERROR_CHECK(nrf_drv_spi_transfer(&spi, address, sizeof(address), m_rx_buf, sizeof(m_rx_buf)));

sizeof(address) is always 4 even if the passed array is (say) 5.

(edited to correct typos)

RE: Problem with SPI communication read/write for MS5540C.

vishal — Thu, 29 Aug 2019 14:13:54 GMT

[quote userid="75734" url="~/f/nordic-q-a/51565/problem-with-spi-communication-read-write-for-ms5540c/206909"]Can you point to where in the sensor's datasheet that is stated?[/quote]

I have attached screen shot below:

Also the sensor does not have SS_pin

Thanks......

RE: Problem with SPI communication read/write for MS5540C.

Simonr — Thu, 29 Aug 2019 13:58:47 GMT

Can you point to where in the sensor's datasheet that is stated? I'm not able to find it. In any case, I think you should try using the SS_Pin to see if it helps, the SS_Pin usually has to be handled somehow AFAIK.

Best regards,

Simon

RE: Problem with SPI communication read/write for MS5540C.

vishal — Thu, 29 Aug 2019 12:20:21 GMT

Okay please set it in public mode to get fast solution.

The sensor datasheet said no need to handle or use SS_Pin so that's why I have comment it.

Thanks for your quick response..!!

RE: Problem with SPI communication read/write for MS5540C.

Simonr — Thu, 29 Aug 2019 11:51:36 GMT

Hi Vishal

First of all, I think this case would be better to post publically, as there are multiple people in the community with experience with different sensors and such. So if you want, I can set it to public mode for you.

I see that you've commented out the SS_PIN in your configuration, but I can't see where you handle it manually, please indulge me.

Best regards,

Simon