Issues with Floating-Point Arithmetic in nRF52832 Project using Segger Embedded Studio

[siltvm]

Recently, we started developing our new project using the nRF52832 as our controller.

We are using Segger Embedded Studio (5.42a) as our IDE.

My question is: Are there any issues while using float variable arithmetic operations in Segger Embedded Studio? Are there any additional changes needed to make in the IDE while using float variables?

Our project involves reading advertising packets from nearby temperature sensor beacons, extracting the temperature from these packets, and printing it to the log console. When we take the bits for temperature and humidity, convert them, and then calculate the actual temperature and humidity using float variables, we are getting junk values or unexpected results.

SDK: nRF5 SDK 17.0.2

IDE: Segger Embedded Studio 5.42a

Controller: nRF52832

[Hieu]

Hi siltvm,

There is no issue with float calculation on the nRF52832 with nRF5 SDK v17.0.2. The SDK version was released and tested with SES v5.40, but 5.42a should still work.

How are you printing the floating number to console? Could it be you are having the kind of issue discussed in this case  How to print floats with NRF_LOG macros?

Can you use the debugger to check if the values of the relevant float variables were correct?

Hieu

[siltvm]

Hieu ,

iam used printf instead of "NRF_LOG" .

i will attach the code for better understanding .  

 case BLE_GAP_EVT_ADV_REPORT:
       
          {
            // Extract advertising report data
            ble_gap_evt_adv_report_t const * p_adv_report = &p_gap_evt->params.adv_report;

            // Define the target MAC address
           // uint8_t target_mac[6] = {0xF4, 0x12, 0xFA, 0xE6, 0x4F, 0x0E};

           //if (memcmp(p_adv_report->peer_addr.addr, target_mac, sizeof(target_mac)) == 0)
    
            // Log the advertising device's address
             if ((p_adv_report->peer_addr.addr[5] == 0xBC) && (p_adv_report->peer_addr.addr[4] == 0x57)&&(p_adv_report->peer_addr.addr[3] == 0x29)&&(p_adv_report->peer_addr.addr[2] == 0x00)&&(p_adv_report->peer_addr.addr[1] == 0x93)&&(p_adv_report->peer_addr.addr[0] == 0xDA))
         {
            NRF_LOG_INFO("Device Address: %02X:%02X:%02X:%02X:%02X:%02X",
                         p_adv_report->peer_addr.addr[5],
                         p_adv_report->peer_addr.addr[4],
                         p_adv_report->peer_addr.addr[3],
                         p_adv_report->peer_addr.addr[2],
                         p_adv_report->peer_addr.addr[1],
                         p_adv_report->peer_addr.addr[0]);
            printf("Device Address: %02X:%02X:%02X:%02X:%02X:%02X\n",
                   p_adv_report->peer_addr.addr[5],
                   p_adv_report->peer_addr.addr[4],
                   p_adv_report->peer_addr.addr[3],
                   p_adv_report->peer_addr.addr[2],
                   p_adv_report->peer_addr.addr[1],
                   p_adv_report->peer_addr.addr[0]);

            //// Log the advertising data
            //NRF_LOG_INFO("Advertising Data (Length: %d):", p_adv_report->data.len);
          
            NRF_LOG_HEXDUMP_INFO(p_adv_report->data.p_data, p_adv_report->data.len);
            printf("\nAdvertising Data (Length: %d): ", p_adv_report->data.len);
            for (uint32_t i = 0; i < p_adv_report->data.len; i++)
            {
               printf("%02X ", p_adv_report->data.p_data[i]);
            }
            printf("\n");
            //temp_humidity((char *)p_adv_report->data.p_data);
            char ch_humidity[8];
            char temperature[8];
            printf("17th pos: %02X ", p_adv_report->data.p_data[17]);
            printf("18th pos: %02X ", p_adv_report->data.p_data[18]);
            int temp = (int)p_adv_report->data.p_data[17]+ (int)(p_adv_report->data.p_data[16]<< 8);
            printf("temp : %d\t",temp);
            int calcTempScaled = (temp * 100) / 256;
            printf("calcTemp (scaled): %d.%02d\n", calcTempScaled / 100, calcTempScaled % 100);
            float calcTemp = (float)temp / 256.0f;
            printf("calcTemp: %.2f\n", calcTemp);
            //sprintf(temperature," %d.%02d", calcTempScaled / 100, calcTempScaled % 100);
            //int hum = (int)p_adv_report->data.p_data[19] + (int)(p_adv_report->data.p_data[18]<< 8);
            float calcHum = (hum / 256.0f);
            sprintf(ch_humidity,"%6.1f", calcHum);
           //int calcHumScaled = (hum * 100) / 256; // Scale result to represent two decimal places
           // Format humidity as a string
           //sprintf(ch_humidity, "%d.%02d", calcHumScaled / 100, calcHumScaled % 100);

           // printf("Temperature: %s °C\t humidity : %s\n", temperature,ch_humidity);
            nrf_delay_ms(100);
          }
        }
        break;

is there any need of enabling "FPU" in seggar embedded studio . if it is needed how can i enable it ? . 

[Hieu]

siltvm,

No, you don't need to do anything special to enable FPU. It should be enabled by default. 
Regardless, its role is not to enable, but only to speed up floating point arithmetic. Even with FPU disabled, you should still have floating point arithmetic support.

The issue is still likely in printing support. You can enable it in Project Option, following this DevZone post:  Float value not getting printed