EEPROM write issue

[domko99]

Hi. I have an issue with EEPROM writing with sensor MLX90632SLD-DCB-000-SP. I am using framework ESP-IDF and microcontroller ESP32.

The first think I want to say is I think I am 100% able to read EEPROM memory from sensor. The normal mode program is working well and i2c_read and i2c_read32 functions dont return ret<0. Also my sensor should be supporting Extended mode because function mlx90632_init() returns ERANGE.

The problem is when I am trying to use MEDICAL mode or EXTENDED mode with appropiate code, Function mlx90632_read_temp_raw_extended() is returning ret<0. As I know in normal mode there is no need to write to EEPROM memory so the code is working.... But in other modes there is so I assume this can be an issue.

For I2C write I am using this code

int32_t mlx90632_i2c_write(int16_t register_address, uint16_t value){
    uint8_t reg_add[] = { (register_address >> 8) & 0xff, register_address & 0xff, 0, 0};
    i2c_master_write_to_device(I2C_MASTER_NUM, MLX90632_I2C_ADDR,reg_add,sizeof(reg_add),1000 / portTICK_PERIOD_MS);
    uint8_t data[2] = { 0 };
    data[0] = value >> 8;
    data[1] = value;
    uint8_t write_buf[] = { (register_address >> 8) & 0xff, register_address & 0xff , data[1], data[0]};
    i2c_master_write_to_device(I2C_MASTER_NUM, MLX90632_I2C_ADDR,write_buf,sizeof(write_buf),1000 / portTICK_PERIOD_MS);
    return 0;
}

and I think I am doing everything what is said in datasheet, erasing memory before write and then write to memory.

Code which I am using to implement extended mode is:

    // i2c init
    mlx90632_read_eeprom(&PR, &PG, &PO, &PT, &Ea, &Eb, &Fa, &Fb, &Ga, &Gb, &Ha, &Hb, &Ka);
    mlx90632_init();                    // returns ERANGE
    mlx90632_set_meas_type(MLX90632_MTYP_EXTENDED);
    mlx90632_start_measurement();
    mlx90632_read_temp_raw_extended(&ambient_new_raw, &ambient_old_raw, &object_new_raw);  // returns ret<0
    /* Now start calculations (no more i2c accesses) */
    /* Calculate ambient temperature */
    ambient = mlx90632_calc_temp_ambient_extended(ambient_new_raw, ambient_old_raw, 
                                                  PT, PR, PG, PO, Gb);
    /* Get preprocessed temperatures needed for object temperature calculation */
    pre_ambient = mlx90632_preprocess_temp_ambient_extended(ambient_new_raw,
                                                                   ambient_old_raw, Gb);
    pre_object = mlx90632_preprocess_temp_object_extended(object_new_raw, ambient_new_raw,
                                                                 ambient_old_raw, Ka);
    /* Calculate object temperature assuming the reflected temperature equals ambient*/
    object = mlx90632_calc_temp_object_extended(pre_object, pre_ambient, ambient, Ea, Eb, Ga, Fa, Fb, Ha, Hb);

I think no function is missing to successfully execute extended measurement. The return value of object and ambient variables is "nan". Also it is taking a long time to get this output (I mean "nan"). And when I want to execute new code with normal mode measurement I need to reset power supply of external PCB board with MLX90632 and after that sensor is producing good data of temperature. I was also looking at example code available here: https://github.com/melexis/mlx90632-example and there is no special implementation of mlx90632_i2c_write function.

If there is someone who can help to solve this problem or had a similar problem and was able to solve it I would be happy for any help. Thanks.

[Letme]

You didn't mention if you defined BITS_PER_LONG = 32 as a compiler flag (-D)? I think it is a 32bit processor right...

[Letme]

Also if you look at exapmle code for extended mode in readme, you have the start_measurement() function before the read_temp_raw_extended. That could be a reason for the return value (it would be cool to know which it is - might be ETIMEOUT?). So first you can remove that function and second you can print the return code.

[domko99]

Hi. I deleted start_measurement() function and the problem is the same. The return value of read_temp_raw_extended is -116. I think it is 116 so it should be ETIMEOUT. And I didnt define BITS_PER_LONG = 32 as a compiler flag. I just define it in mlx90632.h library #define BITS_PER_LONG 32. I dont know how to do it in esp-idf framework but when it is needed to add a flag for bits per long I will try to find out how. Thanks.

[domko99]

And is it necessary to add BITS_PER_LONG = 32 compiler flag? And is it needed only for source files in melexis component (mlx90632.c and mlx90632_extended_meas.c)? Thanks.

[domko99]

Added BITS_PER_LONG = 32 as compiler flag didnt help. I have got pull-up resistors for I2C wires with value of 10 kOhm (integrated in PCB) and in datasheet there is pull-up resistors value of 1 kOhm. Can be this an issue?

[Letme]

I don't think so, your communication should be fine at least you wrote above you received register values. How about making a longer delay with that mdelay function?

[Letme]

Any updates on this? I hope you solved the issue meanwhile...

[mm66]

Hello, I'm having similar problems, mlx90632_read_temp_raw_extended() is also returning ret<0, also it seems that measurement type does not get set correctly, i call mlx90632_get_meas_type() and it seems it is still set to default even though i set it to mlx90632_set_meas_type(MLX90632_MTYP_EXTENDED) beforehand. I set the BITS_PER_LONG to 32 for my system. I don't think there is a problem with my i2c functions since i can read good temperature values in normal mode from the chip as well as calibration constants.

[Letme]

Your write function has correct endianness for the sensor? Are you able to check stuff with scope?

Based on your comment it seems write is not working.

[mm66]

It does have correct endiannes, I mean as I said, measuring in standard mode works good for me. I can try checking what happens with scope. Can you confirm that eeprom version number 0x8505 supports extended mode?

[Letme]

Yes, I also assume Init function also returns ERANGE right?
If you can measure in standard mode that does not mean write function works as expected - because you do not write much unless you work in burst mode for example. That is why I asked. Can you change refresh_rate? Then with burst mode you would see the difference without the scope....