TMAG5170-Q1: TMAG5170

Hi, Scott.

Thanks for the help. But why the last digit is 7 and not 5?

4^4 + 4 + 1 = 261, 261 - 256 = 5

Or am I calculating the CRC wrong?

Regards,

Alexander

I found a CRC4-ITU C implementation /https://github.com/mozram/CRC4-ITU/blob/master/crc4_itu.c/, but I'm not sure if it is correct.

Would love to see this function from TI.

Regards,

Alexander

CRC is not implemented through subtraction.  Rather, the equation specifies a binary pattern that will be iteratively used to perform a calculation against the rest of the data transfer.  In our case with X^4+X+1 this translates to 0b10011.

Code for a device with a longer polynomial is available as a reference here:

https://www.ti.com/lit/an/sbaa456/sbaa456.pdf

Note that for our TMAG5170 the initial value of the CRC result is 0b1111. 

OK.

1. What is the mode of SPI using TMAG5170?

0, 1, 2, 3?

i don't see it in the manual.

2. I applied the code

// ==========================================================================
// CRC Generation Unit - Linear Feedback Shift Register implementation
// (c) Kay Gorontzi, GHSi.de, distributed under the terms of LGPL
// ==========================================================================
char *MakeCRC(char *BitString)
   {
   static char Res[5];                                 // CRC Result
   char CRC[4];
   int  i;
   char DoInvert;
   
   for (i=0; i<4; ++i)  CRC[i] = 0;                    // Init before calculation
   
   for (i=0; i<strlen(BitString); ++i)
      {
      DoInvert = ('1'==BitString[i]) ^ CRC[3];         // XOR required?

      CRC[3] = CRC[2];
      CRC[2] = CRC[1];
      CRC[1] = CRC[0] ^ DoInvert;
      CRC[0] = DoInvert;
      }
      
   for (i=0; i<4; ++i)  Res[3-i] = CRC[i] ? '1' : '0'; // Convert binary to ASCII
   Res[4] = 0;                                         // Set string terminator

   return(Res);
   }

// A simple test driver:

#include <stdio.h>

int main()
   {
   char *Data, *Result;                                       // Declare two strings

   Data = "1101000101000111";
   Result = MakeCRC(Data);                                    // Calculate CRC
   
   printf("CRC of [%s] is [%s] with P=[10011]\n", Data, Result);
   
   return(0);
   }

of course with 0b1111

and I do not see CRC "7" from x0F00040

3. You made a new microcircuit, analog Melexis MLX90395 /https://www.melexis.com/en/product/MLX90395/Triaxis-3D-micropower-magnetometer/

But why not give users a C code (no SPI, just C code) to quickly learn about your chip?
Just reading and writing registers.

Since 2017, I have been working with a good 32-bit ADS1262 and have not seen any problems (there is a CRC as an option), but here the difficulties are out of the blue.

Regards,

Alexander

Scott,

thanks again for your help and patience.

Yes, with the addition of "0000" to x0F00040 (crcinit of course "1111") we finally got the required number "7" using the function from Dr. Gorontzi.

Now let's try to carry out a number of operations.

1. Read "register 4" 2 times (8400000E).

First reading:           40 7D 83 01

Not bad, PREV_CRC_STAT = 0, ERROR_STAT = 0, reset = 0x7D83, though CRC from Dr. Gorontzi is "0", not "1"

Second reading:      60 7D 83 81

Here is worse, ERROR_STAT = 1 ("ERROR_STAT indicates if there is any error bit flipped in the part").

Can you explain what this means ("error bit flipped in the part")?

---

2. Read "register 4" (8400000E) and "register 0" (8000000F).

First reading:           40 7D 83 01 (As in the first case)

Second reading:      E0 7D 83 81

Why 0x7D83, not 0x000 (for "register 0" reset = 0x0)?

---

3. Ok, let's try to disable CRC and read  "register 4".

0F 00 04 07                SPI returns 40 00 00 0A (CRC OK with function from Dr. Gorontzi)

84 00 00 0E                SPI returns 40 00 00 0A

---

It seems that after the transmission of the first 32 bit command SPI (realy 4x8), some more action is needed.

I tried these examples in mbed os, but I don't think it is essentialI.

Regards,

Alexander