casting 4 chars to long int

I am trying to move a 4 byte char buffer into a 32bit integer. No conversion--just move it.
I have the stretch of code below:
==========================================
char buff[4];
buff[0] = 0x0B; //load buffer with number 184551986
buff[1] = 0x00;
buff[2] = 0x0A;
buff[3] = 0x32;
uint32_t adcout;

adcout = (uint32_t)*buff;
===========================================
looking at acdout in the debugger I would expect to see
0B000A32
but i only see
0000000B
It clearly is only moving the 1st byte. I thought the compiler would take the buff length into account (and the 32 bit int length).

What am I missing here?
Thanks
Fritz

You are dereferencing buff (a char*) and assigning it to your int, so you end up with just the first byte. It should be more like:

Code:

---

adcout = *(uint32_t*)buff;

---

First, char* buff is cast to uint32_t*, then that pointer is dereferenced, yielding the whole integer.

If you want the value 184551986, which is 0x0b000a32, then you probably need to store the bytes in reverse (32, 01, 00, 0b) since your machine is probably little-endian. Look it up if you don't know what that means: Endianness - Wikipedia

Thanks for your prompt response John. I am working on PIC18s with the xc compiler which are little-endian. Darn-I have had my share of reversing bytes before!
Your construct makes sense. I don't know if there is a more streamlined way of doing this (my goal of course is to convert the final ADC readings to float and display them our use them elsewhere in the code).

I closed my shop for the night but will check this out tomorrow
cheers
Fritz

Just making a uint32_t pointer to a randomly aligned char buffer risks alignment exceptions.
Many architectures don't allow you to access multi-byte data types on odd addresses.

Throw in the mix endianess, and the only portable sane way to do this is like

Code:

---

adcout = ((uint32_t)buff[3]<<24) |
        ((uint32_t)buff[2]<<16) |
        ((uint32_t)buff[1]<<8) |
        ((uint32_t)buff[0]);

---

You could do something like this:

Code:

---

union U {
  char buffer[4];
  uint32_t value;
};

union U adcout;

adcout.buffer[0] = 0xb;
adcout.buffer[1] = 0;
adcout.buffer[2] = 0xa;
adcout.buffer[3] = 0x32;

// and read from adcout.value.

---

This too will avoid the problem Salem is warning you about.

But.. you don't need that union or to use a pointer to use the 12 bits ADC from PIC18x:

Code:

---

  uint32_t value = (ADCRESH << 8) | ADCRESL;
  double flt = (double)value / ((1 << 12) - 1);

---

Make sure ADCCON0.FM bit is set.
(Assuming 'double' has 53 bits of precision here).