How to convert four uint16_t to double float IEEE754?

Question

I am working in a Modbus project where I have to read modbus Holding and Input registers using ESP8266, I am using ModbusMaster232 library. I returns a response buffer array of uint16_t. I need to convert the uint16_t to double precision float. A quick test in ESP8266 shows double datatype in ESP8266 stores 8 bytes.Does anybody know how to convert four uint16_t to double in Arduino. If so please share your knowledge.

Answer 1

Here's a code fragment for converting a uint16_t. Expand this to work with your array of ints from the library. Remember that on Arduino, - Uno and Mega, at least - doubles are implemented as floats. You won't get the precision you might be expecting.

double xAsDouble;    // ** only float precision despite 'double' declaration
uint16_t xAsInt;

xDouble = xInt;      // the compiler knows how to make this conversion.

Update: After re-reading your question, I realize you're using an ESP8266. My comments about Uno's float precision may not apply.

Update 2: After some discussion, I've learned that the Modbus return is not 4 integers, but a double being passed as 4 words, which need to be reassembled to recreate the original double. The solution will depend on the order in which the words comprising the original double are sent.

Update 3: I suggested the OP try using this general scheme:

// A union lets you look at memory as either (in this case) 4 ints or 1 double
union {
   uint16_t data_out[4];
   double x;
} buffer;

// Assign the data from modbus using the 'int' way of viewing menmory.
// Note I might have the ordering wrong but this is the idea:
buffer.data_out[i] = data_in[i];

Then reading 'buffer.x' would get the 4 words of memory as a double.

... After trying both orderings, the OP reports he is obtaining the expected result,