Another change I might add to this pull request is to not make read_68 return anything at all and not check for a PEC mismatch. Reasoning for that is all the functions that use read_68 they calculate the PEC again for each IC in the daisy chain and update the pec_match variable which read_68 does not.

I tried it and found that it would reduce the program size by an additional 68 bytes. Bringing the total reduction to 198 bytes.

Then we can define the read_68 in a more narrow way: it only reads all the received data and stores it into an array and nothing else.

It would look like this (reduced from 40 lines to 16 lines of code):

void read_68( uint8_t total_ic, // Number of ICs in the system 
				uint8_t tx_cmd[2], // The command to be transmitted 
				uint8_t *rx_data // Data to be read
				)
{
	uint8_t cmd[4];
	uint16_t cmd_pec;
	
	cmd[0] = tx_cmd[0];
	cmd[1] = tx_cmd[1];
	cmd_pec = pec15_calc(2, cmd);
	cmd[2] = (uint8_t)(cmd_pec >> 8);
	cmd[3] = (uint8_t)(cmd_pec);
	
	cs_low(CS_PIN);
	spi_write_read(cmd, 4, rx_data, (NUM_RX_BYT*total_ic));         //Transmits the command and reads the configuration data of all ICs on the daisy chain into rx_data[] array
	cs_high(CS_PIN);                                         
}

I did some testing today with the hardware. Execution time drops from 158μs to around 148μs (This is with only one LTC6811). Not that significant but still an improvement along side the memory reduction. Everything seemed to work as expected.