AVR Freaks

Line1: The ADMUX register contains essentially three things. Top bits set the reference against which the ADC input voltage is measured. There is one bit that can adjust the position of the result to the left or right that sits "in the middle" and then the bottom bits select which of a number of sources are fed into the ADC using a "multiplexer". In your example the left/right adjust bit is not used so ADLAR will be 0 which means it will assume the default of right adjustment. The "(ADC_VREF_TYPE & 0xFF)" is a very curious piece of code. ADMUX is an 8 bit register so the "& 0xFF" would appear to have no effect whatsoever - usually you might use & with some bit pattern smaller than the whole 8 bits to isolate some value into a fixed range but & 0xFF really has no effect. Even if ADC_VREF_TYPE had been defined to more than 8 bits it still wouldn't really have an effect as the write to the 8 bit register would truncate it anyway. The place you might usually us "& 0xmask" in an ADC read routine when setting ADMUX is when you want to limit the chosen input (ie "adc_input" in this example). The MUX bits in ADMUX for mega8 are just the lower 4 bits so you might have (adc_input & 0x0F) to ensure that only the lower 4 bits "get through" but rather curiously that is not done. Presumably it is up to the user to ensure that they never pass an adc_input value to this routine outside the range 0x00..0x0F (0..15)? Finally the "|" in the middle is simply "OR" which joins together two groups of bits. So in the higher bits you have ADC_VREF_TYPE and in the lower bits you have the MUX channel selection. All is then written together in to the ADMUX register.

Line6: this is simply returning the result. The function itself has a "unsigned int" (that is uint16_t) return type so it will return 16 bits. Those will be the contents of the ADCH and ADCL registers combined. While it is possible to read ADCH/ADCL separately then combine them together in to the full 16 bits most AVR C compilers define some kind of "psuedo register" that is 16 bits wide and means "ADCH and ADCL together". In Codevision they clearly chose to call this ADCW (meaning ADC Word where "Word" is a fuzzy word in computing terms that in this case means 16 bits). Of course the ADC isn't really 16 bits, it is just 10 bits. That is where the ADLAR bit I mentioned comes into play. If it is not set (default) then the 10 bit reading occupies the lower 10 bits of the 16 bit ADCW. So it will return 0..1023. If the ADLAR bit had been set to Left adjust the result it would occupy the upper 10 of the 16 bits. But you usually only do that when you are only really interested in 8 bit reafdins. In that case you want the upper 8 of the 10 bits. So by moving all ten across the ADCH:ADCL boundary towards the upper end (that is ADCH) then if you just pick up ADCH with ADLAR=1 then you get the top 8 bits as intended.