PIC18F46K22 reading ADC value

[Khansokhua]

[gaugeguy]

Because you do not have a 16 bit A/D. You have a 10 bit A/D that can be left justified or right justified.

[PCM programmer]

In other words, 1023 x 64 = 65472.

[Khansokhua]

1111111111000000=65472
0000001111111111=1023

[temtronic]

right..
this code..
adc_value=(adc_value/(4360));

probably won't work as expected when the ADC result is 'right justified'.

[Khansokhua]

[PCM programmer]

[Ttelmah]

Also, the two reads of the ADC. It would take half the time to read this
once. Since these happen _before_ the timer value is set, they add
a lot of time. The ADC takes 32 cycles of the CPU clock per ADC cycle.
It takes 13 cycles of this to read the ADC. So 104 instruction times
for the read. Two reads, gives over 200 cycles of delay here. In fact it
will give wrong values by reading twice in the same line of code. There
needs to be Tacq between successive reads of the ADC. This is badly
wrong...

Several things can be done to help here.
First at 4MHz, the optimum ADC divisor is FOSC/4, not FOSC/32. This
will make the readings 8* faster., so CLK_DIV_4.
If you look in the data sheet at 4MHz, the /32 selection has the note '3',
which says "for faster conversion times the selection of another clock
source is recommended'...
Then, you really should not be reading the ADC both inside and outside the
interrupt. Take the reading outside the interrupt only, and store the readings
when they are done into a global variable. Use this in the interrupt.
Also set the ADC to automatically perform the acquisition.
ADC_TAD_MUL_8. in the setup. This then ensured that Tacq is correctly
allowed whenever an acquisition is done. You may well find with these
changes the ADC is a lot more reliable.