Adc giving widely flactuating values

[tamersoss]

Hi freinds ...
Here is a code i made to measure temperature using a thermocouple and a two stage opamp amplifier ... Using my avo the amplifier gives 0.03 V at 0.c to 3.13 V at 300 0.c , I uses a pic16f877a , the problem is that i get on my pc a great fluctuation of values which is about 40 increments ... I don't know the impedance of my analog circuit ... How can i measure it ? Please tell me if the settings has something wrong ... Can you explain the effect of adjusting sampling time ? ... Many thanks for caring ... Bye

DEFINE OSC 20
trisa=255
DEFINE ADC_BITS 10
DEFINE ADC_CLOCK 3
DEFINE ADC_SAMPLEUS 100
temp var word

adcon1=%10000000
loop :
adcin 0,temp 'measure temperature
serout portB.7,4,["C",temp.byte0,temp.BYTE1] 'send result to PC
pause 20
goto loop

[mister_e]

Could you post your schematic here?

[tamersoss]

see the .doc file

[mister_e]

In theory, the impedance of your circuit is the impedance of your LM324 output. Few tenth of ohm from what i remind. So it shouldn't be the problem.

What happen if you place a 0.1uF between pin 3 of your LM324 and GND?

Can you monitor the LM324 output with a scope and post a screen-shot of it here?

[tamersoss]

It didn't help ... The flactuation is the same ... I only want tell you that i made a mdification on the schematic which is that i replced the 70 k by a 100 k resistor

[sougata]

Hi,

If you are using your VDD as Vref for the ADC then it is important that your power supply is clean and steady. Because that effects your overall conversion accuracy. General remedies are decoupling caps on the VDD-VSS pins of your PIC and a stable source.

As steve mentioned there could be noise from the op-amps output too. So decouplers for them as well.

[skimask]

It didn't help ... The flactuation is the same ... I only want tell you that i made a mdification on the schematic which is that i replced the 70 k by a 100 k resistor

Have you tried using a bit of software averaging? Read the input a bunch of times, sum the total of the reads, divide by the number of reads, use that number. I know it's not a perfect solution, but it's always worked for me (read that as I'm compensating for bad circuit design )
JDG