Go Down

Topic: thermistor not consistant (Read 599 times) previous topic - next topic

garcara

I have a voltage divider (very standard) setup with a meat temperature prob that I calibrated.

Here is my output once a second:

Code: [Select]

72.19
68.31
62.82
68.31
64.23
65.61
62.82
68.31
68.31
65.61
66.97
65.61
64.23
68.31
66.97


No long wires or anything, reading on the right analog pin and all. Any ideas?

retrolefty

What are the resistance values of the fixed resistor and the variable thermistor. Could be they aren't providing a 10,000 ohm or less output impedence needed for best results?


garcara

My fixed resister is 10k, not sure what value you want from the thermistor?

Temp (F) - Resistance:
38 - 266,487
210 - 20,176
140 - 61,538

retrolefty

#3
Mar 09, 2012, 02:16 pm Last Edit: Mar 09, 2012, 02:18 pm by retrolefty Reason: 1
The easiest thing you could try is a little software filtering. Read the analog value 8 times in a row ( without delaying between reads) adding the value into a single variable and after the 8 additions, divide by 8 and print out the results, then reset the variable and start another 8 reads, add, divide, display, reset, lather, rinse, repeat. See if that helps or not. On the hardware side one might consider a op-amp buffer to isolate the rather high impedance value of the thermistor from the input impedance of the A/D circuitry in the AVR chip if the software averaging is not effective.

Lefty

garcara

Thanks Lefty,

If I go the op amp route, how do I determine what my cut-off frequency should be?

retrolefty

#5
Mar 09, 2012, 06:11 pm Last Edit: Mar 09, 2012, 06:18 pm by retrolefty Reason: 1

Thanks Lefty,

If I go the op amp route, how do I determine what my cut-off frequency should be?


I would just try it as a simple non-inverting gain of one opamp buffer circuit first. If that doesn't give consistent numbers then you can always convert the circuit to a low pass filter circuit. Cut-off frequency can be as low as you want as temperature is a pretty slow changing process I would think low Hz values would be fine.

What we are trying to determine is if you value variations are predominantly caused by the higher output impedance of your divider or from just signal noise or a combination of both. So try the simple buffer first in my opinion.

Lefty

garcara

Thanks so much, I will give it a try.

Go Up