analog readings go up w/o reason after few hours

Forum 2005-2010 (read only) Hardware Interfacing
1

Hi, i have a problem reading analog input from an arduino clone (solarbotics ardweeny).

The schematics is very simple.

Three 2% 10K NTC thermistors are connected to the A0, A1 and A2 pins respectively, as in the following diagram:

+5V ----- [thermistor] ----analog pin-----[ 10k resistor] --- GND

I am using the +5V and GND pins of the board for input voltage and ground. Supply is from a PC USB port via sparkfun ftdi dongle.

I read the input voltage several times, perform several calculations to derive the temperature and compute an average.

The thing works correctly and within the tolerance of the thermistors for a few hours, then suddenly the readings from the three pins seem to jump together, and I record higher temperature.

There is no actual change in the temperature, and the jump happens at the same time for all three thermistors, i.e. it is not a problem of the environment, or the sensors. The temperature ranges measured are well within the tolerance of the equipment.

Resetting the board several times seems to fix the problem for a while.

I tried to search the forum, but frankly I am lost at what to search for and haven't found anything useful. Has anyone experienced a similar problem? Any ideas where to start looking?

Thanks in advance,

2

The set-up looks o.k. You normally connect the R and the NTC vice versa, but that is no issue. Also this devider makes you independant of the actual value of the supply voltage, as it cancels out with the ADC reference.

So everything should be fine...
What do you mean by SEVERAL resets???

If one reset helps, it could be a hidden software issue, most notorious are overflows of timer related variables....

3

Thanks for your comment.

By "resetting it several times" I mean sometimes it gets back to normal measurements the first time, sometimes it takes several resets, after which the value drops to the expected values. That is why I don't think it is a software issue.

I've double-checked the software, and there is nothing that would overflow -- I am not using timers in the temperature calculations.

Since it happens randomly I haven't yet be able to actually measure the voltages "after", so I am now waiting for the defect to occur again.

As I said, I am totally out of ideas as to what is wrong -- being a hardware newbie doesn't help :slight_smile:

4

Aref is generally not decoupled, which can cause similar effects. Can you connect a 100nF cap between Aref und ground ? It will do no harm, but might help...

5

Thanks. It didn't help much. Maybe it is a faulty contact somewhere, I'll resolder and try again ...

6

Use a DMM and measure the voltages output at the takeoff points to the arduino, if they stay put and the arduino reads differently, it will give you some clues.

I'd be checking the USB output voltage stays put as well......

7

The supply voltage is absolutely irrellevant, as it is also the ADC reference.

8

Confusing thermistors in voltage dividers with those damn LM35's......

I'll stick with DS1820s. :slight_smile:

9

Thanks.

In this case, however, there's no confusion as to the kind of sensor. Also, I am looking for clues to the possible source of the error, not a replacement sensor.

10

I'd swap the microcontroller as a first step, probably not so easy on an ardweeny. Easy when you have several boards and chips floating around.

11

turns out the problem was a bad connection somewhere in my soldering, not related to the sensor, or to the ardweeny.

the dog fixed it once accidentally by brushing against the device, re-soldering seems to have fixed it completely.

thanks again for the help.