External Temperature sensors (LM35) and crude readings

we've several QAIs connected to temp' sensors (*) ...

Help says 0-5V 8bit, 'though one or two of ours are 0-10V ...

ours are set to 2% resolution, and we get readings in 0.4degC steps ...

have thought a few times of adjusting the sensitivity, but not done so yet - ISTR our sensors are good to 0.5degC, so that might be why we chose what we did - can't remember, some time ago now !


(*) Rapid, clip-on temperature sensors / thermistors ...

The LM35 has a resolution of 10mV/C. The standard (0-3.3V) QAI has a resolution of ~13mV/step. So you will at best be able to get a little above 1C resolution. The scaling of the mapping curve in the analogue signal object will also have to be adjusted since the default is 1:1 i.e 10% change equates to 10(C) change whereas 10% change of signal actually means 0.33V which is equivalent to a 0.33/10mV = 33C change on the LM35. So you have to change the quadratic parameter b to 3.3 and c to 2 (because I think the LM35 starts with 2C=0mV). If you want better resolution then a thermistor arrangement should yield ~0.5C.

Thanks Chris, I've actually used those from rapidonline before (based on your suggestion) and they have worked well. I´ve used the clip on type. This time I wanted to test the LM35 instead since I want to measure temperature in a water tank. I thought LM35 would be more secure for this.
/Glenn

Thanks Karam, I used a QAI with 10V which I thought would give a better resolution.
Anyway I go with the thermistors instead.

/Glenn

Thermistors in noisy environment

I tested the thermistor instead and while it work fine "on land", it give wrong readings when I put it in the tank. I would expect it to give a reading of 24C which should be the pool temperature. Instead it switches between 62C and 68C every second or so.

The tank is quite "noisy environment" with chemicals (chlor) and heavy pumps and the cable between thermistor and QAI is ap 6meters.
Perhaps that could be the reason

So, I guess it is back to the proved and tested clip on thermistors. Only problem is that all pipes are in plastic... Perhaps contact paste? Any ideas anyone?

/Glenn

not tried this, but ... how about using the first sensor in my list, the one that's on a lead, set inside a piece of earthed copper pipe in the water, to provide an electrical shield against the electrical noise ... ??

The resolution is 8 bit in all cases. If you use a QAI with a 10V range you will actually reduce the measurement resolution because you will be using less of the full range for the same voltage output from the sensor. As I undertsand it the LM35 in basic temperature configuration gives out 10mV/C starting with 0V at 2C. You also need a 4-20V supply for the LM35 so I guess for a standard QAI this may mean using the IDRANet 12-15V power bus connection (with a local capacitor across the supply pins at the LM35). A thernistor obviously needs to form part of a voltage divider at he QAI input meaning you need to add a suitable resistor e.g 10k to 0V. I'd need to know in more detail something about the thermistor and resistor values that you are using, but whereas a long cable may encourage noise pickup (which you can reduce by adding a capacitor at the input to the QAI) it should not significantly affect the average value since the cable resistance should be pretty small relative to the thermistor and resistor values. Are you immersing the sensor? If so then perhaps you are getting leakage/conductance across the connections in that area which I would expect to lower the resistance and therefore appear as a higher reading.