LM35 testing

How can i test LM35 temperature sensor?

I need it for my design in data gathering part. I need to test it in different temperature.

What suggestion do you have on what testing will I do.

My project is getting the temperature(room) and display it in LCD wirelessly using Arduino and Zigbee..

Please help me in my testing. By the way I already do it and it work properly. I can now detect temperature and display it to LCD but not wirelessly. I just don't know if its accurate.

My first idea on testing it is buying a digital room termometer. But some people says that my design is much more accurate than that. Or is there a way that i can compare it to Room termometer?

Please give me idea on how to test my design.

I guess you just have to buy an accurate thermometer. You need a known good reference to test anything.

One such reference is icy water, I think that's near as damit to 0C. And boiling water is 100C by definition at sea level.


Rob

Graynomad: I guess you just have to buy an accurate thermometer. You need a known good reference to test anything.

One such reference is icy water, I think that's near as damit to 0C. And boiling water is 100C by definition at sea level.


Rob

Oh. But i need to measure room temperature. Is it ok if i use icy water or boiling water in my testing of my design.

By the way. I forgot to mention about the other part of my design. It is that i have a fan that will change its speed depending on the temperature. It has 4 speed level. So maybe i should test four different temperature measurement. How can i come up with that?

Thanks for helping me

Looking at the LM35's data sheet it's about .25C linear between 0 and 100C. The graph cuts the 0% error line at 0 and 100 and peaks at +.25C off at about 50C.

So I would say that calibrating at 0 and 100 is good for at least .25C and you could improve that with software.

Otherwise all I can think of is you get a really accurate thermometer.

i have a fan that will change its speed depending on the temperature. It has 4 speed level.

I don't know what you are cooling but I'd say .25C has to be good enough, probably 5C is OK.

Just how accurate do you need? What are you cooling?


Rob

Graynomad: Looking at the LM35's data sheet it's about .25C linear between 0 and 100C. The graph cuts the 0% error line at 0 and 100 and peaks at +.25C off at about 50C.

So I would say that calibrating at 0 and 100 is good for at least .25C and you could improve that with software.

Otherwise all I can think of is you get a really accurate thermometer.

i have a fan that will change its speed depending on the temperature. It has 4 speed level.

I don't know what you are cooling but I'd say .25C has to be good enough, probably 5C is OK.

Just how accurate do you need? What are you cooling?


Rob

My fan will cool a serverrack. But the temperature sensor will be place in the ceiling. Therefore the temperature sensor will detect the server room temperature. The temperature from sensor will be send wirelessly to the fan.

Can you help me in the exact thing that i will do. Sorry for my stupidity. But thanks for helping me :)

My fan will cool a serverrack.

Then I would think accuracy was the last of your problems, calibrating against any half-decent thermometer would do.

But the temperature sensor will be place in the ceiling.

Normally you sense at the location that needs cooling, just be careful that there's no reason the relationship between the ceiling temp and the server temp is not linear. EG if there's a window that allows a breeze to flow along the ceiling the sensor will think it's nice and cool while the server is cooking.

Also you have to look at the temp difference between the ceiling and the server box, no point calibrating for the ceiling of the server is always 20C hotter.

BTW I won't be online for the next few days or even a week, so if there are any more questions maybe someone else can answer.


Rob

Graynomad:

My fan will cool a serverrack.

Then I would think accuracy was the last of your problems, calibrating against any half-decent thermometer would do.

But the temperature sensor will be place in the ceiling.

Normally you sense at the location that needs cooling, just be careful that there's no reason the relationship between the ceiling temp and the server temp is not linear. EG if there's a window that allows a breeze to flow along the ceiling the sensor will think it's nice and cool while the server is cooking.

Also you have to look at the temp difference between the ceiling and the server box, no point calibrating for the ceiling of the server is always 20C hotter.

BTW I won't be online for the next few days or even a week, so if there are any more questions maybe someone else can answer.


Rob

This design project is already accepted :)

I read a research that some serverrack has holes in the top of the rack. Then the temperature of server rack will go directly to the top of the room.

I just need to get an idea how can i test my temperature sensor if its accurate. Please help me.

do you have suggested tables. or suggested temperature to compare. Because i think i need to get 4 different temperature and i don't know how to get these measurement. My other idea is pump a hair blower to my sensor. Or some dry ice I think. Because my design prototype will be placed in a box. I don't know what kind of box yet. But i will represent the server room.

My device will run depending on the temperature . If the temperature is 13C and below. The fan will be off. If the temperature is between 13C and 18, Fan will be level 1. If the temperature is between 19 and 22, Fan will be level 2. And 23 and above. The fan will be level 3 which is the max level. I already make a design that will run this things. All i need now is the data gathering for my thesis.

I based the temperature because i read a research that a "SERVER ROOM TEMPERATURE" should range from 13C to 22C.

PS: I will be using 4 temperature sensor.(4 LM35 to get the average temperature)

My other idea is pump a hair blower to my sensor

Sounds good to me, a hair dryer and a thermometer and you're done. Although 13C might be a problem.

The thing is the sensor is pretty linear, it doesn't matter at what temp you calibrate it for such a non-critical application. Get it right at 21 and it will still be good at 31 and 11.

IIRC that sensor puts out 10mV/C, so at 100C you will only get 1v from it. Using an Ardiuno ADC you have approx 5mV resolution so your temp accuracy will be .5C, there are ways to get better but that should be good enough.


Rob

Graynomad:

My other idea is pump a hair blower to my sensor

Sounds good to me, a hair dryer and a thermometer and you’re done. Although 13C might be a problem.

The thing is the sensor is pretty linear, it doesn’t matter at what temp you calibrate it for such a non-critical application. Get it right at 21 and it will still be good at 31 and 11.

IIRC that sensor puts out 10mV/C, so at 100C you will only get 1v from it. Using an Ardiuno ADC you have approx 5mV resolution so your temp accuracy will be .5C, there are ways to get better but that should be good enough.


Rob

Sorry but i can’t really understand things about voltage something.

I’m now here at philippines. And when i run my device it detects 27C to 28C which means i need to cool it down.

And when i run my device it detects 27C to 28C

So you already have the temp sensor working?


Rob

Graynomad:

And when i run my device it detects 27C to 28C

So you already have the temp sensor working?


Rob

Yup its already working together with the fan. But its not yet in wireless communication

So compare the values you are getting with a thermometer. Job done.


Rob

Graynomad: So compare the values you are getting with a thermometer. Job done.


Rob

But how can I get the temperature measurement that i need. Because i need to test. If my fan will work in that different temperature measurement.

If you mean heating the inside the server box then the hair dryer is as good a way as any I guess to heat up an enclosure.

If you mean to test that your program trips at the right point while testing outside the box then same story.

Unless we have a communication problem I think it’s been covered.


Rob

Graynomad:
If you mean heating the inside the server box then the hair dryer is as good a way as any I guess to heat up an enclosure.

If you mean to test that your program trips at the right point while testing outside the box then same story.

Unless we have a communication problem I think it’s been covered.


Rob

I mean getting tempearture measurement like between 13 and 19. and so on. Like i post before.

between 13 and 19

I imagine those temps would be difficult to come by in the Philippines.

Stick the sensor in a fridge, do the test at 2AM :), get a box of dry ice like you mentioned before and hold the sensor close until it gets to 13 degrees, use your imagination.


Rob

Graynomad:

between 13 and 19

I imagine those temps would be difficult to come by in the Philippines.

Stick the sensor in a fridge, do the test at 2AM :), get a box of dry ice like you mentioned before and hold the sensor close until it gets to 13 degrees, use your imagination.


Rob

Sir i have a question. If I get in my device a temperature measurement of 28C and in digital thermometer 26C. Can I just edit my formula in the Arduino IDE and make it subtract by 2C. Example Temp - 2. So it can match the temperature measurement in the digital thermometer

Yep, you can do what you like to the reading. If it is a constant 2 off then subtract 2 as you say.

If the error changes over the range you can have a lookup table and subtract different values according to the reading.


Rob

Graynomad:
Yep, you can do what you like to the reading. If it is a constant 2 off then subtract 2 as you say.

If the error changes over the range you can have a lookup table and subtract different values according to the reading.


Rob

Sir thanks for your help. I really appreciate it.

I’ll just match the reading of my device from a digital room thermometer to tell in my data gathering that it is accurate. Is it ok if i do that?

Yes, and no. Do you have any reason to think that your digital thermometer is accurate? Your only options, as already discussed are to get an accurate thermometer to calibrate against, or use some other known temp e.g. freezing & boiling (adjusted for altitude) water. Having said that, for your purposes, does it actually matter if your measurements are a degree or two off? For minimizing power consumption of course, perhaps it does.

It's the classic problem of telling time with two watches - how do you know if either is right?