how to measuring room temperature accuratly

[abolfazlk873]

hi everyone,

i am intend to build a thermometer for house to controlling speed of cooling system.

The sensor that i used for this porpose is LM35 but it is not able to measuring room temperature accuratly, i don't know reason of this. when the cooling system is worked and we are chilled, the themperature that sensor is displayed was steady.

first i think that the position of sensor is not correct and for this reason i am reposit the sensor but still the sensivity of sensor not changed. just when i posit the sensor in AC duct the sesor work correct. this is due to intensity of air flow.

important thing is i should posit the sensor in the main conrolling box and because this i can't posit the sensor in the ac duct.

i am searching for the ways to eliminate this problem. I am grateful for your guidance.

 

[KlausST]

Hi,

The sensor that i used for this porpose is LM35 but it is not able to measuring room temperature accuratly, ..

For sure the LM35 is able to measure room temperature.

i don't know reason of this.

neither do we. Because you don´t give any error description.
How do you know it does not work?

Please tell about your application. Best if you use pictures and schematics.
Then tell how you measure the LM35 signal.
Then tell what you expect... and what does not work like expected. (give values)

Klaus

 

[Audioguru]

It is normal for you to feel chilled when cool air is blowing your skin heat away from you. Adjust the AC system so it does not cause drafts on you.

 

[c_mitra]

I presume that you mean the room air temperature. The temperature of the air within a room is not constant and it varies from point to point. For example, the temp will be high near the ceiling and low near the floor (cold air stay near the floor and hot air rises to the top). The temp will be higher (or lower, depending on the outside temp) near the windows and lower near the inside walls. It is impossible to "stir" the air enough to get a const temp (within 0.1C) but with care it is possible to get a const temp within 1-2C.

The sensor you are using is an absolute temp device; it does not need a reference temp. It should be able to get temp accurate within 0.1C (+/- 0.1 or at most +/- 0.2)- if not change the device.

The sensor must be protected from draft; place it in the centre of a perforated tube so that it is in contact with open air.

The connecting wires must be very thin- they carry lots of heat along with them. support with device with foam and not with thick wires.

The temp you feel is not the same as the temp the device sees.

 

[abolfazlk873]

schematic is shown below:

program is right because in the ac duct, sensor is work well due to to high density of air flow.
but i need to sensing temperature in the corner of room that doesn't sufficient air flow.
the example behavior of sensor:
when cooling system is off, the temperature was shown is 33.
when cooling system is on and we are chilled, the temperature of sensor just 1 degree is reduced.

 

[c_mitra]

when cooling system is off, the temperature was shown is 33.
when cooling system is on and we are chilled, the temperature of sensor just 1 degree is reduced.

Where the sensor is located? Is it measuring the temp of the air at the inlet?

 

[andre_luis]

when cooling system is off, the temperature was shown is 33.
when cooling system is on and we are chilled, the temperature of sensor just 1 degree is reduced.

You migh be aware that the temperature measured in presence of air flux is not necessarily the the same measured in 'static' conditions of the air, particularly when the sensor itself has a self heating characteristic, although quite small. The faster the wind, more heat is carried from the sensor.

Another thing that you could consider in order to keep the measured value against instantaneous variations, is to perform either an algebraic average of seferal samples, and also tie the sensor to a bulky metal ( thermically well conductive ) to minimize the amplitude of variations.

 

[abolfazlk873]

i am making two structure for the sensor.
1. the first structure, sensor is placed on the heatsink and heatsink is placed at the room.

2. the second structure, sensor is placed behind the frame. the frame was rimosed for flowing temperature of room in to frame.

 

[Warpspeed]

Depends what you mean by accurately.

LM35 accuracy will be within one degree, but it will be very slow to respond to changes in air temperature because of its large thermal mass and poor thermal coupling to the air.
Its quite a good choice for a digital room thermometer, but possibly not such a good choice for use as an actual temperature controller.

A small bead thermistor usually will not measure to within one degree accuracy, but it will respond very fast to small changes, often within one second.
That can be an advantage or a disadvantage, as the heating/cooling system will be short cycling and going on and off very frequently which can be pretty punishing to the heating/cooling equipment.

Its not easy to strike a good compromise, but the key to it is selecting the right sensor time constant, and experimenting with placement within the room.

 

[c_mitra]

If you tie the sensor to a large heat sink, your response time will be very long. You will not see the temp of the air.

With an air conditioner in the room, the temperature of the room is never going to be uniform; colder air is being injected, it is allowed to mix, heat is allowed to be lost from the room and warmer air is being taken out and a fixed proportion of the room air is being replaced continuously.

The most convenient location for the sensor for the room temperature is the place where the air is being removed from the room. It tells the compressor the amount of cooling to be used on this mass of air.

 

[abolfazlk873]

I am really admire when I see the excellent operation of thermostats such as Honeywell vs my designed thermostat.
that thermostat really good operation in the same conditions that i tested my designed thermostat, but i misknow what is reason of this.

 

[c_mitra]

There are two aspects:

1. You can maintain a low differential of the input and output temperatures using a variable frequency drive for the compressor- that will provide a pleasant experience.

2. You can play with the hysteresis settings- the temps at which the compressor turns on and off and also the air flow. If the difference of temp at which the compressor turns on and off is more than a degree, you need to slow down the fan so that the compressor runs will be longer in time.

There are other ways but they are not related to electronics.

 

[andre_luis]

If you tie the sensor to a large heat sink, your response time will be very long. You will not see the temp of the air

It is fact, but for a pratical utilization we have to consider that the purpose on this application is supposed to not necessarily show the instantaneous temperature saw by the probe ( sensor+heatsink ) but the average temperature, and this apparatus makes some kind of 'integration', such as a low pass filtering, because we should not expect the room temperature arising so fast. Note that if this value is used to determine the trigger point of a thermostat, even considering its intrinsic histeresis, would be not desirable to have instantaneous oscillations in the temperature around the average.

 

[c_mitra]

It is fact, but for a pratical utilization we have to consider that the purpose on this application is supposed to not necessarily show the instantaneous temperature saw by the probe ( sensor+heatsink ) but the average temperature, and this apparatus makes some kind of 'integration', such as a low pass filtering, because we should not expect the room temperature arising so fast. Note that if this value is used to determine the trigger point of a thermostat, even considering its intrinsic histeresis, would be not desirable to have instantaneous oscillations in the temperature around the average.

I agree. This also gives some kind of built-in hysteresis.

However, I do not have the numbers and I am just doing a rough estimate- if the heat sink is 100 gm in weight, it will have about 1000 times heat capacity compared to air - or about 100 kg of air. Just for a rough estimate, 100 kg of air is about 100 m3 (the air present in a room of size 10m X 5m X 2m (pretty big room).

To have fast response, we use thermistor based thermometers that have the effective sensor mass close to about 1-10 mg of mass. They respond in less than a minute (about 5-10 s).

The bare sensor (including the casing) will be about 500 mg and will have a fair response time (perhaps 5-10 m). These values are rough guess values.

 

[andre_luis]

I do not have the numbers and I am just doing a rough estimate- if the heat sink is 100 gm in weight

In the presence of air blowing, the heatsink mass possibly would have a smaller impact when compared to the effect of the rapid diffusion of the temperature inside. Perhaps such a bulky sink as shown in the previous picture is not so necessary, since its major goal is to increase the efficiency of the thermal coupling between the air with the sensor ( the self-heating displacement of temperature on LM35 is almost negligible, less than 0,1ºC ).