LM35 as a room temperature sensor (any other good alternatives)?

Discussion in 'The Projects Forum' started by yaantey, Oct 25, 2011.

  1. yaantey

    Thread Starter Active Member

    Oct 7, 2011
    Hey people,
    Im doing an diploma project in which I need to make a room temperature sensor? After some hours research I have found LM35. I would like to know:

    If LM35 is a good sensor for the mentioned purpose?
    Also, whats the importance of the quiescent current in a sensor? From my understand its good to have a low quiescent current sensor to make project power efficient.
    Importance of gain in a sensor? From my understand this will make the sensor more accurate by giving large output per variation of degree celcius.
    Importance of output impedance in a sensor? Which I don't really know.
    Also, is using a digital sensor or analogue sensor better for this application?
    As LM35 doesn't use single supply according to datasheet, so what is single supply?
    Is there an alternative sensor that could be easy to implement or better accuracy or requires low power?

    Hope someone can give me genuine help to understand the mentioned questions.

  2. Wendy


    Mar 24, 2008
    It is a good alternative, you can make a thermometer with the device if you so choose.

    The less current the better. Current will induce heating in the device, which will skew the readings.

    This device is already calibrated in °C, no gain needed other than to trip other circuitry.

    A single supply is one battery. Dual supplies are two batteries, with the connection brought out and called ground. Op amps use ± power supplies. Some circuits, like motherboards, require many voltages, each is considered a separate power supply. The fact you don't know this (it is extremely basic) suggests you are not studying electronics. You would help yourself and us if you give us some back ground on what you are studying.



    The datasheet gives you all the information you need, including the fact that for room temperature you don't need a dual power supply. It even gives you the circuitry needed to use it.

    This chip is about as simple as it gets. Try to duplicate the circuits you see in both links and get a feel for it.
    Last edited: Oct 25, 2011
  3. crutschow


    Mar 14, 2008
    The LM35 is a good sensor. National also recommends the LM94022. You might look at that.

    Quiescent current determines the power dissipated, and thus the self heating of the sensor. Efficiency is only a factor if you are using battery power.

    Certainly a higher gain will give better resolution and noise rejection, but that can limit the dynamic range (before you run out of signal range).

    The output impedance determines what load you can drive without affecting accuracy.

    Whether you want analog or digital output depends upon what you want to do with the signal. If you want a digital readout then a digital output device is generally better. Digital is also easier to send (less problem with noise) if there is a long distance between the sensor and receiver.
  4. RiJoRI

    Well-Known Member

    Aug 15, 2007
    A digital temp sensor would be the Dallas Semiconductor DS1620.

  5. ErnieM

    AAC Fanatic!

    Apr 24, 2011

    Actually you did not state what the sensor is to drive, what uses the output. So no one could comment.

    Well it can't use more power then you can provide. A secondary effect is if it takes too much current it can self heat. The LM35 should have neither problem.

    Don't confuse accuracy and resolution. The untouched output of the sensor has the most accuracy. This output thru an amplifier would have a better resolution but a poorer accuracy.

    That depends on what is using the output.

    Again, that depends on how the output is used. If you have an analog comparator it is a good choice. If you have a computer based control then you would prefer a digital output.

    It can use a single supply. Check figure 6 of Nat Semi's data sheet.

    Billions and billions. (OK, well, many.)
  6. Adjuster

    Late Member

    Dec 26, 2010
    High gain in a sensor may help accuracy if there are significant extraneous signals present, such as noise or DC offsets from the circuit handling its output, or electrical interference from the environment.

    The sensor gain only needs to be high enough so that these extra sources of error will be small in comparison with the errors inherent in the detector itself. How much gain is required to assure this will depend on the performance of the rest of the circuitry, and perhaps on its surroundings.
  7. hgmjr

    Retired Moderator

    Jan 28, 2005
    You did not mention whether you are planning to use a microcontroller as part of your design.

    Are you? If so, do you know which one yet?

  8. yaantey

    Thread Starter Active Member

    Oct 7, 2011
    I am using a microcontroller, which is the arduino uno board that comes with ATmega328 microcontroller. This is an exercise to learn how to interface sensor with microcontroller and to display the readings on a display. The microcontroller it self has a 10bit ADC. So, do you think its better I use a digital temperatur sensor instead of an analogue sensor.
  9. pilko

    Senior Member

    Dec 8, 2008
    I am using a microcontroller to read six temperature sensors and control a heat pump,oil furnace and wood furnace.I have used thermisters and several other analog sensors,including the LM35.
    The main problem I encounted was errors due to electrical noise,especially over distances greater than 10 meters (even with filtering,and shielded and twisted-pair cables).
    I now use Dallas DS18B20's with great success.

  10. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
  11. wayneh


    Sep 9, 2010
    Probably; it's very easy and simple to use. Since it will output +0.25V at 25°C, you can use it for room temperature with a single supply.
    Efficiency isn't a concern - it's about self-heating. This is a concern when the device may be insulated from the material being measured. If that material is calm air, that itself is insulation and the IC can warm some amount, introducing a high bias to the result.
    Not terribly important, IMHO. I wanted to use the LM34 for this reason (since 1. farenheit degrees are smaller, meaning you get more voltage per degree and 2., 0°F is colder than 0°C). But the LM35 is MUCH easier to find.
    Versatility. The lower impedance, the more things can be used to transmit and read the signal. High impedance outputs require more elaborate design considerations, to avoid altering the signal.
    Can't answer - as noted it depends what you want.
    You need a dual supply only to produce negative voltages at temperatures below °C. If you stay in positive temps, you can use just a single supply (only positive voltages).
    I did a fairly long search into this area and arrived at the LM35 and the digital option noted by others. There may be better things out there, but these are both widely used and available.
  12. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    Since you have a pretty powerful micro there you may consider a digital output device. These things give you a direct temperature reading so while perhaps a little harder to interface you get a direct useable output.

    When using say the LM35 and you get that .25V output at 25°C you have to send it to an analog to digital converter. For a 5V reference and 10 bits, you get a base resolution of 5V/1023 = 4.8mV/bit. For a device that changes 10mV/°C the bit output has a resolution of

    (10mV/°C) / (4.8mV/bit) = 2 bits/°C

    With a lower reference voltage or by adding gain to the sensor output this resolution can go way up, though both methods decrease the accuracy.


    Resolution means how small a change can be detected.

    Accuracy means how precise that detectable small change is.
  13. wayneh


    Sep 9, 2010
    Sorry to nitpick but resolution and precision bear on the same thing - how repeatable a measurement is. Accuracy refers to how close a measurement technique comes to the "real" value, whether there is a bias in one direction or the other. It's possible to have great precision and very poor accuracy (a stopped watch, for instance) but you can rarely get good accuracy without good precision. Of course even the stopped watch is always precise and even perfectly accurate twice a day.
  14. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    By all means, nitpick away. But I could see that discussion going into several pages as we all talk past each other, so best to start a new topic and not hijack our new friend's thread.

    I agree to your definition of accuracy. My statement of "accuracy" was not a "precise" as I intended it.

    I do not agree that resolution and precision mean the same thing. While precision is a measure of how repeatable a measurement is, resolution is a measure of how small a change is detectable.

    (From Wikipedia, "Accuracy and precision")