Hello 
I'm still on the process of developing the PCB for our IoT-based fire alarm system and as of now I've been looking for possible options to improve the system.
I'm using LM35 (TO-92) as a fixed temp. type sensor and I'm quite thinking that it might not respond that fast during ambient temperature changes. According to the datasheet, I could use an epoxy or a small heat fin to the sensor to solve this.
Is it reasonable to do so? If yes, how (since I'm scared I might damage the sensor)?
rayrayrayyourboat:
(since I'm scared I might damage the sensor)?
I think there is a lot more to be scared about. Surely, there are better ways to detect a fire than a temperature sensor.
kindly explain your project....maybe i can suggest...
What speed of response are you wanting?
@Nick_Pyner
I'm also using an ionization smoke sensor (HIS-07) which is quite sensitive. D'you think these should suffice?
@chandantaluja
I'm developing a WSN-based wireless fire and panic alarm using the Arduino Pro Mini, XBee, LM35, HIS-07 with some couple o' modules like the TP4056 and MT3068. Some resistors, regulators, ICs, caps, buzzers and LEDs and batteries is all. Basically it's mostly made up of ready-made components to be placed on a PCB.
I think just about less than a minute. 1 and a half at most. The LM35 has 3-4 mins. before reading the final value in still air. It kind of bothers me.
Make a bare copper/constantan thermocouple using the thinnest wire you can find.
A tiny 100k glass bead thermistor (NTC), as used in hotbeds, has verry little mass.
And can be easily read by an Arduino.
Less than $0.50 on ebay.
Don't just measure absolute temperature.
Trigger an alarm when the A/D value changes too much in e.g. 10 seconds.
Leo..
@mikb55
Hmmmm I'll try? Thanks anyway. 
@Wawa
Okay. I'll order right away. Seen ones online and yeah, cheap. thanks too. But can I just use the 10K ones? They're much more accessible here in our area
10k means self-heating is 10x as bad as the 100k version.
You can fix that by attaching something to the NTC, but that increases the thermal mass.
If it's cheap enough, you could just try.
Leo..
Never made a heat detector for a fire alarm, but a small (low mass) sensors seem the right choice.
Use a 100k pull up (or pull down) resistor with the 100k thermistor.
Just read the A/D value from the pin at one second intervals, and see how much it changes.
Sensor should be about 0.1 degree C per A/D value at room temp.
Add a ~100n ceramic capacitor from pin to ground if readings are not stable.
Leo..
@Wawa
Thank you for replying always. I just can't add karma on mobile