Detecting a Candle Flame

Thread Starter

Yasitha Kasthuri

Joined Nov 8, 2015
5
I want to detect a candle flame from a distance of 30 cm.
What is the best sensor that is available to do this? (Sharp IR, UV, etc.)
What softwares can I use to simulate such a sensor?

Any help is greatly appreciated. Thanx!
 

Attachments

blocco a spirale

Joined Jun 18, 2008
1,546
IR sensor with a lens in front of it. This is the kind of thing you really need to build and prototype; a simulation (whatever that would be, I don't know?) wouldn't tell you anything useful. i.e. nothing you could reliably transfer to the real world.
 
Last edited:

wayneh

Joined Sep 9, 2010
16,114
Even a solar cell from a garden light could easily detect a candle at that short distance. But the devil is in the details: interfering light sources, power supply, etc.

You'll get better feedback if you offer more detail. What's in the picture?
 

Thread Starter

Yasitha Kasthuri

Joined Nov 8, 2015
5
It's a robotic competiton . There are 9 candle holder and only 4 of them will be lit.
The robot have to find that four and put the flame out.
This is my first robotic project and kinda nada on sensor knowledge.

Thnx for the feedback . The competiton arena is sun lit.
So will a sharp ir sensor work?
 

blocco a spirale

Joined Jun 18, 2008
1,546
It's a robotic competiton . There are 9 candle holder and only 4 of them will be lit.
The robot have to find that four and put the flame out.
This is my first robotic project and kinda nada on sensor knowledge.

Thnx for the feedback . The competiton arena is sun lit.
So will a sharp ir sensor work?
It looks simple enough.

Have you got a datasheet for the Sharp sensor, I believe they make more than one type.
 
Last edited:

wayneh

Joined Sep 9, 2010
16,114
The competiton arena is sun lit.
So will a sharp ir sensor work?
That's a lot more challenging. You will have to discriminate between the candle (the signal) and ambient, varying sunlight (the noise).

Most of the IR sensors you'll find expect to receive a modulated signal (from a matched transmitter) and won't respond to a candle. You need a "naked" IR sensor. But then you'll need some smarts to sort signal from noise.
 

crutschow

Joined Mar 14, 2008
23,527
One way would be to mechanically scan the scene horizontally on either side of the robots center-line with a Si detector and detect the added brightness from the candle flame.
AC coupling will block most of the signal from the ambient light which is a DC level.
You then have the robot orient itself so that the candle bright spot is centered in the field-of-view and it will be facing the flame.

I worked on a celestial airplane navigation system that used sensitive CCDs to detect stars in full daylight. It could detect the light from the stars added to the large background light from the clear sky. Rather amazing system.
 

Thread Starter

Yasitha Kasthuri

Joined Nov 8, 2015
5
That's a lot more challenging. You will have to discriminate between the candle (the signal) and ambient, varying sunlight (the noise).

Most of the IR sensors you'll find expect to receive a modulated signal (from a matched transmitter) and won't respond to a candle. You need a "naked" IR sensor. But then you'll need some smarts to sort signal from noise.
Thnx a lot for the input .. Can you explain more what you meant by "some smarts to sort signals from noise"?
 

Thread Starter

Yasitha Kasthuri

Joined Nov 8, 2015
5
One way would be to mechanically scan the scene horizontally on either side of the robots center-line with a Si detector and detect the added brightness from the candle flame.
AC coupling will block most of the signal from the ambient light which is a DC level.
You then have the robot orient itself so that the candle bright spot is centered in the field-of-view and it will be facing the flame.

I worked on a celestial airplane navigation system that used sensitive CCDs to detect stars in full daylight. It could detect the light from the stars added to the large background light from the clear sky. Rather amazing system.
Hey, Thank you very much for the idea. First I am gonna learn what a si detector is. I hope that it would be better than the ir sensing method.
 

Thread Starter

Yasitha Kasthuri

Joined Nov 8, 2015
5
I have seen a great sensor from Hamamatsu that can differentiate between a flame and sunlight. The basic search term is "UVTRON". Here is someone with a demo board:

http://www.active-robots.com/uvtron-flame-detector-package

You can find the general Hamamatsu sensors here:

http://www.hamamatsu.com/jp/en/product/category/3100/3007/index.html

and Linear tech has a sample circuit in their app note an59.
Hey, Thanx for the input. I saw the UVTRON to. It seems to be the perfect sensor for my project. But it's is too expensive. That is why I'm looking for alternatives.
 

crutschow

Joined Mar 14, 2008
23,527
You make be able to use a non-contact temperature sensor like this
......................
The problem is that, as seen here, a candle flame outputs very little energy in the 4μm to 16μm spectrum that the temperature sensor is sensitive to. It's designed to detect temperatures near or slightly above ambient.
A candle output peaks at around 0.7μm which can be readily detected by a silicon diode or transistor detector.
 
Top