Can i use an LDR to detect presence of a flame

craigcurtin

Senior Member
Guys,

I have an oil burner (based on a babington ball design) that i wish to start automating.

Basically the burner will occasionally go out (for any number of reasons), whihc has no effect other than that the oil pump continues to run for no good reason.

Stage 1 of my project is to try and see if a suitably positioned LDR (attached to a Picaxe) would be able to detect the presence/absence of flame and shutdown a relay to the oil pump.

Stage 2 will be to hook it up with a SimpleLan which will then send me an SMS - but that is another story !

So my thought is to get a 1/2" copper tube about 150mm long, mount the LDR at one end (which is sealed, with just the wires coming through (possibly something there to let any excess heat escape)

The other end of the tube is close to the babington ball and pointed directly at the flame (which shoots down a tube so there would be little ambient light).

Sample some numbers during operation to see if there is a threshold which is a definitive cutoff number.

Does this sound feasible or have i missed something here ?

Craig
 

moxhamj

New Member
That will work. Just measure the resistance of the LDR with the flame on and off and then put a resistor of similar value in series with the LDR. The LDR will be biased roughly at 2.5V.
 

premelec

Senior Member
You'll get a whole lot of flicker.. so use a capacitor to smooth that AND if your fire is lit with a 20KV transformer arc you'll have a whole lot of electrical noise to fend against! - sorry I'm old school and on gas so I don't know the ball structure - perhaps that go rid of the HV transformer arc...
 

craigcurtin

Senior Member
Thanks for the quick replies

I will give it a go with the LDR and resistor thanks Dr.

Premelec, Nope unfortunately at the moment the babington is lit using a MAPP torch (it burns Waste Veggie Oil) which takes a big flame to get the initial heat up and then get to the stable combustion phase.

If i can get this monitoring and alerting happening OK, then the next stage is to move to an autostart with Natural Gas - then i will have to investigate the best way to light that flame.

I was planning on doing samples every 10 seconds or so and then get rid of spurious ones to remove some of the flicker issues - A babington fire does not flicker much - it more roars like an angry dragon !!

Craig
 

moxhamj

New Member
The capacitor value wouldn't be critical but maybe a 4u7 or something across the LDR. LDRs have a fairly slow response time anyway.
 

gengis

New Member
It is feasible to do what you want. Assuming nothing in there is cherry red and going to indicate it is flaming when it isn't burning

You hint in your post: "Basically the burner will occasionally go out (for any number of reasons), which has no effect other than that the oil pump continues to run for no good reason." that this thing already senses the flame to tell the pump or valve to stop sending oil into an unlit burner. Can it turn off the pump or signal your electronics?

Another possible pitfall - you never want to send oil into a hot burner chamber without venting it first - so there should be a combustion air blower running and/or time delay. There's a possibility that the parts of the burner will be above the flashpoint and sending oil in will vaporize a quantity (heavy white smoke) followed by an explosion.

There's probably some commercial device that will sense the flame and do it more safely than your photocell - normally thermocouples are used or flame ionization (flames conduct electricity - so one method is to put a high temp electrode along with a ceramic insulator into the path of the flame - flame cleans electrode insulator (spark plug may work)) a 200 Volt bias supply is enough to send current through a flame. "flame ionization detector"

I'd thoroughly research this first; you may be building a liability . . .

There are commercial photocell flame detectors, so it has already been done. Don't know how they handle soot or stay clean though. Heat may destroy the typical hobby photocells some use solder in the connections - maybe the commercial ones are hardened against it - gotta figure there'd be a lot of infrared heating even at some distance.
 

steliosm

Senior Member
I think PIR sensors (the one that can detect human movement) can also detect fire.
A good PIR sensor will give you a digital output, so no LDR and ADC is needed.

As for the SimpleLan part, make sure you are using a PicAxe chip with plenty of memory (e.g. 18X or 28X1). You are probably going to use a mail2sms gateway. As you can see here: http://www.steliosm.net/picaxelan.html the email functionality uses a great deal of memory space.
 
Last edited:

Dippy

Moderator
Not so sure about PIR+digital output. It may depend how flickery the flame is. I reckon you'll get on/offy results as they are designed to detect changes. Go and stand in front of one and stand still, after a while it won't detect you anymore.
 

steliosm

Senior Member
Dippy, if you stand still the PIR wont be able to detect you. The way those device work is by catching the Passive infrared Light (http://en.wikipedia.org/wiki/Passive_infrared_sensor). When a man moves the muscles create this Infrared light. When a man stops moving, his body doesn't emit an Infrared light.

Most of the PIR sensors are equipped with a small IC in order to control a NC output and keep it on for a few seconds so the alarm system can see it while probing all the sensors.

I had a PIR sensor connected to a 08m and I was using interrupts to sound a small piezzo buzzer whenever I was crossing it's 'area'.
 

Dippy

Moderator
Stelios: "When a man moves the muscles create this Infrared light. When a man stops moving, his body doesn't emit an Infrared light."

Blimey, where did you get that? Sorry, that is SO wrong.
You always emit infra-red. OK when you waggle your muscles or run around you get hotter and emit more.

I've used pyro IRs many many times and they (the ones designed for people-detection) are designed to detect changes e.g. person moving and usu focussed by the fresnel poly lenses. When you stand still the I/R detected from body stops changing rapidly and the sensor 'gets used' to the IR received. In these case the 'digital' o/p types just switch off.

If you don't believe me then try it.
 

steliosm

Senior Member
So, according to what your are saying, if you point the infrared sensor to a person standing still you get an alarm or not?
I mean, stand stil in the room and have some one turn the sensor until it can 'see' you. Will you get an alarm then? Based on what you're saying you should get an alarm. Did you get one?
 

Dippy

Moderator
I don't think you have understood me. Maybe I didn't put it clearly.
I apologise for a long boring post in advance.


You, or me, emit longish inra-red light , call it heat radiation if you like. We always will until we reach zero. We eat food, we burn it, we exercise, we get hot. School physics.

The commercial I/R (security) type detectors are designed to be sensitive to CHANGEs in IR.
They usually have dual or quad element sensors within. They have fresnel lenses on the front to focus (cf 'magnify') the IR from your body. Also the fresnel lens is designed so that your IR light / heat is focussed in an on/offy way - Data Sheets show this as 'beams' of sensitivity. (I know they are not beams per se but they are drawn that way to show the principle).

If you stand still in front of one for a short period of time they will STOP detecting you as there has been NO detectable change in I/R on the sensing elements. Or, another way, there is no difference in IR falling on or between the dual or quad elements within the sensor.

In a dual element version if one element is suddenly subject to IR but the other isn't then it assumes something is moving. If they both get nominally the same level of IR then they think that no IR emitting body is moving - hence they don't false alarm (hopefully) as ambient temp changes. The actual sensitivity will depend on the element type and the processing electronics.

If you wore an insulated very shiny suit and space helmet you could probably be a good burglar as long the place only used PIR. :)

Alternatively you could buy a micropile IR sensor and it will give an output based on your body heat. e.g. an infra-red point-and-shoot thermometer.

"So, according to what your are saying, if you point the infrared sensor to a person standing still you get an alarm or not?" - if the sensor is still and they stand absolutely still for a few seconds the device (assuming it being a commercial PIR detector) will not detect you. If it does then it is not much good as a burglar alarm.

I mean, stand stil in the room and have some one turn the sensor until it can 'see' you. Will you get an alarm then? Based on what you're saying you should get an alarm. Did you get one?
If you stand still and the sensor is moved (such that you move relatively across it's field of view) then it will detect you as it senses a moving heat source. I made a scanning passive IR detector once and could 'see' myself on a 'scope screen. I could even see my arms waggling.

Anyway, this has digressed from flame detection enough now, eh?
 

Michael 2727

Senior Member
PIRs work by detecting the change in IR between
2 X PIR elements, the Fresnel Lens helps to make
small movements in the transmitted IR (Body Heat)
swing across the face of the PIR Elements more readily.
The 2 PIR elements are connected to an OP Amp which
detects the imbalance in the inputs.
PIRs are elements tuned to detect IR Radiation very
close to the human body +/- a few Deg.

Hence if you creep up very slowly and/or directly in front
of a PIR or you have just come in out of freezing
weather you are less likely to cause a rapid change
between the 2 PIR elements and won't be detected.

I was a Security Tech once upon a time :p
 

hippy

Technical Support
Staff member
Connect a PIR up to a buzzer and play "walk across the room without setting it off" and you have an instant party game. It's the modern day equivalent of 'Grasshopper' in Kung Fu walking on the rice paper carpet without tearing it.

Takes a bit of practice but it can be done. Selling the swag after is the hard part :)
 

Dippy

Moderator
"When you can walk it's length and leave no trace, you will have learned." - or something like that.
"When you can take the pebble from my hand, it will be time for you to leave."
Classics.
 

hippy

Technical Support
Staff member
When you lever a PICAXE out of a socket using only your fingers, and it flips round, 'bites', and leaves a neat set of two rows of holes in your thumb ( with blood gently oozing from each ), think back to carrying the couldren of hot coals and remember you too can have the marks of a true Kung Fu warrior. The holes don't last too long, but the painful memory does :)
 

WHITEKNUCKLES

New Member
How did we ever manage before TV?

Older technology Flame Failure LDRs for direct connection to relays.
Protected behind single heat absorbing glass, looking down the length of 12 foot long atomised oil flame.

Dave
 

Attachments

papaof2

Senior Member
LDRs for oil flames because the flames are yellow/red and the LDRs used have peak response in the red/yellow region.

UV detectors for gas flames because they are blue to nearly clear and give off UV.

Some more reference materials:

http://articles.jimtrade.com/1/20.htm - about "gas rod" detectors (an ionization detector) for gas flames

http://www.indiamart.com/paragonauto/electronic-switches.html#flame-failure-detector-timer - detailed description of a commercial flame failure unit with all the safety features (restart, timeout on restart, etc)

http://stroco.dana6.dk/data/images/medieupload/35mgb.pdf - 1MB PDF on operation & maintenance of an oil burner, with a cutaway diagram of the placement of the oil nozzle, igniter, and photcell

John
 

demonicpicaxeguy

Senior Member
why not ust stretch a piece of nichrome wire across where the flame should be, then making use of the picaxe adc surely you should see a difference between a flame and no flame, regardless of fuel, unless the wire melts of course.....
 

andrew_qld

Senior Member
With an LDR, you would have to experiment think, to find the one that responds the best to the "colour" of the flame.

I think the easiest thing would be to put a thermister (or other industrial temperature sensor) into the flame and feed it to a picaxe analogue input.

Before anyone says that there would be too much hystersis (ie it would take time to "cool" after the flame went out), you would see a sharp temperature drop when the flame went out and you could get the picaxe to monitor that rate of change.

Andrew
 

craigcurtin

Senior Member
Thanks for all that - i think

Hey Guys, Thanks for all the input once more - especially the digressions into PIR and Kung Fu moves. !!

I will start with an LDR, then can always fallback on a Thermistor if required

I response to one of the posts re spontaneous combustion etc of the oil - babingtons work by atomizing the oil running over a spherical surface (usually using compressed air) as the atomizing agent. In the event the flame goes out the combustion tube will not spontaneously ignite the oil again. My goal with this is to turn off the solenoid that controls the air compressor and also the pump that delivers the oil.

WIll give feedback as i continue to test

regards

Craig
 

KMoffett

Senior Member
I'd go along with Andrew, but buy a replacement furnace pilot light thermocouple. These are relatively cheap, meant to go directly in the flame, and output a voltage that could be read by a PICAXE ADC10 input (5mV steps). The output of these are either 250mV or 750mV in the flame, and near 0mV at room temp, but enough to easily distinguish large temp (flame/no flame) changes.

Ken
 

premelec

Senior Member
To build on the idea of a thermocouple... I have extracted solenoids and their thermocouple from old gas water heaters - there is a spring loaded valve to cut off the gas when the thermocouple is not hot and this assembly could be used to hit a oil pump cutoff microswitch - with some sort of latch to determine that it _should_ be hot rather than the pump is off. Aside from this there are thermopile 'millivolt' type thermocouple arrays used which produce a much higher voltage than the single thermocouple which would be easier to detect a voltage from... also used in the heating industry - these are just a bunch of single thermocouples in series in a fatter array.
 

craigcurtin

Senior Member
Thanks Guys,

I have a solar hot water shop around the corner who always have old gas and water heaters out the back i will check them out

Craig
 
Top