Multiple thermocouples for flame detection

[Buriedcode]

Hi,

I have been asked by a colleague about using multiple thermocouples for flame detection in an outdoor gas burner.  The trouble is, there is nowhere to mount the thermocouple in the flame (its quite a wide flame) only to one side.  When a thermocouple safety valve was fitted, when the wind blew the flame would no longer be on the sensor, so the valve cut out - as it should if it detects no flame.  There is no provision for a pilot light either.

So, using 3 or more should guarantee at least one is in the flame, regardless of where the wind is blowing.  But of course I can't just 'OR' these together with diodes given the output is ~25mV.  Would connecting them in series be a bad idea?  As the connector uses the mounting as its negative connection, it would be fiddly and non-standard but I would prefer to not have an active circuit as that would just require power, an IP rated enclosure and added complication. 

I also suggested tilt switches in series so that the gas cuts out if the burner stand falls or tips.  That should mean if the flame gets blown out, or the stand tilts/tips/falls the gas will be cut off.

[Vgkid]

What about using a precision comparator, and OR gating those.

[LaserSteve]

Have you considered simply paralleling the TCs?  Then you can perhaps still use the same safety valve...
If the TCs are reasonably matched the output will be the average of the two outputs within reason...


Steve

[joeqsmith]

Why use TC's if you just need to know if the flame is burning and not the temperature?   Can you instead just use FID? 

[NiHaoMike]

Look up "flame rectification".

[Buriedcode]

Ok well, I was trying to keep things simple.

At the moment, there is one thermocouple controlling a valve - a 'flame supervision device'.  The user opens the valve, and ignites, once the thermocouple heats, it powers a small solenoid which holds the valve open.  Good so far.  This also means no power supply, extra circuitry etc..  its also very common in gas cookers, and outdoor patio heaters, so the valve and thermocouples are cheap. Of course I'm not particularly interested in absolute temperatures, just need to know if the gas is 'lit'. http://www.upperplumbers.co.uk/images/Text/Gas_safety/Thermo-electric-flame-supervision-device.jpg

I considered putting them in parallel, but I thought this would give an 'average' voltage between two thermo's.. so if there is one either side of the flame, and the wind blows so the flame is on one, that produces enough voltage to hold the valve open, but.. as its in parallel with the other, which is relatively cold, the paralleled output is half way between these two voltages (assuming they are well matched), which may not be enough to hold the valve open.  In series, there is the resistance of one in series with the other, but means that the voltage should be enough.  Series is a hassle as these things come with a copper rod as the negative, and a wire as the positive, so an enclosure to make the connections would either have to be plastic, or a metal enclosure with insulating bushing.

I assume an FID is a flame ionization detector. Seems like a lot of analogue which would require portable power.  There is no pilot light, the burner is essentially a large shallow 'bowl', with the flame front significantly higher than the gas outlet - that's what makes placement of the Thermocouple so annoying.  Any other detector would require the same placement (flame rectification requires something *in* the flame no?).

I'll pick up a few and set them up in series, and then parallel and see what happens.  I'm surprised there aren't examples of using multiple thermo's in this way, but then again most commercial devices either have a pilot light, or designed around the thermocouple being in the flame.

[Gyro]

The thermocouple and associated hold-in solenoid in the flame failure device work at surprisingly high current (for a single thermocouple anyway) - 10s of mA, but VERY low emf, a few mV at most, hence the solid copper construction of the 'lead' between them, it's a very low impedance circuit, the solenoid is a few turns of quite thick wire. It's very unlikely that you would be able to break in to it and hook up several thermocouples in parallel without adding excessive resistance and losing all your emf in the process.

More importantly you need to consider that the flame failure device is definitely a safety item and you really shouldn't mess with it. Series connecting several thermocouples could well delay the gas shutoff by far too long. You could maybe try re-positioning it a bit closer to the flame, but it sounds as if it's serving its function - guarding against a flame that's in danger of blowing out.

[tszaboo]

I would replace the thermocouple with an RTD. They can be connected in series, the output will be the average* of the temperature and it is simpler to measure. PT100 or PT1000.
Maybe you can parallel them, but I believe calculation will be more difficult in that case.

*not really, but close enough.

[Richard Head]

Why don't you use a non-contact IR sensor?

[Gyro]

I'm not sure if I understand the full picture here - I was assuming that the flame failure cutoff valve is an integral part of the burner, as delivered???

If so, then whatever flame detection is used with it needs to work with that cutoff valve, ie. the thermocouple, you can't bypass it or modify it.  If it's a later add-on, then of course you can consider other options.

[Marco]

The thermocouple and associated hold-in solenoid in the flame failure device work at surprisingly high current (for a single thermocouple anyway) - 10s of mA, but VERY low emf, a few mV at most, hence the solid copper construction of the 'lead' between them, it's a very low impedance circuit, the solenoid is a few turns of quite thick wire. It's very unlikely that you would be able to break in to it and hook up several thermocouples in parallel without adding excessive resistance and losing all your emf in the process.

He wants to connect them in series, ie. a thermopile. If one thermocouple with the flame centered can drive the solenoid then three thermocouples in series certainly can, triple the voltage can easily overcome triple the output resistance. If the flame is off center one of the thermocouples will actually generate a larger signal, this will probably be enough to offset the increased output resistance as well ... hard to say for sure though.

[Buriedcode]

The thermocouple and associated hold-in solenoid in the flame failure device work at surprisingly high current (for a single thermocouple anyway) - 10s of mA, but VERY low emf, a few mV at most, hence the solid copper construction of the 'lead' between them, it's a very low impedance circuit, the solenoid is a few turns of quite thick wire. It's very unlikely that you would be able to break in to it and hook up several thermocouples in parallel without adding excessive resistance and losing all your emf in the process.

More importantly you need to consider that the flame failure device is definitely a safety item and you really shouldn't mess with it. Series connecting several thermocouples could well delay the gas shutoff by far too long. You could maybe try re-positioning it a bit closer to the flame, but it sounds as if it's serving its function - guarding against a flame that's in danger of blowing out.

This.  It was not part of the original burner, but added later as a mandatory safety feature (the company runs 'events' and these gas burners are part of the show, someone is always on site with them).  You also highlighted what I suspected... very low output voltage (~25mV) driving a very low resistance solenoid (<1ohm).  This also makes adding a tilt switch troublesome as mercury free ones tend to have >1 ohm resistance.  With that said I'm fairly sure there are larger tilt switches that use microswitches and a 'weight' rated for @3A @ 125VAC that should have minimum voltage drop.

I agree that one should leave this system in place, perhaps with an industrial tilt switch in series, as its directly controlling the valve with no extra electronics to go wrong (failure from low battery, connector corrosion etc..).  However, if we simply cannot get the thermocouple close enough to the flame for reliable operation, I fear I may have to use active circuitry, and that is where I'll have to be clever about it.

Simply measuring the voltage of several thermocouples and controlling the valve output is one way, but even driving the solenoid will be inefficient (30mV @ say, 30mA?) unless I use the thermocouples power, to power the solenoid. With several thermocouples connected to the drains of MOSFETs, that all have their sources commoned: a magnitude selector, that connects the thermocouple with the highest output to the solenoid. I suppose in effect, groups of ideal diodes.   Would have to have a make-before-break selection, with rather tight timing because a break would trip the valve closed, and a sudden addition of another thermo (whose output is just starting to overtake the previous highest) connected in parallel would also cause a minor glitch..

It would of course still need a power supply, and a rather high voltage (>5V) one to make sure the FET's are fully turned on, and to keep their on-resistance down, but as the switching will be very low duty (hopefully) the actual power consumption could be low - at least as far as driving the switches is concerned.  Comparator circuits would still draw a few mA so a PP3 would probably get ~100 hours out of it. Whilst its cool, its probably wise to not power this with a solar cell and a rechargeable batt.

The above idea is pretty much the 'worst case' scenario.  I'm hoping to see the actual burner at the weekend, and try to convince the owner do drill a hole in the basin for the thermo, negating all of this.

Cheers for the help!

[Buriedcode]

He wants to connect them in series, ie. a thermopile. If one thermocouple with the flame centered can drive the solenoid then three thermocouples in series certainly can, triple the voltage can easily overcome triple the output resistance. If the flame is off center one of the thermocouples will actually generate a larger signal, this will probably be enough to offset the increased output resistance as well ... hard to say for sure though.

That was my original thought.  A bit of a bugger connecting them, as the heavy copper 'wire' (harder to bend than a coat hanger) is one terminal, so would require insulation to any junction box.  I doubt 3 thermocouples in series at ambient temperature would have enough oomph to hold the valve open - either all heated to reasonable temperature or one full temp.  The lag in cooling from one to heating or another as the flame dances between them should prevent any sudden dips in power to the solenoid, and its a passive solution, albeit one I'm still not overly comfortable with.

Once I get them to play with we'll see - playing with fire is usually fun...

[Gyro]

Yes, hopefully it would work, but as you say, a pain to do. You'd want to connect them fairly remotely from the flame and there's the issue of non-shorting mounting the thermocouples and routing the 'wires'. There might be other unforseen problems like mineral insulation and whether the center conductor is solderable.

Hopefully you can avoid all this with the hole drilling persuasion. Good luck.

[max_torque]

Would some for of "windshield" not be a better, simpler option???

[Buriedcode]

Yes, hopefully it would work, but as you say, a pain to do. You'd want to connect them fairly remotely from the flame and there's the issue of non-shorting mounting the thermocouples and routing the 'wires'. There might be other unforseen problems like mineral insulation and whether the center conductor is solderable.

Hopefully you can avoid all this with the hole drilling persuasion. Good luck.

Well, it seems mounting the thermocouple in the 'flame' is trickier than I thought. Here's the burner with a thermocouple controlled valve.

As you can see the flame front is generally outside of the shield, and can easily be blown away from the thermocouple.  The colleague who asked thought '6', which I think is overkill, 3 might be more appropriate.  I am now leaning towards a different form of sensor, one that can tolerate being in a flame, but is more of a sensor than something used for power.  Its a pity because as I said it would be cool to use the thermocouples current for the solenoid, leaving any powered circuitry to be pure control - low power.

Then of course there's the matter of rain.  With 3 thermo's arranged around the outside, they would need some form of hood above them in the case of 'very light rain' where its not enough to extinguish the flame, but enough to significantly cool down the thermo's.

Absolute bugger of a problem this is, and I can see it ballooning into complexity and becoming somewhat unreliable.  Also, there are 12 of these to modify!  Whilst I have the thermocouple here, I shall do some voltage/current measurements to see what the solenoid needs, and what the thermocouple outputs.  I suspect the holding current for the valve wont' be much.  This is why I love low powwe electronics, dont' ahve to deal with the mechanics

[CatalinaWOW]

You might try some mechanical solutions, since the problem is fundamentally "mechanical".  Like mounting the tip of the thermocouple in thermal contact with the perforated shield.  There might be enough heat transfer in the shield along with added thermal mass to deal with the problem.  A heat pipe might be used to bring heat from the entire circumference to the thermocouple.  I am assuming the the thermocouple is already mounted downwind for the prevailing winds.  Finally you could plumb the same gas source to a separate, shielded flame (perhaps inside the flame shield) where you would mount the single thermocouple.  This latter solution would be a lot of work, but likely less than many of the solutions that have been proposed here, and once implemented should be quite reliable.  Could even be set up to act as a pilot for the primary flame.

I doubt that paralleling will work.  The added sensors provide a low impedance path in parallel with the solenoid coil and so would steal current from the "high" thermocouple.  A series connection would work in principle, but requires attention to each connection to avoid causing reverse thermal junctions and killing the whole thing.  As you say a royal PITA.

[Gyro]

Arrgh, Not what I was visualising as a burner - that's more of an Olympic flame! I thought you were talking about something like a cooker ring with a nice controlled hot blue flame.

I agree, inside the shield would probably be a better location.

P.S. I wouldn't want his gas bill if there are 12 of those! 

[Buriedcode]

Yeah its not you're average patio heater, its for special effects and events (pretty much dancing yellow flame that just...well...burns... gas).

Tomorrow I'll run some tests, CatalinaWoW, excellent point! While the flame won't always be inside the shield, the shield is bound to heat up a fair bit, and stay hot where-ever the flame goes.  Its jsut a question of whether that temp is high enough to hold the valve open - something that came only be done by testing.  THe fitting around the thermocouple won't fit into the holes on the underside of the shield, and the customer (its just easier to call him that) isn't keen on machining it.  So in true DIY style, I was thinking... steel cable ties.  To just hold it against the shield and hope the thermal contact is hot enough.  If not, I'll remove the collar and try to mount it inside.

A pilot light was my preferred option, as you pointed out would require an extra fitting, and I think the pressure is regulated to be quite low.  I guess the only think I can do is mount it, and test it. Fun!

[CatalinaWOW]

Another approach that might help would be to apply bias in some form to the solenoid.  A permanent magnet, or an adjustment of the spring in the solenoid.  I have done this is in the past, and don't really recommend it for a variety of reasons.  Very fiddly, though stable once achieved.  Totally unacceptable from a safety standpoint in most applications where these valves are used, but in reality might be OK for the application that you are chasing, since escaping gas will not be trapped and the situation is not un-monitored.  If the gas mains are off when no one is there I personally wouldn't worry about the safety.

[LaserSteve]

Yikes, poor initial burner design..
In my gas furnace, the pilot flame is directed to the thermocouple. Perhaps he needs a internal pilot flame, out of the wind....
All that CO2 generated must attract mosquitos..

Steve

[splin]

How about a TEG (thermo electric generator)? It could be mounted to the case with the cool side air cooled or even cooled by the incoming gas pipe? The response time would probably be rather slow but it could be improved by mounting it offset from the case so the hot side is heated by radiation rather than heat conducted through the relatively large thermal mass casing. This would probably also allow you to use a cheap Peltier cooler rather than a TEG, but you'd have to be careful of the temperature rating which can be 80C max on standard devices.

They have relatively high outputs which can be increased by using larger devices or more in parallel, but depending on how you mount and cool it, you might have the opposite problem and need to reduce the output with a parallel resistor. It would need careful setup to ensure it cuts out reliably and fast enough in different weather conditions, especially wind strength and direction. Using the gas pipe to cool it would help a lot in keeping the cold side at a more consistent temperature than relying on air cooling.

You would need to seal them carefully to prevent moisture getting in but at least it should fail safe which would be more difficult to ensure with an electronic solution.

[Buriedcode]

hi splin,

The 80C max temp of those devices would be troublesome, as the flame is bound to hit whatever I use to sense temperature at some point.  Unless maybe its on the underside, but even then, >80C.

I think I would have to order the thermos anyway, because that has the fitting for the solenoid valve.  If I did use one, hopefully I could find one with lower output voltage.  Also, it'll be outside, most likely in all weathers (sure rain will kill the flame, but I think these are stored outside too..).  It is something to think about though, I might pick one up for a tinker..

[johnwa]

Instead of trying to parallel the sensors, what about paralleling multiple valves? If this is not practical, then would it be possible to modify the solenoid to have three independent windings? You would need to maintain the ampere-turns and wire cross-sectional area for each winding, so the magnetic circuit might need to be modified.

Probably not a very practical idea overall, but it should still be fairly fail-safe.

[LaserSteve]

There is always solar blind UV flame detection, but that is a tad expensive for a backyard project. Unless your client is Scarlet Johansson.  UV flame detector tubes look like a little neon lamp, and are basically a gas filled photodiode. Commercial modules are 139$ each, but you might need three or four to cover a burner optically...

Aka UVTRON R2868 , in its simplest form, you use it in a  relaxation oscillator that pulses when the flame is on...

Honeywell makes complete detectors, Hamamatsu makes the raw tubes.



Steve