Keeping An Eye On The Water Heater Pilot Light

[WJCarpenter]’s gas water heater uses a small pilot light that needs to stay burning permanently to ignite the main burners as required. Four or five times a year, the pilot light goes out and needs to be manually lit. This involves an expedition from the upstairs bathroom to the basement, always in the early morning, after having spent a few fruitless minutes waiting for hot water. Having grown tired of this exercise, [WJCarpenter] built Water Watcher, a pilot light monitoring system with some ESPs and a light sensor.

Water Watcher consists of an ESP8266 connected to a light sensor taped to the inspection window of the water heater. It reports the status of the pilot light over MQTT to an ESP32-based M5 Atom Matrix in the main bedroom, which displays it using a 5×5 RGB matrix, as demonstrated after the break. Both ESPs run ESPHome, so programming is as easy as giving it a YAML config file. [WJCarpenter] tested a few different light sensors, until he found the TSL2591, which is sensitive to the right wavelengths and has enough dynamic range for watching a pilot light.

This might not be a complicated hack, but we do not doubt that it reduces frustration a bit on those fateful mornings. Be sure to check out the Water Watcher project page, it’s an entertaining read!

44 thoughts on “Keeping An Eye On The Water Heater Pilot Light

  1. Lovely hack.

    Normally when I start to see my pilot light go out every couple of months, it’s time to replace the thermocouple which is usually pretty quick and under $10. Then I don’t have to worry about the pilot light for years to come.

    1. Lived for a decade plus in a house where the pilot went out every windstorm. The flue configuration and orientation of the house to prevailing winds did it, but there was no fixing the problem short of major construction. It happens, in that case exacerbated by tree removals that changed the way the wind hit the house (the trees were 100+ year old and planted as windbreak in the first place)

      I don’t miss that house.

    2. Aren’t there new burners with a catalytic mesh, that don’t need a pilot light because it’s technically always burning just by having a tiny bit of gas flow? You can dial those way down without the burner going out.

  2. My pilot light was driving me nuts, going out seemingly at random… well the clothes drier is right next to the hot water tank and I began to suspect that when the door of the drier was opened too enthusiastically, it was blowing out the pilot… so I limited the travel of the door, and it’s been fine for many months now, touch wood.

    1. If the air change is having such an effect I’d suspect that there is not enough ventilation in that room.

      How close is this drier to the water heater that is probably not good with a gravity venting appliance anyway..

      1. It’s not air change, it’s like how to blow out a candle with a chessboard. Bottom of the drier door is only a foot off the floor, I guess it probably wouldn’t have done it if we had pedestals, but the washer is top load, so not really a pedestal candidate.

        Ventilation is abundant, there’s a FAG furnace there too with ventilation to code, which wasn’t even running the times it went out through April to October.

  3. Had a similar problem, caused by excessive carbon buildup on the thermocouple used for sensing the pilot was lit. Three hours to tear that section of the water heater apart thanks all the stuff in the way due to low NOx regulations and all the gaskets that had to be replaced (cost a fair amount more than $10 as well). Life in lovely Taxifornia.

    This hack reminds me of a similar one I did on my first car as a teen. Had a headlamp do out which I could recover by giving it a proper thump to temporarily weld the broken filiament ends together, good enough for another few days. Didn’t have a lot of cash on hand for a new sealed beam (it’s been a while since I was a teen) but had an phototransistor, LED, resistor, and wire in the parts bin to make a remote monitor.

    1. >> Had a headlamp do out … parts bin to make a remote monitor

      I was once in a car (maybe big Detroit iron from the 70’s?) that had little plastic light pipes in front fenders that came out the top into a little backwards-facing chrome fitting allowed you so see the state of the headlights from inside the cabin.

      Thought it was cool, if a bit over-the-top. Headlamps musta sucked in the 70’s, I guess

      1. Not true. Perhaps they don’t generate as much as cars, but they do produce NOx. At least, there are regulations in my 90% natural gas heated country because of central heating gas burner emissions.

  4. I actually have an ESP module based boiler telemetry system – it has thermistors on the domestic hot water and hydronic supply/returns to the (hot water) boiler. Also has an ultrasonic rangefinder to measure the depth of the fuel oil in the tank. This was also inspired by a cold shower…

    1. Just you, or how many other bodies? Any appliances need hot water?

      Spitballing my numbers on a tanked water heater, natural gas, occasional warm water use in washer, cold most loads. I think I’m around 400k BTU in 6 months with around 3.5 bodies and I’m seeing 432k BTU per 20lb of propane.

  5. I wonder how hard it would be to just go the next step, and maybe re-purpose say the “spark module” (don’t know the name) from like a pilot-less gas range-top to make a spark gap to relight the pilot.

    I mean you could even use one of the piezoelectric-spark cigarette/candle lighters, if you could figure out a reliable way to “press down” the the button for the piezo spark clicker, but that starts getting a bit too Rube Goldberg for reliability.

    1. i’ve repaired a few diesel heaters that use high energy spark modules to ignite the jet of fuel , often used in hospitals in africa for steam in areas with bad power. They make quite a odd start up sound, first you hear the high crackle of the spark generator then he high wine of the high pressure pump as it builds and then the valve/injector opens and it transitions to a jet engine with lightning sounds then the coil kicks out and its just the jet engine noise. we had a young technician who had just been trained on installs of these units and he decided to try one outside of the safety chamber that is the boiler itself, after he bypassed every fail safe he proceeded to fire it up and it did what jet engines do, moved darn quick from his location to the wall. thankfully the power was pulled and that stops the fuel injector so no further operation occurred but a good reminder that fail safe’s and junior technicians don’t mix.

    2. >> make a spark gap to relight the pilot.

      You don’t really need a spark, my natural gas furnace has a simple ceramic heating element that does the job.

      It gets energized a few seconds before the gas is turned on, and gets yellow hot. Apparently, that is enough for initial ignition, and it’s an off the shelf part that only needs a relay to mains power.

      Interestingly, there’s some kind of mechanical interlock on the valve so that the main gas doesn’t flow until a probe tip is hot, indicating an ignition source is available. I’ve seen the same thing in normal piloted appliances, where the probe tip extends into the pilot flame.

      I’ve always wondered just what the mechanism is, anyone know?

    3. It already has that piezoelectric sparker for manually relighting the pilot. So, it’s a simple matter of building a robot arm to …. Hmm, there we go again!

      You would need to coordinate the pushing of at least one button no matter what. When the pilot light goes out, the gas supply to the pilot light is cut off. The thermostatic controller has a button (they call it a reset button) that you have to hold down while pushing the piezoelectric sparker button. Then you have to keep holding the reset button for 30-60 seconds after the pilot is lit, else the gas supply is cut off again and the pilot light goes out.

      Any robot arm that I built would not be able to handle that. I’m sure the robot would fail within 30 seconds. :-)

  6. The newest water heaters I’ve seen for gas have LED’s glowing and an electronic control but no battery or line hookup. Thermocouple pilot powered I guess. It takes more than one junction to light up a LED.

  7. Are pilot lights still a thing stateside? Not had boilers with them for a good 10-15 years here in the UK (and if your boiler is older than that, you’re well worth upgrading for the efficiency savings!)

    1. I’m in the US and I remember my parents’ and grandparents appliances having them. I’m in my second house, this one being a duplex and have yet to own anything with a pilot light. Gas water heaters and gas stoves on both sides, they all start themselves as needed by electric spark.

      It’s news to me that you can still get appliances with pilot lights here. Is there an advantage?

      1. thinking about it pilot light keeps a bit of heat under the tank does that affect longevity (good or bad?)

        I imagine the pilot light burning could be safer in some situations and less safe in others

        the only concrete con I can come up with is that you are constantly wasting a bit of fuel with a standing pilot

      2. We’ve had problems keeping the pilot lit on our Nat. gas fireplace.
        One technician told us to clean the glass window from time to time as the pilot light (when working) leaves a residue on it.

        We hope to someday replace the pilot with an on demand ignitor.

      3. The water heaters are completely independent of mains electricity, well, mine is, so in event of power outage, knocking out the heating blower, at least you could fill hot water bottles.

    2. Unless your boiler is so old that safe use is questionable I can’t see how spending money to upgrade could possibly pay off in the 10-15 years these things are intended to last especially with the declining quality

  8. As the ultimate goal is hot watet, I’d probably add a temperature sensor attached to the outside of the water tank wall inside the insulation. A 1-wire DS18B20 is cheap and very easy to interface. Temperature modified display patterns are the obvious next step.

  9. If a pilot light goes out that often, you need to call a repair man. I used to be that repair man and the gadget you are trying to sell is a joke. Tape it to the window 🤣😂 And most water heaters today have electric ignition. My goal was no return call. And all water heaters that I worked on, stayed working for years afterward. Not bragging, just there is not a lot to a water heater. Clean pilot, replace thermocouple and your good to go for years to come.

  10. In my previous house in Kansas, the pilot light on the heater would get blown out every time I turned on the attic fan before opening a window. House was tight enough to pull air down the flue and blow it out. House built 2001.

  11. people with newer hotwater heaters dont realise that they TOO go out, maybe even more often! the pilot auto re-light cycle is only a few mins (10mins?) and the water stays hot way longer then that…

    its when the pilot auto re-light relay goes click, click, click-whoosh, click-whoosh, chick-whoosh, click, click, click, and gives up after 8 tries.

    its when the heating goes out due to wind you notice much more, as your house is ALREADY 1 degree “cool” by the time heat is demanded, then the temp falls even further as it attemps another 8 tries and just looks pathetic.

    older heaters actually work in a wind storm.

  12. Some of the comments suggest the water heater might be “out of tune” if the pilot light goes out, and it either needs adjusting or it might even need replacing. That could be; I’m no Water Heaterologist.


    — There is a pretty strong correlation between gusty conditions and the pilot going out. Not every gusty condition blows out the pilot, but when the pilot goes out, it has been gusty. (I don’t think it’s the speed of the wind that matters. It’s the unevenness of the gusting.)

    — The water heater is a few years old now, but the first time this happened was no more than a month or two after it was installed. That’s when I talked to the Water Heater Guy who told me the exhaust stack was not tall enough above the roof. Now, he might have been blowing smoke (clean, natural gas smoke!), but the theory sounded reasonable.

    Several times, I’ve tried to find info about this via web searching, but I never found an article that said “if your stack is too short, it can cause your pilot light to get blown out”. Does anybody have any real know-how about that?

    1. I know that stacks, flues, chimneys should be 3ft above top of roof and everything around them to draw steadily, so if it’s shorter than that, or halfway down the side of a dormer or something, that’s probably it.

      Even then, if you’ve got a significantly tall tree in one direction, when the wind is blowing from that direction, you might get something best envisioned as the “waterfall” of wind off the top of it on the backside. The strength and exact position of that downdraft varies with how strong the wind is. I think you’re usually clear if it’s twice as far away as it is tall above your chimney, i.e. 25ft higher, needs to be 50+ ft away.

      1. I got out the manual and found this requirement in the installation instructions:

        — 2 feet above anything within 10 feet
        — 3 feet above the surface of the roof to the bottom of the vent cap

        Mine is within a few feet of a vertical wall and less than 1 foot from the roof to the bottom of the vent cap, so it loses on both counts. It at least vindicates the comments from the Water Heater Guy who told me it was too short.

        Adding some kind of vent pipe extension to get to 3 feet above the roof doesn’t sound too bad, but getting above that vertical wall would need another 10-12 feet of extension sticking straight up out of the roof. Maybe I could pretend it’s a flagpole!

        1. So it would definitely be worth going up to 3ft, whether it would be worth going up to clear the wall depends on local prevailing winds. Typically, if the wall is to your east or north, it wouldn’t be a problem 80% of the time, but if it’s to your west or south, then that’s a very frequent problem. Possibly, one of those tilting hat type flue caps, or one of the ones with a fin to turn it away from the wind might help.

  13. Very nice project. I’ve done something similar for a radiator heating boiler but settled on just using a 2nd thermocouple in the flame as the the sensor. I’m going to try to adopt the TSL2591 as the sensing element. I too started with an optical flame sensor and did some research and found the spectrum sensitivity inadequate but didn’t go as far as the author did. Cudos to him.
    In my case I alarmed the pilot light because the boiler thermocouples fail every 12-24 months. For all the folks that argue for solving the original problem, sometimes it’s just not that easy. Cleaning, multiple technicians, replacing the gas valve, a long talk with Honeywell (thermocouple) representatives all yielded no real solution. Hard to debug something when you need to wait a long time for a real failure. The Honeywell folks agree the lifetime should be longer but fall back to all eventually fail excuse. I’ve been told the only solution is tearing out the whole thing but that is like blowing up the house to get rid of mice. I suspect the thermocouple overheats when the boiler fires and over time metal fatigue takes over. I have measured the failing thermocouples and they tend to be in the low 20s of millivolts rather than the 25-35mv range of newer ones. Changing thermocouples isn’t difficult and one day the system will get changed out, in the meantime it’s the notification of failure and a quick replacement the monitor is trying to solve.
    I’d like to try the optical sensor because my duplex thermocouple can be a pain. I have been surprised at how sensitive the placement of the monitoring thermocouple is. Interestingly an improper placement can result in the monitoring thermocouple lowering its output to the below 20mv (causing an alarm state) during the time the boiler is firing. Probably overheating again and kinda feeds the hypothesis about failure mechanisms. I did put hysteresis in the program to delay alarming but I’ve had to change the timing depending on the thermocouple itself, the placement and the weather conditions. Overall it works, but I tweak the program almost every season, so it’s not ideal. Perhaps the optical method will allow some distance for the sensor and it will be more consistent.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.