Pilot Light Monitor for Hot Water Heater

We live out on the plains where it gets windy. The pilot light on our water heater frequently blows out and we get to take cold showers. I had purchased a flame sensor, thinking I could just point it towards the flame and then turn on an alarm if it no longer sensed a flame. It seems the flame sensor only senses an orange flame and the pilot light is blue. There is a very slight change in the analog signal, but not big enough to be reliable and it's very tricky to point to achieve this change.

I'm no electrical engineer, so I don't really know how the sensor itself works. Is there a way to alter the frequency that the sensor is sensing? Or does someone else have a brilliant idea I'm missing.

I'm sure someone will say get a new water heater. It is only 2 years old and due to certain circumstances with the install a replacement is quite cost prohibitive.

The sensor I purchased is here:
http://www.amazon.com/gp/product/B00AFSEC2Y/ref=oh_aui_detailpage_o02_s00?ie=UTF8&psc=1

I wired one of these to a furnace plilotlight gas supply line (which is hot as long as the flame is burning).

It looks like a standard IR receiver. If it is then the answer is "no, there is no way to change the frequency."

Try adding a "shield" to the sensor to reduce the amount of ambient light that can reach the sensor. Something like a soda straw.

According to NIST, the infrared emission for natural gas is at a considerable longer wavelength so the sensor may just never work reliably.

LarryD:
I wired one of these to a furnace plilotlight gas supply line (which is hot as long as the flame is burning).
Waterproof 1-Wire DS18B20 Digital temperature sensor : ID 381 : Adafruit Industries, Unique & fun DIY electronics and kits

Is it inside where the burner is or somewhere outside the burner. I could place another thermocouple in the flame and sense that, but thought the flame sensor was a simpler solution. I do have 4 additional max31855 boards, but no fiberglass jacketed thermocouple wire.

We're rural, so it's propane.

Water heaters with built-in flame sensors usually use a thermocouple positioned right in the flame. This generates a few millivolts, which is enough to power the extremely efficient circuit.

Even at the temperature of a flame, I think the thermocouple will still need amplification to be fed to the Arduino's analog input but you don't need super precision.

The optical flame sensor is neat but you can't change its wavelength.

It's looking like a secondary temp sensor of some type is the best solution. Thanks for the input guys. If someone has another brilliant idea, please post.

Post #1
Clamp one to supply line.

Have you tried the digital output with different settings of the potentiometer? Apparently that should give a clear on or off.

The heat from the pilot flame creates an electrical current in the thermopile circuit that operates a small spring-loaded electromagnet in the gas valve. The electromagnet holds the main gas supply line interrupter open, allowing gas to flow to the pilot and main burner. If the pilot goes out, the electromagnet will no longer receive a current and the spring-loaded interrupter will shut off the gas supply.

You may be able to sense the magnetic field with a Hall sensor.

The amazon's one has 760nm to 1100nm

Plan A: Analog UV Light Sensor

240nm-380nm

Plan B. matching pair of photodiode with no attenuate.

http://forum.arduino.cc/index.php?topic=162837.msg1216731#msg1216731

Visible Photo Diode enhanced for the Blue/Green Spectrum, 450nm-550nm

Plan C: 67MM Goja Full Blue Color Filter + Typical silicon photodiode spectral response

Color Filter

http://forum.arduino.cc/index.php?topic=162837.msg1216714#msg1216714

Typical silicon photodiode spectral response

http://forum.arduino.cc/index.php?topic=162837.msg1216700#msg1216700

Select whatever Filter base on wavelenth you need.

Use soda straw to reduce detection angle as well filter or narrow band photodiode to reduce the amount of ambient light.