# Switching a millivolt gas valve with a low-voltage digital timer

I have one of those old gas heaters that has a millivolt valve powered by a thermopile. I want to control it with the brain of a digital programmable outlet timer (minus the AC switching components - this is just the controller board) so that I can program it to come on for an hour in the morning before I wake up, without having it running all night.

The valve runs on 0.5v, and when connected draws a current of 75mA. To actuate the valve, you simply have to make a circuit between two contacts.

The timer is powered by a 1.5v AAA battery. When it switches on, it makes a circuit between two contacts. I thought that I could simply wire these two contacts to the heater valve contacts to actuate the valve. Unfortunately, the amount of current that makes it through the timer is only 1.85mA, which is not enough to actuate the valve!

I'd like to keep this setup as energy-efficient as possible so I don't have to connect a power cable or swap batteries all the time. The valve is great because it's powered by the heater, and the controller is great because it draws like 5µA from a single AAA battery. Is there a way to use transistors or other components to give the valve the 75mA that it needs?

Thanks!!

Update #1:

simulate this circuit – Schematic created using CircuitLab

I've tried simulating the MOSFET circuit on the left, but even when I set the gate threshold to 0.2v, I can only get the drain-source current up to 75mA if I pump like 100v into the gate, which obviously won't work. The circuit on the right would be great if I can find a relay with a really low coil voltage that consumes very little current. Any ideas for either one?

Update #2:

I modified dmitryvm's schematic to eliminate the 1.5V battery and add a resistor to represent powering the timer. I also added a switch that will hopefully work in practice when I use the timer as a switch. Here's the new schematic:

simulate this circuit

I did some testing of the transistor, and it seems to work, although I had to crank up the voltage to 10V because my multimeter's resolution is 10µA.

Update #3:

Success!! I discovered that the timer actually puts out 1.5v when switched on, so I was able to trigger the FDN337N with it. What a great little MOSFET! Thanks to dmitryvm for the recommendation. I even made my own PCB. The timer is now programmed and connected to my heater. Here are photos of everything:

Ugly but functional Eagle schematic

Eagle PCB layout

Timer and PCB

Timer connected directly to heater (temporary until I have time to hook it up in series with the thermostat).

Thanks again to dmitryvm and everyone else who contributed. This is my first time posting on StackExchange, and I am really happy that this project was successful.

• Sketch out what you think you need as a block diagram. Give links to the thermopile and valve too. May 1, 2016 at 9:27
• Done! Your thoughts? May 7, 2016 at 3:51
• How can you get more than 70 odd mA with a supply of half a volt and 7 ohm load. Just do ohms law. May 7, 2016 at 8:43
• Is the valve part of the heater safety circuit? If so, it's designed to fail safe. The thermopile can't give an output if there is no flame and the valve will shut the gas off. Once you introduce a battery you have a risk of a wrong-side failure mode. e.g., MOSFET gate breakdown allows current to the valve. May 7, 2016 at 10:44

Since the timer is powered by a 1.5v AAA, your can expect end-of-life cell voltage around 1 V. So the valve control circuit should be able to deal with input voltage range 1 - 1.5 V. Not sure that your can achieve this goal with a single MOSFET circuit; there are many power MOSFETs with low resistance at Vgs > 2.5 V, but it may be hard to find a power MOSFET with low resistance at Vgs = 1 V.

To use widely available MOSFETs with low resistance at Vgs > 2.5 V (e.g. FDN337N, IRF6201, IRL6342 etc) your can consider the following circuit:

simulate this circuit – Schematic created using CircuitLab

In this circuit, expected life of the CR2032 cell should be on the order of ten years.

Update: the working implementation, as I can see, looks like the circuit below, but without R2 and R3 (the role of these resistors is explained below):

simulate this circuit

Inside the dashed box is my guess about the timer internals.

I feel the need to repeat the warning from the comment by transistor:

The thermopile can't give an output if there is no flame and the valve will shut the gas off. Once you introduce a battery you have a risk of a wrong-side failure mode. e.g., MOSFET gate breakdown allows current to the valve.

That's why R2 is recommended.

Also, as I have said earlier in the comments, 1 megaohm resistor between the gate and the source (R3 on the schematic) is also recommended. If the timer output is implemented as an open collector or open drain (as I have guessed), then the MOSFET may spontaneously turn on without the signal from the timer. R3 prevents this possibility.

Please keep in mind that the MOSFET may be turned on even with picoampere leakage current, which may happen due to change in humidity or temperature, if there is no DC path between the gate and the source. 1M resistor provides such path in any possible case.

And the final note: while FDN337N works fine with the fresh 1.5 V battery, but it may fail to work with the partially discharged battery. This may be checked by an experiment.

• Wow, awesome! Thank you, this is exactly what I was hoping for. Would you explain the theory around triggering the 2N3906? If it's too complicated to explain here, would you tell me what to read up on? I'm really curious how it works. May 7, 2016 at 18:22
• And actually, do you think it would be possible to eliminate the 1.5V battery and run the timer off the 3V battery by converting the voltage to 1.5V? The timer only draws like 5µA, so a 225mAh battery would still last 5 years if it could be done efficiently. May 7, 2016 at 19:43
• As for triggering 2N3906, nothing really complicated here. 2N3906 acts as a BJT switch (google "BJT switch" for more details). In short: BJTs are typically considered as current-controlled devices, as opposed to MOSFETs, which are voltage controlled; turn-on base-emitter voltage of virtually any silicon BJT lies in the range 0.5...0.8 V (depends on collector current and temperature). Thus, it is possible to control a BJT switch with a low voltage signal (below 1 V), provided that the base current will be sufficient to meet the saturation condition. For a typical BJT, Ic/Ib < 10 will be OK. May 7, 2016 at 19:57
• 2N3906 is a p-n-p transistor. To turn it on, the base potential must be less than the emitter potential. The timer does this job. R1 limits the base current to prevent unnecessary waste of energy. R2 pulls the base to the emitter to guarantee turn-off in the absence of the timer signal. C1 is optional, but highly recommended to filter out noise and spikes and to prevent (possible) HF generation. R3 pulls the gate to the source while Q1 is closed. Higher values of R3 are possible, but not recommended due to leakage current. Even with R3=1M, the CR2032 cell will probably outlive the heater. May 7, 2016 at 19:58
• I have removed a comment about the elimination of 1.5 V; not so easy to do in this circuit. May 7, 2016 at 20:30

If you want to keep power consumption low you should insert a little buck regulator between battery and valve. This will be controlled by your timer and will reduce voltage of batteries without losses so current drawn from battery will be low. For switching from thermopile probably you have to find some low gate voltage Mosfet or a reed relay because transistors have voltage drops too high.

• Any ideas on which MOSFET to get? I'd like to avoid powering the valve with the battery directly, because the thermopile is designed to stop actuating the valve if the pilot light goes out. If something goes wrong and the valve gets actuated by the battery when the pilot light has been blown out, it could fill the room with gas. May 7, 2016 at 3:50
• Did you verify if there is a voltage elevator inside the timer? If there is a display probably you will find also some voltage higher than 1.5 V so maybe you could employ it for driving mosfets. Also you should provide some device for lighting the pilot burner. May 9, 2016 at 6:46