Methane Gas Sensor Board

I’ve opened up a combination carbon-monoxide and natural gas alarm that was reaching its end of life. (First, I replaced it with a new alarm.)

Gas sensor add-in board.

At the bottom of the main board is a small daughterboard that contains the actual gas sensor and some interface electronics. Initially, I hypothesized that this board could be added or removed to change the detector model from a CO alarm to a combination CO/gas alarm. This would allow them to mass produce the main board (reducing cost), and they could easily vary the quantity of combination detectors produced based on consumer demand.

To test this hypothesis, I desoldered the methane gas board. The unit powered on just fine. It displayed 0 ppm and flashed a decimal point every second to show that it was operational. Then, after about a minute...

Gas detection displayed on LED.

The lack of the gas detector board causes this model to show an alert and blast my ears. Ouch.

Part of me is reassured that a non-functioning gas sensor would result in an alert from the detector. That’s a good design. On the other hand, it probably should display “Err” instead, so that the consumer won’t call the fire department.

Natural Gas Sensor Circuitry

The daughterboard is attached to the motherboard with a five-pin connector. Does the daughterboard have a microcontroller on it that communicates to the main microcontrollers via a serial interface? Or, is the sensor value buffered and transmitted via an analog value? No.

Gas detector board.

1. Figaro TGS 2611 methane gas sensor.
2. Precision resistor (3570 Ω 1% tolerance) in series with the sensor to create a voltage divider. Likely ±100 ppm or ±50 ppm metal film to reduce changes due to temperature.
3. Wow. A thermistor. This intentionally changes resistance with temperature in a specifically known and controlled way. It may attempt to compensate for temperature, it may be a heat alarm (not a smoke alarm), or it may detect whether the heater in the gas detector is active.
4. The resistors, thermistor, and gas sensor are hooked up to the inputs of a 339 quad comparator. This type of chip outputs either disconnected or low (GND) depending on the comparison between two voltages.
5. Since the 339 chip can’t output a high (+5V) signal, 100 kΩ resistors are used to pull-up the outputs. Thus, the output becomes high (5V) or low (0V).
6. One of the gas detection outputs is fed through a 100 ohm resistor to a capacitor.
7. The capacitor and resistor average or smooth the initial comparator output. This eliminates spikes.
8. The smoothed output is fed into another input of the comparator, and compared against 50% of the power supply voltage value by using two 100 kΩ resistors as a voltage divider. This converts the analog value to a digital value.
9. The gas sensor uses about 60 mA. To save power and increase sensor longevity between readings, the power to the comparator and gas sensor can be turned on and off via the transistor.

The circuit outputs only digital values (true or false) from the comparator. So, the main microcontroller is unable to display the gas PPM value since it can’t read it directly.

Why not have the gas sensor and thermistor analog signals go directly to a microcontroller for processing? In doing so...

• The ppm (parts per million) detected by the sensor could be displayed.
• The firmware could be modified to be less sensitive to reduce false alarms, without having to re-engineer the resistor and capacitor values. Some simple software could perform the averaging and spike removal.
• The capacitor, comparator, pull-up resistors, and other resistors could be eliminated.

Neither of the PIC microcontrollers used in the device have analog-to-digital converter pins. At the time the CO & gas detector was designed, someone probably computed the cost of a microcontroller with analog inputs versus the cost of the components on the daughterboard.

Quick Test of a Gas Sensor

The Figaro datasheet indicates that the gas sensor is sensitive to methane, as well as hydrogen and ethanol. I’m pretty sure the electrolysis rust removal apparatus generates hydrogen gas. However, I’d need to set it up, and I might need to capture the hydrogen to trigger a response.

A more convenient gas is ethanol -- also known as ethyl alcohol. I suppose a glass of beer or wine might do, however I simply held the Nighthawk gas sensor to an open container of ethyl rubbing alcohol from the medicine cabinet. In about 15 seconds, the pin on the detector board flipped from high to low. (FYI: Ethanol won’t work for sensors that include a filter.)

So, it looks like I’ve salvaged a gas sensor out of the device. Let’s see what else I can get from it.