In continuation of Carbon Monoxide teardowns, we now have a Kidde KN-COEG-3 detector designed to handle both carbon monoxide and natural gas in residential use. Unlike the previous target (First Alert FCD2BT), this one never worked that well, causing frequent “GAS” false panics and piercing alarms, so a teardown is a natural progression for this model!
In this design, a separate wall pack is used to deal with AC mains, supplying 9V AC to the unit. The supply can either stay in the case or be used remotely.
Taking one screw out, we get the case to open:
Next to it is a small SMT board holding the natural gas sensor from Figaro:
Now on to the main board details:
Taking the label off, reveals yet another Microchip part:
Not much is on the back of the board, just somewhat messy TH joints done with noclean solder as a post-operation.
And a bt of a crappy soldering on the AC input pins
And finally a block diagram of my limited reverse engineering effort together with a brief description of how this thing works:
The power supply is very straightforward an AC wallpack feeding a rectifier and a 5V regulator for the logic. Battery voltage is diode OR-ed in. The logic section is split into two microcontrollers. The older PIC16CR54C serves as a decoder for the 7-segment LED display. It received data serially over two lines (clk and data at about 58Khz) from the main micro, this time PIC16F628A. There is the usual R46C101 piezo horn driver, with enable line connected to the main micro. Its LED drive facilities are not used.
Sensing is done via two separate sensors. Natural gas detection is handled by Figaro TGS2611 sensor. The output is fed into an LM339 comparator and then into the main microcontroller. Carbon Monoxide is sensed by a Kidde branded sensor that appears to be electrochemical in nature, with only two pins coming out of it. The signal is converted from current to voltage by a low bias current Microchip MCP602 opamp and fed into the micro. Of interest is a MOSFET across the sensor, it is intended to keep the sensor shorted when no measurement is performed. That prevents sensor from damaging itself when not being used. I’ve not been able to locate the datasheet for the exact sensor used, but this Figaro model seems very similar. Note the 1-2nA/ppm output current, necessitating guard traces around opamp input node.The Figaro appnote is here
A nice discussion thread for these teardowns is located over at EEVBlog. Thanks to the folks there for the feedback!