Laser Frequency Comb - circuits

In addition to the hardware and process design of the Laser Frequency Comb (LFC), SteveO asked us to design a redundant "kill switch" mechanism of some sort to protect the instrument in the event of the heaters running out of control and cooking things. The controller came with an internal mechanical relay that would open a switch in the event of over-heating or sensor failure (if the thermistors got unplugged, for instance). We added a second relay: single pull double throw, which we wired so that it would latch open in the event of a failure.

I looked through my circuits textbooks and the mighty internets to no avail. Eventually I came up with this, with help from Amir Mirfakhrai who traced the current around and around until he found the mistakes:

When the system is powered on, the momentary reset-switch must be manually pushed ‘on’ to power our added redundant relay. Once powered, the switches change so the heaters get power and the alarm lamp turns off. The redundant relay remains powered (latched) with the other side of the lamp switch. The controller's solid state relay (ssr) controls heater power by opening and closing with a variable duty cycle. The controller's internal mechanical relay opens in the event of an alarm, killing power to the redundant relay, illuminating the alarm lamp and stopping the heaters. Our redundant relay remains unpowered until the manual reset is pushed to restart the system.

Here is a photo of the controller box half-opened while we were fixing some wiring.

The controller, solid state relay mounted on an aluminum heat sink cooled by a 12V DC fan with its 12V DC transformer, in-line fuses and wiring fit neatly into the box I made. The added redundant relay and its more powerful 20V DC power supply had to be strapped on top with some cable ties. The reset switch and alarm lamp are mounted on the top of the box as well.

The alarm lamp is illuminated in this photo, which means the heaters were turned off. Indeed, the actual temperature of 23.05 deg C is much colder than the set point of 28 deg C. We were testing to see how cold the system could get.