It has been several weeks since my last post. Since that time, I adapted a real time signal viewer to display the messages transmitted by the wireless controller I programmed and installed to replace the existing PIC controller, which I carelessly blew up. The viewer displays 19 of the parameters transmitted by the controller. This real time or recorded view was helpful to figuring out what was going on, much better than examining huge message log files.
As shown on earlier entries on the wiki, the combustion chamber temperature rises upon ignition and gas, then reaches a steady state where it remains for a few minutes. The interior case temperature continues to rise during this time. When the interior temperature reaches about 40°C, the combustion temperature starts a sudden rise toward about 165°C, much hotter. However, before it gets there, combustion ceases for no apparent reason. As the temperature falls, it is still not possible to restart combustion using a combination of gas and igniter commands until the case temperature declines to about 30°C.
The programmed algorithm uses only the combustion temperature probe value. The case temperature and humidity is transmitted but not used for control. The controller is programmed with two combustion temperatures, a low and a high. The high temperature is intended as a safety feature, temporarily shutting the gas flow and cooling the chamber. However, I set the temperature to the highest value that would stop combustion before it failed. That value was 140°C. I introduced a new algorithm that, at that temperature, enters the startup state. The startup state turns off the gas and waits for the trap to cool down to the startup maximum temperature, now 40°C, then starts the trap.
This works during the hot summer days, but the unit oscillates on and off constantly until the outside temperature drops and the unit stays on for the evening and night until it gets hot the next day. The downside is that the trap is operating only about 50% of the time during the day, which is not very effective in catching the fewer mosquitos flying around in the heat. It also repeatedly cycles the igniter, consuming its useful life more rapidly.
Under the theory that the combustion stops because the gas flow is interrupted at high temperature, I installed a 40mm miniature 40mm 12 volt fan to blow fresh air from the switch opening onto the valve assembly. This fan has an output of 4.1 cfm, which seems laughable compared to the existing fan. Using sealing tape to create an air intake duct from the switch opening in the case, the fan was directed on the valve. There was no effect.
But it still had to be the gas valve. What else would stop combustion? I disassembled the valve, and cleaned the orifice and interior, which was not obviously clogged, and reassembled. No difference.
Today, I used a T9 bit to unscrew then remove the active part of the valve, and taped over the port with sealing tape to bypass it, using the tank valve instead (dangerous), and restarted. The trap worked perfectly! The combustion temperature rose to 130°C, and the interior temperature to 50°C, and stayed in a perfect steady state! This is definitive proof that the valve was causing both the temperature rise (runaway) and the stoppage, because neither occurred after bypassing the valve.
So either the coil was opening up, or the magnetic plunger was losing its magnetism at high temperature. I removed the coil, removed the contacts from the connector, inspected, cleaned, and measured the coil to be 200.4Ω, regardless of any vibration or wiggling. Both connectors had gold plated contacts in good condition. I sprayed DeOxit contact cleaner anyway. I took the plunger assembly from another valve, installed it into the Defender, and plugged the coil back into the controller, moving the connected in and out to dislodge any possible contamination on the male connector.
After one hour, the combustion probe temperature is a very steady 130°C, and the case temperature is a pretty hot 55°C. That is 15°C hotter than the previous failure point. Its working!
Changing two variables leaves the question: is it the coil / connector, or is it the valve mechanism that made the difference? After a day or so, I will restore the original mechanism and see if that works, and will know for sure.
The end is in sight. I hope I can find the screws to the case. Stay tuned.