t1d, your post brings up several issues that I've been wondering about.
I have indeed spent a good bit of time watching the Youtube videos on all these toaster oven projects, and I still don't understand how they deal with the lag that must be involved with any temperature change. (You said your oven overshoots during reflow.) I see everyone using feedback from the thermocouple as input for some kind of PID controller, but my understanding is that PIDs need to be *tuned*. And with one exception, I don't see any provision for adapting the controller to the particular oven being used, or its level of added insulation, or added heating elements.
The exception I believe is the Controleo, which has a "Learn" mode. It turns on the empty oven, then shuts it off at 100C, but then follows the temperature as it continues to rise, until it cools back down. That information supposedly gives the controller the tuning information it needs. But that controller is over $100, and I just don't have that kind of budget.
The other thing that strikes me is that anyone using the oven has to know how to place the thermocouple, because everything depends on that.
Given the variety of people who may be using the oven in my situation, with varying levels of understanding about the process, I came up with this idea for locking down the heating profile, and eliminating operator error:
I would establish a total ON/OFF cycle time of, say, 5 seconds. For each such cycle the average power level would be determined by the duty cycle. The 5 seconds would be divided by 100, giving duty cycle granularity of 50 ms. The present setting would be shown on a 2-digit display, and would be set using a rotary encoder.
All the tuning would be done by me ahead of time. Using a TM-902C thermocouple/display and a sample board, I would experimentally develop a sequence of duty cycle settings that produces the proper curve for leaded paste, such as:
10 cycles at 50%
14 cycles at 70%
etc.
As I develop the sequence, it would be loaded into the "controller", which would then simply execute that sequence when you push the button. It would be the same sequence, and the same heating profile, every time. The user can follow the temperature with the TM-902C if he wants to, but he doesn't have to. And the thermocouple is not even connected to the controller. There would be a buzzer indicating when to open the door.
Of course there's nothing magic about 5 seconds. It could be anything that works. Also, I'm not sure whether board size or parts composition would affect the reflow period, but as long as it doesn't overshoot, I hope a somewhat longer period should work for pretty much anything. I'm also not sure about adjusting for a warmer initial temperature in case multiple boards are done.
I'll keep an eye out for an Ebay bargain on a Crydom or similar SSR. And I may experiment a bit with a regular relay to see if that might work.
I'm also looking at the $8 Walmart Mainstays single-burner hotplate, with a steel circular saw blade on top. In theory, it could be controlled by the same process.