brettski74 / solder-reflow-plate Goto Github PK

Thermal imaging of the hot plate shows a large discrepency between the temperature in the centre of the hot plate and its edge. Given that the temperature sensor is off to the side of the hot plate, it's temperature is likely to be even lower. Experiment with different ways of improving the accuracy of the temperature measurements. Possible strategies are:

1. Software based - apply a correction function to estimate the average or peak temperature of the hotplate for a give temperature reading at the sensor. Maybe a linear or a piecewise linear approximation could be used.
2. Hardware and software based - we already have what could be a copper RTD in place on the hot plate. Could we use that to measure temperature? If the resistance of the hotplate traces could be estimated in real time by measuring the voltage across it and current through it, this could be used to estimate the average temperature of the hot plate and since it would be measuring temperature across the entire hot plate, it might be more accurate than a thermal sensor placed off to the side of the hotplate. The question is whether this can be done reliably and accurately enough to be usable.

There are numerous calls to main_menu() littered throughout the code. Since these calls all originate in functions that were themselves called from main_menu(), we have some infinite recursion going on. While it seems unlikely that the reflow plate would continue running long enough to exhaust the stack, it's unlikely that anything good will result if these recursive calls do eventually exhaust the stack. Code should be refactored to use the loop() function and operate as a state machine.

The heat function doesn't seem able to return a non-zero value. As a result, heating cycles always end with a message saying that the cycle was cancelled. The function needs a return 1 statement at the end.

Current code uses hard-coded PWM settings to try to follow a proper reflow profile, but these don't necessarily have any correlation to what the temperature is actually doing over time. A PID control strategy should allow for more accurate control of temperatures and rates of temperature rise/fall, provided that we have accurate temperature measurements to begin with. It should also allow the device to be able to be calibrated to work for different power supplies and plate designs.

Temperature readings on an unpowered heating plate seem relatively accurate and around room temperature (eg. 24 celsius), but once power is applied, temperature readings initially go down substantially - as low as perhaps 10 celsius. They then seem to lag far behind the actual hot plate temperature resulting in the hotplate far exceeding the desired maximum temperature. Need to investigate:

1. Why temperature goes down when hotplate power is applied.
2. Whether temperature lag is due to some problem or simply the difference in temperature between the centre and edge of the hot plate.