The last two weeks have been very productive for the team. As mentioned in more detail in Caleb’s report, after fixing the issue which was repeatedly killing gate driver ICs, the circuit immediately proceeded to kill a MOSFET. This issue was resolved very effectively by including two clamp diodes on the gate driver output, and tying all grounds. Furthermore, circuit reliability is almost entirely verified. With this issue fixed, we attempted to proceed with integration testing. Throughout the testing, we managed to get small arcs to metal objects within a few millimeters of the topload, however nothing strong, long, or consistent. Furthermore, this only occurred when the oscillator frequency was set to 3.54 MHz. Not only is this frequency over double the original target frequency, it is also too much for the MOSFET to handle at the full 48V, and the MOSFET saw some overheating, though we noticed and shut it down before any damage was done.
Upon further testing of the coil’s resonance frequencies independently, it was realized that the self capacitance of the secondary coil was far too low. The topload was replaced with a steel ball measuring approximately 80mm in diameter, which brought the measured resonance frequency (adjusted for probe capacitance) down to 1.48MHz. This of course also called for a redesign of the primary coil. This was further necessitated by the looser coiling on the primary being affected by frequent movement, and therefore the inductance no longer being within spec. We swapped from a PVC former to a 3D-printed one, which included channels cut out for the primary coil wire to be held into. The final design also includes a tophat which secures the wires in these channels. To meet the new target resonance frequency, the primary turn count was increased to exactly 7.75 turns. This resulted in a new primary resonance frequency of 1.425 MHZ, which is well within oscillator adjustability range.
There are two remaining risks to the project. First is thermal management, specifically in the MOSFET. Caleb is attempting to find some solution to this issue, however we are not entirely certain it will be an issue with everything now retuned to a more appropriate frequency. The other risk is that it simply fails to function as intended. This is something we have been combatting consistently with the weeks of unit testing and fixing that we have been performing. By tomorrow, we will have a fully assembled product.