On Thu, Aug 30, 2012 at 3:38 PM, David Roberson <[email protected]> wrote:
I wanted to mention one observation that is fairly important. If you set > the upper turn around timing extremely critically, it is possible to get a > very large COP. The reason is that the time constants associated with the > thermal resistance and capacitance become quite large. The timing is as > critical as it is large however and the system is balanced upon a sharp > edge. It typically does not take long for the positive feedback to > dominate and the curve begins a rapid decent. > It sounds like your model suggests that it is fairly easy to have a power excursion that sinters the substrate if the device is operated at too high a temperature. I wonder whether this is behind Defkalion's using discrete spikes spikes in the input power rather than a continuous drive. Perhaps they find this eliminates some of the feedback problem. Are you including a stochastic component in the temperature as a function of the input power? If you do, I suspect the model will have to be operated at a lower average temperature than if the model were purely deterministic. Eric
