From: David Roberson Actually the characteristic curves suggest that the input power acts like a bias that stands behind the incremental behavior. If that bias is quickly removed then there should exist a point of operation that is located ahead of the dangerous region. Unless some strong memory exists, I can imagine that the process would reverse as we all hope.
Dave, Given what you say above – what about the possibility of a higher level of control simply by use of pulsed power (at very low duty)? For instance, if we know from prior experiment that 100 watts of DC will eventually lead to large gain but at the risk of thermal runaway, and we also know that quenching begins almost immediately with removal of power (unless the system has already progressed to instability) – then it would seem that low duty pulsing with the same net power will provide better control against a runaway. (that is the premise but I have not data to back it). In effect, as an alternative to 100 watts DC, it would be possible to design and construct a pulsed power supply that will provide something like 2000 watt pulses at 5% duty. The net power in is the same, but 95% of the time there is no power. The frequency can be long but the idea is to alternate short sharp pulses with long delays. Is there any reason in your model to suggest that this approach is valid? Jones