Bulldog147 - you are assuming there is a correlation between the TCase sensor (on the motherboard) and the core sensor, in the CPU.
Quite apart from the issue of scaling (10 degrees is 15% of 67 degrees, but only 10% of 100 degrees) there is also the issue of heat transfer from core to TCase sensor.
Lets say the CPU is idling at 40C and you load it 100%, the loaded core temperature leaps up immediately to 60C and then, as the CPU cooler struggles to dump the heat, eventually equilibrates at about 70C after several minutes.
The TCase sensor does not react immediately because the heat has to travel from the centre of the CPU to the sensor on the motherboard. So your cores could well be over 100C and throttling and the TCase sensor wouldn't ever catch up because the CPU would shut the PC down.
I think if you are going to try and assert a safe core temperature vs. TCase then you would need to know the position of the TCase sensor for each motherboard, the thermal conduction properties of the CPU casing, CPU socket and motherboard and finally, establish the lag time from the core temperature stabilizing to the TCase sensor temperature stabilizing.
In the Pentium 4 days we had a program called Throttlewatch. The motherboard sensors didn't actually give a good indication of when the CPU was overheating, so you used Throttlewatch to tell you when it was throttling and that basically meant you had overcooked the overclock (no pun intended). In those days we had motherboard temperatures of 85-90C and we only cut back voltages or FSBs when THrottlewatch said we had to because the CPU would tell you when it was overheating, not the motherboard. That is still the same today. Temperatures are useful for establishing how good your cooling system is, and arbitrarily picking a 10 degree differential at the top end of an uncalibrated thermal sensor range isn't a good way to decide what's 'safe' for your CPU.
Be honest, when did you last hear of anyone destroying an Intel CPU with heat? Volts, I absolutely agree should be monitored and controlled because with 45nm CPUs are very sensitive to electron migration damage, but not heat.
I think you're doing a good thing trying to establish this, and I look forward to what you come up with next. Maybe look at how accurate the voltage settings are vs. a voltmeter on the motherboard? You'll be horrified how inaccurate they are!
In the meanwhile, I'm going to let my CPUs protect themselves heatwise.
