In the entry below, Frank meant to add some commentary explaining that he thinks this long time constant is anomalously long.
At the 40m, it is much shorter. The insulation there is much thinner, and the can is somewhat shorter, but I think the main effect is that the steady state temperature of this can is too low. The temperature setpoint should be set to 35-40 C.
Then the time constant can be remeasured by doing a small step (~1 deg). We can't assume linearity over 10 deg - what we care about is the small signal time constant at the quiesscent temperature. Although the heating time and the cooling time will be slightly different, they will be comparable for small steps.
If the quiesscent temperature is set to be equal to the room temperature, it is insanity. We need to count on the room acting as a conductive and convective cold load since we only have a heater for an actuator and not a cooler.