OK, now this issue has got me to go digging around in the firmware,
and a number of issues and answers are coming to light. I'll talk about extruders in the build thread, but as for this item, my conclusion is that I will run the heatbed on a standalone pid controller, looped with a thermocouple, and then add a thermistor to talk back to ramps. The output to the heater will be set low and will operate a status indicator. That way the ramps is happy, and the heat bed is far more reliable, and accurate. This will make a difference in calculations, more than in real time operation, but the overall advantage far outweighs the cost.