Alan, thank you very much!!! It's amazing how obvious something can seem after someone else points it out. As an engineer with almost 50 years experience in analog signal conditioning and temperature measurement (including thermistors at YSI), I should have caught this.
The thermistor is heating up on start-up, not due to self-heat (it is dissipating only about 0.1 milliwatts) but due to poor thermal design. As it turns out, a modification we are already planning should solve the problem.
In our present advanced-prototype design the surface-mount thermistor is on the same board as the charger circuit, and not really close to the batteries. The charge current regulator (LM317T) and two diodes, which collectively dissipate about 1.5 watts, are also surface mount components on the same board, about 1-1/2 inches away from the thermistor. When I fan the board the problem goes away (until I stop fanning).
We're already planning to relocate the thermistor next to the batteries, off the circuit board, and have a thermistor on order. Thank you very much for your help.
Regards,
Harry Trietley, JH Technology, Inc.