Obtained two thermistors Vishay NTCALUG01T103FL and NTCALUG02A103F161L from Digikey which were supposed to be beta 3984 and 3435 respectively. Except when I read the spec sheet for the first one I missed that all of the units were 3984
except one, which was also 3435, and guess which one I clicked on? Anyway the two thermistors are actually slightly different physically, most evidently by the first having much thicker wires coming out, whereas the wires for the .*161L unit are very thin, almost as small as headphone wires. To minimize temperature variation measurements were made in the (closed) garage, on top of the (long inactive) dryer. First a sealed furnace filter was put down then two brown paper shopping bags (so that they thermistors would track air temperature rather than the case temperature of the heavy metal dryer), then these two thermistors and the one from the BM1500 were placed close together. Those and the wires with clips to attach to them were left in the garage at all times so that they would be at the same temperature. At semiregular intervals I would go out to the garage, quickly measure the resistance of each thermistor by attaching one end of each prewarmed clip wire to the thermistor and the other to the UT210E multimeter lead, and then measure the temperature of the same spot on a bag (brown paper, no ink) near them with an IR thermometer from about the same distance. When the temperature peaked and then started falling (by a fraction of a degree) in the late afternoon measurements were stopped. Data were then plotted using gnuplot along with the theoretical curves for a perfect thermistor 10K@25C with beta 3435 and 3950. Here is that plot:
This shows that the BatteryMinder 1500 sensor is 10k@25C, beta 3950 (or very close to it, possibly 3984). Two types of experimental errors are visible. In one the IR thermometer measurement is off, and that drags all 3 data plots off the theoretical lines. See for instance the points at 36.1 C. I believe the thermometer was having systematic problems operating near 40C, which is why all the points there diverged a little from the theoretical curve. Additionally, the tiny wires on the .*161L thermistor seemed to sometimes make a poor connection, adding a little bit of resistance. I don't know if that was a purely mechanical thing or if there was some sort of coating on it, perhaps a thin layer of insulation left after it was stripped.
The thermistor that came with the charger is embedded in a block of plastic, and it responds somewhat slowly to temperature changes. However, these lug thermistors are much smaller and the lug is metal and conducts heat well. When air blows over them their resistance changes swiftly. For that reason, if one is going to be used with the BatteryMinder instead of the supplied thermistor the lug needs to be fastened tightly, for a good thermal connection, to one of the battery posts or possibly the thick cables coming off them. (Ideally it would be embedded in the battery case, as is done with lithium batteries, but doing that after the fact with a lead acid battery is more likely to break the battery than it is to give better temperature measurements. I strongly suspect that if they are not firmly attached their readings could easily be off 2 or 3 degrees C, or more. That wouldn't be the end of the world (look at the temperature curves for float/charge vs. temperature), but if one is going to temperature compensate the charger, it really should be for the temperature of the battery, which is the only temperature which matters.
