Thanks for that- this had been causing me some consternation prior to actually receiving the unit and seeing how it worked.
Also, a note on the use of "other" temp sensors:
I'm awaiting the delivery of a MAC motor from Cell-Man. These motors are fitted with an LM35 temperature sensor, which is different from either the LM335 or the NTC thermistor that Justin suggests. Specifically, thermistors and LM335s act like resistors or zener diodes, which conduct current to ground depending on temperature. As such the CA-3 is fitted with a pullup resistor on the "NTC" pad. Although the part number seems quite similar (LM35 vs. LM335), these devices are quite different.
The LM35 is an active device, which outputs a positive voltage which varies with temperature. Cell_Man selected this device over the passive devices which Justin suggested to enable folks without CAs to use a simple voltmeter to measure temperature (the LM35 outputs 10mv / °C, so it reads nicely on even a simple $5 voltmeter.) However the internal architecture of the LM35 is such that it won't work properly with a pullup voltage applied to its output- it can only source current, not sink it.
So, to use one of these devices, you need to remove the pullup resistor (R17) from the CA board. It's a tiny little surface-mount part, but shouldn't be too difficult if you have a pair of irons to lift it up tweezer-style:
Having removed this resistor, you can pick up 5v at the adjacent pad to supply operating power to the LM35, and then calibrate the software for 0 degrees = 0 volts and 100 degrees = 1 volt.EDIT:
The NTC input on the CA tends to float high when unconnected, or when connected to a device which cannot sink current to ground on its output such as the LM35. With the circuit configured as described above, I saw false-high readings of 100 degrees or more, just from the stray voltages floating around the circuit.
This has been solved by simply placing a pull-down resistor between the NTC pad and ground, to eliminate the stray voltage. I've tested the circuit with a 390 ohm resistor (simply because it's the first small-value I happened to pull out of the pile), and it is now giving me a believable indication of 22-23 degrees while resting at room temperature. I haven't load-tested the bike yet to see how it performs as the motor warms up (I'm tethered to my workbench at the moment owing to the lack of a functional battery, and I found out the hard way that using your rear brake as a dyno load doesn't work for very long.)
Too low of a resistor value will negatively impact the performance of the circuit, as the output of the LM35 is relatively weak- it can only source 10ma on its output. I may experiment with larger resistor values to see if I can still hold down the float voltages while pulling less current from the LM35, though even at 390 ohms I'm loading it pretty lightly- less than 4ma at 150 degrees (1.5v).
The concept of "average" requires that 50% of the population be below it.