View Single Post
Old 04-29-2011, 09:41 AM   #196 (permalink)
Freyguy
EV Builder
 
Freyguy's Avatar
 
Join Date: Nov 2010
Location: Macungie, PA
Posts: 157
Thanks: 15
Thanked 64 Times in 40 Posts
Quote:
Originally Posted by harlequin2 View Post
I've just connected the thermistor for Tmot to my board and it reads just on 3 deg high. That is, showing 27 deg C when my fairly accurate thermometer says 23.7. Is this about the best accuracy one can expect? Or is there something that needs trimming somewhere?
Further, I have connected the one for Tbat also and it reads the same 27 deg - as one would hope!

Now here's an interesting thing! The increase in temp is explained by the fact that my desk light is only a metre or so away and the IR it puts out is enough to warm the thermistors by a couple of degrees! If I shade them from the light, they drop back to about the same as my thermometer. Quite amazing.
Anyway, I guess they are more than accurate enough for the purpose.
Hey there harlequin2! Yeah, isn't that interesting with the IR? Imagine if the thermistors were black instead of blue
This board was not designed with test equipment accuracy in mind. Just accurate enough to know what's going on with various aspects of the car. There are trims for traction and accessory battery voltages since these need a bit more accuracy.
The thermistors are cheap 2% parts. The resistor divider on the board uses 1% parts. Raw temp has 8-bit accuracy (0.4%) and the conversion from F to Celcius uses 9/16 vs 5/8 (1/2 deg C error). And behind it all is the ADC reference which I chose to use the +5V supply versus a dedicated voltage reference (would have lost an ADC port in the process). I tried to keep this tight though which is why I spec'ed in a 2% regulator vs 5%. So if you stack up all the errors worst case is around +/-6% on temp.
  Reply With Quote